/src/CMake/Utilities/cmexpat/lib/xmlparse.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* a30d2613dcfdef81475a9d1a349134d2d42722172fdaa7d5bb12ed2aa74b9596 (2.4.6+) |
2 | | __ __ _ |
3 | | ___\ \/ /_ __ __ _| |_ |
4 | | / _ \\ /| '_ \ / _` | __| |
5 | | | __// \| |_) | (_| | |_ |
6 | | \___/_/\_\ .__/ \__,_|\__| |
7 | | |_| XML parser |
8 | | |
9 | | Copyright (c) 1997-2000 Thai Open Source Software Center Ltd |
10 | | Copyright (c) 2000 Clark Cooper <coopercc@users.sourceforge.net> |
11 | | Copyright (c) 2000-2006 Fred L. Drake, Jr. <fdrake@users.sourceforge.net> |
12 | | Copyright (c) 2001-2002 Greg Stein <gstein@users.sourceforge.net> |
13 | | Copyright (c) 2002-2016 Karl Waclawek <karl@waclawek.net> |
14 | | Copyright (c) 2005-2009 Steven Solie <steven@solie.ca> |
15 | | Copyright (c) 2016 Eric Rahm <erahm@mozilla.com> |
16 | | Copyright (c) 2016-2022 Sebastian Pipping <sebastian@pipping.org> |
17 | | Copyright (c) 2016 Gaurav <g.gupta@samsung.com> |
18 | | Copyright (c) 2016 Thomas Beutlich <tc@tbeu.de> |
19 | | Copyright (c) 2016 Gustavo Grieco <gustavo.grieco@imag.fr> |
20 | | Copyright (c) 2016 Pascal Cuoq <cuoq@trust-in-soft.com> |
21 | | Copyright (c) 2016 Ed Schouten <ed@nuxi.nl> |
22 | | Copyright (c) 2017-2018 Rhodri James <rhodri@wildebeest.org.uk> |
23 | | Copyright (c) 2017 Václav Slavík <vaclav@slavik.io> |
24 | | Copyright (c) 2017 Viktor Szakats <commit@vsz.me> |
25 | | Copyright (c) 2017 Chanho Park <chanho61.park@samsung.com> |
26 | | Copyright (c) 2017 Rolf Eike Beer <eike@sf-mail.de> |
27 | | Copyright (c) 2017 Hans Wennborg <hans@chromium.org> |
28 | | Copyright (c) 2018 Anton Maklakov <antmak.pub@gmail.com> |
29 | | Copyright (c) 2018 Benjamin Peterson <benjamin@python.org> |
30 | | Copyright (c) 2018 Marco Maggi <marco.maggi-ipsu@poste.it> |
31 | | Copyright (c) 2018 Mariusz Zaborski <oshogbo@vexillium.org> |
32 | | Copyright (c) 2019 David Loffredo <loffredo@steptools.com> |
33 | | Copyright (c) 2019-2020 Ben Wagner <bungeman@chromium.org> |
34 | | Copyright (c) 2019 Vadim Zeitlin <vadim@zeitlins.org> |
35 | | Copyright (c) 2021 Dong-hee Na <donghee.na@python.org> |
36 | | Copyright (c) 2022 Samanta Navarro <ferivoz@riseup.net> |
37 | | Licensed under the MIT license: |
38 | | |
39 | | Permission is hereby granted, free of charge, to any person obtaining |
40 | | a copy of this software and associated documentation files (the |
41 | | "Software"), to deal in the Software without restriction, including |
42 | | without limitation the rights to use, copy, modify, merge, publish, |
43 | | distribute, sublicense, and/or sell copies of the Software, and to permit |
44 | | persons to whom the Software is furnished to do so, subject to the |
45 | | following conditions: |
46 | | |
47 | | The above copyright notice and this permission notice shall be included |
48 | | in all copies or substantial portions of the Software. |
49 | | |
50 | | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, |
51 | | EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF |
52 | | MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN |
53 | | NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, |
54 | | DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR |
55 | | OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE |
56 | | USE OR OTHER DEALINGS IN THE SOFTWARE. |
57 | | */ |
58 | | |
59 | | #define XML_BUILDING_EXPAT 1 |
60 | | |
61 | | #include <expat_config.h> |
62 | | |
63 | | #if ! defined(_GNU_SOURCE) |
64 | | # define _GNU_SOURCE 1 /* syscall prototype */ |
65 | | #endif |
66 | | |
67 | | #ifdef _WIN32 |
68 | | /* force stdlib to define rand_s() */ |
69 | | # if ! defined(_CRT_RAND_S) |
70 | | # define _CRT_RAND_S |
71 | | # endif |
72 | | #endif |
73 | | |
74 | | #include <stddef.h> |
75 | | #include <string.h> /* memset(), memcpy() */ |
76 | | #include <assert.h> |
77 | | #include <limits.h> /* UINT_MAX */ |
78 | | #include <stdio.h> /* fprintf */ |
79 | | #include <stdlib.h> /* getenv, rand_s */ |
80 | | #include <stdint.h> /* uintptr_t */ |
81 | | #include <math.h> /* isnan */ |
82 | | |
83 | | #ifdef _WIN32 |
84 | | # define getpid GetCurrentProcessId |
85 | | #else |
86 | | # include <sys/time.h> /* gettimeofday() */ |
87 | | # include <sys/types.h> /* getpid() */ |
88 | | # include <unistd.h> /* getpid() */ |
89 | | # include <fcntl.h> /* O_RDONLY */ |
90 | | # include <errno.h> |
91 | | #endif |
92 | | |
93 | | #ifdef _WIN32 |
94 | | # include "winconfig.h" |
95 | | #endif |
96 | | |
97 | | #include "ascii.h" |
98 | | #include "expat.h" |
99 | | #include "siphash.h" |
100 | | |
101 | | #if defined(HAVE_GETRANDOM) || defined(HAVE_SYSCALL_GETRANDOM) |
102 | | # if defined(HAVE_GETRANDOM) |
103 | | # include <sys/random.h> /* getrandom */ |
104 | | # else |
105 | | # include <unistd.h> /* syscall */ |
106 | | # include <sys/syscall.h> /* SYS_getrandom */ |
107 | | # endif |
108 | | # if ! defined(GRND_NONBLOCK) |
109 | | # define GRND_NONBLOCK 0x0001 |
110 | | # endif /* defined(GRND_NONBLOCK) */ |
111 | | #endif /* defined(HAVE_GETRANDOM) || defined(HAVE_SYSCALL_GETRANDOM) */ |
112 | | |
113 | | #if defined(HAVE_LIBBSD) \ |
114 | | && (defined(HAVE_ARC4RANDOM_BUF) || defined(HAVE_ARC4RANDOM)) |
115 | | # include <bsd/stdlib.h> |
116 | | #endif |
117 | | |
118 | | #if defined(_WIN32) && ! defined(LOAD_LIBRARY_SEARCH_SYSTEM32) |
119 | | # define LOAD_LIBRARY_SEARCH_SYSTEM32 0x00000800 |
120 | | #endif |
121 | | |
122 | | #if ! defined(HAVE_GETRANDOM) && ! defined(HAVE_SYSCALL_GETRANDOM) \ |
123 | | && ! defined(HAVE_ARC4RANDOM_BUF) && ! defined(HAVE_ARC4RANDOM) \ |
124 | | && ! defined(XML_DEV_URANDOM) && ! defined(_WIN32) \ |
125 | | && ! defined(XML_POOR_ENTROPY) |
126 | | # error You do not have support for any sources of high quality entropy \ |
127 | | enabled. For end user security, that is probably not what you want. \ |
128 | | \ |
129 | | Your options include: \ |
130 | | * Linux >=3.17 + glibc >=2.25 (getrandom): HAVE_GETRANDOM, \ |
131 | | * Linux >=3.17 + glibc (including <2.25) (syscall SYS_getrandom): HAVE_SYSCALL_GETRANDOM, \ |
132 | | * BSD / macOS >=10.7 (arc4random_buf): HAVE_ARC4RANDOM_BUF, \ |
133 | | * BSD / macOS (including <10.7) (arc4random): HAVE_ARC4RANDOM, \ |
134 | | * libbsd (arc4random_buf): HAVE_ARC4RANDOM_BUF + HAVE_LIBBSD, \ |
135 | | * libbsd (arc4random): HAVE_ARC4RANDOM + HAVE_LIBBSD, \ |
136 | | * Linux (including <3.17) / BSD / macOS (including <10.7) (/dev/urandom): XML_DEV_URANDOM, \ |
137 | | * Windows >=Vista (rand_s): _WIN32. \ |
138 | | \ |
139 | | If insist on not using any of these, bypass this error by defining \ |
140 | | XML_POOR_ENTROPY; you have been warned. \ |
141 | | \ |
142 | | If you have reasons to patch this detection code away or need changes \ |
143 | | to the build system, please open a bug. Thank you! |
144 | | #endif |
145 | | |
146 | | #ifdef XML_UNICODE |
147 | | # define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX |
148 | | # define XmlConvert XmlUtf16Convert |
149 | | # define XmlGetInternalEncoding XmlGetUtf16InternalEncoding |
150 | | # define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS |
151 | | # define XmlEncode XmlUtf16Encode |
152 | | # define MUST_CONVERT(enc, s) (! (enc)->isUtf16 || (((uintptr_t)(s)) & 1)) |
153 | | typedef unsigned short ICHAR; |
154 | | #else |
155 | | # define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX |
156 | 19.9M | # define XmlConvert XmlUtf8Convert |
157 | 7.48k | # define XmlGetInternalEncoding XmlGetUtf8InternalEncoding |
158 | | # define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS |
159 | 754k | # define XmlEncode XmlUtf8Encode |
160 | 959k | # define MUST_CONVERT(enc, s) (! (enc)->isUtf8) |
161 | | typedef char ICHAR; |
162 | | #endif |
163 | | |
164 | | #ifndef XML_NS |
165 | | |
166 | 0 | # define XmlInitEncodingNS XmlInitEncoding |
167 | 0 | # define XmlInitUnknownEncodingNS XmlInitUnknownEncoding |
168 | | # undef XmlGetInternalEncodingNS |
169 | 0 | # define XmlGetInternalEncodingNS XmlGetInternalEncoding |
170 | 0 | # define XmlParseXmlDeclNS XmlParseXmlDecl |
171 | | |
172 | | #endif |
173 | | |
174 | | #ifdef XML_UNICODE |
175 | | |
176 | | # ifdef XML_UNICODE_WCHAR_T |
177 | | # define XML_T(x) (const wchar_t) x |
178 | | # define XML_L(x) L##x |
179 | | # else |
180 | | # define XML_T(x) (const unsigned short)x |
181 | | # define XML_L(x) x |
182 | | # endif |
183 | | |
184 | | #else |
185 | | |
186 | 2.01M | # define XML_T(x) x |
187 | 7.43k | # define XML_L(x) x |
188 | | |
189 | | #endif |
190 | | |
191 | | /* Round up n to be a multiple of sz, where sz is a power of 2. */ |
192 | 289k | #define ROUND_UP(n, sz) (((n) + ((sz)-1)) & ~((sz)-1)) |
193 | | |
194 | | /* Do safe (NULL-aware) pointer arithmetic */ |
195 | 37.3k | #define EXPAT_SAFE_PTR_DIFF(p, q) (((p) && (q)) ? ((p) - (q)) : 0) |
196 | | |
197 | | #include "internal.h" |
198 | | #include "xmltok.h" |
199 | | #include "xmlrole.h" |
200 | | |
201 | | typedef const XML_Char *KEY; |
202 | | |
203 | | typedef struct { |
204 | | KEY name; |
205 | | } NAMED; |
206 | | |
207 | | typedef struct { |
208 | | NAMED **v; |
209 | | unsigned char power; |
210 | | size_t size; |
211 | | size_t used; |
212 | | const XML_Memory_Handling_Suite *mem; |
213 | | } HASH_TABLE; |
214 | | |
215 | | static size_t keylen(KEY s); |
216 | | |
217 | | static void copy_salt_to_sipkey(XML_Parser parser, struct sipkey *key); |
218 | | |
219 | | /* For probing (after a collision) we need a step size relative prime |
220 | | to the hash table size, which is a power of 2. We use double-hashing, |
221 | | since we can calculate a second hash value cheaply by taking those bits |
222 | | of the first hash value that were discarded (masked out) when the table |
223 | | index was calculated: index = hash & mask, where mask = table->size - 1. |
224 | | We limit the maximum step size to table->size / 4 (mask >> 2) and make |
225 | | it odd, since odd numbers are always relative prime to a power of 2. |
226 | | */ |
227 | | #define SECOND_HASH(hash, mask, power) \ |
228 | 91.5k | ((((hash) & ~(mask)) >> ((power)-1)) & ((mask) >> 2)) |
229 | | #define PROBE_STEP(hash, mask, power) \ |
230 | 91.5k | ((unsigned char)((SECOND_HASH(hash, mask, power)) | 1)) |
231 | | |
232 | | typedef struct { |
233 | | NAMED **p; |
234 | | NAMED **end; |
235 | | } HASH_TABLE_ITER; |
236 | | |
237 | 1.20M | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ |
238 | 7.48k | #define INIT_DATA_BUF_SIZE 1024 |
239 | 7.84k | #define INIT_ATTS_SIZE 16 |
240 | 0 | #define INIT_ATTS_VERSION 0xFFFFFFFF |
241 | 29.9k | #define INIT_BLOCK_SIZE 1024 |
242 | 7.47k | #define INIT_BUFFER_SIZE 1024 |
243 | | |
244 | 0 | #define EXPAND_SPARE 24 |
245 | | |
246 | | typedef struct binding { |
247 | | struct prefix *prefix; |
248 | | struct binding *nextTagBinding; |
249 | | struct binding *prevPrefixBinding; |
250 | | const struct attribute_id *attId; |
251 | | XML_Char *uri; |
252 | | int uriLen; |
253 | | int uriAlloc; |
254 | | } BINDING; |
255 | | |
256 | | typedef struct prefix { |
257 | | const XML_Char *name; |
258 | | BINDING *binding; |
259 | | } PREFIX; |
260 | | |
261 | | typedef struct { |
262 | | const XML_Char *str; |
263 | | const XML_Char *localPart; |
264 | | const XML_Char *prefix; |
265 | | int strLen; |
266 | | int uriLen; |
267 | | int prefixLen; |
268 | | } TAG_NAME; |
269 | | |
270 | | /* TAG represents an open element. |
271 | | The name of the element is stored in both the document and API |
272 | | encodings. The memory buffer 'buf' is a separately-allocated |
273 | | memory area which stores the name. During the XML_Parse()/ |
274 | | XMLParseBuffer() when the element is open, the memory for the 'raw' |
275 | | version of the name (in the document encoding) is shared with the |
276 | | document buffer. If the element is open across calls to |
277 | | XML_Parse()/XML_ParseBuffer(), the buffer is re-allocated to |
278 | | contain the 'raw' name as well. |
279 | | |
280 | | A parser re-uses these structures, maintaining a list of allocated |
281 | | TAG objects in a free list. |
282 | | */ |
283 | | typedef struct tag { |
284 | | struct tag *parent; /* parent of this element */ |
285 | | const char *rawName; /* tagName in the original encoding */ |
286 | | int rawNameLength; |
287 | | TAG_NAME name; /* tagName in the API encoding */ |
288 | | char *buf; /* buffer for name components */ |
289 | | char *bufEnd; /* end of the buffer */ |
290 | | BINDING *bindings; |
291 | | } TAG; |
292 | | |
293 | | typedef struct { |
294 | | const XML_Char *name; |
295 | | const XML_Char *textPtr; |
296 | | int textLen; /* length in XML_Chars */ |
297 | | int processed; /* # of processed bytes - when suspended */ |
298 | | const XML_Char *systemId; |
299 | | const XML_Char *base; |
300 | | const XML_Char *publicId; |
301 | | const XML_Char *notation; |
302 | | XML_Bool open; |
303 | | XML_Bool is_param; |
304 | | XML_Bool is_internal; /* true if declared in internal subset outside PE */ |
305 | | } ENTITY; |
306 | | |
307 | | typedef struct { |
308 | | enum XML_Content_Type type; |
309 | | enum XML_Content_Quant quant; |
310 | | const XML_Char *name; |
311 | | int firstchild; |
312 | | int lastchild; |
313 | | int childcnt; |
314 | | int nextsib; |
315 | | } CONTENT_SCAFFOLD; |
316 | | |
317 | 0 | #define INIT_SCAFFOLD_ELEMENTS 32 |
318 | | |
319 | | typedef struct block { |
320 | | struct block *next; |
321 | | int size; |
322 | | XML_Char s[1]; |
323 | | } BLOCK; |
324 | | |
325 | | typedef struct { |
326 | | BLOCK *blocks; |
327 | | BLOCK *freeBlocks; |
328 | | const XML_Char *end; |
329 | | XML_Char *ptr; |
330 | | XML_Char *start; |
331 | | const XML_Memory_Handling_Suite *mem; |
332 | | } STRING_POOL; |
333 | | |
334 | | /* The XML_Char before the name is used to determine whether |
335 | | an attribute has been specified. */ |
336 | | typedef struct attribute_id { |
337 | | XML_Char *name; |
338 | | PREFIX *prefix; |
339 | | XML_Bool maybeTokenized; |
340 | | XML_Bool xmlns; |
341 | | } ATTRIBUTE_ID; |
342 | | |
343 | | typedef struct { |
344 | | const ATTRIBUTE_ID *id; |
345 | | XML_Bool isCdata; |
346 | | const XML_Char *value; |
347 | | } DEFAULT_ATTRIBUTE; |
348 | | |
349 | | typedef struct { |
350 | | unsigned long version; |
351 | | unsigned long hash; |
352 | | const XML_Char *uriName; |
353 | | } NS_ATT; |
354 | | |
355 | | typedef struct { |
356 | | const XML_Char *name; |
357 | | PREFIX *prefix; |
358 | | const ATTRIBUTE_ID *idAtt; |
359 | | int nDefaultAtts; |
360 | | int allocDefaultAtts; |
361 | | DEFAULT_ATTRIBUTE *defaultAtts; |
362 | | } ELEMENT_TYPE; |
363 | | |
364 | | typedef struct { |
365 | | HASH_TABLE generalEntities; |
366 | | HASH_TABLE elementTypes; |
367 | | HASH_TABLE attributeIds; |
368 | | HASH_TABLE prefixes; |
369 | | STRING_POOL pool; |
370 | | STRING_POOL entityValuePool; |
371 | | /* false once a parameter entity reference has been skipped */ |
372 | | XML_Bool keepProcessing; |
373 | | /* true once an internal or external PE reference has been encountered; |
374 | | this includes the reference to an external subset */ |
375 | | XML_Bool hasParamEntityRefs; |
376 | | XML_Bool standalone; |
377 | | #ifdef XML_DTD |
378 | | /* indicates if external PE has been read */ |
379 | | XML_Bool paramEntityRead; |
380 | | HASH_TABLE paramEntities; |
381 | | #endif /* XML_DTD */ |
382 | | PREFIX defaultPrefix; |
383 | | /* === scaffolding for building content model === */ |
384 | | XML_Bool in_eldecl; |
385 | | CONTENT_SCAFFOLD *scaffold; |
386 | | unsigned contentStringLen; |
387 | | unsigned scaffSize; |
388 | | unsigned scaffCount; |
389 | | int scaffLevel; |
390 | | int *scaffIndex; |
391 | | } DTD; |
392 | | |
393 | | typedef struct open_internal_entity { |
394 | | const char *internalEventPtr; |
395 | | const char *internalEventEndPtr; |
396 | | struct open_internal_entity *next; |
397 | | ENTITY *entity; |
398 | | int startTagLevel; |
399 | | XML_Bool betweenDecl; /* WFC: PE Between Declarations */ |
400 | | } OPEN_INTERNAL_ENTITY; |
401 | | |
402 | | enum XML_Account { |
403 | | XML_ACCOUNT_DIRECT, /* bytes directly passed to the Expat parser */ |
404 | | XML_ACCOUNT_ENTITY_EXPANSION, /* intermediate bytes produced during entity |
405 | | expansion */ |
406 | | XML_ACCOUNT_NONE /* i.e. do not account, was accounted already */ |
407 | | }; |
408 | | |
409 | | #ifdef XML_DTD |
410 | | typedef unsigned long long XmlBigCount; |
411 | | typedef struct accounting { |
412 | | XmlBigCount countBytesDirect; |
413 | | XmlBigCount countBytesIndirect; |
414 | | int debugLevel; |
415 | | float maximumAmplificationFactor; // >=1.0 |
416 | | unsigned long long activationThresholdBytes; |
417 | | } ACCOUNTING; |
418 | | |
419 | | typedef struct entity_stats { |
420 | | unsigned int countEverOpened; |
421 | | unsigned int currentDepth; |
422 | | unsigned int maximumDepthSeen; |
423 | | int debugLevel; |
424 | | } ENTITY_STATS; |
425 | | #endif /* XML_DTD */ |
426 | | |
427 | | typedef enum XML_Error PTRCALL Processor(XML_Parser parser, const char *start, |
428 | | const char *end, const char **endPtr); |
429 | | |
430 | | static Processor prologProcessor; |
431 | | static Processor prologInitProcessor; |
432 | | static Processor contentProcessor; |
433 | | static Processor cdataSectionProcessor; |
434 | | #ifdef XML_DTD |
435 | | static Processor ignoreSectionProcessor; |
436 | | static Processor externalParEntProcessor; |
437 | | static Processor externalParEntInitProcessor; |
438 | | static Processor entityValueProcessor; |
439 | | static Processor entityValueInitProcessor; |
440 | | #endif /* XML_DTD */ |
441 | | static Processor epilogProcessor; |
442 | | static Processor errorProcessor; |
443 | | static Processor externalEntityInitProcessor; |
444 | | static Processor externalEntityInitProcessor2; |
445 | | static Processor externalEntityInitProcessor3; |
446 | | static Processor externalEntityContentProcessor; |
447 | | static Processor internalEntityProcessor; |
448 | | |
449 | | static enum XML_Error handleUnknownEncoding(XML_Parser parser, |
450 | | const XML_Char *encodingName); |
451 | | static enum XML_Error processXmlDecl(XML_Parser parser, int isGeneralTextEntity, |
452 | | const char *s, const char *next); |
453 | | static enum XML_Error initializeEncoding(XML_Parser parser); |
454 | | static enum XML_Error doProlog(XML_Parser parser, const ENCODING *enc, |
455 | | const char *s, const char *end, int tok, |
456 | | const char *next, const char **nextPtr, |
457 | | XML_Bool haveMore, XML_Bool allowClosingDoctype, |
458 | | enum XML_Account account); |
459 | | static enum XML_Error processInternalEntity(XML_Parser parser, ENTITY *entity, |
460 | | XML_Bool betweenDecl); |
461 | | static enum XML_Error doContent(XML_Parser parser, int startTagLevel, |
462 | | const ENCODING *enc, const char *start, |
463 | | const char *end, const char **endPtr, |
464 | | XML_Bool haveMore, enum XML_Account account); |
465 | | static enum XML_Error doCdataSection(XML_Parser parser, const ENCODING *, |
466 | | const char **startPtr, const char *end, |
467 | | const char **nextPtr, XML_Bool haveMore, |
468 | | enum XML_Account account); |
469 | | #ifdef XML_DTD |
470 | | static enum XML_Error doIgnoreSection(XML_Parser parser, const ENCODING *, |
471 | | const char **startPtr, const char *end, |
472 | | const char **nextPtr, XML_Bool haveMore); |
473 | | #endif /* XML_DTD */ |
474 | | |
475 | | static void freeBindings(XML_Parser parser, BINDING *bindings); |
476 | | static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *, |
477 | | const char *s, TAG_NAME *tagNamePtr, |
478 | | BINDING **bindingsPtr, |
479 | | enum XML_Account account); |
480 | | static enum XML_Error addBinding(XML_Parser parser, PREFIX *prefix, |
481 | | const ATTRIBUTE_ID *attId, const XML_Char *uri, |
482 | | BINDING **bindingsPtr); |
483 | | static int defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, XML_Bool isCdata, |
484 | | XML_Bool isId, const XML_Char *dfltValue, |
485 | | XML_Parser parser); |
486 | | static enum XML_Error storeAttributeValue(XML_Parser parser, const ENCODING *, |
487 | | XML_Bool isCdata, const char *, |
488 | | const char *, STRING_POOL *, |
489 | | enum XML_Account account); |
490 | | static enum XML_Error appendAttributeValue(XML_Parser parser, const ENCODING *, |
491 | | XML_Bool isCdata, const char *, |
492 | | const char *, STRING_POOL *, |
493 | | enum XML_Account account); |
494 | | static ATTRIBUTE_ID *getAttributeId(XML_Parser parser, const ENCODING *enc, |
495 | | const char *start, const char *end); |
496 | | static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); |
497 | | static enum XML_Error storeEntityValue(XML_Parser parser, const ENCODING *enc, |
498 | | const char *start, const char *end, |
499 | | enum XML_Account account); |
500 | | static int reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, |
501 | | const char *start, const char *end); |
502 | | static int reportComment(XML_Parser parser, const ENCODING *enc, |
503 | | const char *start, const char *end); |
504 | | static void reportDefault(XML_Parser parser, const ENCODING *enc, |
505 | | const char *start, const char *end); |
506 | | |
507 | | static const XML_Char *getContext(XML_Parser parser); |
508 | | static XML_Bool setContext(XML_Parser parser, const XML_Char *context); |
509 | | |
510 | | static void FASTCALL normalizePublicId(XML_Char *s); |
511 | | |
512 | | static DTD *dtdCreate(const XML_Memory_Handling_Suite *ms); |
513 | | /* do not call if m_parentParser != NULL */ |
514 | | static void dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms); |
515 | | static void dtdDestroy(DTD *p, XML_Bool isDocEntity, |
516 | | const XML_Memory_Handling_Suite *ms); |
517 | | static int dtdCopy(XML_Parser oldParser, DTD *newDtd, const DTD *oldDtd, |
518 | | const XML_Memory_Handling_Suite *ms); |
519 | | static int copyEntityTable(XML_Parser oldParser, HASH_TABLE *, STRING_POOL *, |
520 | | const HASH_TABLE *); |
521 | | static NAMED *lookup(XML_Parser parser, HASH_TABLE *table, KEY name, |
522 | | size_t createSize); |
523 | | static void FASTCALL hashTableInit(HASH_TABLE *, |
524 | | const XML_Memory_Handling_Suite *ms); |
525 | | static void FASTCALL hashTableClear(HASH_TABLE *); |
526 | | static void FASTCALL hashTableDestroy(HASH_TABLE *); |
527 | | static void FASTCALL hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *); |
528 | | static NAMED *FASTCALL hashTableIterNext(HASH_TABLE_ITER *); |
529 | | |
530 | | static void FASTCALL poolInit(STRING_POOL *, |
531 | | const XML_Memory_Handling_Suite *ms); |
532 | | static void FASTCALL poolClear(STRING_POOL *); |
533 | | static void FASTCALL poolDestroy(STRING_POOL *); |
534 | | static XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, |
535 | | const char *ptr, const char *end); |
536 | | static XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, |
537 | | const char *ptr, const char *end); |
538 | | static XML_Bool FASTCALL poolGrow(STRING_POOL *pool); |
539 | | static const XML_Char *FASTCALL poolCopyString(STRING_POOL *pool, |
540 | | const XML_Char *s); |
541 | | static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, |
542 | | int n); |
543 | | static const XML_Char *FASTCALL poolAppendString(STRING_POOL *pool, |
544 | | const XML_Char *s); |
545 | | |
546 | | static int FASTCALL nextScaffoldPart(XML_Parser parser); |
547 | | static XML_Content *build_model(XML_Parser parser); |
548 | | static ELEMENT_TYPE *getElementType(XML_Parser parser, const ENCODING *enc, |
549 | | const char *ptr, const char *end); |
550 | | |
551 | | static XML_Char *copyString(const XML_Char *s, |
552 | | const XML_Memory_Handling_Suite *memsuite); |
553 | | |
554 | | static unsigned long generate_hash_secret_salt(XML_Parser parser); |
555 | | static XML_Bool startParsing(XML_Parser parser); |
556 | | |
557 | | static XML_Parser parserCreate(const XML_Char *encodingName, |
558 | | const XML_Memory_Handling_Suite *memsuite, |
559 | | const XML_Char *nameSep, DTD *dtd); |
560 | | |
561 | | static void parserInit(XML_Parser parser, const XML_Char *encodingName); |
562 | | |
563 | | #ifdef XML_DTD |
564 | | static float accountingGetCurrentAmplification(XML_Parser rootParser); |
565 | | static void accountingReportStats(XML_Parser originParser, const char *epilog); |
566 | | static void accountingOnAbort(XML_Parser originParser); |
567 | | static void accountingReportDiff(XML_Parser rootParser, |
568 | | unsigned int levelsAwayFromRootParser, |
569 | | const char *before, const char *after, |
570 | | ptrdiff_t bytesMore, int source_line, |
571 | | enum XML_Account account); |
572 | | static XML_Bool accountingDiffTolerated(XML_Parser originParser, int tok, |
573 | | const char *before, const char *after, |
574 | | int source_line, |
575 | | enum XML_Account account); |
576 | | |
577 | | static void entityTrackingReportStats(XML_Parser parser, ENTITY *entity, |
578 | | const char *action, int sourceLine); |
579 | | static void entityTrackingOnOpen(XML_Parser parser, ENTITY *entity, |
580 | | int sourceLine); |
581 | | static void entityTrackingOnClose(XML_Parser parser, ENTITY *entity, |
582 | | int sourceLine); |
583 | | |
584 | | static XML_Parser getRootParserOf(XML_Parser parser, |
585 | | unsigned int *outLevelDiff); |
586 | | #endif /* XML_DTD */ |
587 | | |
588 | | static unsigned long getDebugLevel(const char *variableName, |
589 | | unsigned long defaultDebugLevel); |
590 | | |
591 | 32.1k | #define poolStart(pool) ((pool)->start) |
592 | | #define poolEnd(pool) ((pool)->ptr) |
593 | 178k | #define poolLength(pool) ((pool)->ptr - (pool)->start) |
594 | 628 | #define poolChop(pool) ((void)--(pool->ptr)) |
595 | 151k | #define poolLastChar(pool) (((pool)->ptr)[-1]) |
596 | 10.5M | #define poolDiscard(pool) ((pool)->ptr = (pool)->start) |
597 | 100k | #define poolFinish(pool) ((pool)->start = (pool)->ptr) |
598 | | #define poolAppendChar(pool, c) \ |
599 | 23.7M | (((pool)->ptr == (pool)->end && ! poolGrow(pool)) \ |
600 | 23.7M | ? 0 \ |
601 | 23.7M | : ((*((pool)->ptr)++ = c), 1)) |
602 | | |
603 | | struct XML_ParserStruct { |
604 | | /* The first member must be m_userData so that the XML_GetUserData |
605 | | macro works. */ |
606 | | void *m_userData; |
607 | | void *m_handlerArg; |
608 | | char *m_buffer; |
609 | | const XML_Memory_Handling_Suite m_mem; |
610 | | /* first character to be parsed */ |
611 | | const char *m_bufferPtr; |
612 | | /* past last character to be parsed */ |
613 | | char *m_bufferEnd; |
614 | | /* allocated end of m_buffer */ |
615 | | const char *m_bufferLim; |
616 | | XML_Index m_parseEndByteIndex; |
617 | | const char *m_parseEndPtr; |
618 | | XML_Char *m_dataBuf; |
619 | | XML_Char *m_dataBufEnd; |
620 | | XML_StartElementHandler m_startElementHandler; |
621 | | XML_EndElementHandler m_endElementHandler; |
622 | | XML_CharacterDataHandler m_characterDataHandler; |
623 | | XML_ProcessingInstructionHandler m_processingInstructionHandler; |
624 | | XML_CommentHandler m_commentHandler; |
625 | | XML_StartCdataSectionHandler m_startCdataSectionHandler; |
626 | | XML_EndCdataSectionHandler m_endCdataSectionHandler; |
627 | | XML_DefaultHandler m_defaultHandler; |
628 | | XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler; |
629 | | XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler; |
630 | | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; |
631 | | XML_NotationDeclHandler m_notationDeclHandler; |
632 | | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; |
633 | | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; |
634 | | XML_NotStandaloneHandler m_notStandaloneHandler; |
635 | | XML_ExternalEntityRefHandler m_externalEntityRefHandler; |
636 | | XML_Parser m_externalEntityRefHandlerArg; |
637 | | XML_SkippedEntityHandler m_skippedEntityHandler; |
638 | | XML_UnknownEncodingHandler m_unknownEncodingHandler; |
639 | | XML_ElementDeclHandler m_elementDeclHandler; |
640 | | XML_AttlistDeclHandler m_attlistDeclHandler; |
641 | | XML_EntityDeclHandler m_entityDeclHandler; |
642 | | XML_XmlDeclHandler m_xmlDeclHandler; |
643 | | const ENCODING *m_encoding; |
644 | | INIT_ENCODING m_initEncoding; |
645 | | const ENCODING *m_internalEncoding; |
646 | | const XML_Char *m_protocolEncodingName; |
647 | | XML_Bool m_ns; |
648 | | XML_Bool m_ns_triplets; |
649 | | void *m_unknownEncodingMem; |
650 | | void *m_unknownEncodingData; |
651 | | void *m_unknownEncodingHandlerData; |
652 | | void(XMLCALL *m_unknownEncodingRelease)(void *); |
653 | | PROLOG_STATE m_prologState; |
654 | | Processor *m_processor; |
655 | | enum XML_Error m_errorCode; |
656 | | const char *m_eventPtr; |
657 | | const char *m_eventEndPtr; |
658 | | const char *m_positionPtr; |
659 | | OPEN_INTERNAL_ENTITY *m_openInternalEntities; |
660 | | OPEN_INTERNAL_ENTITY *m_freeInternalEntities; |
661 | | XML_Bool m_defaultExpandInternalEntities; |
662 | | int m_tagLevel; |
663 | | ENTITY *m_declEntity; |
664 | | const XML_Char *m_doctypeName; |
665 | | const XML_Char *m_doctypeSysid; |
666 | | const XML_Char *m_doctypePubid; |
667 | | const XML_Char *m_declAttributeType; |
668 | | const XML_Char *m_declNotationName; |
669 | | const XML_Char *m_declNotationPublicId; |
670 | | ELEMENT_TYPE *m_declElementType; |
671 | | ATTRIBUTE_ID *m_declAttributeId; |
672 | | XML_Bool m_declAttributeIsCdata; |
673 | | XML_Bool m_declAttributeIsId; |
674 | | DTD *m_dtd; |
675 | | const XML_Char *m_curBase; |
676 | | TAG *m_tagStack; |
677 | | TAG *m_freeTagList; |
678 | | BINDING *m_inheritedBindings; |
679 | | BINDING *m_freeBindingList; |
680 | | int m_attsSize; |
681 | | int m_nSpecifiedAtts; |
682 | | int m_idAttIndex; |
683 | | ATTRIBUTE *m_atts; |
684 | | NS_ATT *m_nsAtts; |
685 | | unsigned long m_nsAttsVersion; |
686 | | unsigned char m_nsAttsPower; |
687 | | #ifdef XML_ATTR_INFO |
688 | | XML_AttrInfo *m_attInfo; |
689 | | #endif |
690 | | POSITION m_position; |
691 | | STRING_POOL m_tempPool; |
692 | | STRING_POOL m_temp2Pool; |
693 | | char *m_groupConnector; |
694 | | unsigned int m_groupSize; |
695 | | XML_Char m_namespaceSeparator; |
696 | | XML_Parser m_parentParser; |
697 | | XML_ParsingStatus m_parsingStatus; |
698 | | #ifdef XML_DTD |
699 | | XML_Bool m_isParamEntity; |
700 | | XML_Bool m_useForeignDTD; |
701 | | enum XML_ParamEntityParsing m_paramEntityParsing; |
702 | | #endif |
703 | | unsigned long m_hash_secret_salt; |
704 | | #ifdef XML_DTD |
705 | | ACCOUNTING m_accounting; |
706 | | ENTITY_STATS m_entity_stats; |
707 | | #endif |
708 | | }; |
709 | | |
710 | 2.43M | #define MALLOC(parser, s) (parser->m_mem.malloc_fcn((s))) |
711 | 12.0k | #define REALLOC(parser, p, s) (parser->m_mem.realloc_fcn((p), (s))) |
712 | 2.47M | #define FREE(parser, p) (parser->m_mem.free_fcn((p))) |
713 | | |
714 | | XML_Parser XMLCALL |
715 | 7.48k | XML_ParserCreate(const XML_Char *encodingName) { |
716 | 7.48k | return XML_ParserCreate_MM(encodingName, NULL, NULL); |
717 | 7.48k | } |
718 | | |
719 | | XML_Parser XMLCALL |
720 | 0 | XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) { |
721 | 0 | XML_Char tmp[2] = {nsSep, 0}; |
722 | 0 | return XML_ParserCreate_MM(encodingName, NULL, tmp); |
723 | 0 | } |
724 | | |
725 | | static const XML_Char implicitContext[] |
726 | | = {ASCII_x, ASCII_m, ASCII_l, ASCII_EQUALS, ASCII_h, |
727 | | ASCII_t, ASCII_t, ASCII_p, ASCII_COLON, ASCII_SLASH, |
728 | | ASCII_SLASH, ASCII_w, ASCII_w, ASCII_w, ASCII_PERIOD, |
729 | | ASCII_w, ASCII_3, ASCII_PERIOD, ASCII_o, ASCII_r, |
730 | | ASCII_g, ASCII_SLASH, ASCII_X, ASCII_M, ASCII_L, |
731 | | ASCII_SLASH, ASCII_1, ASCII_9, ASCII_9, ASCII_8, |
732 | | ASCII_SLASH, ASCII_n, ASCII_a, ASCII_m, ASCII_e, |
733 | | ASCII_s, ASCII_p, ASCII_a, ASCII_c, ASCII_e, |
734 | | '\0'}; |
735 | | |
736 | | /* To avoid warnings about unused functions: */ |
737 | | #if ! defined(HAVE_ARC4RANDOM_BUF) && ! defined(HAVE_ARC4RANDOM) |
738 | | |
739 | | # if defined(HAVE_GETRANDOM) || defined(HAVE_SYSCALL_GETRANDOM) |
740 | | |
741 | | /* Obtain entropy on Linux 3.17+ */ |
742 | | static int |
743 | 7.48k | writeRandomBytes_getrandom_nonblock(void *target, size_t count) { |
744 | 7.48k | int success = 0; /* full count bytes written? */ |
745 | 7.48k | size_t bytesWrittenTotal = 0; |
746 | 7.48k | const unsigned int getrandomFlags = GRND_NONBLOCK; |
747 | | |
748 | 7.48k | do { |
749 | 7.48k | void *const currentTarget = (void *)((char *)target + bytesWrittenTotal); |
750 | 7.48k | const size_t bytesToWrite = count - bytesWrittenTotal; |
751 | | |
752 | 7.48k | const int bytesWrittenMore = |
753 | 7.48k | # if defined(HAVE_GETRANDOM) |
754 | 7.48k | getrandom(currentTarget, bytesToWrite, getrandomFlags); |
755 | | # else |
756 | | syscall(SYS_getrandom, currentTarget, bytesToWrite, getrandomFlags); |
757 | | # endif |
758 | | |
759 | 7.48k | if (bytesWrittenMore > 0) { |
760 | 7.48k | bytesWrittenTotal += bytesWrittenMore; |
761 | 7.48k | if (bytesWrittenTotal >= count) |
762 | 7.48k | success = 1; |
763 | 7.48k | } |
764 | 7.48k | } while (! success && (errno == EINTR)); |
765 | | |
766 | 7.48k | return success; |
767 | 7.48k | } |
768 | | |
769 | | # endif /* defined(HAVE_GETRANDOM) || defined(HAVE_SYSCALL_GETRANDOM) */ |
770 | | |
771 | | # if ! defined(_WIN32) && defined(XML_DEV_URANDOM) |
772 | | |
773 | | /* Extract entropy from /dev/urandom */ |
774 | | static int |
775 | 0 | writeRandomBytes_dev_urandom(void *target, size_t count) { |
776 | 0 | int success = 0; /* full count bytes written? */ |
777 | 0 | size_t bytesWrittenTotal = 0; |
778 | |
|
779 | 0 | const int fd = open("/dev/urandom", O_RDONLY); |
780 | 0 | if (fd < 0) { |
781 | 0 | return 0; |
782 | 0 | } |
783 | | |
784 | 0 | do { |
785 | 0 | void *const currentTarget = (void *)((char *)target + bytesWrittenTotal); |
786 | 0 | const size_t bytesToWrite = count - bytesWrittenTotal; |
787 | |
|
788 | 0 | const ssize_t bytesWrittenMore = read(fd, currentTarget, bytesToWrite); |
789 | |
|
790 | 0 | if (bytesWrittenMore > 0) { |
791 | 0 | bytesWrittenTotal += bytesWrittenMore; |
792 | 0 | if (bytesWrittenTotal >= count) |
793 | 0 | success = 1; |
794 | 0 | } |
795 | 0 | } while (! success && (errno == EINTR)); |
796 | |
|
797 | 0 | close(fd); |
798 | 0 | return success; |
799 | 0 | } |
800 | | |
801 | | # endif /* ! defined(_WIN32) && defined(XML_DEV_URANDOM) */ |
802 | | |
803 | | #endif /* ! defined(HAVE_ARC4RANDOM_BUF) && ! defined(HAVE_ARC4RANDOM) */ |
804 | | |
805 | | #if defined(HAVE_ARC4RANDOM) && ! defined(HAVE_ARC4RANDOM_BUF) |
806 | | |
807 | | static void |
808 | | writeRandomBytes_arc4random(void *target, size_t count) { |
809 | | size_t bytesWrittenTotal = 0; |
810 | | |
811 | | while (bytesWrittenTotal < count) { |
812 | | const uint32_t random32 = arc4random(); |
813 | | size_t i = 0; |
814 | | |
815 | | for (; (i < sizeof(random32)) && (bytesWrittenTotal < count); |
816 | | i++, bytesWrittenTotal++) { |
817 | | const uint8_t random8 = (uint8_t)(random32 >> (i * 8)); |
818 | | ((uint8_t *)target)[bytesWrittenTotal] = random8; |
819 | | } |
820 | | } |
821 | | } |
822 | | |
823 | | #endif /* defined(HAVE_ARC4RANDOM) && ! defined(HAVE_ARC4RANDOM_BUF) */ |
824 | | |
825 | | #ifdef _WIN32 |
826 | | |
827 | | /* Provide declaration of rand_s() for MinGW-32 (not 64, which has it), |
828 | | as it didn't declare it in its header prior to version 5.3.0 of its |
829 | | runtime package (mingwrt, containing stdlib.h). The upstream fix |
830 | | was introduced at https://osdn.net/projects/mingw/ticket/39658 . */ |
831 | | # if defined(__MINGW32__) && defined(__MINGW32_VERSION) \ |
832 | | && __MINGW32_VERSION < 5003000L && ! defined(__MINGW64_VERSION_MAJOR) |
833 | | __declspec(dllimport) int rand_s(unsigned int *); |
834 | | # endif |
835 | | |
836 | | /* Obtain entropy on Windows using the rand_s() function which |
837 | | * generates cryptographically secure random numbers. Internally it |
838 | | * uses RtlGenRandom API which is present in Windows XP and later. |
839 | | */ |
840 | | static int |
841 | | writeRandomBytes_rand_s(void *target, size_t count) { |
842 | | size_t bytesWrittenTotal = 0; |
843 | | |
844 | | while (bytesWrittenTotal < count) { |
845 | | unsigned int random32 = 0; |
846 | | size_t i = 0; |
847 | | |
848 | | if (rand_s(&random32)) |
849 | | return 0; /* failure */ |
850 | | |
851 | | for (; (i < sizeof(random32)) && (bytesWrittenTotal < count); |
852 | | i++, bytesWrittenTotal++) { |
853 | | const uint8_t random8 = (uint8_t)(random32 >> (i * 8)); |
854 | | ((uint8_t *)target)[bytesWrittenTotal] = random8; |
855 | | } |
856 | | } |
857 | | return 1; /* success */ |
858 | | } |
859 | | |
860 | | #endif /* _WIN32 */ |
861 | | |
862 | | #if ! defined(HAVE_ARC4RANDOM_BUF) && ! defined(HAVE_ARC4RANDOM) |
863 | | |
864 | | static unsigned long |
865 | 0 | gather_time_entropy(void) { |
866 | | # ifdef _WIN32 |
867 | | FILETIME ft; |
868 | | GetSystemTimeAsFileTime(&ft); /* never fails */ |
869 | | return ft.dwHighDateTime ^ ft.dwLowDateTime; |
870 | | # else |
871 | 0 | struct timeval tv; |
872 | 0 | int gettimeofday_res; |
873 | |
|
874 | 0 | gettimeofday_res = gettimeofday(&tv, NULL); |
875 | |
|
876 | 0 | # if defined(NDEBUG) |
877 | 0 | (void)gettimeofday_res; |
878 | | # else |
879 | | assert(gettimeofday_res == 0); |
880 | | # endif /* defined(NDEBUG) */ |
881 | | |
882 | | /* Microseconds time is <20 bits entropy */ |
883 | 0 | return tv.tv_usec; |
884 | 0 | # endif |
885 | 0 | } |
886 | | |
887 | | #endif /* ! defined(HAVE_ARC4RANDOM_BUF) && ! defined(HAVE_ARC4RANDOM) */ |
888 | | |
889 | | static unsigned long |
890 | 7.48k | ENTROPY_DEBUG(const char *label, unsigned long entropy) { |
891 | 7.48k | if (getDebugLevel("EXPAT_ENTROPY_DEBUG", 0) >= 1u) { |
892 | 0 | fprintf(stderr, "expat: Entropy: %s --> 0x%0*lx (%lu bytes)\n", label, |
893 | 0 | (int)sizeof(entropy) * 2, entropy, (unsigned long)sizeof(entropy)); |
894 | 0 | } |
895 | 7.48k | return entropy; |
896 | 7.48k | } |
897 | | |
898 | | static unsigned long |
899 | 7.48k | generate_hash_secret_salt(XML_Parser parser) { |
900 | 7.48k | unsigned long entropy; |
901 | 7.48k | (void)parser; |
902 | | |
903 | | /* "Failproof" high quality providers: */ |
904 | | #if defined(HAVE_ARC4RANDOM_BUF) |
905 | | arc4random_buf(&entropy, sizeof(entropy)); |
906 | | return ENTROPY_DEBUG("arc4random_buf", entropy); |
907 | | #elif defined(HAVE_ARC4RANDOM) |
908 | | writeRandomBytes_arc4random((void *)&entropy, sizeof(entropy)); |
909 | | return ENTROPY_DEBUG("arc4random", entropy); |
910 | | #else |
911 | | /* Try high quality providers first .. */ |
912 | | # ifdef _WIN32 |
913 | | if (writeRandomBytes_rand_s((void *)&entropy, sizeof(entropy))) { |
914 | | return ENTROPY_DEBUG("rand_s", entropy); |
915 | | } |
916 | | # elif defined(HAVE_GETRANDOM) || defined(HAVE_SYSCALL_GETRANDOM) |
917 | 7.48k | if (writeRandomBytes_getrandom_nonblock((void *)&entropy, sizeof(entropy))) { |
918 | 7.48k | return ENTROPY_DEBUG("getrandom", entropy); |
919 | 7.48k | } |
920 | 0 | # endif |
921 | 0 | # if ! defined(_WIN32) && defined(XML_DEV_URANDOM) |
922 | 0 | if (writeRandomBytes_dev_urandom((void *)&entropy, sizeof(entropy))) { |
923 | 0 | return ENTROPY_DEBUG("/dev/urandom", entropy); |
924 | 0 | } |
925 | 0 | # endif /* ! defined(_WIN32) && defined(XML_DEV_URANDOM) */ |
926 | | /* .. and self-made low quality for backup: */ |
927 | | |
928 | | /* Process ID is 0 bits entropy if attacker has local access */ |
929 | 0 | entropy = gather_time_entropy() ^ getpid(); |
930 | | |
931 | | /* Factors are 2^31-1 and 2^61-1 (Mersenne primes M31 and M61) */ |
932 | 0 | if (sizeof(unsigned long) == 4) { |
933 | 0 | return ENTROPY_DEBUG("fallback(4)", entropy * 2147483647); |
934 | 0 | } else { |
935 | 0 | return ENTROPY_DEBUG("fallback(8)", |
936 | 0 | entropy * (unsigned long)2305843009213693951ULL); |
937 | 0 | } |
938 | 0 | #endif |
939 | 0 | } |
940 | | |
941 | | static unsigned long |
942 | 11.9M | get_hash_secret_salt(XML_Parser parser) { |
943 | 11.9M | if (parser->m_parentParser != NULL) |
944 | 0 | return get_hash_secret_salt(parser->m_parentParser); |
945 | 11.9M | return parser->m_hash_secret_salt; |
946 | 11.9M | } |
947 | | |
948 | | static XML_Bool /* only valid for root parser */ |
949 | 7.48k | startParsing(XML_Parser parser) { |
950 | | /* hash functions must be initialized before setContext() is called */ |
951 | 7.48k | if (parser->m_hash_secret_salt == 0) |
952 | 7.48k | parser->m_hash_secret_salt = generate_hash_secret_salt(parser); |
953 | 7.48k | if (parser->m_ns) { |
954 | | /* implicit context only set for root parser, since child |
955 | | parsers (i.e. external entity parsers) will inherit it |
956 | | */ |
957 | 0 | return setContext(parser, implicitContext); |
958 | 0 | } |
959 | 7.48k | return XML_TRUE; |
960 | 7.48k | } |
961 | | |
962 | | XML_Parser XMLCALL |
963 | | XML_ParserCreate_MM(const XML_Char *encodingName, |
964 | | const XML_Memory_Handling_Suite *memsuite, |
965 | 7.48k | const XML_Char *nameSep) { |
966 | 7.48k | return parserCreate(encodingName, memsuite, nameSep, NULL); |
967 | 7.48k | } |
968 | | |
969 | | static XML_Parser |
970 | | parserCreate(const XML_Char *encodingName, |
971 | | const XML_Memory_Handling_Suite *memsuite, const XML_Char *nameSep, |
972 | 7.48k | DTD *dtd) { |
973 | 7.48k | XML_Parser parser; |
974 | | |
975 | 7.48k | if (memsuite) { |
976 | 0 | XML_Memory_Handling_Suite *mtemp; |
977 | 0 | parser = memsuite->malloc_fcn(sizeof(struct XML_ParserStruct)); |
978 | 0 | if (parser != NULL) { |
979 | 0 | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); |
980 | 0 | mtemp->malloc_fcn = memsuite->malloc_fcn; |
981 | 0 | mtemp->realloc_fcn = memsuite->realloc_fcn; |
982 | 0 | mtemp->free_fcn = memsuite->free_fcn; |
983 | 0 | } |
984 | 7.48k | } else { |
985 | 7.48k | XML_Memory_Handling_Suite *mtemp; |
986 | 7.48k | parser = (XML_Parser)malloc(sizeof(struct XML_ParserStruct)); |
987 | 7.48k | if (parser != NULL) { |
988 | 7.48k | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); |
989 | 7.48k | mtemp->malloc_fcn = malloc; |
990 | 7.48k | mtemp->realloc_fcn = realloc; |
991 | 7.48k | mtemp->free_fcn = free; |
992 | 7.48k | } |
993 | 7.48k | } |
994 | | |
995 | 7.48k | if (! parser) |
996 | 0 | return parser; |
997 | | |
998 | 7.48k | parser->m_buffer = NULL; |
999 | 7.48k | parser->m_bufferLim = NULL; |
1000 | | |
1001 | 7.48k | parser->m_attsSize = INIT_ATTS_SIZE; |
1002 | 7.48k | parser->m_atts |
1003 | 7.48k | = (ATTRIBUTE *)MALLOC(parser, parser->m_attsSize * sizeof(ATTRIBUTE)); |
1004 | 7.48k | if (parser->m_atts == NULL) { |
1005 | 0 | FREE(parser, parser); |
1006 | 0 | return NULL; |
1007 | 0 | } |
1008 | | #ifdef XML_ATTR_INFO |
1009 | | parser->m_attInfo = (XML_AttrInfo *)MALLOC( |
1010 | | parser, parser->m_attsSize * sizeof(XML_AttrInfo)); |
1011 | | if (parser->m_attInfo == NULL) { |
1012 | | FREE(parser, parser->m_atts); |
1013 | | FREE(parser, parser); |
1014 | | return NULL; |
1015 | | } |
1016 | | #endif |
1017 | 7.48k | parser->m_dataBuf |
1018 | 7.48k | = (XML_Char *)MALLOC(parser, INIT_DATA_BUF_SIZE * sizeof(XML_Char)); |
1019 | 7.48k | if (parser->m_dataBuf == NULL) { |
1020 | 0 | FREE(parser, parser->m_atts); |
1021 | | #ifdef XML_ATTR_INFO |
1022 | | FREE(parser, parser->m_attInfo); |
1023 | | #endif |
1024 | 0 | FREE(parser, parser); |
1025 | 0 | return NULL; |
1026 | 0 | } |
1027 | 7.48k | parser->m_dataBufEnd = parser->m_dataBuf + INIT_DATA_BUF_SIZE; |
1028 | | |
1029 | 7.48k | if (dtd) |
1030 | 0 | parser->m_dtd = dtd; |
1031 | 7.48k | else { |
1032 | 7.48k | parser->m_dtd = dtdCreate(&parser->m_mem); |
1033 | 7.48k | if (parser->m_dtd == NULL) { |
1034 | 0 | FREE(parser, parser->m_dataBuf); |
1035 | 0 | FREE(parser, parser->m_atts); |
1036 | | #ifdef XML_ATTR_INFO |
1037 | | FREE(parser, parser->m_attInfo); |
1038 | | #endif |
1039 | 0 | FREE(parser, parser); |
1040 | 0 | return NULL; |
1041 | 0 | } |
1042 | 7.48k | } |
1043 | | |
1044 | 7.48k | parser->m_freeBindingList = NULL; |
1045 | 7.48k | parser->m_freeTagList = NULL; |
1046 | 7.48k | parser->m_freeInternalEntities = NULL; |
1047 | | |
1048 | 7.48k | parser->m_groupSize = 0; |
1049 | 7.48k | parser->m_groupConnector = NULL; |
1050 | | |
1051 | 7.48k | parser->m_unknownEncodingHandler = NULL; |
1052 | 7.48k | parser->m_unknownEncodingHandlerData = NULL; |
1053 | | |
1054 | 7.48k | parser->m_namespaceSeparator = ASCII_EXCL; |
1055 | 7.48k | parser->m_ns = XML_FALSE; |
1056 | 7.48k | parser->m_ns_triplets = XML_FALSE; |
1057 | | |
1058 | 7.48k | parser->m_nsAtts = NULL; |
1059 | 7.48k | parser->m_nsAttsVersion = 0; |
1060 | 7.48k | parser->m_nsAttsPower = 0; |
1061 | | |
1062 | 7.48k | parser->m_protocolEncodingName = NULL; |
1063 | | |
1064 | 7.48k | poolInit(&parser->m_tempPool, &(parser->m_mem)); |
1065 | 7.48k | poolInit(&parser->m_temp2Pool, &(parser->m_mem)); |
1066 | 7.48k | parserInit(parser, encodingName); |
1067 | | |
1068 | 7.48k | if (encodingName && ! parser->m_protocolEncodingName) { |
1069 | 0 | XML_ParserFree(parser); |
1070 | 0 | return NULL; |
1071 | 0 | } |
1072 | | |
1073 | 7.48k | if (nameSep) { |
1074 | 0 | parser->m_ns = XML_TRUE; |
1075 | 0 | parser->m_internalEncoding = XmlGetInternalEncodingNS(); |
1076 | 0 | parser->m_namespaceSeparator = *nameSep; |
1077 | 7.48k | } else { |
1078 | 7.48k | parser->m_internalEncoding = XmlGetInternalEncoding(); |
1079 | 7.48k | } |
1080 | | |
1081 | 7.48k | return parser; |
1082 | 7.48k | } |
1083 | | |
1084 | | static void |
1085 | 7.48k | parserInit(XML_Parser parser, const XML_Char *encodingName) { |
1086 | 7.48k | parser->m_processor = prologInitProcessor; |
1087 | 7.48k | XmlPrologStateInit(&parser->m_prologState); |
1088 | 7.48k | if (encodingName != NULL) { |
1089 | 0 | parser->m_protocolEncodingName = copyString(encodingName, &(parser->m_mem)); |
1090 | 0 | } |
1091 | 7.48k | parser->m_curBase = NULL; |
1092 | 7.48k | XmlInitEncoding(&parser->m_initEncoding, &parser->m_encoding, 0); |
1093 | 7.48k | parser->m_userData = NULL; |
1094 | 7.48k | parser->m_handlerArg = NULL; |
1095 | 7.48k | parser->m_startElementHandler = NULL; |
1096 | 7.48k | parser->m_endElementHandler = NULL; |
1097 | 7.48k | parser->m_characterDataHandler = NULL; |
1098 | 7.48k | parser->m_processingInstructionHandler = NULL; |
1099 | 7.48k | parser->m_commentHandler = NULL; |
1100 | 7.48k | parser->m_startCdataSectionHandler = NULL; |
1101 | 7.48k | parser->m_endCdataSectionHandler = NULL; |
1102 | 7.48k | parser->m_defaultHandler = NULL; |
1103 | 7.48k | parser->m_startDoctypeDeclHandler = NULL; |
1104 | 7.48k | parser->m_endDoctypeDeclHandler = NULL; |
1105 | 7.48k | parser->m_unparsedEntityDeclHandler = NULL; |
1106 | 7.48k | parser->m_notationDeclHandler = NULL; |
1107 | 7.48k | parser->m_startNamespaceDeclHandler = NULL; |
1108 | 7.48k | parser->m_endNamespaceDeclHandler = NULL; |
1109 | 7.48k | parser->m_notStandaloneHandler = NULL; |
1110 | 7.48k | parser->m_externalEntityRefHandler = NULL; |
1111 | 7.48k | parser->m_externalEntityRefHandlerArg = parser; |
1112 | 7.48k | parser->m_skippedEntityHandler = NULL; |
1113 | 7.48k | parser->m_elementDeclHandler = NULL; |
1114 | 7.48k | parser->m_attlistDeclHandler = NULL; |
1115 | 7.48k | parser->m_entityDeclHandler = NULL; |
1116 | 7.48k | parser->m_xmlDeclHandler = NULL; |
1117 | 7.48k | parser->m_bufferPtr = parser->m_buffer; |
1118 | 7.48k | parser->m_bufferEnd = parser->m_buffer; |
1119 | 7.48k | parser->m_parseEndByteIndex = 0; |
1120 | 7.48k | parser->m_parseEndPtr = NULL; |
1121 | 7.48k | parser->m_declElementType = NULL; |
1122 | 7.48k | parser->m_declAttributeId = NULL; |
1123 | 7.48k | parser->m_declEntity = NULL; |
1124 | 7.48k | parser->m_doctypeName = NULL; |
1125 | 7.48k | parser->m_doctypeSysid = NULL; |
1126 | 7.48k | parser->m_doctypePubid = NULL; |
1127 | 7.48k | parser->m_declAttributeType = NULL; |
1128 | 7.48k | parser->m_declNotationName = NULL; |
1129 | 7.48k | parser->m_declNotationPublicId = NULL; |
1130 | 7.48k | parser->m_declAttributeIsCdata = XML_FALSE; |
1131 | 7.48k | parser->m_declAttributeIsId = XML_FALSE; |
1132 | 7.48k | memset(&parser->m_position, 0, sizeof(POSITION)); |
1133 | 7.48k | parser->m_errorCode = XML_ERROR_NONE; |
1134 | 7.48k | parser->m_eventPtr = NULL; |
1135 | 7.48k | parser->m_eventEndPtr = NULL; |
1136 | 7.48k | parser->m_positionPtr = NULL; |
1137 | 7.48k | parser->m_openInternalEntities = NULL; |
1138 | 7.48k | parser->m_defaultExpandInternalEntities = XML_TRUE; |
1139 | 7.48k | parser->m_tagLevel = 0; |
1140 | 7.48k | parser->m_tagStack = NULL; |
1141 | 7.48k | parser->m_inheritedBindings = NULL; |
1142 | 7.48k | parser->m_nSpecifiedAtts = 0; |
1143 | 7.48k | parser->m_unknownEncodingMem = NULL; |
1144 | 7.48k | parser->m_unknownEncodingRelease = NULL; |
1145 | 7.48k | parser->m_unknownEncodingData = NULL; |
1146 | 7.48k | parser->m_parentParser = NULL; |
1147 | 7.48k | parser->m_parsingStatus.parsing = XML_INITIALIZED; |
1148 | | #ifdef XML_DTD |
1149 | | parser->m_isParamEntity = XML_FALSE; |
1150 | | parser->m_useForeignDTD = XML_FALSE; |
1151 | | parser->m_paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; |
1152 | | #endif |
1153 | 7.48k | parser->m_hash_secret_salt = 0; |
1154 | | |
1155 | | #ifdef XML_DTD |
1156 | | memset(&parser->m_accounting, 0, sizeof(ACCOUNTING)); |
1157 | | parser->m_accounting.debugLevel = getDebugLevel("EXPAT_ACCOUNTING_DEBUG", 0u); |
1158 | | parser->m_accounting.maximumAmplificationFactor |
1159 | | = EXPAT_BILLION_LAUGHS_ATTACK_PROTECTION_MAXIMUM_AMPLIFICATION_DEFAULT; |
1160 | | parser->m_accounting.activationThresholdBytes |
1161 | | = EXPAT_BILLION_LAUGHS_ATTACK_PROTECTION_ACTIVATION_THRESHOLD_DEFAULT; |
1162 | | |
1163 | | memset(&parser->m_entity_stats, 0, sizeof(ENTITY_STATS)); |
1164 | | parser->m_entity_stats.debugLevel = getDebugLevel("EXPAT_ENTITY_DEBUG", 0u); |
1165 | | #endif |
1166 | 7.48k | } |
1167 | | |
1168 | | /* moves list of bindings to m_freeBindingList */ |
1169 | | static void FASTCALL |
1170 | 0 | moveToFreeBindingList(XML_Parser parser, BINDING *bindings) { |
1171 | 0 | while (bindings) { |
1172 | 0 | BINDING *b = bindings; |
1173 | 0 | bindings = bindings->nextTagBinding; |
1174 | 0 | b->nextTagBinding = parser->m_freeBindingList; |
1175 | 0 | parser->m_freeBindingList = b; |
1176 | 0 | } |
1177 | 0 | } |
1178 | | |
1179 | | XML_Bool XMLCALL |
1180 | 0 | XML_ParserReset(XML_Parser parser, const XML_Char *encodingName) { |
1181 | 0 | TAG *tStk; |
1182 | 0 | OPEN_INTERNAL_ENTITY *openEntityList; |
1183 | |
|
1184 | 0 | if (parser == NULL) |
1185 | 0 | return XML_FALSE; |
1186 | | |
1187 | 0 | if (parser->m_parentParser) |
1188 | 0 | return XML_FALSE; |
1189 | | /* move m_tagStack to m_freeTagList */ |
1190 | 0 | tStk = parser->m_tagStack; |
1191 | 0 | while (tStk) { |
1192 | 0 | TAG *tag = tStk; |
1193 | 0 | tStk = tStk->parent; |
1194 | 0 | tag->parent = parser->m_freeTagList; |
1195 | 0 | moveToFreeBindingList(parser, tag->bindings); |
1196 | 0 | tag->bindings = NULL; |
1197 | 0 | parser->m_freeTagList = tag; |
1198 | 0 | } |
1199 | | /* move m_openInternalEntities to m_freeInternalEntities */ |
1200 | 0 | openEntityList = parser->m_openInternalEntities; |
1201 | 0 | while (openEntityList) { |
1202 | 0 | OPEN_INTERNAL_ENTITY *openEntity = openEntityList; |
1203 | 0 | openEntityList = openEntity->next; |
1204 | 0 | openEntity->next = parser->m_freeInternalEntities; |
1205 | 0 | parser->m_freeInternalEntities = openEntity; |
1206 | 0 | } |
1207 | 0 | moveToFreeBindingList(parser, parser->m_inheritedBindings); |
1208 | 0 | FREE(parser, parser->m_unknownEncodingMem); |
1209 | 0 | if (parser->m_unknownEncodingRelease) |
1210 | 0 | parser->m_unknownEncodingRelease(parser->m_unknownEncodingData); |
1211 | 0 | poolClear(&parser->m_tempPool); |
1212 | 0 | poolClear(&parser->m_temp2Pool); |
1213 | 0 | FREE(parser, (void *)parser->m_protocolEncodingName); |
1214 | 0 | parser->m_protocolEncodingName = NULL; |
1215 | 0 | parserInit(parser, encodingName); |
1216 | 0 | dtdReset(parser->m_dtd, &parser->m_mem); |
1217 | 0 | return XML_TRUE; |
1218 | 0 | } |
1219 | | |
1220 | | enum XML_Status XMLCALL |
1221 | 0 | XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) { |
1222 | 0 | if (parser == NULL) |
1223 | 0 | return XML_STATUS_ERROR; |
1224 | | /* Block after XML_Parse()/XML_ParseBuffer() has been called. |
1225 | | XXX There's no way for the caller to determine which of the |
1226 | | XXX possible error cases caused the XML_STATUS_ERROR return. |
1227 | | */ |
1228 | 0 | if (parser->m_parsingStatus.parsing == XML_PARSING |
1229 | 0 | || parser->m_parsingStatus.parsing == XML_SUSPENDED) |
1230 | 0 | return XML_STATUS_ERROR; |
1231 | | |
1232 | | /* Get rid of any previous encoding name */ |
1233 | 0 | FREE(parser, (void *)parser->m_protocolEncodingName); |
1234 | |
|
1235 | 0 | if (encodingName == NULL) |
1236 | | /* No new encoding name */ |
1237 | 0 | parser->m_protocolEncodingName = NULL; |
1238 | 0 | else { |
1239 | | /* Copy the new encoding name into allocated memory */ |
1240 | 0 | parser->m_protocolEncodingName = copyString(encodingName, &(parser->m_mem)); |
1241 | 0 | if (! parser->m_protocolEncodingName) |
1242 | 0 | return XML_STATUS_ERROR; |
1243 | 0 | } |
1244 | 0 | return XML_STATUS_OK; |
1245 | 0 | } |
1246 | | |
1247 | | XML_Parser XMLCALL |
1248 | | XML_ExternalEntityParserCreate(XML_Parser oldParser, const XML_Char *context, |
1249 | 0 | const XML_Char *encodingName) { |
1250 | 0 | XML_Parser parser = oldParser; |
1251 | 0 | DTD *newDtd = NULL; |
1252 | 0 | DTD *oldDtd; |
1253 | 0 | XML_StartElementHandler oldStartElementHandler; |
1254 | 0 | XML_EndElementHandler oldEndElementHandler; |
1255 | 0 | XML_CharacterDataHandler oldCharacterDataHandler; |
1256 | 0 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler; |
1257 | 0 | XML_CommentHandler oldCommentHandler; |
1258 | 0 | XML_StartCdataSectionHandler oldStartCdataSectionHandler; |
1259 | 0 | XML_EndCdataSectionHandler oldEndCdataSectionHandler; |
1260 | 0 | XML_DefaultHandler oldDefaultHandler; |
1261 | 0 | XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler; |
1262 | 0 | XML_NotationDeclHandler oldNotationDeclHandler; |
1263 | 0 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler; |
1264 | 0 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler; |
1265 | 0 | XML_NotStandaloneHandler oldNotStandaloneHandler; |
1266 | 0 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler; |
1267 | 0 | XML_SkippedEntityHandler oldSkippedEntityHandler; |
1268 | 0 | XML_UnknownEncodingHandler oldUnknownEncodingHandler; |
1269 | 0 | XML_ElementDeclHandler oldElementDeclHandler; |
1270 | 0 | XML_AttlistDeclHandler oldAttlistDeclHandler; |
1271 | 0 | XML_EntityDeclHandler oldEntityDeclHandler; |
1272 | 0 | XML_XmlDeclHandler oldXmlDeclHandler; |
1273 | 0 | ELEMENT_TYPE *oldDeclElementType; |
1274 | |
|
1275 | 0 | void *oldUserData; |
1276 | 0 | void *oldHandlerArg; |
1277 | 0 | XML_Bool oldDefaultExpandInternalEntities; |
1278 | 0 | XML_Parser oldExternalEntityRefHandlerArg; |
1279 | | #ifdef XML_DTD |
1280 | | enum XML_ParamEntityParsing oldParamEntityParsing; |
1281 | | int oldInEntityValue; |
1282 | | #endif |
1283 | 0 | XML_Bool oldns_triplets; |
1284 | | /* Note that the new parser shares the same hash secret as the old |
1285 | | parser, so that dtdCopy and copyEntityTable can lookup values |
1286 | | from hash tables associated with either parser without us having |
1287 | | to worry which hash secrets each table has. |
1288 | | */ |
1289 | 0 | unsigned long oldhash_secret_salt; |
1290 | | |
1291 | | /* Validate the oldParser parameter before we pull everything out of it */ |
1292 | 0 | if (oldParser == NULL) |
1293 | 0 | return NULL; |
1294 | | |
1295 | | /* Stash the original parser contents on the stack */ |
1296 | 0 | oldDtd = parser->m_dtd; |
1297 | 0 | oldStartElementHandler = parser->m_startElementHandler; |
1298 | 0 | oldEndElementHandler = parser->m_endElementHandler; |
1299 | 0 | oldCharacterDataHandler = parser->m_characterDataHandler; |
1300 | 0 | oldProcessingInstructionHandler = parser->m_processingInstructionHandler; |
1301 | 0 | oldCommentHandler = parser->m_commentHandler; |
1302 | 0 | oldStartCdataSectionHandler = parser->m_startCdataSectionHandler; |
1303 | 0 | oldEndCdataSectionHandler = parser->m_endCdataSectionHandler; |
1304 | 0 | oldDefaultHandler = parser->m_defaultHandler; |
1305 | 0 | oldUnparsedEntityDeclHandler = parser->m_unparsedEntityDeclHandler; |
1306 | 0 | oldNotationDeclHandler = parser->m_notationDeclHandler; |
1307 | 0 | oldStartNamespaceDeclHandler = parser->m_startNamespaceDeclHandler; |
1308 | 0 | oldEndNamespaceDeclHandler = parser->m_endNamespaceDeclHandler; |
1309 | 0 | oldNotStandaloneHandler = parser->m_notStandaloneHandler; |
1310 | 0 | oldExternalEntityRefHandler = parser->m_externalEntityRefHandler; |
1311 | 0 | oldSkippedEntityHandler = parser->m_skippedEntityHandler; |
1312 | 0 | oldUnknownEncodingHandler = parser->m_unknownEncodingHandler; |
1313 | 0 | oldElementDeclHandler = parser->m_elementDeclHandler; |
1314 | 0 | oldAttlistDeclHandler = parser->m_attlistDeclHandler; |
1315 | 0 | oldEntityDeclHandler = parser->m_entityDeclHandler; |
1316 | 0 | oldXmlDeclHandler = parser->m_xmlDeclHandler; |
1317 | 0 | oldDeclElementType = parser->m_declElementType; |
1318 | |
|
1319 | 0 | oldUserData = parser->m_userData; |
1320 | 0 | oldHandlerArg = parser->m_handlerArg; |
1321 | 0 | oldDefaultExpandInternalEntities = parser->m_defaultExpandInternalEntities; |
1322 | 0 | oldExternalEntityRefHandlerArg = parser->m_externalEntityRefHandlerArg; |
1323 | | #ifdef XML_DTD |
1324 | | oldParamEntityParsing = parser->m_paramEntityParsing; |
1325 | | oldInEntityValue = parser->m_prologState.inEntityValue; |
1326 | | #endif |
1327 | 0 | oldns_triplets = parser->m_ns_triplets; |
1328 | | /* Note that the new parser shares the same hash secret as the old |
1329 | | parser, so that dtdCopy and copyEntityTable can lookup values |
1330 | | from hash tables associated with either parser without us having |
1331 | | to worry which hash secrets each table has. |
1332 | | */ |
1333 | 0 | oldhash_secret_salt = parser->m_hash_secret_salt; |
1334 | |
|
1335 | | #ifdef XML_DTD |
1336 | | if (! context) |
1337 | | newDtd = oldDtd; |
1338 | | #endif /* XML_DTD */ |
1339 | | |
1340 | | /* Note that the magical uses of the pre-processor to make field |
1341 | | access look more like C++ require that `parser' be overwritten |
1342 | | here. This makes this function more painful to follow than it |
1343 | | would be otherwise. |
1344 | | */ |
1345 | 0 | if (parser->m_ns) { |
1346 | 0 | XML_Char tmp[2] = {parser->m_namespaceSeparator, 0}; |
1347 | 0 | parser = parserCreate(encodingName, &parser->m_mem, tmp, newDtd); |
1348 | 0 | } else { |
1349 | 0 | parser = parserCreate(encodingName, &parser->m_mem, NULL, newDtd); |
1350 | 0 | } |
1351 | |
|
1352 | 0 | if (! parser) |
1353 | 0 | return NULL; |
1354 | | |
1355 | 0 | parser->m_startElementHandler = oldStartElementHandler; |
1356 | 0 | parser->m_endElementHandler = oldEndElementHandler; |
1357 | 0 | parser->m_characterDataHandler = oldCharacterDataHandler; |
1358 | 0 | parser->m_processingInstructionHandler = oldProcessingInstructionHandler; |
1359 | 0 | parser->m_commentHandler = oldCommentHandler; |
1360 | 0 | parser->m_startCdataSectionHandler = oldStartCdataSectionHandler; |
1361 | 0 | parser->m_endCdataSectionHandler = oldEndCdataSectionHandler; |
1362 | 0 | parser->m_defaultHandler = oldDefaultHandler; |
1363 | 0 | parser->m_unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler; |
1364 | 0 | parser->m_notationDeclHandler = oldNotationDeclHandler; |
1365 | 0 | parser->m_startNamespaceDeclHandler = oldStartNamespaceDeclHandler; |
1366 | 0 | parser->m_endNamespaceDeclHandler = oldEndNamespaceDeclHandler; |
1367 | 0 | parser->m_notStandaloneHandler = oldNotStandaloneHandler; |
1368 | 0 | parser->m_externalEntityRefHandler = oldExternalEntityRefHandler; |
1369 | 0 | parser->m_skippedEntityHandler = oldSkippedEntityHandler; |
1370 | 0 | parser->m_unknownEncodingHandler = oldUnknownEncodingHandler; |
1371 | 0 | parser->m_elementDeclHandler = oldElementDeclHandler; |
1372 | 0 | parser->m_attlistDeclHandler = oldAttlistDeclHandler; |
1373 | 0 | parser->m_entityDeclHandler = oldEntityDeclHandler; |
1374 | 0 | parser->m_xmlDeclHandler = oldXmlDeclHandler; |
1375 | 0 | parser->m_declElementType = oldDeclElementType; |
1376 | 0 | parser->m_userData = oldUserData; |
1377 | 0 | if (oldUserData == oldHandlerArg) |
1378 | 0 | parser->m_handlerArg = parser->m_userData; |
1379 | 0 | else |
1380 | 0 | parser->m_handlerArg = parser; |
1381 | 0 | if (oldExternalEntityRefHandlerArg != oldParser) |
1382 | 0 | parser->m_externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; |
1383 | 0 | parser->m_defaultExpandInternalEntities = oldDefaultExpandInternalEntities; |
1384 | 0 | parser->m_ns_triplets = oldns_triplets; |
1385 | 0 | parser->m_hash_secret_salt = oldhash_secret_salt; |
1386 | 0 | parser->m_parentParser = oldParser; |
1387 | | #ifdef XML_DTD |
1388 | | parser->m_paramEntityParsing = oldParamEntityParsing; |
1389 | | parser->m_prologState.inEntityValue = oldInEntityValue; |
1390 | | if (context) { |
1391 | | #endif /* XML_DTD */ |
1392 | 0 | if (! dtdCopy(oldParser, parser->m_dtd, oldDtd, &parser->m_mem) |
1393 | 0 | || ! setContext(parser, context)) { |
1394 | 0 | XML_ParserFree(parser); |
1395 | 0 | return NULL; |
1396 | 0 | } |
1397 | 0 | parser->m_processor = externalEntityInitProcessor; |
1398 | | #ifdef XML_DTD |
1399 | | } else { |
1400 | | /* The DTD instance referenced by parser->m_dtd is shared between the |
1401 | | document's root parser and external PE parsers, therefore one does not |
1402 | | need to call setContext. In addition, one also *must* not call |
1403 | | setContext, because this would overwrite existing prefix->binding |
1404 | | pointers in parser->m_dtd with ones that get destroyed with the external |
1405 | | PE parser. This would leave those prefixes with dangling pointers. |
1406 | | */ |
1407 | | parser->m_isParamEntity = XML_TRUE; |
1408 | | XmlPrologStateInitExternalEntity(&parser->m_prologState); |
1409 | | parser->m_processor = externalParEntInitProcessor; |
1410 | | } |
1411 | | #endif /* XML_DTD */ |
1412 | 0 | return parser; |
1413 | 0 | } |
1414 | | |
1415 | | static void FASTCALL |
1416 | 1.22M | destroyBindings(BINDING *bindings, XML_Parser parser) { |
1417 | 1.22M | for (;;) { |
1418 | 1.22M | BINDING *b = bindings; |
1419 | 1.22M | if (! b) |
1420 | 1.22M | break; |
1421 | 0 | bindings = b->nextTagBinding; |
1422 | 0 | FREE(parser, b->uri); |
1423 | 0 | FREE(parser, b); |
1424 | 0 | } |
1425 | 1.22M | } |
1426 | | |
1427 | | void XMLCALL |
1428 | 7.48k | XML_ParserFree(XML_Parser parser) { |
1429 | 7.48k | TAG *tagList; |
1430 | 7.48k | OPEN_INTERNAL_ENTITY *entityList; |
1431 | 7.48k | if (parser == NULL) |
1432 | 0 | return; |
1433 | | /* free m_tagStack and m_freeTagList */ |
1434 | 7.48k | tagList = parser->m_tagStack; |
1435 | 1.21M | for (;;) { |
1436 | 1.21M | TAG *p; |
1437 | 1.21M | if (tagList == NULL) { |
1438 | 7.76k | if (parser->m_freeTagList == NULL) |
1439 | 7.48k | break; |
1440 | 289 | tagList = parser->m_freeTagList; |
1441 | 289 | parser->m_freeTagList = NULL; |
1442 | 289 | } |
1443 | 1.20M | p = tagList; |
1444 | 1.20M | tagList = tagList->parent; |
1445 | 1.20M | FREE(parser, p->buf); |
1446 | 1.20M | destroyBindings(p->bindings, parser); |
1447 | 1.20M | FREE(parser, p); |
1448 | 1.20M | } |
1449 | | /* free m_openInternalEntities and m_freeInternalEntities */ |
1450 | 7.48k | entityList = parser->m_openInternalEntities; |
1451 | 9.01k | for (;;) { |
1452 | 9.01k | OPEN_INTERNAL_ENTITY *openEntity; |
1453 | 9.01k | if (entityList == NULL) { |
1454 | 8.06k | if (parser->m_freeInternalEntities == NULL) |
1455 | 7.48k | break; |
1456 | 583 | entityList = parser->m_freeInternalEntities; |
1457 | 583 | parser->m_freeInternalEntities = NULL; |
1458 | 583 | } |
1459 | 1.53k | openEntity = entityList; |
1460 | 1.53k | entityList = entityList->next; |
1461 | 1.53k | FREE(parser, openEntity); |
1462 | 1.53k | } |
1463 | | |
1464 | 7.48k | destroyBindings(parser->m_freeBindingList, parser); |
1465 | 7.48k | destroyBindings(parser->m_inheritedBindings, parser); |
1466 | 7.48k | poolDestroy(&parser->m_tempPool); |
1467 | 7.48k | poolDestroy(&parser->m_temp2Pool); |
1468 | 7.48k | FREE(parser, (void *)parser->m_protocolEncodingName); |
1469 | | #ifdef XML_DTD |
1470 | | /* external parameter entity parsers share the DTD structure |
1471 | | parser->m_dtd with the root parser, so we must not destroy it |
1472 | | */ |
1473 | | if (! parser->m_isParamEntity && parser->m_dtd) |
1474 | | #else |
1475 | 7.48k | if (parser->m_dtd) |
1476 | 7.48k | #endif /* XML_DTD */ |
1477 | 7.48k | dtdDestroy(parser->m_dtd, (XML_Bool)! parser->m_parentParser, |
1478 | 7.48k | &parser->m_mem); |
1479 | 7.48k | FREE(parser, (void *)parser->m_atts); |
1480 | | #ifdef XML_ATTR_INFO |
1481 | | FREE(parser, (void *)parser->m_attInfo); |
1482 | | #endif |
1483 | 7.48k | FREE(parser, parser->m_groupConnector); |
1484 | 7.48k | FREE(parser, parser->m_buffer); |
1485 | 7.48k | FREE(parser, parser->m_dataBuf); |
1486 | 7.48k | FREE(parser, parser->m_nsAtts); |
1487 | 7.48k | FREE(parser, parser->m_unknownEncodingMem); |
1488 | 7.48k | if (parser->m_unknownEncodingRelease) |
1489 | 0 | parser->m_unknownEncodingRelease(parser->m_unknownEncodingData); |
1490 | 7.48k | FREE(parser, parser); |
1491 | 7.48k | } |
1492 | | |
1493 | | void XMLCALL |
1494 | 0 | XML_UseParserAsHandlerArg(XML_Parser parser) { |
1495 | 0 | if (parser != NULL) |
1496 | 0 | parser->m_handlerArg = parser; |
1497 | 0 | } |
1498 | | |
1499 | | enum XML_Error XMLCALL |
1500 | 0 | XML_UseForeignDTD(XML_Parser parser, XML_Bool useDTD) { |
1501 | 0 | if (parser == NULL) |
1502 | 0 | return XML_ERROR_INVALID_ARGUMENT; |
1503 | | #ifdef XML_DTD |
1504 | | /* block after XML_Parse()/XML_ParseBuffer() has been called */ |
1505 | | if (parser->m_parsingStatus.parsing == XML_PARSING |
1506 | | || parser->m_parsingStatus.parsing == XML_SUSPENDED) |
1507 | | return XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING; |
1508 | | parser->m_useForeignDTD = useDTD; |
1509 | | return XML_ERROR_NONE; |
1510 | | #else |
1511 | 0 | UNUSED_P(useDTD); |
1512 | 0 | return XML_ERROR_FEATURE_REQUIRES_XML_DTD; |
1513 | 0 | #endif |
1514 | 0 | } |
1515 | | |
1516 | | void XMLCALL |
1517 | 0 | XML_SetReturnNSTriplet(XML_Parser parser, int do_nst) { |
1518 | 0 | if (parser == NULL) |
1519 | 0 | return; |
1520 | | /* block after XML_Parse()/XML_ParseBuffer() has been called */ |
1521 | 0 | if (parser->m_parsingStatus.parsing == XML_PARSING |
1522 | 0 | || parser->m_parsingStatus.parsing == XML_SUSPENDED) |
1523 | 0 | return; |
1524 | 0 | parser->m_ns_triplets = do_nst ? XML_TRUE : XML_FALSE; |
1525 | 0 | } |
1526 | | |
1527 | | void XMLCALL |
1528 | 7.48k | XML_SetUserData(XML_Parser parser, void *p) { |
1529 | 7.48k | if (parser == NULL) |
1530 | 0 | return; |
1531 | 7.48k | if (parser->m_handlerArg == parser->m_userData) |
1532 | 7.48k | parser->m_handlerArg = parser->m_userData = p; |
1533 | 0 | else |
1534 | 0 | parser->m_userData = p; |
1535 | 7.48k | } |
1536 | | |
1537 | | enum XML_Status XMLCALL |
1538 | 0 | XML_SetBase(XML_Parser parser, const XML_Char *p) { |
1539 | 0 | if (parser == NULL) |
1540 | 0 | return XML_STATUS_ERROR; |
1541 | 0 | if (p) { |
1542 | 0 | p = poolCopyString(&parser->m_dtd->pool, p); |
1543 | 0 | if (! p) |
1544 | 0 | return XML_STATUS_ERROR; |
1545 | 0 | parser->m_curBase = p; |
1546 | 0 | } else |
1547 | 0 | parser->m_curBase = NULL; |
1548 | 0 | return XML_STATUS_OK; |
1549 | 0 | } |
1550 | | |
1551 | | const XML_Char *XMLCALL |
1552 | 0 | XML_GetBase(XML_Parser parser) { |
1553 | 0 | if (parser == NULL) |
1554 | 0 | return NULL; |
1555 | 0 | return parser->m_curBase; |
1556 | 0 | } |
1557 | | |
1558 | | int XMLCALL |
1559 | 0 | XML_GetSpecifiedAttributeCount(XML_Parser parser) { |
1560 | 0 | if (parser == NULL) |
1561 | 0 | return -1; |
1562 | 0 | return parser->m_nSpecifiedAtts; |
1563 | 0 | } |
1564 | | |
1565 | | int XMLCALL |
1566 | 0 | XML_GetIdAttributeIndex(XML_Parser parser) { |
1567 | 0 | if (parser == NULL) |
1568 | 0 | return -1; |
1569 | 0 | return parser->m_idAttIndex; |
1570 | 0 | } |
1571 | | |
1572 | | #ifdef XML_ATTR_INFO |
1573 | | const XML_AttrInfo *XMLCALL |
1574 | | XML_GetAttributeInfo(XML_Parser parser) { |
1575 | | if (parser == NULL) |
1576 | | return NULL; |
1577 | | return parser->m_attInfo; |
1578 | | } |
1579 | | #endif |
1580 | | |
1581 | | void XMLCALL |
1582 | | XML_SetElementHandler(XML_Parser parser, XML_StartElementHandler start, |
1583 | 7.48k | XML_EndElementHandler end) { |
1584 | 7.48k | if (parser == NULL) |
1585 | 0 | return; |
1586 | 7.48k | parser->m_startElementHandler = start; |
1587 | 7.48k | parser->m_endElementHandler = end; |
1588 | 7.48k | } |
1589 | | |
1590 | | void XMLCALL |
1591 | 0 | XML_SetStartElementHandler(XML_Parser parser, XML_StartElementHandler start) { |
1592 | 0 | if (parser != NULL) |
1593 | 0 | parser->m_startElementHandler = start; |
1594 | 0 | } |
1595 | | |
1596 | | void XMLCALL |
1597 | 0 | XML_SetEndElementHandler(XML_Parser parser, XML_EndElementHandler end) { |
1598 | 0 | if (parser != NULL) |
1599 | 0 | parser->m_endElementHandler = end; |
1600 | 0 | } |
1601 | | |
1602 | | void XMLCALL |
1603 | | XML_SetCharacterDataHandler(XML_Parser parser, |
1604 | 7.48k | XML_CharacterDataHandler handler) { |
1605 | 7.48k | if (parser != NULL) |
1606 | 7.48k | parser->m_characterDataHandler = handler; |
1607 | 7.48k | } |
1608 | | |
1609 | | void XMLCALL |
1610 | | XML_SetProcessingInstructionHandler(XML_Parser parser, |
1611 | 0 | XML_ProcessingInstructionHandler handler) { |
1612 | 0 | if (parser != NULL) |
1613 | 0 | parser->m_processingInstructionHandler = handler; |
1614 | 0 | } |
1615 | | |
1616 | | void XMLCALL |
1617 | 0 | XML_SetCommentHandler(XML_Parser parser, XML_CommentHandler handler) { |
1618 | 0 | if (parser != NULL) |
1619 | 0 | parser->m_commentHandler = handler; |
1620 | 0 | } |
1621 | | |
1622 | | void XMLCALL |
1623 | | XML_SetCdataSectionHandler(XML_Parser parser, |
1624 | | XML_StartCdataSectionHandler start, |
1625 | 0 | XML_EndCdataSectionHandler end) { |
1626 | 0 | if (parser == NULL) |
1627 | 0 | return; |
1628 | 0 | parser->m_startCdataSectionHandler = start; |
1629 | 0 | parser->m_endCdataSectionHandler = end; |
1630 | 0 | } |
1631 | | |
1632 | | void XMLCALL |
1633 | | XML_SetStartCdataSectionHandler(XML_Parser parser, |
1634 | 0 | XML_StartCdataSectionHandler start) { |
1635 | 0 | if (parser != NULL) |
1636 | 0 | parser->m_startCdataSectionHandler = start; |
1637 | 0 | } |
1638 | | |
1639 | | void XMLCALL |
1640 | | XML_SetEndCdataSectionHandler(XML_Parser parser, |
1641 | 0 | XML_EndCdataSectionHandler end) { |
1642 | 0 | if (parser != NULL) |
1643 | 0 | parser->m_endCdataSectionHandler = end; |
1644 | 0 | } |
1645 | | |
1646 | | void XMLCALL |
1647 | 0 | XML_SetDefaultHandler(XML_Parser parser, XML_DefaultHandler handler) { |
1648 | 0 | if (parser == NULL) |
1649 | 0 | return; |
1650 | 0 | parser->m_defaultHandler = handler; |
1651 | 0 | parser->m_defaultExpandInternalEntities = XML_FALSE; |
1652 | 0 | } |
1653 | | |
1654 | | void XMLCALL |
1655 | 0 | XML_SetDefaultHandlerExpand(XML_Parser parser, XML_DefaultHandler handler) { |
1656 | 0 | if (parser == NULL) |
1657 | 0 | return; |
1658 | 0 | parser->m_defaultHandler = handler; |
1659 | 0 | parser->m_defaultExpandInternalEntities = XML_TRUE; |
1660 | 0 | } |
1661 | | |
1662 | | void XMLCALL |
1663 | | XML_SetDoctypeDeclHandler(XML_Parser parser, XML_StartDoctypeDeclHandler start, |
1664 | 0 | XML_EndDoctypeDeclHandler end) { |
1665 | 0 | if (parser == NULL) |
1666 | 0 | return; |
1667 | 0 | parser->m_startDoctypeDeclHandler = start; |
1668 | 0 | parser->m_endDoctypeDeclHandler = end; |
1669 | 0 | } |
1670 | | |
1671 | | void XMLCALL |
1672 | | XML_SetStartDoctypeDeclHandler(XML_Parser parser, |
1673 | 0 | XML_StartDoctypeDeclHandler start) { |
1674 | 0 | if (parser != NULL) |
1675 | 0 | parser->m_startDoctypeDeclHandler = start; |
1676 | 0 | } |
1677 | | |
1678 | | void XMLCALL |
1679 | 0 | XML_SetEndDoctypeDeclHandler(XML_Parser parser, XML_EndDoctypeDeclHandler end) { |
1680 | 0 | if (parser != NULL) |
1681 | 0 | parser->m_endDoctypeDeclHandler = end; |
1682 | 0 | } |
1683 | | |
1684 | | void XMLCALL |
1685 | | XML_SetUnparsedEntityDeclHandler(XML_Parser parser, |
1686 | 0 | XML_UnparsedEntityDeclHandler handler) { |
1687 | 0 | if (parser != NULL) |
1688 | 0 | parser->m_unparsedEntityDeclHandler = handler; |
1689 | 0 | } |
1690 | | |
1691 | | void XMLCALL |
1692 | 0 | XML_SetNotationDeclHandler(XML_Parser parser, XML_NotationDeclHandler handler) { |
1693 | 0 | if (parser != NULL) |
1694 | 0 | parser->m_notationDeclHandler = handler; |
1695 | 0 | } |
1696 | | |
1697 | | void XMLCALL |
1698 | | XML_SetNamespaceDeclHandler(XML_Parser parser, |
1699 | | XML_StartNamespaceDeclHandler start, |
1700 | 0 | XML_EndNamespaceDeclHandler end) { |
1701 | 0 | if (parser == NULL) |
1702 | 0 | return; |
1703 | 0 | parser->m_startNamespaceDeclHandler = start; |
1704 | 0 | parser->m_endNamespaceDeclHandler = end; |
1705 | 0 | } |
1706 | | |
1707 | | void XMLCALL |
1708 | | XML_SetStartNamespaceDeclHandler(XML_Parser parser, |
1709 | 0 | XML_StartNamespaceDeclHandler start) { |
1710 | 0 | if (parser != NULL) |
1711 | 0 | parser->m_startNamespaceDeclHandler = start; |
1712 | 0 | } |
1713 | | |
1714 | | void XMLCALL |
1715 | | XML_SetEndNamespaceDeclHandler(XML_Parser parser, |
1716 | 0 | XML_EndNamespaceDeclHandler end) { |
1717 | 0 | if (parser != NULL) |
1718 | 0 | parser->m_endNamespaceDeclHandler = end; |
1719 | 0 | } |
1720 | | |
1721 | | void XMLCALL |
1722 | | XML_SetNotStandaloneHandler(XML_Parser parser, |
1723 | 0 | XML_NotStandaloneHandler handler) { |
1724 | 0 | if (parser != NULL) |
1725 | 0 | parser->m_notStandaloneHandler = handler; |
1726 | 0 | } |
1727 | | |
1728 | | void XMLCALL |
1729 | | XML_SetExternalEntityRefHandler(XML_Parser parser, |
1730 | 0 | XML_ExternalEntityRefHandler handler) { |
1731 | 0 | if (parser != NULL) |
1732 | 0 | parser->m_externalEntityRefHandler = handler; |
1733 | 0 | } |
1734 | | |
1735 | | void XMLCALL |
1736 | 0 | XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) { |
1737 | 0 | if (parser == NULL) |
1738 | 0 | return; |
1739 | 0 | if (arg) |
1740 | 0 | parser->m_externalEntityRefHandlerArg = (XML_Parser)arg; |
1741 | 0 | else |
1742 | 0 | parser->m_externalEntityRefHandlerArg = parser; |
1743 | 0 | } |
1744 | | |
1745 | | void XMLCALL |
1746 | | XML_SetSkippedEntityHandler(XML_Parser parser, |
1747 | 0 | XML_SkippedEntityHandler handler) { |
1748 | 0 | if (parser != NULL) |
1749 | 0 | parser->m_skippedEntityHandler = handler; |
1750 | 0 | } |
1751 | | |
1752 | | void XMLCALL |
1753 | | XML_SetUnknownEncodingHandler(XML_Parser parser, |
1754 | 0 | XML_UnknownEncodingHandler handler, void *data) { |
1755 | 0 | if (parser == NULL) |
1756 | 0 | return; |
1757 | 0 | parser->m_unknownEncodingHandler = handler; |
1758 | 0 | parser->m_unknownEncodingHandlerData = data; |
1759 | 0 | } |
1760 | | |
1761 | | void XMLCALL |
1762 | 0 | XML_SetElementDeclHandler(XML_Parser parser, XML_ElementDeclHandler eldecl) { |
1763 | 0 | if (parser != NULL) |
1764 | 0 | parser->m_elementDeclHandler = eldecl; |
1765 | 0 | } |
1766 | | |
1767 | | void XMLCALL |
1768 | 0 | XML_SetAttlistDeclHandler(XML_Parser parser, XML_AttlistDeclHandler attdecl) { |
1769 | 0 | if (parser != NULL) |
1770 | 0 | parser->m_attlistDeclHandler = attdecl; |
1771 | 0 | } |
1772 | | |
1773 | | void XMLCALL |
1774 | 0 | XML_SetEntityDeclHandler(XML_Parser parser, XML_EntityDeclHandler handler) { |
1775 | 0 | if (parser != NULL) |
1776 | 0 | parser->m_entityDeclHandler = handler; |
1777 | 0 | } |
1778 | | |
1779 | | void XMLCALL |
1780 | 0 | XML_SetXmlDeclHandler(XML_Parser parser, XML_XmlDeclHandler handler) { |
1781 | 0 | if (parser != NULL) |
1782 | 0 | parser->m_xmlDeclHandler = handler; |
1783 | 0 | } |
1784 | | |
1785 | | int XMLCALL |
1786 | | XML_SetParamEntityParsing(XML_Parser parser, |
1787 | 0 | enum XML_ParamEntityParsing peParsing) { |
1788 | 0 | if (parser == NULL) |
1789 | 0 | return 0; |
1790 | | /* block after XML_Parse()/XML_ParseBuffer() has been called */ |
1791 | 0 | if (parser->m_parsingStatus.parsing == XML_PARSING |
1792 | 0 | || parser->m_parsingStatus.parsing == XML_SUSPENDED) |
1793 | 0 | return 0; |
1794 | | #ifdef XML_DTD |
1795 | | parser->m_paramEntityParsing = peParsing; |
1796 | | return 1; |
1797 | | #else |
1798 | 0 | return peParsing == XML_PARAM_ENTITY_PARSING_NEVER; |
1799 | 0 | #endif |
1800 | 0 | } |
1801 | | |
1802 | | int XMLCALL |
1803 | 0 | XML_SetHashSalt(XML_Parser parser, unsigned long hash_salt) { |
1804 | 0 | if (parser == NULL) |
1805 | 0 | return 0; |
1806 | 0 | if (parser->m_parentParser) |
1807 | 0 | return XML_SetHashSalt(parser->m_parentParser, hash_salt); |
1808 | | /* block after XML_Parse()/XML_ParseBuffer() has been called */ |
1809 | 0 | if (parser->m_parsingStatus.parsing == XML_PARSING |
1810 | 0 | || parser->m_parsingStatus.parsing == XML_SUSPENDED) |
1811 | 0 | return 0; |
1812 | 0 | parser->m_hash_secret_salt = hash_salt; |
1813 | 0 | return 1; |
1814 | 0 | } |
1815 | | |
1816 | | enum XML_Status XMLCALL |
1817 | 11.3k | XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) { |
1818 | 11.3k | if ((parser == NULL) || (len < 0) || ((s == NULL) && (len != 0))) { |
1819 | 0 | if (parser != NULL) |
1820 | 0 | parser->m_errorCode = XML_ERROR_INVALID_ARGUMENT; |
1821 | 0 | return XML_STATUS_ERROR; |
1822 | 0 | } |
1823 | 11.3k | switch (parser->m_parsingStatus.parsing) { |
1824 | 0 | case XML_SUSPENDED: |
1825 | 0 | parser->m_errorCode = XML_ERROR_SUSPENDED; |
1826 | 0 | return XML_STATUS_ERROR; |
1827 | 0 | case XML_FINISHED: |
1828 | 0 | parser->m_errorCode = XML_ERROR_FINISHED; |
1829 | 0 | return XML_STATUS_ERROR; |
1830 | 7.48k | case XML_INITIALIZED: |
1831 | 7.48k | if (parser->m_parentParser == NULL && ! startParsing(parser)) { |
1832 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
1833 | 0 | return XML_STATUS_ERROR; |
1834 | 0 | } |
1835 | | /* fall through */ |
1836 | 11.3k | default: |
1837 | 11.3k | parser->m_parsingStatus.parsing = XML_PARSING; |
1838 | 11.3k | } |
1839 | | |
1840 | 11.3k | if (len == 0) { |
1841 | 3.82k | parser->m_parsingStatus.finalBuffer = (XML_Bool)isFinal; |
1842 | 3.82k | if (! isFinal) |
1843 | 2 | return XML_STATUS_OK; |
1844 | 3.82k | parser->m_positionPtr = parser->m_bufferPtr; |
1845 | 3.82k | parser->m_parseEndPtr = parser->m_bufferEnd; |
1846 | | |
1847 | | /* If data are left over from last buffer, and we now know that these |
1848 | | data are the final chunk of input, then we have to check them again |
1849 | | to detect errors based on that fact. |
1850 | | */ |
1851 | 3.82k | parser->m_errorCode |
1852 | 3.82k | = parser->m_processor(parser, parser->m_bufferPtr, |
1853 | 3.82k | parser->m_parseEndPtr, &parser->m_bufferPtr); |
1854 | | |
1855 | 3.82k | if (parser->m_errorCode == XML_ERROR_NONE) { |
1856 | 49 | switch (parser->m_parsingStatus.parsing) { |
1857 | 0 | case XML_SUSPENDED: |
1858 | | /* It is hard to be certain, but it seems that this case |
1859 | | * cannot occur. This code is cleaning up a previous parse |
1860 | | * with no new data (since len == 0). Changing the parsing |
1861 | | * state requires getting to execute a handler function, and |
1862 | | * there doesn't seem to be an opportunity for that while in |
1863 | | * this circumstance. |
1864 | | * |
1865 | | * Given the uncertainty, we retain the code but exclude it |
1866 | | * from coverage tests. |
1867 | | * |
1868 | | * LCOV_EXCL_START |
1869 | | */ |
1870 | 0 | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, |
1871 | 0 | parser->m_bufferPtr, &parser->m_position); |
1872 | 0 | parser->m_positionPtr = parser->m_bufferPtr; |
1873 | 0 | return XML_STATUS_SUSPENDED; |
1874 | | /* LCOV_EXCL_STOP */ |
1875 | 0 | case XML_INITIALIZED: |
1876 | 49 | case XML_PARSING: |
1877 | 49 | parser->m_parsingStatus.parsing = XML_FINISHED; |
1878 | | /* fall through */ |
1879 | 49 | default: |
1880 | 49 | return XML_STATUS_OK; |
1881 | 49 | } |
1882 | 49 | } |
1883 | 3.77k | parser->m_eventEndPtr = parser->m_eventPtr; |
1884 | 3.77k | parser->m_processor = errorProcessor; |
1885 | 3.77k | return XML_STATUS_ERROR; |
1886 | 3.82k | } |
1887 | | #ifndef XML_CONTEXT_BYTES |
1888 | | else if (parser->m_bufferPtr == parser->m_bufferEnd) { |
1889 | | const char *end; |
1890 | | int nLeftOver; |
1891 | | enum XML_Status result; |
1892 | | /* Detect overflow (a+b > MAX <==> b > MAX-a) */ |
1893 | | if ((XML_Size)len > ((XML_Size)-1) / 2 - parser->m_parseEndByteIndex) { |
1894 | | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
1895 | | parser->m_eventPtr = parser->m_eventEndPtr = NULL; |
1896 | | parser->m_processor = errorProcessor; |
1897 | | return XML_STATUS_ERROR; |
1898 | | } |
1899 | | parser->m_parseEndByteIndex += len; |
1900 | | parser->m_positionPtr = s; |
1901 | | parser->m_parsingStatus.finalBuffer = (XML_Bool)isFinal; |
1902 | | |
1903 | | parser->m_errorCode |
1904 | | = parser->m_processor(parser, s, parser->m_parseEndPtr = s + len, &end); |
1905 | | |
1906 | | if (parser->m_errorCode != XML_ERROR_NONE) { |
1907 | | parser->m_eventEndPtr = parser->m_eventPtr; |
1908 | | parser->m_processor = errorProcessor; |
1909 | | return XML_STATUS_ERROR; |
1910 | | } else { |
1911 | | switch (parser->m_parsingStatus.parsing) { |
1912 | | case XML_SUSPENDED: |
1913 | | result = XML_STATUS_SUSPENDED; |
1914 | | break; |
1915 | | case XML_INITIALIZED: |
1916 | | case XML_PARSING: |
1917 | | if (isFinal) { |
1918 | | parser->m_parsingStatus.parsing = XML_FINISHED; |
1919 | | return XML_STATUS_OK; |
1920 | | } |
1921 | | /* fall through */ |
1922 | | default: |
1923 | | result = XML_STATUS_OK; |
1924 | | } |
1925 | | } |
1926 | | |
1927 | | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, end, |
1928 | | &parser->m_position); |
1929 | | nLeftOver = s + len - end; |
1930 | | if (nLeftOver) { |
1931 | | if (parser->m_buffer == NULL |
1932 | | || nLeftOver > parser->m_bufferLim - parser->m_buffer) { |
1933 | | /* avoid _signed_ integer overflow */ |
1934 | | char *temp = NULL; |
1935 | | const int bytesToAllocate = (int)((unsigned)len * 2U); |
1936 | | if (bytesToAllocate > 0) { |
1937 | | temp = (char *)REALLOC(parser, parser->m_buffer, bytesToAllocate); |
1938 | | } |
1939 | | if (temp == NULL) { |
1940 | | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
1941 | | parser->m_eventPtr = parser->m_eventEndPtr = NULL; |
1942 | | parser->m_processor = errorProcessor; |
1943 | | return XML_STATUS_ERROR; |
1944 | | } |
1945 | | parser->m_buffer = temp; |
1946 | | parser->m_bufferLim = parser->m_buffer + bytesToAllocate; |
1947 | | } |
1948 | | memcpy(parser->m_buffer, end, nLeftOver); |
1949 | | } |
1950 | | parser->m_bufferPtr = parser->m_buffer; |
1951 | | parser->m_bufferEnd = parser->m_buffer + nLeftOver; |
1952 | | parser->m_positionPtr = parser->m_bufferPtr; |
1953 | | parser->m_parseEndPtr = parser->m_bufferEnd; |
1954 | | parser->m_eventPtr = parser->m_bufferPtr; |
1955 | | parser->m_eventEndPtr = parser->m_bufferPtr; |
1956 | | return result; |
1957 | | } |
1958 | | #endif /* not defined XML_CONTEXT_BYTES */ |
1959 | 7.47k | else { |
1960 | 7.47k | void *buff = XML_GetBuffer(parser, len); |
1961 | 7.47k | if (buff == NULL) |
1962 | 0 | return XML_STATUS_ERROR; |
1963 | 7.47k | else { |
1964 | 7.47k | memcpy(buff, s, len); |
1965 | 7.47k | return XML_ParseBuffer(parser, len, isFinal); |
1966 | 7.47k | } |
1967 | 7.47k | } |
1968 | 11.3k | } |
1969 | | |
1970 | | enum XML_Status XMLCALL |
1971 | 7.47k | XML_ParseBuffer(XML_Parser parser, int len, int isFinal) { |
1972 | 7.47k | const char *start; |
1973 | 7.47k | enum XML_Status result = XML_STATUS_OK; |
1974 | | |
1975 | 7.47k | if (parser == NULL) |
1976 | 0 | return XML_STATUS_ERROR; |
1977 | 7.47k | switch (parser->m_parsingStatus.parsing) { |
1978 | 0 | case XML_SUSPENDED: |
1979 | 0 | parser->m_errorCode = XML_ERROR_SUSPENDED; |
1980 | 0 | return XML_STATUS_ERROR; |
1981 | 0 | case XML_FINISHED: |
1982 | 0 | parser->m_errorCode = XML_ERROR_FINISHED; |
1983 | 0 | return XML_STATUS_ERROR; |
1984 | 0 | case XML_INITIALIZED: |
1985 | | /* Has someone called XML_GetBuffer successfully before? */ |
1986 | 0 | if (! parser->m_bufferPtr) { |
1987 | 0 | parser->m_errorCode = XML_ERROR_NO_BUFFER; |
1988 | 0 | return XML_STATUS_ERROR; |
1989 | 0 | } |
1990 | | |
1991 | 0 | if (parser->m_parentParser == NULL && ! startParsing(parser)) { |
1992 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
1993 | 0 | return XML_STATUS_ERROR; |
1994 | 0 | } |
1995 | | /* fall through */ |
1996 | 7.47k | default: |
1997 | 7.47k | parser->m_parsingStatus.parsing = XML_PARSING; |
1998 | 7.47k | } |
1999 | | |
2000 | 7.47k | start = parser->m_bufferPtr; |
2001 | 7.47k | parser->m_positionPtr = start; |
2002 | 7.47k | parser->m_bufferEnd += len; |
2003 | 7.47k | parser->m_parseEndPtr = parser->m_bufferEnd; |
2004 | 7.47k | parser->m_parseEndByteIndex += len; |
2005 | 7.47k | parser->m_parsingStatus.finalBuffer = (XML_Bool)isFinal; |
2006 | | |
2007 | 7.47k | parser->m_errorCode = parser->m_processor( |
2008 | 7.47k | parser, start, parser->m_parseEndPtr, &parser->m_bufferPtr); |
2009 | | |
2010 | 7.47k | if (parser->m_errorCode != XML_ERROR_NONE) { |
2011 | 3.65k | parser->m_eventEndPtr = parser->m_eventPtr; |
2012 | 3.65k | parser->m_processor = errorProcessor; |
2013 | 3.65k | return XML_STATUS_ERROR; |
2014 | 3.82k | } else { |
2015 | 3.82k | switch (parser->m_parsingStatus.parsing) { |
2016 | 0 | case XML_SUSPENDED: |
2017 | 0 | result = XML_STATUS_SUSPENDED; |
2018 | 0 | break; |
2019 | 0 | case XML_INITIALIZED: |
2020 | 3.82k | case XML_PARSING: |
2021 | 3.82k | if (isFinal) { |
2022 | 0 | parser->m_parsingStatus.parsing = XML_FINISHED; |
2023 | 0 | return result; |
2024 | 0 | } |
2025 | 3.82k | default:; /* should not happen */ |
2026 | 3.82k | } |
2027 | 3.82k | } |
2028 | | |
2029 | 3.82k | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, |
2030 | 3.82k | parser->m_bufferPtr, &parser->m_position); |
2031 | 3.82k | parser->m_positionPtr = parser->m_bufferPtr; |
2032 | 3.82k | return result; |
2033 | 7.47k | } |
2034 | | |
2035 | | void *XMLCALL |
2036 | 7.47k | XML_GetBuffer(XML_Parser parser, int len) { |
2037 | 7.47k | if (parser == NULL) |
2038 | 0 | return NULL; |
2039 | 7.47k | if (len < 0) { |
2040 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
2041 | 0 | return NULL; |
2042 | 0 | } |
2043 | 7.47k | switch (parser->m_parsingStatus.parsing) { |
2044 | 0 | case XML_SUSPENDED: |
2045 | 0 | parser->m_errorCode = XML_ERROR_SUSPENDED; |
2046 | 0 | return NULL; |
2047 | 0 | case XML_FINISHED: |
2048 | 0 | parser->m_errorCode = XML_ERROR_FINISHED; |
2049 | 0 | return NULL; |
2050 | 7.47k | default:; |
2051 | 7.47k | } |
2052 | | |
2053 | 7.47k | if (len > EXPAT_SAFE_PTR_DIFF(parser->m_bufferLim, parser->m_bufferEnd)) { |
2054 | 7.47k | #ifdef XML_CONTEXT_BYTES |
2055 | 7.47k | int keep; |
2056 | 7.47k | #endif /* defined XML_CONTEXT_BYTES */ |
2057 | | /* Do not invoke signed arithmetic overflow: */ |
2058 | 7.47k | int neededSize = (int)((unsigned)len |
2059 | 7.47k | + (unsigned)EXPAT_SAFE_PTR_DIFF( |
2060 | 7.47k | parser->m_bufferEnd, parser->m_bufferPtr)); |
2061 | 7.47k | if (neededSize < 0) { |
2062 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
2063 | 0 | return NULL; |
2064 | 0 | } |
2065 | 7.47k | #ifdef XML_CONTEXT_BYTES |
2066 | 7.47k | keep = (int)EXPAT_SAFE_PTR_DIFF(parser->m_bufferPtr, parser->m_buffer); |
2067 | 7.47k | if (keep > XML_CONTEXT_BYTES) |
2068 | 0 | keep = XML_CONTEXT_BYTES; |
2069 | | /* Detect and prevent integer overflow */ |
2070 | 7.47k | if (keep > INT_MAX - neededSize) { |
2071 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
2072 | 0 | return NULL; |
2073 | 0 | } |
2074 | 7.47k | neededSize += keep; |
2075 | 7.47k | #endif /* defined XML_CONTEXT_BYTES */ |
2076 | 7.47k | if (neededSize |
2077 | 7.47k | <= EXPAT_SAFE_PTR_DIFF(parser->m_bufferLim, parser->m_buffer)) { |
2078 | 0 | #ifdef XML_CONTEXT_BYTES |
2079 | 0 | if (keep < EXPAT_SAFE_PTR_DIFF(parser->m_bufferPtr, parser->m_buffer)) { |
2080 | 0 | int offset |
2081 | 0 | = (int)EXPAT_SAFE_PTR_DIFF(parser->m_bufferPtr, parser->m_buffer) |
2082 | 0 | - keep; |
2083 | | /* The buffer pointers cannot be NULL here; we have at least some bytes |
2084 | | * in the buffer */ |
2085 | 0 | memmove(parser->m_buffer, &parser->m_buffer[offset], |
2086 | 0 | parser->m_bufferEnd - parser->m_bufferPtr + keep); |
2087 | 0 | parser->m_bufferEnd -= offset; |
2088 | 0 | parser->m_bufferPtr -= offset; |
2089 | 0 | } |
2090 | | #else |
2091 | | if (parser->m_buffer && parser->m_bufferPtr) { |
2092 | | memmove(parser->m_buffer, parser->m_bufferPtr, |
2093 | | EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr)); |
2094 | | parser->m_bufferEnd |
2095 | | = parser->m_buffer |
2096 | | + EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr); |
2097 | | parser->m_bufferPtr = parser->m_buffer; |
2098 | | } |
2099 | | #endif /* not defined XML_CONTEXT_BYTES */ |
2100 | 7.47k | } else { |
2101 | 7.47k | char *newBuf; |
2102 | 7.47k | int bufferSize |
2103 | 7.47k | = (int)EXPAT_SAFE_PTR_DIFF(parser->m_bufferLim, parser->m_bufferPtr); |
2104 | 7.47k | if (bufferSize == 0) |
2105 | 7.47k | bufferSize = INIT_BUFFER_SIZE; |
2106 | 11.7k | do { |
2107 | | /* Do not invoke signed arithmetic overflow: */ |
2108 | 11.7k | bufferSize = (int)(2U * (unsigned)bufferSize); |
2109 | 11.7k | } while (bufferSize < neededSize && bufferSize > 0); |
2110 | 7.47k | if (bufferSize <= 0) { |
2111 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
2112 | 0 | return NULL; |
2113 | 0 | } |
2114 | 7.47k | newBuf = (char *)MALLOC(parser, bufferSize); |
2115 | 7.47k | if (newBuf == 0) { |
2116 | 0 | parser->m_errorCode = XML_ERROR_NO_MEMORY; |
2117 | 0 | return NULL; |
2118 | 0 | } |
2119 | 7.47k | parser->m_bufferLim = newBuf + bufferSize; |
2120 | 7.47k | #ifdef XML_CONTEXT_BYTES |
2121 | 7.47k | if (parser->m_bufferPtr) { |
2122 | 0 | memcpy(newBuf, &parser->m_bufferPtr[-keep], |
2123 | 0 | EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr) |
2124 | 0 | + keep); |
2125 | 0 | FREE(parser, parser->m_buffer); |
2126 | 0 | parser->m_buffer = newBuf; |
2127 | 0 | parser->m_bufferEnd |
2128 | 0 | = parser->m_buffer |
2129 | 0 | + EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr) |
2130 | 0 | + keep; |
2131 | 0 | parser->m_bufferPtr = parser->m_buffer + keep; |
2132 | 7.47k | } else { |
2133 | | /* This must be a brand new buffer with no data in it yet */ |
2134 | 7.47k | parser->m_bufferEnd = newBuf; |
2135 | 7.47k | parser->m_bufferPtr = parser->m_buffer = newBuf; |
2136 | 7.47k | } |
2137 | | #else |
2138 | | if (parser->m_bufferPtr) { |
2139 | | memcpy(newBuf, parser->m_bufferPtr, |
2140 | | EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr)); |
2141 | | FREE(parser, parser->m_buffer); |
2142 | | parser->m_bufferEnd |
2143 | | = newBuf |
2144 | | + EXPAT_SAFE_PTR_DIFF(parser->m_bufferEnd, parser->m_bufferPtr); |
2145 | | } else { |
2146 | | /* This must be a brand new buffer with no data in it yet */ |
2147 | | parser->m_bufferEnd = newBuf; |
2148 | | } |
2149 | | parser->m_bufferPtr = parser->m_buffer = newBuf; |
2150 | | #endif /* not defined XML_CONTEXT_BYTES */ |
2151 | 7.47k | } |
2152 | 7.47k | parser->m_eventPtr = parser->m_eventEndPtr = NULL; |
2153 | 7.47k | parser->m_positionPtr = NULL; |
2154 | 7.47k | } |
2155 | 7.47k | return parser->m_bufferEnd; |
2156 | 7.47k | } |
2157 | | |
2158 | | enum XML_Status XMLCALL |
2159 | 0 | XML_StopParser(XML_Parser parser, XML_Bool resumable) { |
2160 | 0 | if (parser == NULL) |
2161 | 0 | return XML_STATUS_ERROR; |
2162 | 0 | switch (parser->m_parsingStatus.parsing) { |
2163 | 0 | case XML_SUSPENDED: |
2164 | 0 | if (resumable) { |
2165 | 0 | parser->m_errorCode = XML_ERROR_SUSPENDED; |
2166 | 0 | return XML_STATUS_ERROR; |
2167 | 0 | } |
2168 | 0 | parser->m_parsingStatus.parsing = XML_FINISHED; |
2169 | 0 | break; |
2170 | 0 | case XML_FINISHED: |
2171 | 0 | parser->m_errorCode = XML_ERROR_FINISHED; |
2172 | 0 | return XML_STATUS_ERROR; |
2173 | 0 | default: |
2174 | 0 | if (resumable) { |
2175 | | #ifdef XML_DTD |
2176 | | if (parser->m_isParamEntity) { |
2177 | | parser->m_errorCode = XML_ERROR_SUSPEND_PE; |
2178 | | return XML_STATUS_ERROR; |
2179 | | } |
2180 | | #endif |
2181 | 0 | parser->m_parsingStatus.parsing = XML_SUSPENDED; |
2182 | 0 | } else |
2183 | 0 | parser->m_parsingStatus.parsing = XML_FINISHED; |
2184 | 0 | } |
2185 | 0 | return XML_STATUS_OK; |
2186 | 0 | } |
2187 | | |
2188 | | enum XML_Status XMLCALL |
2189 | 0 | XML_ResumeParser(XML_Parser parser) { |
2190 | 0 | enum XML_Status result = XML_STATUS_OK; |
2191 | |
|
2192 | 0 | if (parser == NULL) |
2193 | 0 | return XML_STATUS_ERROR; |
2194 | 0 | if (parser->m_parsingStatus.parsing != XML_SUSPENDED) { |
2195 | 0 | parser->m_errorCode = XML_ERROR_NOT_SUSPENDED; |
2196 | 0 | return XML_STATUS_ERROR; |
2197 | 0 | } |
2198 | 0 | parser->m_parsingStatus.parsing = XML_PARSING; |
2199 | |
|
2200 | 0 | parser->m_errorCode = parser->m_processor( |
2201 | 0 | parser, parser->m_bufferPtr, parser->m_parseEndPtr, &parser->m_bufferPtr); |
2202 | |
|
2203 | 0 | if (parser->m_errorCode != XML_ERROR_NONE) { |
2204 | 0 | parser->m_eventEndPtr = parser->m_eventPtr; |
2205 | 0 | parser->m_processor = errorProcessor; |
2206 | 0 | return XML_STATUS_ERROR; |
2207 | 0 | } else { |
2208 | 0 | switch (parser->m_parsingStatus.parsing) { |
2209 | 0 | case XML_SUSPENDED: |
2210 | 0 | result = XML_STATUS_SUSPENDED; |
2211 | 0 | break; |
2212 | 0 | case XML_INITIALIZED: |
2213 | 0 | case XML_PARSING: |
2214 | 0 | if (parser->m_parsingStatus.finalBuffer) { |
2215 | 0 | parser->m_parsingStatus.parsing = XML_FINISHED; |
2216 | 0 | return result; |
2217 | 0 | } |
2218 | 0 | default:; |
2219 | 0 | } |
2220 | 0 | } |
2221 | | |
2222 | 0 | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, |
2223 | 0 | parser->m_bufferPtr, &parser->m_position); |
2224 | 0 | parser->m_positionPtr = parser->m_bufferPtr; |
2225 | 0 | return result; |
2226 | 0 | } |
2227 | | |
2228 | | void XMLCALL |
2229 | 0 | XML_GetParsingStatus(XML_Parser parser, XML_ParsingStatus *status) { |
2230 | 0 | if (parser == NULL) |
2231 | 0 | return; |
2232 | 0 | assert(status != NULL); |
2233 | 0 | *status = parser->m_parsingStatus; |
2234 | 0 | } |
2235 | | |
2236 | | enum XML_Error XMLCALL |
2237 | 7.43k | XML_GetErrorCode(XML_Parser parser) { |
2238 | 7.43k | if (parser == NULL) |
2239 | 0 | return XML_ERROR_INVALID_ARGUMENT; |
2240 | 7.43k | return parser->m_errorCode; |
2241 | 7.43k | } |
2242 | | |
2243 | | XML_Index XMLCALL |
2244 | 0 | XML_GetCurrentByteIndex(XML_Parser parser) { |
2245 | 0 | if (parser == NULL) |
2246 | 0 | return -1; |
2247 | 0 | if (parser->m_eventPtr) |
2248 | 0 | return (XML_Index)(parser->m_parseEndByteIndex |
2249 | 0 | - (parser->m_parseEndPtr - parser->m_eventPtr)); |
2250 | 0 | return -1; |
2251 | 0 | } |
2252 | | |
2253 | | int XMLCALL |
2254 | 0 | XML_GetCurrentByteCount(XML_Parser parser) { |
2255 | 0 | if (parser == NULL) |
2256 | 0 | return 0; |
2257 | 0 | if (parser->m_eventEndPtr && parser->m_eventPtr) |
2258 | 0 | return (int)(parser->m_eventEndPtr - parser->m_eventPtr); |
2259 | 0 | return 0; |
2260 | 0 | } |
2261 | | |
2262 | | const char *XMLCALL |
2263 | 0 | XML_GetInputContext(XML_Parser parser, int *offset, int *size) { |
2264 | 0 | #ifdef XML_CONTEXT_BYTES |
2265 | 0 | if (parser == NULL) |
2266 | 0 | return NULL; |
2267 | 0 | if (parser->m_eventPtr && parser->m_buffer) { |
2268 | 0 | if (offset != NULL) |
2269 | 0 | *offset = (int)(parser->m_eventPtr - parser->m_buffer); |
2270 | 0 | if (size != NULL) |
2271 | 0 | *size = (int)(parser->m_bufferEnd - parser->m_buffer); |
2272 | 0 | return parser->m_buffer; |
2273 | 0 | } |
2274 | | #else |
2275 | | (void)parser; |
2276 | | (void)offset; |
2277 | | (void)size; |
2278 | | #endif /* defined XML_CONTEXT_BYTES */ |
2279 | 0 | return (const char *)0; |
2280 | 0 | } |
2281 | | |
2282 | | XML_Size XMLCALL |
2283 | 7.43k | XML_GetCurrentLineNumber(XML_Parser parser) { |
2284 | 7.43k | if (parser == NULL) |
2285 | 0 | return 0; |
2286 | 7.43k | if (parser->m_eventPtr && parser->m_eventPtr >= parser->m_positionPtr) { |
2287 | 7.42k | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, |
2288 | 7.42k | parser->m_eventPtr, &parser->m_position); |
2289 | 7.42k | parser->m_positionPtr = parser->m_eventPtr; |
2290 | 7.42k | } |
2291 | 7.43k | return parser->m_position.lineNumber + 1; |
2292 | 7.43k | } |
2293 | | |
2294 | | XML_Size XMLCALL |
2295 | 7.43k | XML_GetCurrentColumnNumber(XML_Parser parser) { |
2296 | 7.43k | if (parser == NULL) |
2297 | 0 | return 0; |
2298 | 7.43k | if (parser->m_eventPtr && parser->m_eventPtr >= parser->m_positionPtr) { |
2299 | 7.42k | XmlUpdatePosition(parser->m_encoding, parser->m_positionPtr, |
2300 | 7.42k | parser->m_eventPtr, &parser->m_position); |
2301 | 7.42k | parser->m_positionPtr = parser->m_eventPtr; |
2302 | 7.42k | } |
2303 | 7.43k | return parser->m_position.columnNumber; |
2304 | 7.43k | } |
2305 | | |
2306 | | void XMLCALL |
2307 | 0 | XML_FreeContentModel(XML_Parser parser, XML_Content *model) { |
2308 | 0 | if (parser != NULL) |
2309 | 0 | FREE(parser, model); |
2310 | 0 | } |
2311 | | |
2312 | | void *XMLCALL |
2313 | 0 | XML_MemMalloc(XML_Parser parser, size_t size) { |
2314 | 0 | if (parser == NULL) |
2315 | 0 | return NULL; |
2316 | 0 | return MALLOC(parser, size); |
2317 | 0 | } |
2318 | | |
2319 | | void *XMLCALL |
2320 | 0 | XML_MemRealloc(XML_Parser parser, void *ptr, size_t size) { |
2321 | 0 | if (parser == NULL) |
2322 | 0 | return NULL; |
2323 | 0 | return REALLOC(parser, ptr, size); |
2324 | 0 | } |
2325 | | |
2326 | | void XMLCALL |
2327 | 0 | XML_MemFree(XML_Parser parser, void *ptr) { |
2328 | 0 | if (parser != NULL) |
2329 | 0 | FREE(parser, ptr); |
2330 | 0 | } |
2331 | | |
2332 | | void XMLCALL |
2333 | 0 | XML_DefaultCurrent(XML_Parser parser) { |
2334 | 0 | if (parser == NULL) |
2335 | 0 | return; |
2336 | 0 | if (parser->m_defaultHandler) { |
2337 | 0 | if (parser->m_openInternalEntities) |
2338 | 0 | reportDefault(parser, parser->m_internalEncoding, |
2339 | 0 | parser->m_openInternalEntities->internalEventPtr, |
2340 | 0 | parser->m_openInternalEntities->internalEventEndPtr); |
2341 | 0 | else |
2342 | 0 | reportDefault(parser, parser->m_encoding, parser->m_eventPtr, |
2343 | 0 | parser->m_eventEndPtr); |
2344 | 0 | } |
2345 | 0 | } |
2346 | | |
2347 | | const XML_LChar *XMLCALL |
2348 | 7.43k | XML_ErrorString(enum XML_Error code) { |
2349 | 7.43k | switch (code) { |
2350 | 0 | case XML_ERROR_NONE: |
2351 | 0 | return NULL; |
2352 | 3 | case XML_ERROR_NO_MEMORY: |
2353 | 3 | return XML_L("out of memory"); |
2354 | 646 | case XML_ERROR_SYNTAX: |
2355 | 646 | return XML_L("syntax error"); |
2356 | 2.16k | case XML_ERROR_NO_ELEMENTS: |
2357 | 2.16k | return XML_L("no element found"); |
2358 | 2.21k | case XML_ERROR_INVALID_TOKEN: |
2359 | 2.21k | return XML_L("not well-formed (invalid token)"); |
2360 | 1.13k | case XML_ERROR_UNCLOSED_TOKEN: |
2361 | 1.13k | return XML_L("unclosed token"); |
2362 | 208 | case XML_ERROR_PARTIAL_CHAR: |
2363 | 208 | return XML_L("partial character"); |
2364 | 162 | case XML_ERROR_TAG_MISMATCH: |
2365 | 162 | return XML_L("mismatched tag"); |
2366 | 136 | case XML_ERROR_DUPLICATE_ATTRIBUTE: |
2367 | 136 | return XML_L("duplicate attribute"); |
2368 | 17 | case XML_ERROR_JUNK_AFTER_DOC_ELEMENT: |
2369 | 17 | return XML_L("junk after document element"); |
2370 | 4 | case XML_ERROR_PARAM_ENTITY_REF: |
2371 | 4 | return XML_L("illegal parameter entity reference"); |
2372 | 164 | case XML_ERROR_UNDEFINED_ENTITY: |
2373 | 164 | return XML_L("undefined entity"); |
2374 | 16 | case XML_ERROR_RECURSIVE_ENTITY_REF: |
2375 | 16 | return XML_L("recursive entity reference"); |
2376 | 43 | case XML_ERROR_ASYNC_ENTITY: |
2377 | 43 | return XML_L("asynchronous entity"); |
2378 | 72 | case XML_ERROR_BAD_CHAR_REF: |
2379 | 72 | return XML_L("reference to invalid character number"); |
2380 | 2 | case XML_ERROR_BINARY_ENTITY_REF: |
2381 | 2 | return XML_L("reference to binary entity"); |
2382 | 2 | case XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF: |
2383 | 2 | return XML_L("reference to external entity in attribute"); |
2384 | 2 | case XML_ERROR_MISPLACED_XML_PI: |
2385 | 2 | return XML_L("XML or text declaration not at start of entity"); |
2386 | 41 | case XML_ERROR_UNKNOWN_ENCODING: |
2387 | 41 | return XML_L("unknown encoding"); |
2388 | 1 | case XML_ERROR_INCORRECT_ENCODING: |
2389 | 1 | return XML_L("encoding specified in XML declaration is incorrect"); |
2390 | 148 | case XML_ERROR_UNCLOSED_CDATA_SECTION: |
2391 | 148 | return XML_L("unclosed CDATA section"); |
2392 | 0 | case XML_ERROR_EXTERNAL_ENTITY_HANDLING: |
2393 | 0 | return XML_L("error in processing external entity reference"); |
2394 | 0 | case XML_ERROR_NOT_STANDALONE: |
2395 | 0 | return XML_L("document is not standalone"); |
2396 | 0 | case XML_ERROR_UNEXPECTED_STATE: |
2397 | 0 | return XML_L("unexpected parser state - please send a bug report"); |
2398 | 0 | case XML_ERROR_ENTITY_DECLARED_IN_PE: |
2399 | 0 | return XML_L("entity declared in parameter entity"); |
2400 | 0 | case XML_ERROR_FEATURE_REQUIRES_XML_DTD: |
2401 | 0 | return XML_L("requested feature requires XML_DTD support in Expat"); |
2402 | 0 | case XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING: |
2403 | 0 | return XML_L("cannot change setting once parsing has begun"); |
2404 | | /* Added in 1.95.7. */ |
2405 | 0 | case XML_ERROR_UNBOUND_PREFIX: |
2406 | 0 | return XML_L("unbound prefix"); |
2407 | | /* Added in 1.95.8. */ |
2408 | 0 | case XML_ERROR_UNDECLARING_PREFIX: |
2409 | 0 | return XML_L("must not undeclare prefix"); |
2410 | 0 | case XML_ERROR_INCOMPLETE_PE: |
2411 | 0 | return XML_L("incomplete markup in parameter entity"); |
2412 | 230 | case XML_ERROR_XML_DECL: |
2413 | 230 | return XML_L("XML declaration not well-formed"); |
2414 | 0 | case XML_ERROR_TEXT_DECL: |
2415 | 0 | return XML_L("text declaration not well-formed"); |
2416 | 14 | case XML_ERROR_PUBLICID: |
2417 | 14 | return XML_L("illegal character(s) in public id"); |
2418 | 0 | case XML_ERROR_SUSPENDED: |
2419 | 0 | return XML_L("parser suspended"); |
2420 | 0 | case XML_ERROR_NOT_SUSPENDED: |
2421 | 0 | return XML_L("parser not suspended"); |
2422 | 0 | case XML_ERROR_ABORTED: |
2423 | 0 | return XML_L("parsing aborted"); |
2424 | 0 | case XML_ERROR_FINISHED: |
2425 | 0 | return XML_L("parsing finished"); |
2426 | 0 | case XML_ERROR_SUSPEND_PE: |
2427 | 0 | return XML_L("cannot suspend in external parameter entity"); |
2428 | | /* Added in 2.0.0. */ |
2429 | 0 | case XML_ERROR_RESERVED_PREFIX_XML: |
2430 | 0 | return XML_L( |
2431 | 0 | "reserved prefix (xml) must not be undeclared or bound to another namespace name"); |
2432 | 0 | case XML_ERROR_RESERVED_PREFIX_XMLNS: |
2433 | 0 | return XML_L("reserved prefix (xmlns) must not be declared or undeclared"); |
2434 | 0 | case XML_ERROR_RESERVED_NAMESPACE_URI: |
2435 | 0 | return XML_L( |
2436 | 0 | "prefix must not be bound to one of the reserved namespace names"); |
2437 | | /* Added in 2.2.5. */ |
2438 | 0 | case XML_ERROR_INVALID_ARGUMENT: /* Constant added in 2.2.1, already */ |
2439 | 0 | return XML_L("invalid argument"); |
2440 | | /* Added in 2.3.0. */ |
2441 | 0 | case XML_ERROR_NO_BUFFER: |
2442 | 0 | return XML_L( |
2443 | 0 | "a successful prior call to function XML_GetBuffer is required"); |
2444 | | /* Added in 2.4.0. */ |
2445 | 0 | case XML_ERROR_AMPLIFICATION_LIMIT_BREACH: |
2446 | 0 | return XML_L( |
2447 | 7.43k | "limit on input amplification factor (from DTD and entities) breached"); |
2448 | 7.43k | } |
2449 | 0 | return NULL; |
2450 | 7.43k | } |
2451 | | |
2452 | | const XML_LChar *XMLCALL |
2453 | 0 | XML_ExpatVersion(void) { |
2454 | | /* V1 is used to string-ize the version number. However, it would |
2455 | | string-ize the actual version macro *names* unless we get them |
2456 | | substituted before being passed to V1. CPP is defined to expand |
2457 | | a macro, then rescan for more expansions. Thus, we use V2 to expand |
2458 | | the version macros, then CPP will expand the resulting V1() macro |
2459 | | with the correct numerals. */ |
2460 | | /* ### I'm assuming cpp is portable in this respect... */ |
2461 | |
|
2462 | 0 | #define V1(a, b, c) XML_L(#a) XML_L(".") XML_L(#b) XML_L(".") XML_L(#c) |
2463 | 0 | #define V2(a, b, c) XML_L("expat_") V1(a, b, c) |
2464 | |
|
2465 | 0 | return V2(XML_MAJOR_VERSION, XML_MINOR_VERSION, XML_MICRO_VERSION); |
2466 | |
|
2467 | 0 | #undef V1 |
2468 | 0 | #undef V2 |
2469 | 0 | } |
2470 | | |
2471 | | XML_Expat_Version XMLCALL |
2472 | 0 | XML_ExpatVersionInfo(void) { |
2473 | 0 | XML_Expat_Version version; |
2474 | |
|
2475 | 0 | version.major = XML_MAJOR_VERSION; |
2476 | 0 | version.minor = XML_MINOR_VERSION; |
2477 | 0 | version.micro = XML_MICRO_VERSION; |
2478 | |
|
2479 | 0 | return version; |
2480 | 0 | } |
2481 | | |
2482 | | const XML_Feature *XMLCALL |
2483 | 0 | XML_GetFeatureList(void) { |
2484 | 0 | static const XML_Feature features[] = { |
2485 | 0 | {XML_FEATURE_SIZEOF_XML_CHAR, XML_L("sizeof(XML_Char)"), |
2486 | 0 | sizeof(XML_Char)}, |
2487 | 0 | {XML_FEATURE_SIZEOF_XML_LCHAR, XML_L("sizeof(XML_LChar)"), |
2488 | 0 | sizeof(XML_LChar)}, |
2489 | | #ifdef XML_UNICODE |
2490 | | {XML_FEATURE_UNICODE, XML_L("XML_UNICODE"), 0}, |
2491 | | #endif |
2492 | | #ifdef XML_UNICODE_WCHAR_T |
2493 | | {XML_FEATURE_UNICODE_WCHAR_T, XML_L("XML_UNICODE_WCHAR_T"), 0}, |
2494 | | #endif |
2495 | | #ifdef XML_DTD |
2496 | | {XML_FEATURE_DTD, XML_L("XML_DTD"), 0}, |
2497 | | #endif |
2498 | 0 | #ifdef XML_CONTEXT_BYTES |
2499 | 0 | {XML_FEATURE_CONTEXT_BYTES, XML_L("XML_CONTEXT_BYTES"), |
2500 | 0 | XML_CONTEXT_BYTES}, |
2501 | 0 | #endif |
2502 | | #ifdef XML_MIN_SIZE |
2503 | | {XML_FEATURE_MIN_SIZE, XML_L("XML_MIN_SIZE"), 0}, |
2504 | | #endif |
2505 | | #ifdef XML_NS |
2506 | | {XML_FEATURE_NS, XML_L("XML_NS"), 0}, |
2507 | | #endif |
2508 | | #ifdef XML_LARGE_SIZE |
2509 | | {XML_FEATURE_LARGE_SIZE, XML_L("XML_LARGE_SIZE"), 0}, |
2510 | | #endif |
2511 | | #ifdef XML_ATTR_INFO |
2512 | | {XML_FEATURE_ATTR_INFO, XML_L("XML_ATTR_INFO"), 0}, |
2513 | | #endif |
2514 | | #ifdef XML_DTD |
2515 | | /* Added in Expat 2.4.0. */ |
2516 | | {XML_FEATURE_BILLION_LAUGHS_ATTACK_PROTECTION_MAXIMUM_AMPLIFICATION_DEFAULT, |
2517 | | XML_L("XML_BLAP_MAX_AMP"), |
2518 | | (long int) |
2519 | | EXPAT_BILLION_LAUGHS_ATTACK_PROTECTION_MAXIMUM_AMPLIFICATION_DEFAULT}, |
2520 | | {XML_FEATURE_BILLION_LAUGHS_ATTACK_PROTECTION_ACTIVATION_THRESHOLD_DEFAULT, |
2521 | | XML_L("XML_BLAP_ACT_THRES"), |
2522 | | EXPAT_BILLION_LAUGHS_ATTACK_PROTECTION_ACTIVATION_THRESHOLD_DEFAULT}, |
2523 | | #endif |
2524 | 0 | {XML_FEATURE_END, NULL, 0}}; |
2525 | |
|
2526 | 0 | return features; |
2527 | 0 | } |
2528 | | |
2529 | | #ifdef XML_DTD |
2530 | | XML_Bool XMLCALL |
2531 | | XML_SetBillionLaughsAttackProtectionMaximumAmplification( |
2532 | | XML_Parser parser, float maximumAmplificationFactor) { |
2533 | | if ((parser == NULL) || (parser->m_parentParser != NULL) |
2534 | | || isnan(maximumAmplificationFactor) |
2535 | | || (maximumAmplificationFactor < 1.0f)) { |
2536 | | return XML_FALSE; |
2537 | | } |
2538 | | parser->m_accounting.maximumAmplificationFactor = maximumAmplificationFactor; |
2539 | | return XML_TRUE; |
2540 | | } |
2541 | | |
2542 | | XML_Bool XMLCALL |
2543 | | XML_SetBillionLaughsAttackProtectionActivationThreshold( |
2544 | | XML_Parser parser, unsigned long long activationThresholdBytes) { |
2545 | | if ((parser == NULL) || (parser->m_parentParser != NULL)) { |
2546 | | return XML_FALSE; |
2547 | | } |
2548 | | parser->m_accounting.activationThresholdBytes = activationThresholdBytes; |
2549 | | return XML_TRUE; |
2550 | | } |
2551 | | #endif /* XML_DTD */ |
2552 | | |
2553 | | /* Initially tag->rawName always points into the parse buffer; |
2554 | | for those TAG instances opened while the current parse buffer was |
2555 | | processed, and not yet closed, we need to store tag->rawName in a more |
2556 | | permanent location, since the parse buffer is about to be discarded. |
2557 | | */ |
2558 | | static XML_Bool |
2559 | 2.08k | storeRawNames(XML_Parser parser) { |
2560 | 2.08k | TAG *tag = parser->m_tagStack; |
2561 | 291k | while (tag) { |
2562 | 289k | int bufSize; |
2563 | 289k | int nameLen = sizeof(XML_Char) * (tag->name.strLen + 1); |
2564 | 289k | size_t rawNameLen; |
2565 | 289k | char *rawNameBuf = tag->buf + nameLen; |
2566 | | /* Stop if already stored. Since m_tagStack is a stack, we can stop |
2567 | | at the first entry that has already been copied; everything |
2568 | | below it in the stack is already been accounted for in a |
2569 | | previous call to this function. |
2570 | | */ |
2571 | 289k | if (tag->rawName == rawNameBuf) |
2572 | 0 | break; |
2573 | | /* For re-use purposes we need to ensure that the |
2574 | | size of tag->buf is a multiple of sizeof(XML_Char). |
2575 | | */ |
2576 | 289k | rawNameLen = ROUND_UP(tag->rawNameLength, sizeof(XML_Char)); |
2577 | | /* Detect and prevent integer overflow. */ |
2578 | 289k | if (rawNameLen > (size_t)INT_MAX - nameLen) |
2579 | 0 | return XML_FALSE; |
2580 | 289k | bufSize = nameLen + (int)rawNameLen; |
2581 | 289k | if (bufSize > tag->bufEnd - tag->buf) { |
2582 | 995 | char *temp = (char *)REALLOC(parser, tag->buf, bufSize); |
2583 | 995 | if (temp == NULL) |
2584 | 0 | return XML_FALSE; |
2585 | | /* if tag->name.str points to tag->buf (only when namespace |
2586 | | processing is off) then we have to update it |
2587 | | */ |
2588 | 995 | if (tag->name.str == (XML_Char *)tag->buf) |
2589 | 995 | tag->name.str = (XML_Char *)temp; |
2590 | | /* if tag->name.localPart is set (when namespace processing is on) |
2591 | | then update it as well, since it will always point into tag->buf |
2592 | | */ |
2593 | 995 | if (tag->name.localPart) |
2594 | 0 | tag->name.localPart |
2595 | 0 | = (XML_Char *)temp + (tag->name.localPart - (XML_Char *)tag->buf); |
2596 | 995 | tag->buf = temp; |
2597 | 995 | tag->bufEnd = temp + bufSize; |
2598 | 995 | rawNameBuf = temp + nameLen; |
2599 | 995 | } |
2600 | 289k | memcpy(rawNameBuf, tag->rawName, tag->rawNameLength); |
2601 | 289k | tag->rawName = rawNameBuf; |
2602 | 289k | tag = tag->parent; |
2603 | 289k | } |
2604 | 2.08k | return XML_TRUE; |
2605 | 2.08k | } |
2606 | | |
2607 | | static enum XML_Error PTRCALL |
2608 | | contentProcessor(XML_Parser parser, const char *start, const char *end, |
2609 | 5.83k | const char **endPtr) { |
2610 | 5.83k | enum XML_Error result = doContent( |
2611 | 5.83k | parser, 0, parser->m_encoding, start, end, endPtr, |
2612 | 5.83k | (XML_Bool)! parser->m_parsingStatus.finalBuffer, XML_ACCOUNT_DIRECT); |
2613 | 5.83k | if (result == XML_ERROR_NONE) { |
2614 | 2.08k | if (! storeRawNames(parser)) |
2615 | 0 | return XML_ERROR_NO_MEMORY; |
2616 | 2.08k | } |
2617 | 5.83k | return result; |
2618 | 5.83k | } |
2619 | | |
2620 | | static enum XML_Error PTRCALL |
2621 | | externalEntityInitProcessor(XML_Parser parser, const char *start, |
2622 | 0 | const char *end, const char **endPtr) { |
2623 | 0 | enum XML_Error result = initializeEncoding(parser); |
2624 | 0 | if (result != XML_ERROR_NONE) |
2625 | 0 | return result; |
2626 | 0 | parser->m_processor = externalEntityInitProcessor2; |
2627 | 0 | return externalEntityInitProcessor2(parser, start, end, endPtr); |
2628 | 0 | } |
2629 | | |
2630 | | static enum XML_Error PTRCALL |
2631 | | externalEntityInitProcessor2(XML_Parser parser, const char *start, |
2632 | 0 | const char *end, const char **endPtr) { |
2633 | 0 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ |
2634 | 0 | int tok = XmlContentTok(parser->m_encoding, start, end, &next); |
2635 | 0 | switch (tok) { |
2636 | 0 | case XML_TOK_BOM: |
2637 | | #ifdef XML_DTD |
2638 | | if (! accountingDiffTolerated(parser, tok, start, next, __LINE__, |
2639 | | XML_ACCOUNT_DIRECT)) { |
2640 | | accountingOnAbort(parser); |
2641 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
2642 | | } |
2643 | | #endif /* XML_DTD */ |
2644 | | |
2645 | | /* If we are at the end of the buffer, this would cause the next stage, |
2646 | | i.e. externalEntityInitProcessor3, to pass control directly to |
2647 | | doContent (by detecting XML_TOK_NONE) without processing any xml text |
2648 | | declaration - causing the error XML_ERROR_MISPLACED_XML_PI in doContent. |
2649 | | */ |
2650 | 0 | if (next == end && ! parser->m_parsingStatus.finalBuffer) { |
2651 | 0 | *endPtr = next; |
2652 | 0 | return XML_ERROR_NONE; |
2653 | 0 | } |
2654 | 0 | start = next; |
2655 | 0 | break; |
2656 | 0 | case XML_TOK_PARTIAL: |
2657 | 0 | if (! parser->m_parsingStatus.finalBuffer) { |
2658 | 0 | *endPtr = start; |
2659 | 0 | return XML_ERROR_NONE; |
2660 | 0 | } |
2661 | 0 | parser->m_eventPtr = start; |
2662 | 0 | return XML_ERROR_UNCLOSED_TOKEN; |
2663 | 0 | case XML_TOK_PARTIAL_CHAR: |
2664 | 0 | if (! parser->m_parsingStatus.finalBuffer) { |
2665 | 0 | *endPtr = start; |
2666 | 0 | return XML_ERROR_NONE; |
2667 | 0 | } |
2668 | 0 | parser->m_eventPtr = start; |
2669 | 0 | return XML_ERROR_PARTIAL_CHAR; |
2670 | 0 | } |
2671 | 0 | parser->m_processor = externalEntityInitProcessor3; |
2672 | 0 | return externalEntityInitProcessor3(parser, start, end, endPtr); |
2673 | 0 | } |
2674 | | |
2675 | | static enum XML_Error PTRCALL |
2676 | | externalEntityInitProcessor3(XML_Parser parser, const char *start, |
2677 | 0 | const char *end, const char **endPtr) { |
2678 | 0 | int tok; |
2679 | 0 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ |
2680 | 0 | parser->m_eventPtr = start; |
2681 | 0 | tok = XmlContentTok(parser->m_encoding, start, end, &next); |
2682 | | /* Note: These bytes are accounted later in: |
2683 | | - processXmlDecl |
2684 | | - externalEntityContentProcessor |
2685 | | */ |
2686 | 0 | parser->m_eventEndPtr = next; |
2687 | |
|
2688 | 0 | switch (tok) { |
2689 | 0 | case XML_TOK_XML_DECL: { |
2690 | 0 | enum XML_Error result; |
2691 | 0 | result = processXmlDecl(parser, 1, start, next); |
2692 | 0 | if (result != XML_ERROR_NONE) |
2693 | 0 | return result; |
2694 | 0 | switch (parser->m_parsingStatus.parsing) { |
2695 | 0 | case XML_SUSPENDED: |
2696 | 0 | *endPtr = next; |
2697 | 0 | return XML_ERROR_NONE; |
2698 | 0 | case XML_FINISHED: |
2699 | 0 | return XML_ERROR_ABORTED; |
2700 | 0 | default: |
2701 | 0 | start = next; |
2702 | 0 | } |
2703 | 0 | } break; |
2704 | 0 | case XML_TOK_PARTIAL: |
2705 | 0 | if (! parser->m_parsingStatus.finalBuffer) { |
2706 | 0 | *endPtr = start; |
2707 | 0 | return XML_ERROR_NONE; |
2708 | 0 | } |
2709 | 0 | return XML_ERROR_UNCLOSED_TOKEN; |
2710 | 0 | case XML_TOK_PARTIAL_CHAR: |
2711 | 0 | if (! parser->m_parsingStatus.finalBuffer) { |
2712 | 0 | *endPtr = start; |
2713 | 0 | return XML_ERROR_NONE; |
2714 | 0 | } |
2715 | 0 | return XML_ERROR_PARTIAL_CHAR; |
2716 | 0 | } |
2717 | 0 | parser->m_processor = externalEntityContentProcessor; |
2718 | 0 | parser->m_tagLevel = 1; |
2719 | 0 | return externalEntityContentProcessor(parser, start, end, endPtr); |
2720 | 0 | } |
2721 | | |
2722 | | static enum XML_Error PTRCALL |
2723 | | externalEntityContentProcessor(XML_Parser parser, const char *start, |
2724 | 0 | const char *end, const char **endPtr) { |
2725 | 0 | enum XML_Error result |
2726 | 0 | = doContent(parser, 1, parser->m_encoding, start, end, endPtr, |
2727 | 0 | (XML_Bool)! parser->m_parsingStatus.finalBuffer, |
2728 | 0 | XML_ACCOUNT_ENTITY_EXPANSION); |
2729 | 0 | if (result == XML_ERROR_NONE) { |
2730 | 0 | if (! storeRawNames(parser)) |
2731 | 0 | return XML_ERROR_NO_MEMORY; |
2732 | 0 | } |
2733 | 0 | return result; |
2734 | 0 | } |
2735 | | |
2736 | | static enum XML_Error |
2737 | | doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, |
2738 | | const char *s, const char *end, const char **nextPtr, |
2739 | 26.9k | XML_Bool haveMore, enum XML_Account account) { |
2740 | | /* save one level of indirection */ |
2741 | 26.9k | DTD *const dtd = parser->m_dtd; |
2742 | | |
2743 | 26.9k | const char **eventPP; |
2744 | 26.9k | const char **eventEndPP; |
2745 | 26.9k | if (enc == parser->m_encoding) { |
2746 | 5.83k | eventPP = &parser->m_eventPtr; |
2747 | 5.83k | eventEndPP = &parser->m_eventEndPtr; |
2748 | 21.0k | } else { |
2749 | 21.0k | eventPP = &(parser->m_openInternalEntities->internalEventPtr); |
2750 | 21.0k | eventEndPP = &(parser->m_openInternalEntities->internalEventEndPtr); |
2751 | 21.0k | } |
2752 | 26.9k | *eventPP = s; |
2753 | | |
2754 | 3.37M | for (;;) { |
2755 | 3.37M | const char *next = s; /* XmlContentTok doesn't always set the last arg */ |
2756 | 3.37M | int tok = XmlContentTok(enc, s, end, &next); |
2757 | | #ifdef XML_DTD |
2758 | | const char *accountAfter |
2759 | | = ((tok == XML_TOK_TRAILING_RSQB) || (tok == XML_TOK_TRAILING_CR)) |
2760 | | ? (haveMore ? s /* i.e. 0 bytes */ : end) |
2761 | | : next; |
2762 | | if (! accountingDiffTolerated(parser, tok, s, accountAfter, __LINE__, |
2763 | | account)) { |
2764 | | accountingOnAbort(parser); |
2765 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
2766 | | } |
2767 | | #endif |
2768 | 3.37M | *eventEndPP = next; |
2769 | 3.37M | switch (tok) { |
2770 | 18 | case XML_TOK_TRAILING_CR: |
2771 | 18 | if (haveMore) { |
2772 | 9 | *nextPtr = s; |
2773 | 9 | return XML_ERROR_NONE; |
2774 | 9 | } |
2775 | 9 | *eventEndPP = end; |
2776 | 9 | if (parser->m_characterDataHandler) { |
2777 | 9 | XML_Char c = 0xA; |
2778 | 9 | parser->m_characterDataHandler(parser->m_handlerArg, &c, 1); |
2779 | 9 | } else if (parser->m_defaultHandler) |
2780 | 0 | reportDefault(parser, enc, s, end); |
2781 | | /* We are at the end of the final buffer, should we check for |
2782 | | XML_SUSPENDED, XML_FINISHED? |
2783 | | */ |
2784 | 9 | if (startTagLevel == 0) |
2785 | 9 | return XML_ERROR_NO_ELEMENTS; |
2786 | 0 | if (parser->m_tagLevel != startTagLevel) |
2787 | 0 | return XML_ERROR_ASYNC_ENTITY; |
2788 | 0 | *nextPtr = end; |
2789 | 0 | return XML_ERROR_NONE; |
2790 | 18.2k | case XML_TOK_NONE: |
2791 | 18.2k | if (haveMore) { |
2792 | 1.20k | *nextPtr = s; |
2793 | 1.20k | return XML_ERROR_NONE; |
2794 | 1.20k | } |
2795 | 17.0k | if (startTagLevel > 0) { |
2796 | 15.8k | if (parser->m_tagLevel != startTagLevel) |
2797 | 31 | return XML_ERROR_ASYNC_ENTITY; |
2798 | 15.8k | *nextPtr = s; |
2799 | 15.8k | return XML_ERROR_NONE; |
2800 | 15.8k | } |
2801 | 1.20k | return XML_ERROR_NO_ELEMENTS; |
2802 | 994 | case XML_TOK_INVALID: |
2803 | 994 | *eventPP = next; |
2804 | 994 | return XML_ERROR_INVALID_TOKEN; |
2805 | 1.33k | case XML_TOK_PARTIAL: |
2806 | 1.33k | if (haveMore) { |
2807 | 606 | *nextPtr = s; |
2808 | 606 | return XML_ERROR_NONE; |
2809 | 606 | } |
2810 | 732 | return XML_ERROR_UNCLOSED_TOKEN; |
2811 | 166 | case XML_TOK_PARTIAL_CHAR: |
2812 | 166 | if (haveMore) { |
2813 | 83 | *nextPtr = s; |
2814 | 83 | return XML_ERROR_NONE; |
2815 | 83 | } |
2816 | 83 | return XML_ERROR_PARTIAL_CHAR; |
2817 | 29.7k | case XML_TOK_ENTITY_REF: { |
2818 | 29.7k | const XML_Char *name; |
2819 | 29.7k | ENTITY *entity; |
2820 | 29.7k | XML_Char ch = (XML_Char)XmlPredefinedEntityName( |
2821 | 29.7k | enc, s + enc->minBytesPerChar, next - enc->minBytesPerChar); |
2822 | 29.7k | if (ch) { |
2823 | | #ifdef XML_DTD |
2824 | | /* NOTE: We are replacing 4-6 characters original input for 1 character |
2825 | | * so there is no amplification and hence recording without |
2826 | | * protection. */ |
2827 | | accountingDiffTolerated(parser, tok, (char *)&ch, |
2828 | | ((char *)&ch) + sizeof(XML_Char), __LINE__, |
2829 | | XML_ACCOUNT_ENTITY_EXPANSION); |
2830 | | #endif /* XML_DTD */ |
2831 | 377 | if (parser->m_characterDataHandler) |
2832 | 377 | parser->m_characterDataHandler(parser->m_handlerArg, &ch, 1); |
2833 | 0 | else if (parser->m_defaultHandler) |
2834 | 0 | reportDefault(parser, enc, s, next); |
2835 | 377 | break; |
2836 | 377 | } |
2837 | 29.3k | name = poolStoreString(&dtd->pool, enc, s + enc->minBytesPerChar, |
2838 | 29.3k | next - enc->minBytesPerChar); |
2839 | 29.3k | if (! name) |
2840 | 0 | return XML_ERROR_NO_MEMORY; |
2841 | 29.3k | entity = (ENTITY *)lookup(parser, &dtd->generalEntities, name, 0); |
2842 | 29.3k | poolDiscard(&dtd->pool); |
2843 | | /* First, determine if a check for an existing declaration is needed; |
2844 | | if yes, check that the entity exists, and that it is internal, |
2845 | | otherwise call the skipped entity or default handler. |
2846 | | */ |
2847 | 29.3k | if (! dtd->hasParamEntityRefs || dtd->standalone) { |
2848 | 17.9k | if (! entity) |
2849 | 117 | return XML_ERROR_UNDEFINED_ENTITY; |
2850 | 17.8k | else if (! entity->is_internal) |
2851 | 0 | return XML_ERROR_ENTITY_DECLARED_IN_PE; |
2852 | 17.9k | } else if (! entity) { |
2853 | 8.09k | if (parser->m_skippedEntityHandler) |
2854 | 0 | parser->m_skippedEntityHandler(parser->m_handlerArg, name, 0); |
2855 | 8.09k | else if (parser->m_defaultHandler) |
2856 | 0 | reportDefault(parser, enc, s, next); |
2857 | 8.09k | break; |
2858 | 8.09k | } |
2859 | 21.1k | if (entity->open) |
2860 | 1 | return XML_ERROR_RECURSIVE_ENTITY_REF; |
2861 | 21.1k | if (entity->notation) |
2862 | 1 | return XML_ERROR_BINARY_ENTITY_REF; |
2863 | 21.1k | if (entity->textPtr) { |
2864 | 21.0k | enum XML_Error result; |
2865 | 21.0k | if (! parser->m_defaultExpandInternalEntities) { |
2866 | 0 | if (parser->m_skippedEntityHandler) |
2867 | 0 | parser->m_skippedEntityHandler(parser->m_handlerArg, entity->name, |
2868 | 0 | 0); |
2869 | 0 | else if (parser->m_defaultHandler) |
2870 | 0 | reportDefault(parser, enc, s, next); |
2871 | 0 | break; |
2872 | 0 | } |
2873 | 21.0k | result = processInternalEntity(parser, entity, XML_FALSE); |
2874 | 21.0k | if (result != XML_ERROR_NONE) |
2875 | 622 | return result; |
2876 | 21.0k | } else if (parser->m_externalEntityRefHandler) { |
2877 | 0 | const XML_Char *context; |
2878 | 0 | entity->open = XML_TRUE; |
2879 | 0 | context = getContext(parser); |
2880 | 0 | entity->open = XML_FALSE; |
2881 | 0 | if (! context) |
2882 | 0 | return XML_ERROR_NO_MEMORY; |
2883 | 0 | if (! parser->m_externalEntityRefHandler( |
2884 | 0 | parser->m_externalEntityRefHandlerArg, context, entity->base, |
2885 | 0 | entity->systemId, entity->publicId)) |
2886 | 0 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
2887 | 0 | poolDiscard(&parser->m_tempPool); |
2888 | 64 | } else if (parser->m_defaultHandler) |
2889 | 0 | reportDefault(parser, enc, s, next); |
2890 | 20.5k | break; |
2891 | 21.1k | } |
2892 | 1.33M | case XML_TOK_START_TAG_NO_ATTS: |
2893 | | /* fall through */ |
2894 | 1.34M | case XML_TOK_START_TAG_WITH_ATTS: { |
2895 | 1.34M | TAG *tag; |
2896 | 1.34M | enum XML_Error result; |
2897 | 1.34M | XML_Char *toPtr; |
2898 | 1.34M | if (parser->m_freeTagList) { |
2899 | 134k | tag = parser->m_freeTagList; |
2900 | 134k | parser->m_freeTagList = parser->m_freeTagList->parent; |
2901 | 1.20M | } else { |
2902 | 1.20M | tag = (TAG *)MALLOC(parser, sizeof(TAG)); |
2903 | 1.20M | if (! tag) |
2904 | 0 | return XML_ERROR_NO_MEMORY; |
2905 | 1.20M | tag->buf = (char *)MALLOC(parser, INIT_TAG_BUF_SIZE); |
2906 | 1.20M | if (! tag->buf) { |
2907 | 0 | FREE(parser, tag); |
2908 | 0 | return XML_ERROR_NO_MEMORY; |
2909 | 0 | } |
2910 | 1.20M | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; |
2911 | 1.20M | } |
2912 | 1.34M | tag->bindings = NULL; |
2913 | 1.34M | tag->parent = parser->m_tagStack; |
2914 | 1.34M | parser->m_tagStack = tag; |
2915 | 1.34M | tag->name.localPart = NULL; |
2916 | 1.34M | tag->name.prefix = NULL; |
2917 | 1.34M | tag->rawName = s + enc->minBytesPerChar; |
2918 | 1.34M | tag->rawNameLength = XmlNameLength(enc, tag->rawName); |
2919 | 1.34M | ++parser->m_tagLevel; |
2920 | 1.34M | { |
2921 | 1.34M | const char *rawNameEnd = tag->rawName + tag->rawNameLength; |
2922 | 1.34M | const char *fromPtr = tag->rawName; |
2923 | 1.34M | toPtr = (XML_Char *)tag->buf; |
2924 | 1.35M | for (;;) { |
2925 | 1.35M | int bufSize; |
2926 | 1.35M | int convLen; |
2927 | 1.35M | const enum XML_Convert_Result convert_res |
2928 | 1.35M | = XmlConvert(enc, &fromPtr, rawNameEnd, (ICHAR **)&toPtr, |
2929 | 1.35M | (ICHAR *)tag->bufEnd - 1); |
2930 | 1.35M | convLen = (int)(toPtr - (XML_Char *)tag->buf); |
2931 | 1.35M | if ((fromPtr >= rawNameEnd) |
2932 | 1.35M | || (convert_res == XML_CONVERT_INPUT_INCOMPLETE)) { |
2933 | 1.34M | tag->name.strLen = convLen; |
2934 | 1.34M | break; |
2935 | 1.34M | } |
2936 | 9.78k | bufSize = (int)(tag->bufEnd - tag->buf) << 1; |
2937 | 9.78k | { |
2938 | 9.78k | char *temp = (char *)REALLOC(parser, tag->buf, bufSize); |
2939 | 9.78k | if (temp == NULL) |
2940 | 0 | return XML_ERROR_NO_MEMORY; |
2941 | 9.78k | tag->buf = temp; |
2942 | 9.78k | tag->bufEnd = temp + bufSize; |
2943 | 9.78k | toPtr = (XML_Char *)temp + convLen; |
2944 | 9.78k | } |
2945 | 9.78k | } |
2946 | 1.34M | } |
2947 | 1.34M | tag->name.str = (XML_Char *)tag->buf; |
2948 | 1.34M | *toPtr = XML_T('\0'); |
2949 | 1.34M | result |
2950 | 1.34M | = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings), account); |
2951 | 1.34M | if (result) |
2952 | 192 | return result; |
2953 | 1.34M | if (parser->m_startElementHandler) |
2954 | 1.34M | parser->m_startElementHandler(parser->m_handlerArg, tag->name.str, |
2955 | 1.34M | (const XML_Char **)parser->m_atts); |
2956 | 0 | else if (parser->m_defaultHandler) |
2957 | 0 | reportDefault(parser, enc, s, next); |
2958 | 1.34M | poolClear(&parser->m_tempPool); |
2959 | 1.34M | break; |
2960 | 1.34M | } |
2961 | 5.71k | case XML_TOK_EMPTY_ELEMENT_NO_ATTS: |
2962 | | /* fall through */ |
2963 | 14.5k | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: { |
2964 | 14.5k | const char *rawName = s + enc->minBytesPerChar; |
2965 | 14.5k | enum XML_Error result; |
2966 | 14.5k | BINDING *bindings = NULL; |
2967 | 14.5k | XML_Bool noElmHandlers = XML_TRUE; |
2968 | 14.5k | TAG_NAME name; |
2969 | 14.5k | name.str = poolStoreString(&parser->m_tempPool, enc, rawName, |
2970 | 14.5k | rawName + XmlNameLength(enc, rawName)); |
2971 | 14.5k | if (! name.str) |
2972 | 0 | return XML_ERROR_NO_MEMORY; |
2973 | 14.5k | poolFinish(&parser->m_tempPool); |
2974 | 14.5k | result = storeAtts(parser, enc, s, &name, &bindings, |
2975 | 14.5k | XML_ACCOUNT_NONE /* token spans whole start tag */); |
2976 | 14.5k | if (result != XML_ERROR_NONE) { |
2977 | 11 | freeBindings(parser, bindings); |
2978 | 11 | return result; |
2979 | 11 | } |
2980 | 14.5k | poolFinish(&parser->m_tempPool); |
2981 | 14.5k | if (parser->m_startElementHandler) { |
2982 | 14.5k | parser->m_startElementHandler(parser->m_handlerArg, name.str, |
2983 | 14.5k | (const XML_Char **)parser->m_atts); |
2984 | 14.5k | noElmHandlers = XML_FALSE; |
2985 | 14.5k | } |
2986 | 14.5k | if (parser->m_endElementHandler) { |
2987 | 14.5k | if (parser->m_startElementHandler) |
2988 | 14.5k | *eventPP = *eventEndPP; |
2989 | 14.5k | parser->m_endElementHandler(parser->m_handlerArg, name.str); |
2990 | 14.5k | noElmHandlers = XML_FALSE; |
2991 | 14.5k | } |
2992 | 14.5k | if (noElmHandlers && parser->m_defaultHandler) |
2993 | 0 | reportDefault(parser, enc, s, next); |
2994 | 14.5k | poolClear(&parser->m_tempPool); |
2995 | 14.5k | freeBindings(parser, bindings); |
2996 | 14.5k | } |
2997 | 14.5k | if ((parser->m_tagLevel == 0) |
2998 | 14.5k | && (parser->m_parsingStatus.parsing != XML_FINISHED)) { |
2999 | 72 | if (parser->m_parsingStatus.parsing == XML_SUSPENDED) |
3000 | 0 | parser->m_processor = epilogProcessor; |
3001 | 72 | else |
3002 | 72 | return epilogProcessor(parser, next, end, nextPtr); |
3003 | 72 | } |
3004 | 14.4k | break; |
3005 | 135k | case XML_TOK_END_TAG: |
3006 | 135k | if (parser->m_tagLevel == startTagLevel) |
3007 | 1 | return XML_ERROR_ASYNC_ENTITY; |
3008 | 135k | else { |
3009 | 135k | int len; |
3010 | 135k | const char *rawName; |
3011 | 135k | TAG *tag = parser->m_tagStack; |
3012 | 135k | parser->m_tagStack = tag->parent; |
3013 | 135k | tag->parent = parser->m_freeTagList; |
3014 | 135k | parser->m_freeTagList = tag; |
3015 | 135k | rawName = s + enc->minBytesPerChar * 2; |
3016 | 135k | len = XmlNameLength(enc, rawName); |
3017 | 135k | if (len != tag->rawNameLength |
3018 | 135k | || memcmp(tag->rawName, rawName, len) != 0) { |
3019 | 162 | *eventPP = rawName; |
3020 | 162 | return XML_ERROR_TAG_MISMATCH; |
3021 | 162 | } |
3022 | 134k | --parser->m_tagLevel; |
3023 | 134k | if (parser->m_endElementHandler) { |
3024 | 134k | const XML_Char *localPart; |
3025 | 134k | const XML_Char *prefix; |
3026 | 134k | XML_Char *uri; |
3027 | 134k | localPart = tag->name.localPart; |
3028 | 134k | if (parser->m_ns && localPart) { |
3029 | | /* localPart and prefix may have been overwritten in |
3030 | | tag->name.str, since this points to the binding->uri |
3031 | | buffer which gets re-used; so we have to add them again |
3032 | | */ |
3033 | 0 | uri = (XML_Char *)tag->name.str + tag->name.uriLen; |
3034 | | /* don't need to check for space - already done in storeAtts() */ |
3035 | 0 | while (*localPart) |
3036 | 0 | *uri++ = *localPart++; |
3037 | 0 | prefix = (XML_Char *)tag->name.prefix; |
3038 | 0 | if (parser->m_ns_triplets && prefix) { |
3039 | 0 | *uri++ = parser->m_namespaceSeparator; |
3040 | 0 | while (*prefix) |
3041 | 0 | *uri++ = *prefix++; |
3042 | 0 | } |
3043 | 0 | *uri = XML_T('\0'); |
3044 | 0 | } |
3045 | 134k | parser->m_endElementHandler(parser->m_handlerArg, tag->name.str); |
3046 | 134k | } else if (parser->m_defaultHandler) |
3047 | 0 | reportDefault(parser, enc, s, next); |
3048 | 134k | while (tag->bindings) { |
3049 | 0 | BINDING *b = tag->bindings; |
3050 | 0 | if (parser->m_endNamespaceDeclHandler) |
3051 | 0 | parser->m_endNamespaceDeclHandler(parser->m_handlerArg, |
3052 | 0 | b->prefix->name); |
3053 | 0 | tag->bindings = tag->bindings->nextTagBinding; |
3054 | 0 | b->nextTagBinding = parser->m_freeBindingList; |
3055 | 0 | parser->m_freeBindingList = b; |
3056 | 0 | b->prefix->binding = b->prevPrefixBinding; |
3057 | 0 | } |
3058 | 134k | if ((parser->m_tagLevel == 0) |
3059 | 134k | && (parser->m_parsingStatus.parsing != XML_FINISHED)) { |
3060 | 7 | if (parser->m_parsingStatus.parsing == XML_SUSPENDED) |
3061 | 0 | parser->m_processor = epilogProcessor; |
3062 | 7 | else |
3063 | 7 | return epilogProcessor(parser, next, end, nextPtr); |
3064 | 7 | } |
3065 | 134k | } |
3066 | 134k | break; |
3067 | 134k | case XML_TOK_CHAR_REF: { |
3068 | 32.5k | int n = XmlCharRefNumber(enc, s); |
3069 | 32.5k | if (n < 0) |
3070 | 62 | return XML_ERROR_BAD_CHAR_REF; |
3071 | 32.4k | if (parser->m_characterDataHandler) { |
3072 | 32.4k | XML_Char buf[XML_ENCODE_MAX]; |
3073 | 32.4k | parser->m_characterDataHandler(parser->m_handlerArg, buf, |
3074 | 32.4k | XmlEncode(n, (ICHAR *)buf)); |
3075 | 32.4k | } else if (parser->m_defaultHandler) |
3076 | 0 | reportDefault(parser, enc, s, next); |
3077 | 32.4k | } break; |
3078 | 2 | case XML_TOK_XML_DECL: |
3079 | 2 | return XML_ERROR_MISPLACED_XML_PI; |
3080 | 852k | case XML_TOK_DATA_NEWLINE: |
3081 | 852k | if (parser->m_characterDataHandler) { |
3082 | 852k | XML_Char c = 0xA; |
3083 | 852k | parser->m_characterDataHandler(parser->m_handlerArg, &c, 1); |
3084 | 852k | } else if (parser->m_defaultHandler) |
3085 | 0 | reportDefault(parser, enc, s, next); |
3086 | 852k | break; |
3087 | 1.99k | case XML_TOK_CDATA_SECT_OPEN: { |
3088 | 1.99k | enum XML_Error result; |
3089 | 1.99k | if (parser->m_startCdataSectionHandler) |
3090 | 0 | parser->m_startCdataSectionHandler(parser->m_handlerArg); |
3091 | | /* BEGIN disabled code */ |
3092 | | /* Suppose you doing a transformation on a document that involves |
3093 | | changing only the character data. You set up a defaultHandler |
3094 | | and a characterDataHandler. The defaultHandler simply copies |
3095 | | characters through. The characterDataHandler does the |
3096 | | transformation and writes the characters out escaping them as |
3097 | | necessary. This case will fail to work if we leave out the |
3098 | | following two lines (because & and < inside CDATA sections will |
3099 | | be incorrectly escaped). |
3100 | | |
3101 | | However, now we have a start/endCdataSectionHandler, so it seems |
3102 | | easier to let the user deal with this. |
3103 | | */ |
3104 | 1.99k | else if (0 && parser->m_characterDataHandler) |
3105 | 0 | parser->m_characterDataHandler(parser->m_handlerArg, parser->m_dataBuf, |
3106 | 0 | 0); |
3107 | | /* END disabled code */ |
3108 | 1.99k | else if (parser->m_defaultHandler) |
3109 | 0 | reportDefault(parser, enc, s, next); |
3110 | 1.99k | result |
3111 | 1.99k | = doCdataSection(parser, enc, &next, end, nextPtr, haveMore, account); |
3112 | 1.99k | if (result != XML_ERROR_NONE) |
3113 | 104 | return result; |
3114 | 1.88k | else if (! next) { |
3115 | 111 | parser->m_processor = cdataSectionProcessor; |
3116 | 111 | return result; |
3117 | 111 | } |
3118 | 1.99k | } break; |
3119 | 4.63k | case XML_TOK_TRAILING_RSQB: |
3120 | 4.63k | if (haveMore) { |
3121 | 14 | *nextPtr = s; |
3122 | 14 | return XML_ERROR_NONE; |
3123 | 14 | } |
3124 | 4.62k | if (parser->m_characterDataHandler) { |
3125 | 4.62k | if (MUST_CONVERT(enc, s)) { |
3126 | 3 | ICHAR *dataPtr = (ICHAR *)parser->m_dataBuf; |
3127 | 3 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)parser->m_dataBufEnd); |
3128 | 3 | parser->m_characterDataHandler( |
3129 | 3 | parser->m_handlerArg, parser->m_dataBuf, |
3130 | 3 | (int)(dataPtr - (ICHAR *)parser->m_dataBuf)); |
3131 | 3 | } else |
3132 | 4.62k | parser->m_characterDataHandler( |
3133 | 4.62k | parser->m_handlerArg, (XML_Char *)s, |
3134 | 4.62k | (int)((XML_Char *)end - (XML_Char *)s)); |
3135 | 4.62k | } else if (parser->m_defaultHandler) |
3136 | 0 | reportDefault(parser, enc, s, end); |
3137 | | /* We are at the end of the final buffer, should we check for |
3138 | | XML_SUSPENDED, XML_FINISHED? |
3139 | | */ |
3140 | 4.62k | if (startTagLevel == 0) { |
3141 | 14 | *eventPP = end; |
3142 | 14 | return XML_ERROR_NO_ELEMENTS; |
3143 | 14 | } |
3144 | 4.61k | if (parser->m_tagLevel != startTagLevel) { |
3145 | 11 | *eventPP = end; |
3146 | 11 | return XML_ERROR_ASYNC_ENTITY; |
3147 | 11 | } |
3148 | 4.59k | *nextPtr = end; |
3149 | 4.59k | return XML_ERROR_NONE; |
3150 | 938k | case XML_TOK_DATA_CHARS: { |
3151 | 938k | XML_CharacterDataHandler charDataHandler = parser->m_characterDataHandler; |
3152 | 938k | if (charDataHandler) { |
3153 | 938k | if (MUST_CONVERT(enc, s)) { |
3154 | 39.6k | for (;;) { |
3155 | 39.6k | ICHAR *dataPtr = (ICHAR *)parser->m_dataBuf; |
3156 | 39.6k | const enum XML_Convert_Result convert_res = XmlConvert( |
3157 | 39.6k | enc, &s, next, &dataPtr, (ICHAR *)parser->m_dataBufEnd); |
3158 | 39.6k | *eventEndPP = s; |
3159 | 39.6k | charDataHandler(parser->m_handlerArg, parser->m_dataBuf, |
3160 | 39.6k | (int)(dataPtr - (ICHAR *)parser->m_dataBuf)); |
3161 | 39.6k | if ((convert_res == XML_CONVERT_COMPLETED) |
3162 | 39.6k | || (convert_res == XML_CONVERT_INPUT_INCOMPLETE)) |
3163 | 39.4k | break; |
3164 | 262 | *eventPP = s; |
3165 | 262 | } |
3166 | 39.4k | } else |
3167 | 899k | charDataHandler(parser->m_handlerArg, (XML_Char *)s, |
3168 | 899k | (int)((XML_Char *)next - (XML_Char *)s)); |
3169 | 938k | } else if (parser->m_defaultHandler) |
3170 | 0 | reportDefault(parser, enc, s, next); |
3171 | 938k | } break; |
3172 | 5.28k | case XML_TOK_PI: |
3173 | 5.28k | if (! reportProcessingInstruction(parser, enc, s, next)) |
3174 | 0 | return XML_ERROR_NO_MEMORY; |
3175 | 5.28k | break; |
3176 | 5.28k | case XML_TOK_COMMENT: |
3177 | 621 | if (! reportComment(parser, enc, s, next)) |
3178 | 0 | return XML_ERROR_NO_MEMORY; |
3179 | 621 | break; |
3180 | 621 | default: |
3181 | | /* All of the tokens produced by XmlContentTok() have their own |
3182 | | * explicit cases, so this default is not strictly necessary. |
3183 | | * However it is a useful safety net, so we retain the code and |
3184 | | * simply exclude it from the coverage tests. |
3185 | | * |
3186 | | * LCOV_EXCL_START |
3187 | | */ |
3188 | 0 | if (parser->m_defaultHandler) |
3189 | 0 | reportDefault(parser, enc, s, next); |
3190 | 0 | break; |
3191 | | /* LCOV_EXCL_STOP */ |
3192 | 3.37M | } |
3193 | 3.35M | *eventPP = s = next; |
3194 | 3.35M | switch (parser->m_parsingStatus.parsing) { |
3195 | 0 | case XML_SUSPENDED: |
3196 | 0 | *nextPtr = next; |
3197 | 0 | return XML_ERROR_NONE; |
3198 | 0 | case XML_FINISHED: |
3199 | 0 | return XML_ERROR_ABORTED; |
3200 | 3.35M | default:; |
3201 | 3.35M | } |
3202 | 3.35M | } |
3203 | | /* not reached */ |
3204 | 26.9k | } |
3205 | | |
3206 | | /* This function does not call free() on the allocated memory, merely |
3207 | | * moving it to the parser's m_freeBindingList where it can be freed or |
3208 | | * reused as appropriate. |
3209 | | */ |
3210 | | static void |
3211 | 14.5k | freeBindings(XML_Parser parser, BINDING *bindings) { |
3212 | 14.5k | while (bindings) { |
3213 | 0 | BINDING *b = bindings; |
3214 | | |
3215 | | /* m_startNamespaceDeclHandler will have been called for this |
3216 | | * binding in addBindings(), so call the end handler now. |
3217 | | */ |
3218 | 0 | if (parser->m_endNamespaceDeclHandler) |
3219 | 0 | parser->m_endNamespaceDeclHandler(parser->m_handlerArg, b->prefix->name); |
3220 | |
|
3221 | 0 | bindings = bindings->nextTagBinding; |
3222 | 0 | b->nextTagBinding = parser->m_freeBindingList; |
3223 | 0 | parser->m_freeBindingList = b; |
3224 | 0 | b->prefix->binding = b->prevPrefixBinding; |
3225 | 0 | } |
3226 | 14.5k | } |
3227 | | |
3228 | | /* Precondition: all arguments must be non-NULL; |
3229 | | Purpose: |
3230 | | - normalize attributes |
3231 | | - check attributes for well-formedness |
3232 | | - generate namespace aware attribute names (URI, prefix) |
3233 | | - build list of attributes for startElementHandler |
3234 | | - default attributes |
3235 | | - process namespace declarations (check and report them) |
3236 | | - generate namespace aware element name (URI, prefix) |
3237 | | */ |
3238 | | static enum XML_Error |
3239 | | storeAtts(XML_Parser parser, const ENCODING *enc, const char *attStr, |
3240 | | TAG_NAME *tagNamePtr, BINDING **bindingsPtr, |
3241 | 1.35M | enum XML_Account account) { |
3242 | 1.35M | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
3243 | 1.35M | ELEMENT_TYPE *elementType; |
3244 | 1.35M | int nDefaultAtts; |
3245 | 1.35M | const XML_Char **appAtts; /* the attribute list for the application */ |
3246 | 1.35M | int attIndex = 0; |
3247 | 1.35M | int prefixLen; |
3248 | 1.35M | int i; |
3249 | 1.35M | int n; |
3250 | 1.35M | XML_Char *uri; |
3251 | 1.35M | int nPrefixes = 0; |
3252 | 1.35M | BINDING *binding; |
3253 | 1.35M | const XML_Char *localPart; |
3254 | | |
3255 | | /* lookup the element type name */ |
3256 | 1.35M | elementType |
3257 | 1.35M | = (ELEMENT_TYPE *)lookup(parser, &dtd->elementTypes, tagNamePtr->str, 0); |
3258 | 1.35M | if (! elementType) { |
3259 | 9.21k | const XML_Char *name = poolCopyString(&dtd->pool, tagNamePtr->str); |
3260 | 9.21k | if (! name) |
3261 | 0 | return XML_ERROR_NO_MEMORY; |
3262 | 9.21k | elementType = (ELEMENT_TYPE *)lookup(parser, &dtd->elementTypes, name, |
3263 | 9.21k | sizeof(ELEMENT_TYPE)); |
3264 | 9.21k | if (! elementType) |
3265 | 0 | return XML_ERROR_NO_MEMORY; |
3266 | 9.21k | if (parser->m_ns && ! setElementTypePrefix(parser, elementType)) |
3267 | 0 | return XML_ERROR_NO_MEMORY; |
3268 | 9.21k | } |
3269 | 1.35M | nDefaultAtts = elementType->nDefaultAtts; |
3270 | | |
3271 | | /* get the attributes from the tokenizer */ |
3272 | 1.35M | n = XmlGetAttributes(enc, attStr, parser->m_attsSize, parser->m_atts); |
3273 | | |
3274 | | /* Detect and prevent integer overflow */ |
3275 | 1.35M | if (n > INT_MAX - nDefaultAtts) { |
3276 | 0 | return XML_ERROR_NO_MEMORY; |
3277 | 0 | } |
3278 | | |
3279 | 1.35M | if (n + nDefaultAtts > parser->m_attsSize) { |
3280 | 120 | int oldAttsSize = parser->m_attsSize; |
3281 | 120 | ATTRIBUTE *temp; |
3282 | | #ifdef XML_ATTR_INFO |
3283 | | XML_AttrInfo *temp2; |
3284 | | #endif |
3285 | | |
3286 | | /* Detect and prevent integer overflow */ |
3287 | 120 | if ((nDefaultAtts > INT_MAX - INIT_ATTS_SIZE) |
3288 | 120 | || (n > INT_MAX - (nDefaultAtts + INIT_ATTS_SIZE))) { |
3289 | 0 | return XML_ERROR_NO_MEMORY; |
3290 | 0 | } |
3291 | | |
3292 | 120 | parser->m_attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; |
3293 | | |
3294 | | /* Detect and prevent integer overflow. |
3295 | | * The preprocessor guard addresses the "always false" warning |
3296 | | * from -Wtype-limits on platforms where |
3297 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3298 | | #if UINT_MAX >= SIZE_MAX |
3299 | | if ((unsigned)parser->m_attsSize > (size_t)(-1) / sizeof(ATTRIBUTE)) { |
3300 | | parser->m_attsSize = oldAttsSize; |
3301 | | return XML_ERROR_NO_MEMORY; |
3302 | | } |
3303 | | #endif |
3304 | | |
3305 | 120 | temp = (ATTRIBUTE *)REALLOC(parser, (void *)parser->m_atts, |
3306 | 120 | parser->m_attsSize * sizeof(ATTRIBUTE)); |
3307 | 120 | if (temp == NULL) { |
3308 | 0 | parser->m_attsSize = oldAttsSize; |
3309 | 0 | return XML_ERROR_NO_MEMORY; |
3310 | 0 | } |
3311 | 120 | parser->m_atts = temp; |
3312 | | #ifdef XML_ATTR_INFO |
3313 | | /* Detect and prevent integer overflow. |
3314 | | * The preprocessor guard addresses the "always false" warning |
3315 | | * from -Wtype-limits on platforms where |
3316 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3317 | | # if UINT_MAX >= SIZE_MAX |
3318 | | if ((unsigned)parser->m_attsSize > (size_t)(-1) / sizeof(XML_AttrInfo)) { |
3319 | | parser->m_attsSize = oldAttsSize; |
3320 | | return XML_ERROR_NO_MEMORY; |
3321 | | } |
3322 | | # endif |
3323 | | |
3324 | | temp2 = (XML_AttrInfo *)REALLOC(parser, (void *)parser->m_attInfo, |
3325 | | parser->m_attsSize * sizeof(XML_AttrInfo)); |
3326 | | if (temp2 == NULL) { |
3327 | | parser->m_attsSize = oldAttsSize; |
3328 | | return XML_ERROR_NO_MEMORY; |
3329 | | } |
3330 | | parser->m_attInfo = temp2; |
3331 | | #endif |
3332 | 120 | if (n > oldAttsSize) |
3333 | 108 | XmlGetAttributes(enc, attStr, n, parser->m_atts); |
3334 | 120 | } |
3335 | | |
3336 | 1.35M | appAtts = (const XML_Char **)parser->m_atts; |
3337 | 1.38M | for (i = 0; i < n; i++) { |
3338 | 25.4k | ATTRIBUTE *currAtt = &parser->m_atts[i]; |
3339 | | #ifdef XML_ATTR_INFO |
3340 | | XML_AttrInfo *currAttInfo = &parser->m_attInfo[i]; |
3341 | | #endif |
3342 | | /* add the name and value to the attribute list */ |
3343 | 25.4k | ATTRIBUTE_ID *attId |
3344 | 25.4k | = getAttributeId(parser, enc, currAtt->name, |
3345 | 25.4k | currAtt->name + XmlNameLength(enc, currAtt->name)); |
3346 | 25.4k | if (! attId) |
3347 | 0 | return XML_ERROR_NO_MEMORY; |
3348 | | #ifdef XML_ATTR_INFO |
3349 | | currAttInfo->nameStart |
3350 | | = parser->m_parseEndByteIndex - (parser->m_parseEndPtr - currAtt->name); |
3351 | | currAttInfo->nameEnd |
3352 | | = currAttInfo->nameStart + XmlNameLength(enc, currAtt->name); |
3353 | | currAttInfo->valueStart = parser->m_parseEndByteIndex |
3354 | | - (parser->m_parseEndPtr - currAtt->valuePtr); |
3355 | | currAttInfo->valueEnd = parser->m_parseEndByteIndex |
3356 | | - (parser->m_parseEndPtr - currAtt->valueEnd); |
3357 | | #endif |
3358 | | /* Detect duplicate attributes by their QNames. This does not work when |
3359 | | namespace processing is turned on and different prefixes for the same |
3360 | | namespace are used. For this case we have a check further down. |
3361 | | */ |
3362 | 25.4k | if ((attId->name)[-1]) { |
3363 | 136 | if (enc == parser->m_encoding) |
3364 | 134 | parser->m_eventPtr = parser->m_atts[i].name; |
3365 | 136 | return XML_ERROR_DUPLICATE_ATTRIBUTE; |
3366 | 136 | } |
3367 | 25.3k | (attId->name)[-1] = 1; |
3368 | 25.3k | appAtts[attIndex++] = attId->name; |
3369 | 25.3k | if (! parser->m_atts[i].normalized) { |
3370 | 8.32k | enum XML_Error result; |
3371 | 8.32k | XML_Bool isCdata = XML_TRUE; |
3372 | | |
3373 | | /* figure out whether declared as other than CDATA */ |
3374 | 8.32k | if (attId->maybeTokenized) { |
3375 | 966 | int j; |
3376 | 1.29k | for (j = 0; j < nDefaultAtts; j++) { |
3377 | 925 | if (attId == elementType->defaultAtts[j].id) { |
3378 | 594 | isCdata = elementType->defaultAtts[j].isCdata; |
3379 | 594 | break; |
3380 | 594 | } |
3381 | 925 | } |
3382 | 966 | } |
3383 | | |
3384 | | /* normalize the attribute value */ |
3385 | 8.32k | result = storeAttributeValue( |
3386 | 8.32k | parser, enc, isCdata, parser->m_atts[i].valuePtr, |
3387 | 8.32k | parser->m_atts[i].valueEnd, &parser->m_tempPool, account); |
3388 | 8.32k | if (result) |
3389 | 67 | return result; |
3390 | 8.25k | appAtts[attIndex] = poolStart(&parser->m_tempPool); |
3391 | 8.25k | poolFinish(&parser->m_tempPool); |
3392 | 16.9k | } else { |
3393 | | /* the value did not need normalizing */ |
3394 | 16.9k | appAtts[attIndex] = poolStoreString(&parser->m_tempPool, enc, |
3395 | 16.9k | parser->m_atts[i].valuePtr, |
3396 | 16.9k | parser->m_atts[i].valueEnd); |
3397 | 16.9k | if (appAtts[attIndex] == 0) |
3398 | 0 | return XML_ERROR_NO_MEMORY; |
3399 | 16.9k | poolFinish(&parser->m_tempPool); |
3400 | 16.9k | } |
3401 | | /* handle prefixed attribute names */ |
3402 | 25.2k | if (attId->prefix) { |
3403 | 0 | if (attId->xmlns) { |
3404 | | /* deal with namespace declarations here */ |
3405 | 0 | enum XML_Error result = addBinding(parser, attId->prefix, attId, |
3406 | 0 | appAtts[attIndex], bindingsPtr); |
3407 | 0 | if (result) |
3408 | 0 | return result; |
3409 | 0 | --attIndex; |
3410 | 0 | } else { |
3411 | | /* deal with other prefixed names later */ |
3412 | 0 | attIndex++; |
3413 | 0 | nPrefixes++; |
3414 | 0 | (attId->name)[-1] = 2; |
3415 | 0 | } |
3416 | 0 | } else |
3417 | 25.2k | attIndex++; |
3418 | 25.2k | } |
3419 | | |
3420 | | /* set-up for XML_GetSpecifiedAttributeCount and XML_GetIdAttributeIndex */ |
3421 | 1.35M | parser->m_nSpecifiedAtts = attIndex; |
3422 | 1.35M | if (elementType->idAtt && (elementType->idAtt->name)[-1]) { |
3423 | 484 | for (i = 0; i < attIndex; i += 2) |
3424 | 484 | if (appAtts[i] == elementType->idAtt->name) { |
3425 | 307 | parser->m_idAttIndex = i; |
3426 | 307 | break; |
3427 | 307 | } |
3428 | 307 | } else |
3429 | 1.35M | parser->m_idAttIndex = -1; |
3430 | | |
3431 | | /* do attribute defaulting */ |
3432 | 1.35M | for (i = 0; i < nDefaultAtts; i++) { |
3433 | 4.01k | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + i; |
3434 | 4.01k | if (! (da->id->name)[-1] && da->value) { |
3435 | 1.27k | if (da->id->prefix) { |
3436 | 0 | if (da->id->xmlns) { |
3437 | 0 | enum XML_Error result = addBinding(parser, da->id->prefix, da->id, |
3438 | 0 | da->value, bindingsPtr); |
3439 | 0 | if (result) |
3440 | 0 | return result; |
3441 | 0 | } else { |
3442 | 0 | (da->id->name)[-1] = 2; |
3443 | 0 | nPrefixes++; |
3444 | 0 | appAtts[attIndex++] = da->id->name; |
3445 | 0 | appAtts[attIndex++] = da->value; |
3446 | 0 | } |
3447 | 1.27k | } else { |
3448 | 1.27k | (da->id->name)[-1] = 1; |
3449 | 1.27k | appAtts[attIndex++] = da->id->name; |
3450 | 1.27k | appAtts[attIndex++] = da->value; |
3451 | 1.27k | } |
3452 | 1.27k | } |
3453 | 4.01k | } |
3454 | 1.35M | appAtts[attIndex] = 0; |
3455 | | |
3456 | | /* expand prefixed attribute names, check for duplicates, |
3457 | | and clear flags that say whether attributes were specified */ |
3458 | 1.35M | i = 0; |
3459 | 1.35M | if (nPrefixes) { |
3460 | 0 | int j; /* hash table index */ |
3461 | 0 | unsigned long version = parser->m_nsAttsVersion; |
3462 | | |
3463 | | /* Detect and prevent invalid shift */ |
3464 | 0 | if (parser->m_nsAttsPower >= sizeof(unsigned int) * 8 /* bits per byte */) { |
3465 | 0 | return XML_ERROR_NO_MEMORY; |
3466 | 0 | } |
3467 | | |
3468 | 0 | unsigned int nsAttsSize = 1u << parser->m_nsAttsPower; |
3469 | 0 | unsigned char oldNsAttsPower = parser->m_nsAttsPower; |
3470 | | /* size of hash table must be at least 2 * (# of prefixed attributes) */ |
3471 | 0 | if ((nPrefixes << 1) |
3472 | 0 | >> parser->m_nsAttsPower) { /* true for m_nsAttsPower = 0 */ |
3473 | 0 | NS_ATT *temp; |
3474 | | /* hash table size must also be a power of 2 and >= 8 */ |
3475 | 0 | while (nPrefixes >> parser->m_nsAttsPower++) |
3476 | 0 | ; |
3477 | 0 | if (parser->m_nsAttsPower < 3) |
3478 | 0 | parser->m_nsAttsPower = 3; |
3479 | | |
3480 | | /* Detect and prevent invalid shift */ |
3481 | 0 | if (parser->m_nsAttsPower >= sizeof(nsAttsSize) * 8 /* bits per byte */) { |
3482 | | /* Restore actual size of memory in m_nsAtts */ |
3483 | 0 | parser->m_nsAttsPower = oldNsAttsPower; |
3484 | 0 | return XML_ERROR_NO_MEMORY; |
3485 | 0 | } |
3486 | | |
3487 | 0 | nsAttsSize = 1u << parser->m_nsAttsPower; |
3488 | | |
3489 | | /* Detect and prevent integer overflow. |
3490 | | * The preprocessor guard addresses the "always false" warning |
3491 | | * from -Wtype-limits on platforms where |
3492 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3493 | | #if UINT_MAX >= SIZE_MAX |
3494 | | if (nsAttsSize > (size_t)(-1) / sizeof(NS_ATT)) { |
3495 | | /* Restore actual size of memory in m_nsAtts */ |
3496 | | parser->m_nsAttsPower = oldNsAttsPower; |
3497 | | return XML_ERROR_NO_MEMORY; |
3498 | | } |
3499 | | #endif |
3500 | |
|
3501 | 0 | temp = (NS_ATT *)REALLOC(parser, parser->m_nsAtts, |
3502 | 0 | nsAttsSize * sizeof(NS_ATT)); |
3503 | 0 | if (! temp) { |
3504 | | /* Restore actual size of memory in m_nsAtts */ |
3505 | 0 | parser->m_nsAttsPower = oldNsAttsPower; |
3506 | 0 | return XML_ERROR_NO_MEMORY; |
3507 | 0 | } |
3508 | 0 | parser->m_nsAtts = temp; |
3509 | 0 | version = 0; /* force re-initialization of m_nsAtts hash table */ |
3510 | 0 | } |
3511 | | /* using a version flag saves us from initializing m_nsAtts every time */ |
3512 | 0 | if (! version) { /* initialize version flags when version wraps around */ |
3513 | 0 | version = INIT_ATTS_VERSION; |
3514 | 0 | for (j = nsAttsSize; j != 0;) |
3515 | 0 | parser->m_nsAtts[--j].version = version; |
3516 | 0 | } |
3517 | 0 | parser->m_nsAttsVersion = --version; |
3518 | | |
3519 | | /* expand prefixed names and check for duplicates */ |
3520 | 0 | for (; i < attIndex; i += 2) { |
3521 | 0 | const XML_Char *s = appAtts[i]; |
3522 | 0 | if (s[-1] == 2) { /* prefixed */ |
3523 | 0 | ATTRIBUTE_ID *id; |
3524 | 0 | const BINDING *b; |
3525 | 0 | unsigned long uriHash; |
3526 | 0 | struct siphash sip_state; |
3527 | 0 | struct sipkey sip_key; |
3528 | |
|
3529 | 0 | copy_salt_to_sipkey(parser, &sip_key); |
3530 | 0 | sip24_init(&sip_state, &sip_key); |
3531 | |
|
3532 | 0 | ((XML_Char *)s)[-1] = 0; /* clear flag */ |
3533 | 0 | id = (ATTRIBUTE_ID *)lookup(parser, &dtd->attributeIds, s, 0); |
3534 | 0 | if (! id || ! id->prefix) { |
3535 | | /* This code is walking through the appAtts array, dealing |
3536 | | * with (in this case) a prefixed attribute name. To be in |
3537 | | * the array, the attribute must have already been bound, so |
3538 | | * has to have passed through the hash table lookup once |
3539 | | * already. That implies that an entry for it already |
3540 | | * exists, so the lookup above will return a pointer to |
3541 | | * already allocated memory. There is no opportunaity for |
3542 | | * the allocator to fail, so the condition above cannot be |
3543 | | * fulfilled. |
3544 | | * |
3545 | | * Since it is difficult to be certain that the above |
3546 | | * analysis is complete, we retain the test and merely |
3547 | | * remove the code from coverage tests. |
3548 | | */ |
3549 | 0 | return XML_ERROR_NO_MEMORY; /* LCOV_EXCL_LINE */ |
3550 | 0 | } |
3551 | 0 | b = id->prefix->binding; |
3552 | 0 | if (! b) |
3553 | 0 | return XML_ERROR_UNBOUND_PREFIX; |
3554 | | |
3555 | 0 | for (j = 0; j < b->uriLen; j++) { |
3556 | 0 | const XML_Char c = b->uri[j]; |
3557 | 0 | if (! poolAppendChar(&parser->m_tempPool, c)) |
3558 | 0 | return XML_ERROR_NO_MEMORY; |
3559 | 0 | } |
3560 | | |
3561 | 0 | sip24_update(&sip_state, b->uri, b->uriLen * sizeof(XML_Char)); |
3562 | |
|
3563 | 0 | while (*s++ != XML_T(ASCII_COLON)) |
3564 | 0 | ; |
3565 | |
|
3566 | 0 | sip24_update(&sip_state, s, keylen(s) * sizeof(XML_Char)); |
3567 | |
|
3568 | 0 | do { /* copies null terminator */ |
3569 | 0 | if (! poolAppendChar(&parser->m_tempPool, *s)) |
3570 | 0 | return XML_ERROR_NO_MEMORY; |
3571 | 0 | } while (*s++); |
3572 | | |
3573 | 0 | uriHash = (unsigned long)sip24_final(&sip_state); |
3574 | |
|
3575 | 0 | { /* Check hash table for duplicate of expanded name (uriName). |
3576 | | Derived from code in lookup(parser, HASH_TABLE *table, ...). |
3577 | | */ |
3578 | 0 | unsigned char step = 0; |
3579 | 0 | unsigned long mask = nsAttsSize - 1; |
3580 | 0 | j = uriHash & mask; /* index into hash table */ |
3581 | 0 | while (parser->m_nsAtts[j].version == version) { |
3582 | | /* for speed we compare stored hash values first */ |
3583 | 0 | if (uriHash == parser->m_nsAtts[j].hash) { |
3584 | 0 | const XML_Char *s1 = poolStart(&parser->m_tempPool); |
3585 | 0 | const XML_Char *s2 = parser->m_nsAtts[j].uriName; |
3586 | | /* s1 is null terminated, but not s2 */ |
3587 | 0 | for (; *s1 == *s2 && *s1 != 0; s1++, s2++) |
3588 | 0 | ; |
3589 | 0 | if (*s1 == 0) |
3590 | 0 | return XML_ERROR_DUPLICATE_ATTRIBUTE; |
3591 | 0 | } |
3592 | 0 | if (! step) |
3593 | 0 | step = PROBE_STEP(uriHash, mask, parser->m_nsAttsPower); |
3594 | 0 | j < step ? (j += nsAttsSize - step) : (j -= step); |
3595 | 0 | } |
3596 | 0 | } |
3597 | | |
3598 | 0 | if (parser->m_ns_triplets) { /* append namespace separator and prefix */ |
3599 | 0 | parser->m_tempPool.ptr[-1] = parser->m_namespaceSeparator; |
3600 | 0 | s = b->prefix->name; |
3601 | 0 | do { |
3602 | 0 | if (! poolAppendChar(&parser->m_tempPool, *s)) |
3603 | 0 | return XML_ERROR_NO_MEMORY; |
3604 | 0 | } while (*s++); |
3605 | 0 | } |
3606 | | |
3607 | | /* store expanded name in attribute list */ |
3608 | 0 | s = poolStart(&parser->m_tempPool); |
3609 | 0 | poolFinish(&parser->m_tempPool); |
3610 | 0 | appAtts[i] = s; |
3611 | | |
3612 | | /* fill empty slot with new version, uriName and hash value */ |
3613 | 0 | parser->m_nsAtts[j].version = version; |
3614 | 0 | parser->m_nsAtts[j].hash = uriHash; |
3615 | 0 | parser->m_nsAtts[j].uriName = s; |
3616 | |
|
3617 | 0 | if (! --nPrefixes) { |
3618 | 0 | i += 2; |
3619 | 0 | break; |
3620 | 0 | } |
3621 | 0 | } else /* not prefixed */ |
3622 | 0 | ((XML_Char *)s)[-1] = 0; /* clear flag */ |
3623 | 0 | } |
3624 | 0 | } |
3625 | | /* clear flags for the remaining attributes */ |
3626 | 1.38M | for (; i < attIndex; i += 2) |
3627 | 26.2k | ((XML_Char *)(appAtts[i]))[-1] = 0; |
3628 | 1.35M | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) |
3629 | 0 | binding->attId->name[-1] = 0; |
3630 | | |
3631 | 1.35M | if (! parser->m_ns) |
3632 | 1.35M | return XML_ERROR_NONE; |
3633 | | |
3634 | | /* expand the element type name */ |
3635 | 0 | if (elementType->prefix) { |
3636 | 0 | binding = elementType->prefix->binding; |
3637 | 0 | if (! binding) |
3638 | 0 | return XML_ERROR_UNBOUND_PREFIX; |
3639 | 0 | localPart = tagNamePtr->str; |
3640 | 0 | while (*localPart++ != XML_T(ASCII_COLON)) |
3641 | 0 | ; |
3642 | 0 | } else if (dtd->defaultPrefix.binding) { |
3643 | 0 | binding = dtd->defaultPrefix.binding; |
3644 | 0 | localPart = tagNamePtr->str; |
3645 | 0 | } else |
3646 | 0 | return XML_ERROR_NONE; |
3647 | 0 | prefixLen = 0; |
3648 | 0 | if (parser->m_ns_triplets && binding->prefix->name) { |
3649 | 0 | for (; binding->prefix->name[prefixLen++];) |
3650 | 0 | ; /* prefixLen includes null terminator */ |
3651 | 0 | } |
3652 | 0 | tagNamePtr->localPart = localPart; |
3653 | 0 | tagNamePtr->uriLen = binding->uriLen; |
3654 | 0 | tagNamePtr->prefix = binding->prefix->name; |
3655 | 0 | tagNamePtr->prefixLen = prefixLen; |
3656 | 0 | for (i = 0; localPart[i++];) |
3657 | 0 | ; /* i includes null terminator */ |
3658 | | |
3659 | | /* Detect and prevent integer overflow */ |
3660 | 0 | if (binding->uriLen > INT_MAX - prefixLen |
3661 | 0 | || i > INT_MAX - (binding->uriLen + prefixLen)) { |
3662 | 0 | return XML_ERROR_NO_MEMORY; |
3663 | 0 | } |
3664 | | |
3665 | 0 | n = i + binding->uriLen + prefixLen; |
3666 | 0 | if (n > binding->uriAlloc) { |
3667 | 0 | TAG *p; |
3668 | | |
3669 | | /* Detect and prevent integer overflow */ |
3670 | 0 | if (n > INT_MAX - EXPAND_SPARE) { |
3671 | 0 | return XML_ERROR_NO_MEMORY; |
3672 | 0 | } |
3673 | | /* Detect and prevent integer overflow. |
3674 | | * The preprocessor guard addresses the "always false" warning |
3675 | | * from -Wtype-limits on platforms where |
3676 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3677 | | #if UINT_MAX >= SIZE_MAX |
3678 | | if ((unsigned)(n + EXPAND_SPARE) > (size_t)(-1) / sizeof(XML_Char)) { |
3679 | | return XML_ERROR_NO_MEMORY; |
3680 | | } |
3681 | | #endif |
3682 | | |
3683 | 0 | uri = (XML_Char *)MALLOC(parser, (n + EXPAND_SPARE) * sizeof(XML_Char)); |
3684 | 0 | if (! uri) |
3685 | 0 | return XML_ERROR_NO_MEMORY; |
3686 | 0 | binding->uriAlloc = n + EXPAND_SPARE; |
3687 | 0 | memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char)); |
3688 | 0 | for (p = parser->m_tagStack; p; p = p->parent) |
3689 | 0 | if (p->name.str == binding->uri) |
3690 | 0 | p->name.str = uri; |
3691 | 0 | FREE(parser, binding->uri); |
3692 | 0 | binding->uri = uri; |
3693 | 0 | } |
3694 | | /* if m_namespaceSeparator != '\0' then uri includes it already */ |
3695 | 0 | uri = binding->uri + binding->uriLen; |
3696 | 0 | memcpy(uri, localPart, i * sizeof(XML_Char)); |
3697 | | /* we always have a namespace separator between localPart and prefix */ |
3698 | 0 | if (prefixLen) { |
3699 | 0 | uri += i - 1; |
3700 | 0 | *uri = parser->m_namespaceSeparator; /* replace null terminator */ |
3701 | 0 | memcpy(uri + 1, binding->prefix->name, prefixLen * sizeof(XML_Char)); |
3702 | 0 | } |
3703 | 0 | tagNamePtr->str = binding->uri; |
3704 | 0 | return XML_ERROR_NONE; |
3705 | 0 | } |
3706 | | |
3707 | | /* addBinding() overwrites the value of prefix->binding without checking. |
3708 | | Therefore one must keep track of the old value outside of addBinding(). |
3709 | | */ |
3710 | | static enum XML_Error |
3711 | | addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, |
3712 | 0 | const XML_Char *uri, BINDING **bindingsPtr) { |
3713 | 0 | static const XML_Char xmlNamespace[] |
3714 | 0 | = {ASCII_h, ASCII_t, ASCII_t, ASCII_p, ASCII_COLON, |
3715 | 0 | ASCII_SLASH, ASCII_SLASH, ASCII_w, ASCII_w, ASCII_w, |
3716 | 0 | ASCII_PERIOD, ASCII_w, ASCII_3, ASCII_PERIOD, ASCII_o, |
3717 | 0 | ASCII_r, ASCII_g, ASCII_SLASH, ASCII_X, ASCII_M, |
3718 | 0 | ASCII_L, ASCII_SLASH, ASCII_1, ASCII_9, ASCII_9, |
3719 | 0 | ASCII_8, ASCII_SLASH, ASCII_n, ASCII_a, ASCII_m, |
3720 | 0 | ASCII_e, ASCII_s, ASCII_p, ASCII_a, ASCII_c, |
3721 | 0 | ASCII_e, '\0'}; |
3722 | 0 | static const int xmlLen = (int)sizeof(xmlNamespace) / sizeof(XML_Char) - 1; |
3723 | 0 | static const XML_Char xmlnsNamespace[] |
3724 | 0 | = {ASCII_h, ASCII_t, ASCII_t, ASCII_p, ASCII_COLON, ASCII_SLASH, |
3725 | 0 | ASCII_SLASH, ASCII_w, ASCII_w, ASCII_w, ASCII_PERIOD, ASCII_w, |
3726 | 0 | ASCII_3, ASCII_PERIOD, ASCII_o, ASCII_r, ASCII_g, ASCII_SLASH, |
3727 | 0 | ASCII_2, ASCII_0, ASCII_0, ASCII_0, ASCII_SLASH, ASCII_x, |
3728 | 0 | ASCII_m, ASCII_l, ASCII_n, ASCII_s, ASCII_SLASH, '\0'}; |
3729 | 0 | static const int xmlnsLen |
3730 | 0 | = (int)sizeof(xmlnsNamespace) / sizeof(XML_Char) - 1; |
3731 | |
|
3732 | 0 | XML_Bool mustBeXML = XML_FALSE; |
3733 | 0 | XML_Bool isXML = XML_TRUE; |
3734 | 0 | XML_Bool isXMLNS = XML_TRUE; |
3735 | |
|
3736 | 0 | BINDING *b; |
3737 | 0 | int len; |
3738 | | |
3739 | | /* empty URI is only valid for default namespace per XML NS 1.0 (not 1.1) */ |
3740 | 0 | if (*uri == XML_T('\0') && prefix->name) |
3741 | 0 | return XML_ERROR_UNDECLARING_PREFIX; |
3742 | | |
3743 | 0 | if (prefix->name && prefix->name[0] == XML_T(ASCII_x) |
3744 | 0 | && prefix->name[1] == XML_T(ASCII_m) |
3745 | 0 | && prefix->name[2] == XML_T(ASCII_l)) { |
3746 | | /* Not allowed to bind xmlns */ |
3747 | 0 | if (prefix->name[3] == XML_T(ASCII_n) && prefix->name[4] == XML_T(ASCII_s) |
3748 | 0 | && prefix->name[5] == XML_T('\0')) |
3749 | 0 | return XML_ERROR_RESERVED_PREFIX_XMLNS; |
3750 | | |
3751 | 0 | if (prefix->name[3] == XML_T('\0')) |
3752 | 0 | mustBeXML = XML_TRUE; |
3753 | 0 | } |
3754 | | |
3755 | 0 | for (len = 0; uri[len]; len++) { |
3756 | 0 | if (isXML && (len > xmlLen || uri[len] != xmlNamespace[len])) |
3757 | 0 | isXML = XML_FALSE; |
3758 | |
|
3759 | 0 | if (! mustBeXML && isXMLNS |
3760 | 0 | && (len > xmlnsLen || uri[len] != xmlnsNamespace[len])) |
3761 | 0 | isXMLNS = XML_FALSE; |
3762 | | |
3763 | | // NOTE: While Expat does not validate namespace URIs against RFC 3986, |
3764 | | // we have to at least make sure that the XML processor on top of |
3765 | | // Expat (that is splitting tag names by namespace separator into |
3766 | | // 2- or 3-tuples (uri-local or uri-local-prefix)) cannot be confused |
3767 | | // by an attacker putting additional namespace separator characters |
3768 | | // into namespace declarations. That would be ambiguous and not to |
3769 | | // be expected. |
3770 | 0 | if (parser->m_ns && (uri[len] == parser->m_namespaceSeparator)) { |
3771 | 0 | return XML_ERROR_SYNTAX; |
3772 | 0 | } |
3773 | 0 | } |
3774 | 0 | isXML = isXML && len == xmlLen; |
3775 | 0 | isXMLNS = isXMLNS && len == xmlnsLen; |
3776 | |
|
3777 | 0 | if (mustBeXML != isXML) |
3778 | 0 | return mustBeXML ? XML_ERROR_RESERVED_PREFIX_XML |
3779 | 0 | : XML_ERROR_RESERVED_NAMESPACE_URI; |
3780 | | |
3781 | 0 | if (isXMLNS) |
3782 | 0 | return XML_ERROR_RESERVED_NAMESPACE_URI; |
3783 | | |
3784 | 0 | if (parser->m_namespaceSeparator) |
3785 | 0 | len++; |
3786 | 0 | if (parser->m_freeBindingList) { |
3787 | 0 | b = parser->m_freeBindingList; |
3788 | 0 | if (len > b->uriAlloc) { |
3789 | | /* Detect and prevent integer overflow */ |
3790 | 0 | if (len > INT_MAX - EXPAND_SPARE) { |
3791 | 0 | return XML_ERROR_NO_MEMORY; |
3792 | 0 | } |
3793 | | |
3794 | | /* Detect and prevent integer overflow. |
3795 | | * The preprocessor guard addresses the "always false" warning |
3796 | | * from -Wtype-limits on platforms where |
3797 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3798 | | #if UINT_MAX >= SIZE_MAX |
3799 | | if ((unsigned)(len + EXPAND_SPARE) > (size_t)(-1) / sizeof(XML_Char)) { |
3800 | | return XML_ERROR_NO_MEMORY; |
3801 | | } |
3802 | | #endif |
3803 | | |
3804 | 0 | XML_Char *temp = (XML_Char *)REALLOC( |
3805 | 0 | parser, b->uri, sizeof(XML_Char) * (len + EXPAND_SPARE)); |
3806 | 0 | if (temp == NULL) |
3807 | 0 | return XML_ERROR_NO_MEMORY; |
3808 | 0 | b->uri = temp; |
3809 | 0 | b->uriAlloc = len + EXPAND_SPARE; |
3810 | 0 | } |
3811 | 0 | parser->m_freeBindingList = b->nextTagBinding; |
3812 | 0 | } else { |
3813 | 0 | b = (BINDING *)MALLOC(parser, sizeof(BINDING)); |
3814 | 0 | if (! b) |
3815 | 0 | return XML_ERROR_NO_MEMORY; |
3816 | | |
3817 | | /* Detect and prevent integer overflow */ |
3818 | 0 | if (len > INT_MAX - EXPAND_SPARE) { |
3819 | 0 | return XML_ERROR_NO_MEMORY; |
3820 | 0 | } |
3821 | | /* Detect and prevent integer overflow. |
3822 | | * The preprocessor guard addresses the "always false" warning |
3823 | | * from -Wtype-limits on platforms where |
3824 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
3825 | | #if UINT_MAX >= SIZE_MAX |
3826 | | if ((unsigned)(len + EXPAND_SPARE) > (size_t)(-1) / sizeof(XML_Char)) { |
3827 | | return XML_ERROR_NO_MEMORY; |
3828 | | } |
3829 | | #endif |
3830 | | |
3831 | 0 | b->uri |
3832 | 0 | = (XML_Char *)MALLOC(parser, sizeof(XML_Char) * (len + EXPAND_SPARE)); |
3833 | 0 | if (! b->uri) { |
3834 | 0 | FREE(parser, b); |
3835 | 0 | return XML_ERROR_NO_MEMORY; |
3836 | 0 | } |
3837 | 0 | b->uriAlloc = len + EXPAND_SPARE; |
3838 | 0 | } |
3839 | 0 | b->uriLen = len; |
3840 | 0 | memcpy(b->uri, uri, len * sizeof(XML_Char)); |
3841 | 0 | if (parser->m_namespaceSeparator) |
3842 | 0 | b->uri[len - 1] = parser->m_namespaceSeparator; |
3843 | 0 | b->prefix = prefix; |
3844 | 0 | b->attId = attId; |
3845 | 0 | b->prevPrefixBinding = prefix->binding; |
3846 | | /* NULL binding when default namespace undeclared */ |
3847 | 0 | if (*uri == XML_T('\0') && prefix == &parser->m_dtd->defaultPrefix) |
3848 | 0 | prefix->binding = NULL; |
3849 | 0 | else |
3850 | 0 | prefix->binding = b; |
3851 | 0 | b->nextTagBinding = *bindingsPtr; |
3852 | 0 | *bindingsPtr = b; |
3853 | | /* if attId == NULL then we are not starting a namespace scope */ |
3854 | 0 | if (attId && parser->m_startNamespaceDeclHandler) |
3855 | 0 | parser->m_startNamespaceDeclHandler(parser->m_handlerArg, prefix->name, |
3856 | 0 | prefix->binding ? uri : 0); |
3857 | 0 | return XML_ERROR_NONE; |
3858 | 0 | } |
3859 | | |
3860 | | /* The idea here is to avoid using stack for each CDATA section when |
3861 | | the whole file is parsed with one call. |
3862 | | */ |
3863 | | static enum XML_Error PTRCALL |
3864 | | cdataSectionProcessor(XML_Parser parser, const char *start, const char *end, |
3865 | 111 | const char **endPtr) { |
3866 | 111 | enum XML_Error result = doCdataSection( |
3867 | 111 | parser, parser->m_encoding, &start, end, endPtr, |
3868 | 111 | (XML_Bool)! parser->m_parsingStatus.finalBuffer, XML_ACCOUNT_DIRECT); |
3869 | 111 | if (result != XML_ERROR_NONE) |
3870 | 111 | return result; |
3871 | 0 | if (start) { |
3872 | 0 | if (parser->m_parentParser) { /* we are parsing an external entity */ |
3873 | 0 | parser->m_processor = externalEntityContentProcessor; |
3874 | 0 | return externalEntityContentProcessor(parser, start, end, endPtr); |
3875 | 0 | } else { |
3876 | 0 | parser->m_processor = contentProcessor; |
3877 | 0 | return contentProcessor(parser, start, end, endPtr); |
3878 | 0 | } |
3879 | 0 | } |
3880 | 0 | return result; |
3881 | 0 | } |
3882 | | |
3883 | | /* startPtr gets set to non-null if the section is closed, and to null if |
3884 | | the section is not yet closed. |
3885 | | */ |
3886 | | static enum XML_Error |
3887 | | doCdataSection(XML_Parser parser, const ENCODING *enc, const char **startPtr, |
3888 | | const char *end, const char **nextPtr, XML_Bool haveMore, |
3889 | 2.10k | enum XML_Account account) { |
3890 | 2.10k | const char *s = *startPtr; |
3891 | 2.10k | const char **eventPP; |
3892 | 2.10k | const char **eventEndPP; |
3893 | 2.10k | if (enc == parser->m_encoding) { |
3894 | 552 | eventPP = &parser->m_eventPtr; |
3895 | 552 | *eventPP = s; |
3896 | 552 | eventEndPP = &parser->m_eventEndPtr; |
3897 | 1.55k | } else { |
3898 | 1.55k | eventPP = &(parser->m_openInternalEntities->internalEventPtr); |
3899 | 1.55k | eventEndPP = &(parser->m_openInternalEntities->internalEventEndPtr); |
3900 | 1.55k | } |
3901 | 2.10k | *eventPP = s; |
3902 | 2.10k | *startPtr = NULL; |
3903 | | |
3904 | 42.6k | for (;;) { |
3905 | 42.6k | const char *next = s; /* in case of XML_TOK_NONE or XML_TOK_PARTIAL */ |
3906 | 42.6k | int tok = XmlCdataSectionTok(enc, s, end, &next); |
3907 | | #ifdef XML_DTD |
3908 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, account)) { |
3909 | | accountingOnAbort(parser); |
3910 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
3911 | | } |
3912 | | #else |
3913 | 42.6k | UNUSED_P(account); |
3914 | 42.6k | #endif |
3915 | 42.6k | *eventEndPP = next; |
3916 | 42.6k | switch (tok) { |
3917 | 1.77k | case XML_TOK_CDATA_SECT_CLOSE: |
3918 | 1.77k | if (parser->m_endCdataSectionHandler) |
3919 | 0 | parser->m_endCdataSectionHandler(parser->m_handlerArg); |
3920 | | /* BEGIN disabled code */ |
3921 | | /* see comment under XML_TOK_CDATA_SECT_OPEN */ |
3922 | 1.77k | else if (0 && parser->m_characterDataHandler) |
3923 | 0 | parser->m_characterDataHandler(parser->m_handlerArg, parser->m_dataBuf, |
3924 | 0 | 0); |
3925 | | /* END disabled code */ |
3926 | 1.77k | else if (parser->m_defaultHandler) |
3927 | 0 | reportDefault(parser, enc, s, next); |
3928 | 1.77k | *startPtr = next; |
3929 | 1.77k | *nextPtr = next; |
3930 | 1.77k | if (parser->m_parsingStatus.parsing == XML_FINISHED) |
3931 | 0 | return XML_ERROR_ABORTED; |
3932 | 1.77k | else |
3933 | 1.77k | return XML_ERROR_NONE; |
3934 | 23.7k | case XML_TOK_DATA_NEWLINE: |
3935 | 23.7k | if (parser->m_characterDataHandler) { |
3936 | 23.7k | XML_Char c = 0xA; |
3937 | 23.7k | parser->m_characterDataHandler(parser->m_handlerArg, &c, 1); |
3938 | 23.7k | } else if (parser->m_defaultHandler) |
3939 | 0 | reportDefault(parser, enc, s, next); |
3940 | 23.7k | break; |
3941 | 16.7k | case XML_TOK_DATA_CHARS: { |
3942 | 16.7k | XML_CharacterDataHandler charDataHandler = parser->m_characterDataHandler; |
3943 | 16.7k | if (charDataHandler) { |
3944 | 16.7k | if (MUST_CONVERT(enc, s)) { |
3945 | 5.35k | for (;;) { |
3946 | 5.35k | ICHAR *dataPtr = (ICHAR *)parser->m_dataBuf; |
3947 | 5.35k | const enum XML_Convert_Result convert_res = XmlConvert( |
3948 | 5.35k | enc, &s, next, &dataPtr, (ICHAR *)parser->m_dataBufEnd); |
3949 | 5.35k | *eventEndPP = next; |
3950 | 5.35k | charDataHandler(parser->m_handlerArg, parser->m_dataBuf, |
3951 | 5.35k | (int)(dataPtr - (ICHAR *)parser->m_dataBuf)); |
3952 | 5.35k | if ((convert_res == XML_CONVERT_COMPLETED) |
3953 | 5.35k | || (convert_res == XML_CONVERT_INPUT_INCOMPLETE)) |
3954 | 4.68k | break; |
3955 | 671 | *eventPP = s; |
3956 | 671 | } |
3957 | 4.68k | } else |
3958 | 12.0k | charDataHandler(parser->m_handlerArg, (XML_Char *)s, |
3959 | 12.0k | (int)((XML_Char *)next - (XML_Char *)s)); |
3960 | 16.7k | } else if (parser->m_defaultHandler) |
3961 | 0 | reportDefault(parser, enc, s, next); |
3962 | 16.7k | } break; |
3963 | 55 | case XML_TOK_INVALID: |
3964 | 55 | *eventPP = next; |
3965 | 55 | return XML_ERROR_INVALID_TOKEN; |
3966 | 24 | case XML_TOK_PARTIAL_CHAR: |
3967 | 24 | if (haveMore) { |
3968 | 12 | *nextPtr = s; |
3969 | 12 | return XML_ERROR_NONE; |
3970 | 12 | } |
3971 | 12 | return XML_ERROR_PARTIAL_CHAR; |
3972 | 38 | case XML_TOK_PARTIAL: |
3973 | 247 | case XML_TOK_NONE: |
3974 | 247 | if (haveMore) { |
3975 | 99 | *nextPtr = s; |
3976 | 99 | return XML_ERROR_NONE; |
3977 | 99 | } |
3978 | 148 | return XML_ERROR_UNCLOSED_CDATA_SECTION; |
3979 | 0 | default: |
3980 | | /* Every token returned by XmlCdataSectionTok() has its own |
3981 | | * explicit case, so this default case will never be executed. |
3982 | | * We retain it as a safety net and exclude it from the coverage |
3983 | | * statistics. |
3984 | | * |
3985 | | * LCOV_EXCL_START |
3986 | | */ |
3987 | 0 | *eventPP = next; |
3988 | 0 | return XML_ERROR_UNEXPECTED_STATE; |
3989 | | /* LCOV_EXCL_STOP */ |
3990 | 42.6k | } |
3991 | | |
3992 | 40.5k | *eventPP = s = next; |
3993 | 40.5k | switch (parser->m_parsingStatus.parsing) { |
3994 | 0 | case XML_SUSPENDED: |
3995 | 0 | *nextPtr = next; |
3996 | 0 | return XML_ERROR_NONE; |
3997 | 0 | case XML_FINISHED: |
3998 | 0 | return XML_ERROR_ABORTED; |
3999 | 40.5k | default:; |
4000 | 40.5k | } |
4001 | 40.5k | } |
4002 | | /* not reached */ |
4003 | 2.10k | } |
4004 | | |
4005 | | #ifdef XML_DTD |
4006 | | |
4007 | | /* The idea here is to avoid using stack for each IGNORE section when |
4008 | | the whole file is parsed with one call. |
4009 | | */ |
4010 | | static enum XML_Error PTRCALL |
4011 | | ignoreSectionProcessor(XML_Parser parser, const char *start, const char *end, |
4012 | | const char **endPtr) { |
4013 | | enum XML_Error result |
4014 | | = doIgnoreSection(parser, parser->m_encoding, &start, end, endPtr, |
4015 | | (XML_Bool)! parser->m_parsingStatus.finalBuffer); |
4016 | | if (result != XML_ERROR_NONE) |
4017 | | return result; |
4018 | | if (start) { |
4019 | | parser->m_processor = prologProcessor; |
4020 | | return prologProcessor(parser, start, end, endPtr); |
4021 | | } |
4022 | | return result; |
4023 | | } |
4024 | | |
4025 | | /* startPtr gets set to non-null is the section is closed, and to null |
4026 | | if the section is not yet closed. |
4027 | | */ |
4028 | | static enum XML_Error |
4029 | | doIgnoreSection(XML_Parser parser, const ENCODING *enc, const char **startPtr, |
4030 | | const char *end, const char **nextPtr, XML_Bool haveMore) { |
4031 | | const char *next = *startPtr; /* in case of XML_TOK_NONE or XML_TOK_PARTIAL */ |
4032 | | int tok; |
4033 | | const char *s = *startPtr; |
4034 | | const char **eventPP; |
4035 | | const char **eventEndPP; |
4036 | | if (enc == parser->m_encoding) { |
4037 | | eventPP = &parser->m_eventPtr; |
4038 | | *eventPP = s; |
4039 | | eventEndPP = &parser->m_eventEndPtr; |
4040 | | } else { |
4041 | | /* It's not entirely clear, but it seems the following two lines |
4042 | | * of code cannot be executed. The only occasions on which 'enc' |
4043 | | * is not 'encoding' are when this function is called |
4044 | | * from the internal entity processing, and IGNORE sections are an |
4045 | | * error in internal entities. |
4046 | | * |
4047 | | * Since it really isn't clear that this is true, we keep the code |
4048 | | * and just remove it from our coverage tests. |
4049 | | * |
4050 | | * LCOV_EXCL_START |
4051 | | */ |
4052 | | eventPP = &(parser->m_openInternalEntities->internalEventPtr); |
4053 | | eventEndPP = &(parser->m_openInternalEntities->internalEventEndPtr); |
4054 | | /* LCOV_EXCL_STOP */ |
4055 | | } |
4056 | | *eventPP = s; |
4057 | | *startPtr = NULL; |
4058 | | tok = XmlIgnoreSectionTok(enc, s, end, &next); |
4059 | | # ifdef XML_DTD |
4060 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, |
4061 | | XML_ACCOUNT_DIRECT)) { |
4062 | | accountingOnAbort(parser); |
4063 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
4064 | | } |
4065 | | # endif |
4066 | | *eventEndPP = next; |
4067 | | switch (tok) { |
4068 | | case XML_TOK_IGNORE_SECT: |
4069 | | if (parser->m_defaultHandler) |
4070 | | reportDefault(parser, enc, s, next); |
4071 | | *startPtr = next; |
4072 | | *nextPtr = next; |
4073 | | if (parser->m_parsingStatus.parsing == XML_FINISHED) |
4074 | | return XML_ERROR_ABORTED; |
4075 | | else |
4076 | | return XML_ERROR_NONE; |
4077 | | case XML_TOK_INVALID: |
4078 | | *eventPP = next; |
4079 | | return XML_ERROR_INVALID_TOKEN; |
4080 | | case XML_TOK_PARTIAL_CHAR: |
4081 | | if (haveMore) { |
4082 | | *nextPtr = s; |
4083 | | return XML_ERROR_NONE; |
4084 | | } |
4085 | | return XML_ERROR_PARTIAL_CHAR; |
4086 | | case XML_TOK_PARTIAL: |
4087 | | case XML_TOK_NONE: |
4088 | | if (haveMore) { |
4089 | | *nextPtr = s; |
4090 | | return XML_ERROR_NONE; |
4091 | | } |
4092 | | return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */ |
4093 | | default: |
4094 | | /* All of the tokens that XmlIgnoreSectionTok() returns have |
4095 | | * explicit cases to handle them, so this default case is never |
4096 | | * executed. We keep it as a safety net anyway, and remove it |
4097 | | * from our test coverage statistics. |
4098 | | * |
4099 | | * LCOV_EXCL_START |
4100 | | */ |
4101 | | *eventPP = next; |
4102 | | return XML_ERROR_UNEXPECTED_STATE; |
4103 | | /* LCOV_EXCL_STOP */ |
4104 | | } |
4105 | | /* not reached */ |
4106 | | } |
4107 | | |
4108 | | #endif /* XML_DTD */ |
4109 | | |
4110 | | static enum XML_Error |
4111 | 7.48k | initializeEncoding(XML_Parser parser) { |
4112 | 7.48k | const char *s; |
4113 | | #ifdef XML_UNICODE |
4114 | | char encodingBuf[128]; |
4115 | | /* See comments about `protocolEncodingName` in parserInit() */ |
4116 | | if (! parser->m_protocolEncodingName) |
4117 | | s = NULL; |
4118 | | else { |
4119 | | int i; |
4120 | | for (i = 0; parser->m_protocolEncodingName[i]; i++) { |
4121 | | if (i == sizeof(encodingBuf) - 1 |
4122 | | || (parser->m_protocolEncodingName[i] & ~0x7f) != 0) { |
4123 | | encodingBuf[0] = '\0'; |
4124 | | break; |
4125 | | } |
4126 | | encodingBuf[i] = (char)parser->m_protocolEncodingName[i]; |
4127 | | } |
4128 | | encodingBuf[i] = '\0'; |
4129 | | s = encodingBuf; |
4130 | | } |
4131 | | #else |
4132 | 7.48k | s = parser->m_protocolEncodingName; |
4133 | 7.48k | #endif |
4134 | 7.48k | if ((parser->m_ns ? XmlInitEncodingNS : XmlInitEncoding)( |
4135 | 7.48k | &parser->m_initEncoding, &parser->m_encoding, s)) |
4136 | 7.48k | return XML_ERROR_NONE; |
4137 | 0 | return handleUnknownEncoding(parser, parser->m_protocolEncodingName); |
4138 | 7.48k | } |
4139 | | |
4140 | | static enum XML_Error |
4141 | | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *s, |
4142 | 1.56k | const char *next) { |
4143 | 1.56k | const char *encodingName = NULL; |
4144 | 1.56k | const XML_Char *storedEncName = NULL; |
4145 | 1.56k | const ENCODING *newEncoding = NULL; |
4146 | 1.56k | const char *version = NULL; |
4147 | 1.56k | const char *versionend; |
4148 | 1.56k | const XML_Char *storedversion = NULL; |
4149 | 1.56k | int standalone = -1; |
4150 | | |
4151 | | #ifdef XML_DTD |
4152 | | if (! accountingDiffTolerated(parser, XML_TOK_XML_DECL, s, next, __LINE__, |
4153 | | XML_ACCOUNT_DIRECT)) { |
4154 | | accountingOnAbort(parser); |
4155 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
4156 | | } |
4157 | | #endif |
4158 | | |
4159 | 1.56k | if (! (parser->m_ns ? XmlParseXmlDeclNS : XmlParseXmlDecl)( |
4160 | 1.56k | isGeneralTextEntity, parser->m_encoding, s, next, &parser->m_eventPtr, |
4161 | 1.56k | &version, &versionend, &encodingName, &newEncoding, &standalone)) { |
4162 | 230 | if (isGeneralTextEntity) |
4163 | 0 | return XML_ERROR_TEXT_DECL; |
4164 | 230 | else |
4165 | 230 | return XML_ERROR_XML_DECL; |
4166 | 230 | } |
4167 | 1.33k | if (! isGeneralTextEntity && standalone == 1) { |
4168 | 33 | parser->m_dtd->standalone = XML_TRUE; |
4169 | | #ifdef XML_DTD |
4170 | | if (parser->m_paramEntityParsing |
4171 | | == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE) |
4172 | | parser->m_paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; |
4173 | | #endif /* XML_DTD */ |
4174 | 33 | } |
4175 | 1.33k | if (parser->m_xmlDeclHandler) { |
4176 | 0 | if (encodingName != NULL) { |
4177 | 0 | storedEncName = poolStoreString( |
4178 | 0 | &parser->m_temp2Pool, parser->m_encoding, encodingName, |
4179 | 0 | encodingName + XmlNameLength(parser->m_encoding, encodingName)); |
4180 | 0 | if (! storedEncName) |
4181 | 0 | return XML_ERROR_NO_MEMORY; |
4182 | 0 | poolFinish(&parser->m_temp2Pool); |
4183 | 0 | } |
4184 | 0 | if (version) { |
4185 | 0 | storedversion |
4186 | 0 | = poolStoreString(&parser->m_temp2Pool, parser->m_encoding, version, |
4187 | 0 | versionend - parser->m_encoding->minBytesPerChar); |
4188 | 0 | if (! storedversion) |
4189 | 0 | return XML_ERROR_NO_MEMORY; |
4190 | 0 | } |
4191 | 0 | parser->m_xmlDeclHandler(parser->m_handlerArg, storedversion, storedEncName, |
4192 | 0 | standalone); |
4193 | 1.33k | } else if (parser->m_defaultHandler) |
4194 | 0 | reportDefault(parser, parser->m_encoding, s, next); |
4195 | 1.33k | if (parser->m_protocolEncodingName == NULL) { |
4196 | 1.33k | if (newEncoding) { |
4197 | | /* Check that the specified encoding does not conflict with what |
4198 | | * the parser has already deduced. Do we have the same number |
4199 | | * of bytes in the smallest representation of a character? If |
4200 | | * this is UTF-16, is it the same endianness? |
4201 | | */ |
4202 | 1.23k | if (newEncoding->minBytesPerChar != parser->m_encoding->minBytesPerChar |
4203 | 1.23k | || (newEncoding->minBytesPerChar == 2 |
4204 | 1.23k | && newEncoding != parser->m_encoding)) { |
4205 | 1 | parser->m_eventPtr = encodingName; |
4206 | 1 | return XML_ERROR_INCORRECT_ENCODING; |
4207 | 1 | } |
4208 | 1.23k | parser->m_encoding = newEncoding; |
4209 | 1.23k | } else if (encodingName) { |
4210 | 41 | enum XML_Error result; |
4211 | 41 | if (! storedEncName) { |
4212 | 41 | storedEncName = poolStoreString( |
4213 | 41 | &parser->m_temp2Pool, parser->m_encoding, encodingName, |
4214 | 41 | encodingName + XmlNameLength(parser->m_encoding, encodingName)); |
4215 | 41 | if (! storedEncName) |
4216 | 0 | return XML_ERROR_NO_MEMORY; |
4217 | 41 | } |
4218 | 41 | result = handleUnknownEncoding(parser, storedEncName); |
4219 | 41 | poolClear(&parser->m_temp2Pool); |
4220 | 41 | if (result == XML_ERROR_UNKNOWN_ENCODING) |
4221 | 41 | parser->m_eventPtr = encodingName; |
4222 | 41 | return result; |
4223 | 41 | } |
4224 | 1.33k | } |
4225 | | |
4226 | 1.28k | if (storedEncName || storedversion) |
4227 | 0 | poolClear(&parser->m_temp2Pool); |
4228 | | |
4229 | 1.28k | return XML_ERROR_NONE; |
4230 | 1.33k | } |
4231 | | |
4232 | | static enum XML_Error |
4233 | 41 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) { |
4234 | 41 | if (parser->m_unknownEncodingHandler) { |
4235 | 0 | XML_Encoding info; |
4236 | 0 | int i; |
4237 | 0 | for (i = 0; i < 256; i++) |
4238 | 0 | info.map[i] = -1; |
4239 | 0 | info.convert = NULL; |
4240 | 0 | info.data = NULL; |
4241 | 0 | info.release = NULL; |
4242 | 0 | if (parser->m_unknownEncodingHandler(parser->m_unknownEncodingHandlerData, |
4243 | 0 | encodingName, &info)) { |
4244 | 0 | ENCODING *enc; |
4245 | 0 | parser->m_unknownEncodingMem = MALLOC(parser, XmlSizeOfUnknownEncoding()); |
4246 | 0 | if (! parser->m_unknownEncodingMem) { |
4247 | 0 | if (info.release) |
4248 | 0 | info.release(info.data); |
4249 | 0 | return XML_ERROR_NO_MEMORY; |
4250 | 0 | } |
4251 | 0 | enc = (parser->m_ns ? XmlInitUnknownEncodingNS : XmlInitUnknownEncoding)( |
4252 | 0 | parser->m_unknownEncodingMem, info.map, info.convert, info.data); |
4253 | 0 | if (enc) { |
4254 | 0 | parser->m_unknownEncodingData = info.data; |
4255 | 0 | parser->m_unknownEncodingRelease = info.release; |
4256 | 0 | parser->m_encoding = enc; |
4257 | 0 | return XML_ERROR_NONE; |
4258 | 0 | } |
4259 | 0 | } |
4260 | 0 | if (info.release != NULL) |
4261 | 0 | info.release(info.data); |
4262 | 0 | } |
4263 | 41 | return XML_ERROR_UNKNOWN_ENCODING; |
4264 | 41 | } |
4265 | | |
4266 | | static enum XML_Error PTRCALL |
4267 | | prologInitProcessor(XML_Parser parser, const char *s, const char *end, |
4268 | 7.48k | const char **nextPtr) { |
4269 | 7.48k | enum XML_Error result = initializeEncoding(parser); |
4270 | 7.48k | if (result != XML_ERROR_NONE) |
4271 | 0 | return result; |
4272 | 7.48k | parser->m_processor = prologProcessor; |
4273 | 7.48k | return prologProcessor(parser, s, end, nextPtr); |
4274 | 7.48k | } |
4275 | | |
4276 | | #ifdef XML_DTD |
4277 | | |
4278 | | static enum XML_Error PTRCALL |
4279 | | externalParEntInitProcessor(XML_Parser parser, const char *s, const char *end, |
4280 | | const char **nextPtr) { |
4281 | | enum XML_Error result = initializeEncoding(parser); |
4282 | | if (result != XML_ERROR_NONE) |
4283 | | return result; |
4284 | | |
4285 | | /* we know now that XML_Parse(Buffer) has been called, |
4286 | | so we consider the external parameter entity read */ |
4287 | | parser->m_dtd->paramEntityRead = XML_TRUE; |
4288 | | |
4289 | | if (parser->m_prologState.inEntityValue) { |
4290 | | parser->m_processor = entityValueInitProcessor; |
4291 | | return entityValueInitProcessor(parser, s, end, nextPtr); |
4292 | | } else { |
4293 | | parser->m_processor = externalParEntProcessor; |
4294 | | return externalParEntProcessor(parser, s, end, nextPtr); |
4295 | | } |
4296 | | } |
4297 | | |
4298 | | static enum XML_Error PTRCALL |
4299 | | entityValueInitProcessor(XML_Parser parser, const char *s, const char *end, |
4300 | | const char **nextPtr) { |
4301 | | int tok; |
4302 | | const char *start = s; |
4303 | | const char *next = start; |
4304 | | parser->m_eventPtr = start; |
4305 | | |
4306 | | for (;;) { |
4307 | | tok = XmlPrologTok(parser->m_encoding, start, end, &next); |
4308 | | /* Note: Except for XML_TOK_BOM below, these bytes are accounted later in: |
4309 | | - storeEntityValue |
4310 | | - processXmlDecl |
4311 | | */ |
4312 | | parser->m_eventEndPtr = next; |
4313 | | if (tok <= 0) { |
4314 | | if (! parser->m_parsingStatus.finalBuffer && tok != XML_TOK_INVALID) { |
4315 | | *nextPtr = s; |
4316 | | return XML_ERROR_NONE; |
4317 | | } |
4318 | | switch (tok) { |
4319 | | case XML_TOK_INVALID: |
4320 | | return XML_ERROR_INVALID_TOKEN; |
4321 | | case XML_TOK_PARTIAL: |
4322 | | return XML_ERROR_UNCLOSED_TOKEN; |
4323 | | case XML_TOK_PARTIAL_CHAR: |
4324 | | return XML_ERROR_PARTIAL_CHAR; |
4325 | | case XML_TOK_NONE: /* start == end */ |
4326 | | default: |
4327 | | break; |
4328 | | } |
4329 | | /* found end of entity value - can store it now */ |
4330 | | return storeEntityValue(parser, parser->m_encoding, s, end, |
4331 | | XML_ACCOUNT_DIRECT); |
4332 | | } else if (tok == XML_TOK_XML_DECL) { |
4333 | | enum XML_Error result; |
4334 | | result = processXmlDecl(parser, 0, start, next); |
4335 | | if (result != XML_ERROR_NONE) |
4336 | | return result; |
4337 | | /* At this point, m_parsingStatus.parsing cannot be XML_SUSPENDED. For |
4338 | | * that to happen, a parameter entity parsing handler must have attempted |
4339 | | * to suspend the parser, which fails and raises an error. The parser can |
4340 | | * be aborted, but can't be suspended. |
4341 | | */ |
4342 | | if (parser->m_parsingStatus.parsing == XML_FINISHED) |
4343 | | return XML_ERROR_ABORTED; |
4344 | | *nextPtr = next; |
4345 | | /* stop scanning for text declaration - we found one */ |
4346 | | parser->m_processor = entityValueProcessor; |
4347 | | return entityValueProcessor(parser, next, end, nextPtr); |
4348 | | } |
4349 | | /* If we are at the end of the buffer, this would cause XmlPrologTok to |
4350 | | return XML_TOK_NONE on the next call, which would then cause the |
4351 | | function to exit with *nextPtr set to s - that is what we want for other |
4352 | | tokens, but not for the BOM - we would rather like to skip it; |
4353 | | then, when this routine is entered the next time, XmlPrologTok will |
4354 | | return XML_TOK_INVALID, since the BOM is still in the buffer |
4355 | | */ |
4356 | | else if (tok == XML_TOK_BOM && next == end |
4357 | | && ! parser->m_parsingStatus.finalBuffer) { |
4358 | | # ifdef XML_DTD |
4359 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, |
4360 | | XML_ACCOUNT_DIRECT)) { |
4361 | | accountingOnAbort(parser); |
4362 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
4363 | | } |
4364 | | # endif |
4365 | | |
4366 | | *nextPtr = next; |
4367 | | return XML_ERROR_NONE; |
4368 | | } |
4369 | | /* If we get this token, we have the start of what might be a |
4370 | | normal tag, but not a declaration (i.e. it doesn't begin with |
4371 | | "<!"). In a DTD context, that isn't legal. |
4372 | | */ |
4373 | | else if (tok == XML_TOK_INSTANCE_START) { |
4374 | | *nextPtr = next; |
4375 | | return XML_ERROR_SYNTAX; |
4376 | | } |
4377 | | start = next; |
4378 | | parser->m_eventPtr = start; |
4379 | | } |
4380 | | } |
4381 | | |
4382 | | static enum XML_Error PTRCALL |
4383 | | externalParEntProcessor(XML_Parser parser, const char *s, const char *end, |
4384 | | const char **nextPtr) { |
4385 | | const char *next = s; |
4386 | | int tok; |
4387 | | |
4388 | | tok = XmlPrologTok(parser->m_encoding, s, end, &next); |
4389 | | if (tok <= 0) { |
4390 | | if (! parser->m_parsingStatus.finalBuffer && tok != XML_TOK_INVALID) { |
4391 | | *nextPtr = s; |
4392 | | return XML_ERROR_NONE; |
4393 | | } |
4394 | | switch (tok) { |
4395 | | case XML_TOK_INVALID: |
4396 | | return XML_ERROR_INVALID_TOKEN; |
4397 | | case XML_TOK_PARTIAL: |
4398 | | return XML_ERROR_UNCLOSED_TOKEN; |
4399 | | case XML_TOK_PARTIAL_CHAR: |
4400 | | return XML_ERROR_PARTIAL_CHAR; |
4401 | | case XML_TOK_NONE: /* start == end */ |
4402 | | default: |
4403 | | break; |
4404 | | } |
4405 | | } |
4406 | | /* This would cause the next stage, i.e. doProlog to be passed XML_TOK_BOM. |
4407 | | However, when parsing an external subset, doProlog will not accept a BOM |
4408 | | as valid, and report a syntax error, so we have to skip the BOM, and |
4409 | | account for the BOM bytes. |
4410 | | */ |
4411 | | else if (tok == XML_TOK_BOM) { |
4412 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, |
4413 | | XML_ACCOUNT_DIRECT)) { |
4414 | | accountingOnAbort(parser); |
4415 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
4416 | | } |
4417 | | |
4418 | | s = next; |
4419 | | tok = XmlPrologTok(parser->m_encoding, s, end, &next); |
4420 | | } |
4421 | | |
4422 | | parser->m_processor = prologProcessor; |
4423 | | return doProlog(parser, parser->m_encoding, s, end, tok, next, nextPtr, |
4424 | | (XML_Bool)! parser->m_parsingStatus.finalBuffer, XML_TRUE, |
4425 | | XML_ACCOUNT_DIRECT); |
4426 | | } |
4427 | | |
4428 | | static enum XML_Error PTRCALL |
4429 | | entityValueProcessor(XML_Parser parser, const char *s, const char *end, |
4430 | | const char **nextPtr) { |
4431 | | const char *start = s; |
4432 | | const char *next = s; |
4433 | | const ENCODING *enc = parser->m_encoding; |
4434 | | int tok; |
4435 | | |
4436 | | for (;;) { |
4437 | | tok = XmlPrologTok(enc, start, end, &next); |
4438 | | /* Note: These bytes are accounted later in: |
4439 | | - storeEntityValue |
4440 | | */ |
4441 | | if (tok <= 0) { |
4442 | | if (! parser->m_parsingStatus.finalBuffer && tok != XML_TOK_INVALID) { |
4443 | | *nextPtr = s; |
4444 | | return XML_ERROR_NONE; |
4445 | | } |
4446 | | switch (tok) { |
4447 | | case XML_TOK_INVALID: |
4448 | | return XML_ERROR_INVALID_TOKEN; |
4449 | | case XML_TOK_PARTIAL: |
4450 | | return XML_ERROR_UNCLOSED_TOKEN; |
4451 | | case XML_TOK_PARTIAL_CHAR: |
4452 | | return XML_ERROR_PARTIAL_CHAR; |
4453 | | case XML_TOK_NONE: /* start == end */ |
4454 | | default: |
4455 | | break; |
4456 | | } |
4457 | | /* found end of entity value - can store it now */ |
4458 | | return storeEntityValue(parser, enc, s, end, XML_ACCOUNT_DIRECT); |
4459 | | } |
4460 | | start = next; |
4461 | | } |
4462 | | } |
4463 | | |
4464 | | #endif /* XML_DTD */ |
4465 | | |
4466 | | static enum XML_Error PTRCALL |
4467 | | prologProcessor(XML_Parser parser, const char *s, const char *end, |
4468 | 9.22k | const char **nextPtr) { |
4469 | 9.22k | const char *next = s; |
4470 | 9.22k | int tok = XmlPrologTok(parser->m_encoding, s, end, &next); |
4471 | 9.22k | return doProlog(parser, parser->m_encoding, s, end, tok, next, nextPtr, |
4472 | 9.22k | (XML_Bool)! parser->m_parsingStatus.finalBuffer, XML_TRUE, |
4473 | 9.22k | XML_ACCOUNT_DIRECT); |
4474 | 9.22k | } |
4475 | | |
4476 | | static enum XML_Error |
4477 | | doProlog(XML_Parser parser, const ENCODING *enc, const char *s, const char *end, |
4478 | | int tok, const char *next, const char **nextPtr, XML_Bool haveMore, |
4479 | 9.22k | XML_Bool allowClosingDoctype, enum XML_Account account) { |
4480 | | #ifdef XML_DTD |
4481 | | static const XML_Char externalSubsetName[] = {ASCII_HASH, '\0'}; |
4482 | | #endif /* XML_DTD */ |
4483 | 9.22k | static const XML_Char atypeCDATA[] |
4484 | 9.22k | = {ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0'}; |
4485 | 9.22k | static const XML_Char atypeID[] = {ASCII_I, ASCII_D, '\0'}; |
4486 | 9.22k | static const XML_Char atypeIDREF[] |
4487 | 9.22k | = {ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, '\0'}; |
4488 | 9.22k | static const XML_Char atypeIDREFS[] |
4489 | 9.22k | = {ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, ASCII_S, '\0'}; |
4490 | 9.22k | static const XML_Char atypeENTITY[] |
4491 | 9.22k | = {ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, ASCII_Y, '\0'}; |
4492 | 9.22k | static const XML_Char atypeENTITIES[] |
4493 | 9.22k | = {ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, |
4494 | 9.22k | ASCII_I, ASCII_E, ASCII_S, '\0'}; |
4495 | 9.22k | static const XML_Char atypeNMTOKEN[] |
4496 | 9.22k | = {ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, ASCII_E, ASCII_N, '\0'}; |
4497 | 9.22k | static const XML_Char atypeNMTOKENS[] |
4498 | 9.22k | = {ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, |
4499 | 9.22k | ASCII_E, ASCII_N, ASCII_S, '\0'}; |
4500 | 9.22k | static const XML_Char notationPrefix[] |
4501 | 9.22k | = {ASCII_N, ASCII_O, ASCII_T, ASCII_A, ASCII_T, |
4502 | 9.22k | ASCII_I, ASCII_O, ASCII_N, ASCII_LPAREN, '\0'}; |
4503 | 9.22k | static const XML_Char enumValueSep[] = {ASCII_PIPE, '\0'}; |
4504 | 9.22k | static const XML_Char enumValueStart[] = {ASCII_LPAREN, '\0'}; |
4505 | | |
4506 | 9.22k | #ifndef XML_DTD |
4507 | 9.22k | UNUSED_P(account); |
4508 | 9.22k | #endif |
4509 | | |
4510 | | /* save one level of indirection */ |
4511 | 9.22k | DTD *const dtd = parser->m_dtd; |
4512 | | |
4513 | 9.22k | const char **eventPP; |
4514 | 9.22k | const char **eventEndPP; |
4515 | 9.22k | enum XML_Content_Quant quant; |
4516 | | |
4517 | 9.22k | if (enc == parser->m_encoding) { |
4518 | 9.22k | eventPP = &parser->m_eventPtr; |
4519 | 9.22k | eventEndPP = &parser->m_eventEndPtr; |
4520 | 9.22k | } else { |
4521 | 0 | eventPP = &(parser->m_openInternalEntities->internalEventPtr); |
4522 | 0 | eventEndPP = &(parser->m_openInternalEntities->internalEventEndPtr); |
4523 | 0 | } |
4524 | | |
4525 | 928k | for (;;) { |
4526 | 928k | int role; |
4527 | 928k | XML_Bool handleDefault = XML_TRUE; |
4528 | 928k | *eventPP = s; |
4529 | 928k | *eventEndPP = next; |
4530 | 928k | if (tok <= 0) { |
4531 | 5.20k | if (haveMore && tok != XML_TOK_INVALID) { |
4532 | 1.74k | *nextPtr = s; |
4533 | 1.74k | return XML_ERROR_NONE; |
4534 | 1.74k | } |
4535 | 3.46k | switch (tok) { |
4536 | 1.06k | case XML_TOK_INVALID: |
4537 | 1.06k | *eventPP = next; |
4538 | 1.06k | return XML_ERROR_INVALID_TOKEN; |
4539 | 398 | case XML_TOK_PARTIAL: |
4540 | 398 | return XML_ERROR_UNCLOSED_TOKEN; |
4541 | 112 | case XML_TOK_PARTIAL_CHAR: |
4542 | 112 | return XML_ERROR_PARTIAL_CHAR; |
4543 | 98 | case -XML_TOK_PROLOG_S: |
4544 | 98 | tok = -tok; |
4545 | 98 | break; |
4546 | 941 | case XML_TOK_NONE: |
4547 | | #ifdef XML_DTD |
4548 | | /* for internal PE NOT referenced between declarations */ |
4549 | | if (enc != parser->m_encoding |
4550 | | && ! parser->m_openInternalEntities->betweenDecl) { |
4551 | | *nextPtr = s; |
4552 | | return XML_ERROR_NONE; |
4553 | | } |
4554 | | /* WFC: PE Between Declarations - must check that PE contains |
4555 | | complete markup, not only for external PEs, but also for |
4556 | | internal PEs if the reference occurs between declarations. |
4557 | | */ |
4558 | | if (parser->m_isParamEntity || enc != parser->m_encoding) { |
4559 | | if (XmlTokenRole(&parser->m_prologState, XML_TOK_NONE, end, end, enc) |
4560 | | == XML_ROLE_ERROR) |
4561 | | return XML_ERROR_INCOMPLETE_PE; |
4562 | | *nextPtr = s; |
4563 | | return XML_ERROR_NONE; |
4564 | | } |
4565 | | #endif /* XML_DTD */ |
4566 | 941 | return XML_ERROR_NO_ELEMENTS; |
4567 | 848 | default: |
4568 | 848 | tok = -tok; |
4569 | 848 | next = end; |
4570 | 848 | break; |
4571 | 3.46k | } |
4572 | 3.46k | } |
4573 | 923k | role = XmlTokenRole(&parser->m_prologState, tok, s, next, enc); |
4574 | | #ifdef XML_DTD |
4575 | | switch (role) { |
4576 | | case XML_ROLE_INSTANCE_START: // bytes accounted in contentProcessor |
4577 | | case XML_ROLE_XML_DECL: // bytes accounted in processXmlDecl |
4578 | | case XML_ROLE_TEXT_DECL: // bytes accounted in processXmlDecl |
4579 | | break; |
4580 | | default: |
4581 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, account)) { |
4582 | | accountingOnAbort(parser); |
4583 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
4584 | | } |
4585 | | } |
4586 | | #endif |
4587 | 923k | switch (role) { |
4588 | 1.56k | case XML_ROLE_XML_DECL: { |
4589 | 1.56k | enum XML_Error result = processXmlDecl(parser, 0, s, next); |
4590 | 1.56k | if (result != XML_ERROR_NONE) |
4591 | 272 | return result; |
4592 | 1.28k | enc = parser->m_encoding; |
4593 | 1.28k | handleDefault = XML_FALSE; |
4594 | 1.28k | } break; |
4595 | 3.22k | case XML_ROLE_DOCTYPE_NAME: |
4596 | 3.22k | if (parser->m_startDoctypeDeclHandler) { |
4597 | 0 | parser->m_doctypeName |
4598 | 0 | = poolStoreString(&parser->m_tempPool, enc, s, next); |
4599 | 0 | if (! parser->m_doctypeName) |
4600 | 0 | return XML_ERROR_NO_MEMORY; |
4601 | 0 | poolFinish(&parser->m_tempPool); |
4602 | 0 | parser->m_doctypePubid = NULL; |
4603 | 0 | handleDefault = XML_FALSE; |
4604 | 0 | } |
4605 | 3.22k | parser->m_doctypeSysid = NULL; /* always initialize to NULL */ |
4606 | 3.22k | break; |
4607 | 2.96k | case XML_ROLE_DOCTYPE_INTERNAL_SUBSET: |
4608 | 2.96k | if (parser->m_startDoctypeDeclHandler) { |
4609 | 0 | parser->m_startDoctypeDeclHandler( |
4610 | 0 | parser->m_handlerArg, parser->m_doctypeName, parser->m_doctypeSysid, |
4611 | 0 | parser->m_doctypePubid, 1); |
4612 | 0 | parser->m_doctypeName = NULL; |
4613 | 0 | poolClear(&parser->m_tempPool); |
4614 | 0 | handleDefault = XML_FALSE; |
4615 | 0 | } |
4616 | 2.96k | break; |
4617 | | #ifdef XML_DTD |
4618 | | case XML_ROLE_TEXT_DECL: { |
4619 | | enum XML_Error result = processXmlDecl(parser, 1, s, next); |
4620 | | if (result != XML_ERROR_NONE) |
4621 | | return result; |
4622 | | enc = parser->m_encoding; |
4623 | | handleDefault = XML_FALSE; |
4624 | | } break; |
4625 | | #endif /* XML_DTD */ |
4626 | 424 | case XML_ROLE_DOCTYPE_PUBLIC_ID: |
4627 | | #ifdef XML_DTD |
4628 | | parser->m_useForeignDTD = XML_FALSE; |
4629 | | parser->m_declEntity = (ENTITY *)lookup( |
4630 | | parser, &dtd->paramEntities, externalSubsetName, sizeof(ENTITY)); |
4631 | | if (! parser->m_declEntity) |
4632 | | return XML_ERROR_NO_MEMORY; |
4633 | | #endif /* XML_DTD */ |
4634 | 424 | dtd->hasParamEntityRefs = XML_TRUE; |
4635 | 424 | if (parser->m_startDoctypeDeclHandler) { |
4636 | 0 | XML_Char *pubId; |
4637 | 0 | if (! XmlIsPublicId(enc, s, next, eventPP)) |
4638 | 0 | return XML_ERROR_PUBLICID; |
4639 | 0 | pubId = poolStoreString(&parser->m_tempPool, enc, |
4640 | 0 | s + enc->minBytesPerChar, |
4641 | 0 | next - enc->minBytesPerChar); |
4642 | 0 | if (! pubId) |
4643 | 0 | return XML_ERROR_NO_MEMORY; |
4644 | 0 | normalizePublicId(pubId); |
4645 | 0 | poolFinish(&parser->m_tempPool); |
4646 | 0 | parser->m_doctypePubid = pubId; |
4647 | 0 | handleDefault = XML_FALSE; |
4648 | 0 | goto alreadyChecked; |
4649 | 0 | } |
4650 | | /* fall through */ |
4651 | 3.80k | case XML_ROLE_ENTITY_PUBLIC_ID: |
4652 | 3.80k | if (! XmlIsPublicId(enc, s, next, eventPP)) |
4653 | 12 | return XML_ERROR_PUBLICID; |
4654 | 3.79k | alreadyChecked: |
4655 | 3.79k | if (dtd->keepProcessing && parser->m_declEntity) { |
4656 | 656 | XML_Char *tem |
4657 | 656 | = poolStoreString(&dtd->pool, enc, s + enc->minBytesPerChar, |
4658 | 656 | next - enc->minBytesPerChar); |
4659 | 656 | if (! tem) |
4660 | 0 | return XML_ERROR_NO_MEMORY; |
4661 | 656 | normalizePublicId(tem); |
4662 | 656 | parser->m_declEntity->publicId = tem; |
4663 | 656 | poolFinish(&dtd->pool); |
4664 | | /* Don't suppress the default handler if we fell through from |
4665 | | * the XML_ROLE_DOCTYPE_PUBLIC_ID case. |
4666 | | */ |
4667 | 656 | if (parser->m_entityDeclHandler && role == XML_ROLE_ENTITY_PUBLIC_ID) |
4668 | 0 | handleDefault = XML_FALSE; |
4669 | 656 | } |
4670 | 3.79k | break; |
4671 | 3.79k | case XML_ROLE_DOCTYPE_CLOSE: |
4672 | 1.43k | if (allowClosingDoctype != XML_TRUE) { |
4673 | | /* Must not close doctype from within expanded parameter entities */ |
4674 | 0 | return XML_ERROR_INVALID_TOKEN; |
4675 | 0 | } |
4676 | | |
4677 | 1.43k | if (parser->m_doctypeName) { |
4678 | 0 | parser->m_startDoctypeDeclHandler( |
4679 | 0 | parser->m_handlerArg, parser->m_doctypeName, parser->m_doctypeSysid, |
4680 | 0 | parser->m_doctypePubid, 0); |
4681 | 0 | poolClear(&parser->m_tempPool); |
4682 | 0 | handleDefault = XML_FALSE; |
4683 | 0 | } |
4684 | | /* parser->m_doctypeSysid will be non-NULL in the case of a previous |
4685 | | XML_ROLE_DOCTYPE_SYSTEM_ID, even if parser->m_startDoctypeDeclHandler |
4686 | | was not set, indicating an external subset |
4687 | | */ |
4688 | | #ifdef XML_DTD |
4689 | | if (parser->m_doctypeSysid || parser->m_useForeignDTD) { |
4690 | | XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs; |
4691 | | dtd->hasParamEntityRefs = XML_TRUE; |
4692 | | if (parser->m_paramEntityParsing |
4693 | | && parser->m_externalEntityRefHandler) { |
4694 | | ENTITY *entity = (ENTITY *)lookup(parser, &dtd->paramEntities, |
4695 | | externalSubsetName, sizeof(ENTITY)); |
4696 | | if (! entity) { |
4697 | | /* The external subset name "#" will have already been |
4698 | | * inserted into the hash table at the start of the |
4699 | | * external entity parsing, so no allocation will happen |
4700 | | * and lookup() cannot fail. |
4701 | | */ |
4702 | | return XML_ERROR_NO_MEMORY; /* LCOV_EXCL_LINE */ |
4703 | | } |
4704 | | if (parser->m_useForeignDTD) |
4705 | | entity->base = parser->m_curBase; |
4706 | | dtd->paramEntityRead = XML_FALSE; |
4707 | | if (! parser->m_externalEntityRefHandler( |
4708 | | parser->m_externalEntityRefHandlerArg, 0, entity->base, |
4709 | | entity->systemId, entity->publicId)) |
4710 | | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
4711 | | if (dtd->paramEntityRead) { |
4712 | | if (! dtd->standalone && parser->m_notStandaloneHandler |
4713 | | && ! parser->m_notStandaloneHandler(parser->m_handlerArg)) |
4714 | | return XML_ERROR_NOT_STANDALONE; |
4715 | | } |
4716 | | /* if we didn't read the foreign DTD then this means that there |
4717 | | is no external subset and we must reset dtd->hasParamEntityRefs |
4718 | | */ |
4719 | | else if (! parser->m_doctypeSysid) |
4720 | | dtd->hasParamEntityRefs = hadParamEntityRefs; |
4721 | | /* end of DTD - no need to update dtd->keepProcessing */ |
4722 | | } |
4723 | | parser->m_useForeignDTD = XML_FALSE; |
4724 | | } |
4725 | | #endif /* XML_DTD */ |
4726 | 1.43k | if (parser->m_endDoctypeDeclHandler) { |
4727 | 0 | parser->m_endDoctypeDeclHandler(parser->m_handlerArg); |
4728 | 0 | handleDefault = XML_FALSE; |
4729 | 0 | } |
4730 | 1.43k | break; |
4731 | 3.92k | case XML_ROLE_INSTANCE_START: |
4732 | | #ifdef XML_DTD |
4733 | | /* if there is no DOCTYPE declaration then now is the |
4734 | | last chance to read the foreign DTD |
4735 | | */ |
4736 | | if (parser->m_useForeignDTD) { |
4737 | | XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs; |
4738 | | dtd->hasParamEntityRefs = XML_TRUE; |
4739 | | if (parser->m_paramEntityParsing |
4740 | | && parser->m_externalEntityRefHandler) { |
4741 | | ENTITY *entity = (ENTITY *)lookup(parser, &dtd->paramEntities, |
4742 | | externalSubsetName, sizeof(ENTITY)); |
4743 | | if (! entity) |
4744 | | return XML_ERROR_NO_MEMORY; |
4745 | | entity->base = parser->m_curBase; |
4746 | | dtd->paramEntityRead = XML_FALSE; |
4747 | | if (! parser->m_externalEntityRefHandler( |
4748 | | parser->m_externalEntityRefHandlerArg, 0, entity->base, |
4749 | | entity->systemId, entity->publicId)) |
4750 | | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
4751 | | if (dtd->paramEntityRead) { |
4752 | | if (! dtd->standalone && parser->m_notStandaloneHandler |
4753 | | && ! parser->m_notStandaloneHandler(parser->m_handlerArg)) |
4754 | | return XML_ERROR_NOT_STANDALONE; |
4755 | | } |
4756 | | /* if we didn't read the foreign DTD then this means that there |
4757 | | is no external subset and we must reset dtd->hasParamEntityRefs |
4758 | | */ |
4759 | | else |
4760 | | dtd->hasParamEntityRefs = hadParamEntityRefs; |
4761 | | /* end of DTD - no need to update dtd->keepProcessing */ |
4762 | | } |
4763 | | } |
4764 | | #endif /* XML_DTD */ |
4765 | 3.92k | parser->m_processor = contentProcessor; |
4766 | 3.92k | return contentProcessor(parser, s, end, nextPtr); |
4767 | 10.2k | case XML_ROLE_ATTLIST_ELEMENT_NAME: |
4768 | 10.2k | parser->m_declElementType = getElementType(parser, enc, s, next); |
4769 | 10.2k | if (! parser->m_declElementType) |
4770 | 0 | return XML_ERROR_NO_MEMORY; |
4771 | 10.2k | goto checkAttListDeclHandler; |
4772 | 24.6k | case XML_ROLE_ATTRIBUTE_NAME: |
4773 | 24.6k | parser->m_declAttributeId = getAttributeId(parser, enc, s, next); |
4774 | 24.6k | if (! parser->m_declAttributeId) |
4775 | 0 | return XML_ERROR_NO_MEMORY; |
4776 | 24.6k | parser->m_declAttributeIsCdata = XML_FALSE; |
4777 | 24.6k | parser->m_declAttributeType = NULL; |
4778 | 24.6k | parser->m_declAttributeIsId = XML_FALSE; |
4779 | 24.6k | goto checkAttListDeclHandler; |
4780 | 7.02k | case XML_ROLE_ATTRIBUTE_TYPE_CDATA: |
4781 | 7.02k | parser->m_declAttributeIsCdata = XML_TRUE; |
4782 | 7.02k | parser->m_declAttributeType = atypeCDATA; |
4783 | 7.02k | goto checkAttListDeclHandler; |
4784 | 2.92k | case XML_ROLE_ATTRIBUTE_TYPE_ID: |
4785 | 2.92k | parser->m_declAttributeIsId = XML_TRUE; |
4786 | 2.92k | parser->m_declAttributeType = atypeID; |
4787 | 2.92k | goto checkAttListDeclHandler; |
4788 | 220 | case XML_ROLE_ATTRIBUTE_TYPE_IDREF: |
4789 | 220 | parser->m_declAttributeType = atypeIDREF; |
4790 | 220 | goto checkAttListDeclHandler; |
4791 | 201 | case XML_ROLE_ATTRIBUTE_TYPE_IDREFS: |
4792 | 201 | parser->m_declAttributeType = atypeIDREFS; |
4793 | 201 | goto checkAttListDeclHandler; |
4794 | 238 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITY: |
4795 | 238 | parser->m_declAttributeType = atypeENTITY; |
4796 | 238 | goto checkAttListDeclHandler; |
4797 | 330 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES: |
4798 | 330 | parser->m_declAttributeType = atypeENTITIES; |
4799 | 330 | goto checkAttListDeclHandler; |
4800 | 545 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN: |
4801 | 545 | parser->m_declAttributeType = atypeNMTOKEN; |
4802 | 545 | goto checkAttListDeclHandler; |
4803 | 31 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS: |
4804 | 31 | parser->m_declAttributeType = atypeNMTOKENS; |
4805 | 46.4k | checkAttListDeclHandler: |
4806 | 46.4k | if (dtd->keepProcessing && parser->m_attlistDeclHandler) |
4807 | 0 | handleDefault = XML_FALSE; |
4808 | 46.4k | break; |
4809 | 13.6k | case XML_ROLE_ATTRIBUTE_ENUM_VALUE: |
4810 | 14.7k | case XML_ROLE_ATTRIBUTE_NOTATION_VALUE: |
4811 | 14.7k | if (dtd->keepProcessing && parser->m_attlistDeclHandler) { |
4812 | 0 | const XML_Char *prefix; |
4813 | 0 | if (parser->m_declAttributeType) { |
4814 | 0 | prefix = enumValueSep; |
4815 | 0 | } else { |
4816 | 0 | prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE ? notationPrefix |
4817 | 0 | : enumValueStart); |
4818 | 0 | } |
4819 | 0 | if (! poolAppendString(&parser->m_tempPool, prefix)) |
4820 | 0 | return XML_ERROR_NO_MEMORY; |
4821 | 0 | if (! poolAppend(&parser->m_tempPool, enc, s, next)) |
4822 | 0 | return XML_ERROR_NO_MEMORY; |
4823 | 0 | parser->m_declAttributeType = parser->m_tempPool.start; |
4824 | 0 | handleDefault = XML_FALSE; |
4825 | 0 | } |
4826 | 14.7k | break; |
4827 | 14.7k | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: |
4828 | 13.2k | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: |
4829 | 13.2k | if (dtd->keepProcessing) { |
4830 | 13.2k | if (! defineAttribute(parser->m_declElementType, |
4831 | 13.2k | parser->m_declAttributeId, |
4832 | 13.2k | parser->m_declAttributeIsCdata, |
4833 | 13.2k | parser->m_declAttributeIsId, 0, parser)) |
4834 | 0 | return XML_ERROR_NO_MEMORY; |
4835 | 13.2k | if (parser->m_attlistDeclHandler && parser->m_declAttributeType) { |
4836 | 0 | if (*parser->m_declAttributeType == XML_T(ASCII_LPAREN) |
4837 | 0 | || (*parser->m_declAttributeType == XML_T(ASCII_N) |
4838 | 0 | && parser->m_declAttributeType[1] == XML_T(ASCII_O))) { |
4839 | | /* Enumerated or Notation type */ |
4840 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T(ASCII_RPAREN)) |
4841 | 0 | || ! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
4842 | 0 | return XML_ERROR_NO_MEMORY; |
4843 | 0 | parser->m_declAttributeType = parser->m_tempPool.start; |
4844 | 0 | poolFinish(&parser->m_tempPool); |
4845 | 0 | } |
4846 | 0 | *eventEndPP = s; |
4847 | 0 | parser->m_attlistDeclHandler( |
4848 | 0 | parser->m_handlerArg, parser->m_declElementType->name, |
4849 | 0 | parser->m_declAttributeId->name, parser->m_declAttributeType, 0, |
4850 | 0 | role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE); |
4851 | 0 | poolClear(&parser->m_tempPool); |
4852 | 0 | handleDefault = XML_FALSE; |
4853 | 0 | } |
4854 | 13.2k | } |
4855 | 13.2k | break; |
4856 | 13.2k | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: |
4857 | 11.0k | case XML_ROLE_FIXED_ATTRIBUTE_VALUE: |
4858 | 11.0k | if (dtd->keepProcessing) { |
4859 | 11.0k | const XML_Char *attVal; |
4860 | 11.0k | enum XML_Error result = storeAttributeValue( |
4861 | 11.0k | parser, enc, parser->m_declAttributeIsCdata, |
4862 | 11.0k | s + enc->minBytesPerChar, next - enc->minBytesPerChar, &dtd->pool, |
4863 | 11.0k | XML_ACCOUNT_NONE); |
4864 | 11.0k | if (result) |
4865 | 49 | return result; |
4866 | 11.0k | attVal = poolStart(&dtd->pool); |
4867 | 11.0k | poolFinish(&dtd->pool); |
4868 | | /* ID attributes aren't allowed to have a default */ |
4869 | 11.0k | if (! defineAttribute( |
4870 | 11.0k | parser->m_declElementType, parser->m_declAttributeId, |
4871 | 11.0k | parser->m_declAttributeIsCdata, XML_FALSE, attVal, parser)) |
4872 | 0 | return XML_ERROR_NO_MEMORY; |
4873 | 11.0k | if (parser->m_attlistDeclHandler && parser->m_declAttributeType) { |
4874 | 0 | if (*parser->m_declAttributeType == XML_T(ASCII_LPAREN) |
4875 | 0 | || (*parser->m_declAttributeType == XML_T(ASCII_N) |
4876 | 0 | && parser->m_declAttributeType[1] == XML_T(ASCII_O))) { |
4877 | | /* Enumerated or Notation type */ |
4878 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T(ASCII_RPAREN)) |
4879 | 0 | || ! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
4880 | 0 | return XML_ERROR_NO_MEMORY; |
4881 | 0 | parser->m_declAttributeType = parser->m_tempPool.start; |
4882 | 0 | poolFinish(&parser->m_tempPool); |
4883 | 0 | } |
4884 | 0 | *eventEndPP = s; |
4885 | 0 | parser->m_attlistDeclHandler( |
4886 | 0 | parser->m_handlerArg, parser->m_declElementType->name, |
4887 | 0 | parser->m_declAttributeId->name, parser->m_declAttributeType, |
4888 | 0 | attVal, role == XML_ROLE_FIXED_ATTRIBUTE_VALUE); |
4889 | 0 | poolClear(&parser->m_tempPool); |
4890 | 0 | handleDefault = XML_FALSE; |
4891 | 0 | } |
4892 | 11.0k | } |
4893 | 11.0k | break; |
4894 | 11.0k | case XML_ROLE_ENTITY_VALUE: |
4895 | 5.19k | if (dtd->keepProcessing) { |
4896 | 5.19k | enum XML_Error result |
4897 | 5.19k | = storeEntityValue(parser, enc, s + enc->minBytesPerChar, |
4898 | 5.19k | next - enc->minBytesPerChar, XML_ACCOUNT_NONE); |
4899 | 5.19k | if (parser->m_declEntity) { |
4900 | 3.18k | parser->m_declEntity->textPtr = poolStart(&dtd->entityValuePool); |
4901 | 3.18k | parser->m_declEntity->textLen |
4902 | 3.18k | = (int)(poolLength(&dtd->entityValuePool)); |
4903 | 3.18k | poolFinish(&dtd->entityValuePool); |
4904 | 3.18k | if (parser->m_entityDeclHandler) { |
4905 | 0 | *eventEndPP = s; |
4906 | 0 | parser->m_entityDeclHandler( |
4907 | 0 | parser->m_handlerArg, parser->m_declEntity->name, |
4908 | 0 | parser->m_declEntity->is_param, parser->m_declEntity->textPtr, |
4909 | 0 | parser->m_declEntity->textLen, parser->m_curBase, 0, 0, 0); |
4910 | 0 | handleDefault = XML_FALSE; |
4911 | 0 | } |
4912 | 3.18k | } else |
4913 | 2.01k | poolDiscard(&dtd->entityValuePool); |
4914 | 5.19k | if (result != XML_ERROR_NONE) |
4915 | 55 | return result; |
4916 | 5.19k | } |
4917 | 5.13k | break; |
4918 | 5.13k | case XML_ROLE_DOCTYPE_SYSTEM_ID: |
4919 | | #ifdef XML_DTD |
4920 | | parser->m_useForeignDTD = XML_FALSE; |
4921 | | #endif /* XML_DTD */ |
4922 | 267 | dtd->hasParamEntityRefs = XML_TRUE; |
4923 | 267 | if (parser->m_startDoctypeDeclHandler) { |
4924 | 0 | parser->m_doctypeSysid = poolStoreString(&parser->m_tempPool, enc, |
4925 | 0 | s + enc->minBytesPerChar, |
4926 | 0 | next - enc->minBytesPerChar); |
4927 | 0 | if (parser->m_doctypeSysid == NULL) |
4928 | 0 | return XML_ERROR_NO_MEMORY; |
4929 | 0 | poolFinish(&parser->m_tempPool); |
4930 | 0 | handleDefault = XML_FALSE; |
4931 | 0 | } |
4932 | | #ifdef XML_DTD |
4933 | | else |
4934 | | /* use externalSubsetName to make parser->m_doctypeSysid non-NULL |
4935 | | for the case where no parser->m_startDoctypeDeclHandler is set */ |
4936 | | parser->m_doctypeSysid = externalSubsetName; |
4937 | | #endif /* XML_DTD */ |
4938 | 267 | if (! dtd->standalone |
4939 | | #ifdef XML_DTD |
4940 | | && ! parser->m_paramEntityParsing |
4941 | | #endif /* XML_DTD */ |
4942 | 267 | && parser->m_notStandaloneHandler |
4943 | 267 | && ! parser->m_notStandaloneHandler(parser->m_handlerArg)) |
4944 | 0 | return XML_ERROR_NOT_STANDALONE; |
4945 | 267 | #ifndef XML_DTD |
4946 | 267 | break; |
4947 | | #else /* XML_DTD */ |
4948 | | if (! parser->m_declEntity) { |
4949 | | parser->m_declEntity = (ENTITY *)lookup( |
4950 | | parser, &dtd->paramEntities, externalSubsetName, sizeof(ENTITY)); |
4951 | | if (! parser->m_declEntity) |
4952 | | return XML_ERROR_NO_MEMORY; |
4953 | | parser->m_declEntity->publicId = NULL; |
4954 | | } |
4955 | | #endif /* XML_DTD */ |
4956 | | /* fall through */ |
4957 | 4.07k | case XML_ROLE_ENTITY_SYSTEM_ID: |
4958 | 4.07k | if (dtd->keepProcessing && parser->m_declEntity) { |
4959 | 631 | parser->m_declEntity->systemId |
4960 | 631 | = poolStoreString(&dtd->pool, enc, s + enc->minBytesPerChar, |
4961 | 631 | next - enc->minBytesPerChar); |
4962 | 631 | if (! parser->m_declEntity->systemId) |
4963 | 0 | return XML_ERROR_NO_MEMORY; |
4964 | 631 | parser->m_declEntity->base = parser->m_curBase; |
4965 | 631 | poolFinish(&dtd->pool); |
4966 | | /* Don't suppress the default handler if we fell through from |
4967 | | * the XML_ROLE_DOCTYPE_SYSTEM_ID case. |
4968 | | */ |
4969 | 631 | if (parser->m_entityDeclHandler && role == XML_ROLE_ENTITY_SYSTEM_ID) |
4970 | 0 | handleDefault = XML_FALSE; |
4971 | 631 | } |
4972 | 4.07k | break; |
4973 | 4.07k | case XML_ROLE_ENTITY_COMPLETE: |
4974 | 2.53k | if (dtd->keepProcessing && parser->m_declEntity |
4975 | 2.53k | && parser->m_entityDeclHandler) { |
4976 | 0 | *eventEndPP = s; |
4977 | 0 | parser->m_entityDeclHandler( |
4978 | 0 | parser->m_handlerArg, parser->m_declEntity->name, |
4979 | 0 | parser->m_declEntity->is_param, 0, 0, parser->m_declEntity->base, |
4980 | 0 | parser->m_declEntity->systemId, parser->m_declEntity->publicId, 0); |
4981 | 0 | handleDefault = XML_FALSE; |
4982 | 0 | } |
4983 | 2.53k | break; |
4984 | 1.51k | case XML_ROLE_ENTITY_NOTATION_NAME: |
4985 | 1.51k | if (dtd->keepProcessing && parser->m_declEntity) { |
4986 | 244 | parser->m_declEntity->notation |
4987 | 244 | = poolStoreString(&dtd->pool, enc, s, next); |
4988 | 244 | if (! parser->m_declEntity->notation) |
4989 | 0 | return XML_ERROR_NO_MEMORY; |
4990 | 244 | poolFinish(&dtd->pool); |
4991 | 244 | if (parser->m_unparsedEntityDeclHandler) { |
4992 | 0 | *eventEndPP = s; |
4993 | 0 | parser->m_unparsedEntityDeclHandler( |
4994 | 0 | parser->m_handlerArg, parser->m_declEntity->name, |
4995 | 0 | parser->m_declEntity->base, parser->m_declEntity->systemId, |
4996 | 0 | parser->m_declEntity->publicId, parser->m_declEntity->notation); |
4997 | 0 | handleDefault = XML_FALSE; |
4998 | 244 | } else if (parser->m_entityDeclHandler) { |
4999 | 0 | *eventEndPP = s; |
5000 | 0 | parser->m_entityDeclHandler( |
5001 | 0 | parser->m_handlerArg, parser->m_declEntity->name, 0, 0, 0, |
5002 | 0 | parser->m_declEntity->base, parser->m_declEntity->systemId, |
5003 | 0 | parser->m_declEntity->publicId, parser->m_declEntity->notation); |
5004 | 0 | handleDefault = XML_FALSE; |
5005 | 0 | } |
5006 | 244 | } |
5007 | 1.51k | break; |
5008 | 8.46k | case XML_ROLE_GENERAL_ENTITY_NAME: { |
5009 | 8.46k | if (XmlPredefinedEntityName(enc, s, next)) { |
5010 | 82 | parser->m_declEntity = NULL; |
5011 | 82 | break; |
5012 | 82 | } |
5013 | 8.38k | if (dtd->keepProcessing) { |
5014 | 8.38k | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); |
5015 | 8.38k | if (! name) |
5016 | 0 | return XML_ERROR_NO_MEMORY; |
5017 | 8.38k | parser->m_declEntity = (ENTITY *)lookup(parser, &dtd->generalEntities, |
5018 | 8.38k | name, sizeof(ENTITY)); |
5019 | 8.38k | if (! parser->m_declEntity) |
5020 | 0 | return XML_ERROR_NO_MEMORY; |
5021 | 8.38k | if (parser->m_declEntity->name != name) { |
5022 | 4.45k | poolDiscard(&dtd->pool); |
5023 | 4.45k | parser->m_declEntity = NULL; |
5024 | 4.45k | } else { |
5025 | 3.93k | poolFinish(&dtd->pool); |
5026 | 3.93k | parser->m_declEntity->publicId = NULL; |
5027 | 3.93k | parser->m_declEntity->is_param = XML_FALSE; |
5028 | | /* if we have a parent parser or are reading an internal parameter |
5029 | | entity, then the entity declaration is not considered "internal" |
5030 | | */ |
5031 | 3.93k | parser->m_declEntity->is_internal |
5032 | 3.93k | = ! (parser->m_parentParser || parser->m_openInternalEntities); |
5033 | 3.93k | if (parser->m_entityDeclHandler) |
5034 | 0 | handleDefault = XML_FALSE; |
5035 | 3.93k | } |
5036 | 8.38k | } else { |
5037 | 0 | poolDiscard(&dtd->pool); |
5038 | 0 | parser->m_declEntity = NULL; |
5039 | 0 | } |
5040 | 8.38k | } break; |
5041 | 8.38k | case XML_ROLE_PARAM_ENTITY_NAME: |
5042 | | #ifdef XML_DTD |
5043 | | if (dtd->keepProcessing) { |
5044 | | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); |
5045 | | if (! name) |
5046 | | return XML_ERROR_NO_MEMORY; |
5047 | | parser->m_declEntity = (ENTITY *)lookup(parser, &dtd->paramEntities, |
5048 | | name, sizeof(ENTITY)); |
5049 | | if (! parser->m_declEntity) |
5050 | | return XML_ERROR_NO_MEMORY; |
5051 | | if (parser->m_declEntity->name != name) { |
5052 | | poolDiscard(&dtd->pool); |
5053 | | parser->m_declEntity = NULL; |
5054 | | } else { |
5055 | | poolFinish(&dtd->pool); |
5056 | | parser->m_declEntity->publicId = NULL; |
5057 | | parser->m_declEntity->is_param = XML_TRUE; |
5058 | | /* if we have a parent parser or are reading an internal parameter |
5059 | | entity, then the entity declaration is not considered "internal" |
5060 | | */ |
5061 | | parser->m_declEntity->is_internal |
5062 | | = ! (parser->m_parentParser || parser->m_openInternalEntities); |
5063 | | if (parser->m_entityDeclHandler) |
5064 | | handleDefault = XML_FALSE; |
5065 | | } |
5066 | | } else { |
5067 | | poolDiscard(&dtd->pool); |
5068 | | parser->m_declEntity = NULL; |
5069 | | } |
5070 | | #else /* not XML_DTD */ |
5071 | 981 | parser->m_declEntity = NULL; |
5072 | 981 | #endif /* XML_DTD */ |
5073 | 981 | break; |
5074 | 1.23k | case XML_ROLE_NOTATION_NAME: |
5075 | 1.23k | parser->m_declNotationPublicId = NULL; |
5076 | 1.23k | parser->m_declNotationName = NULL; |
5077 | 1.23k | if (parser->m_notationDeclHandler) { |
5078 | 0 | parser->m_declNotationName |
5079 | 0 | = poolStoreString(&parser->m_tempPool, enc, s, next); |
5080 | 0 | if (! parser->m_declNotationName) |
5081 | 0 | return XML_ERROR_NO_MEMORY; |
5082 | 0 | poolFinish(&parser->m_tempPool); |
5083 | 0 | handleDefault = XML_FALSE; |
5084 | 0 | } |
5085 | 1.23k | break; |
5086 | 1.23k | case XML_ROLE_NOTATION_PUBLIC_ID: |
5087 | 1.14k | if (! XmlIsPublicId(enc, s, next, eventPP)) |
5088 | 2 | return XML_ERROR_PUBLICID; |
5089 | 1.14k | if (parser |
5090 | 1.14k | ->m_declNotationName) { /* means m_notationDeclHandler != NULL */ |
5091 | 0 | XML_Char *tem = poolStoreString(&parser->m_tempPool, enc, |
5092 | 0 | s + enc->minBytesPerChar, |
5093 | 0 | next - enc->minBytesPerChar); |
5094 | 0 | if (! tem) |
5095 | 0 | return XML_ERROR_NO_MEMORY; |
5096 | 0 | normalizePublicId(tem); |
5097 | 0 | parser->m_declNotationPublicId = tem; |
5098 | 0 | poolFinish(&parser->m_tempPool); |
5099 | 0 | handleDefault = XML_FALSE; |
5100 | 0 | } |
5101 | 1.14k | break; |
5102 | 1.14k | case XML_ROLE_NOTATION_SYSTEM_ID: |
5103 | 842 | if (parser->m_declNotationName && parser->m_notationDeclHandler) { |
5104 | 0 | const XML_Char *systemId = poolStoreString(&parser->m_tempPool, enc, |
5105 | 0 | s + enc->minBytesPerChar, |
5106 | 0 | next - enc->minBytesPerChar); |
5107 | 0 | if (! systemId) |
5108 | 0 | return XML_ERROR_NO_MEMORY; |
5109 | 0 | *eventEndPP = s; |
5110 | 0 | parser->m_notationDeclHandler( |
5111 | 0 | parser->m_handlerArg, parser->m_declNotationName, parser->m_curBase, |
5112 | 0 | systemId, parser->m_declNotationPublicId); |
5113 | 0 | handleDefault = XML_FALSE; |
5114 | 0 | } |
5115 | 842 | poolClear(&parser->m_tempPool); |
5116 | 842 | break; |
5117 | 359 | case XML_ROLE_NOTATION_NO_SYSTEM_ID: |
5118 | 359 | if (parser->m_declNotationPublicId && parser->m_notationDeclHandler) { |
5119 | 0 | *eventEndPP = s; |
5120 | 0 | parser->m_notationDeclHandler( |
5121 | 0 | parser->m_handlerArg, parser->m_declNotationName, parser->m_curBase, |
5122 | 0 | 0, parser->m_declNotationPublicId); |
5123 | 0 | handleDefault = XML_FALSE; |
5124 | 0 | } |
5125 | 359 | poolClear(&parser->m_tempPool); |
5126 | 359 | break; |
5127 | 644 | case XML_ROLE_ERROR: |
5128 | 644 | switch (tok) { |
5129 | 3 | case XML_TOK_PARAM_ENTITY_REF: |
5130 | | /* PE references in internal subset are |
5131 | | not allowed within declarations. */ |
5132 | 3 | return XML_ERROR_PARAM_ENTITY_REF; |
5133 | 0 | case XML_TOK_XML_DECL: |
5134 | 0 | return XML_ERROR_MISPLACED_XML_PI; |
5135 | 641 | default: |
5136 | 641 | return XML_ERROR_SYNTAX; |
5137 | 644 | } |
5138 | | #ifdef XML_DTD |
5139 | | case XML_ROLE_IGNORE_SECT: { |
5140 | | enum XML_Error result; |
5141 | | if (parser->m_defaultHandler) |
5142 | | reportDefault(parser, enc, s, next); |
5143 | | handleDefault = XML_FALSE; |
5144 | | result = doIgnoreSection(parser, enc, &next, end, nextPtr, haveMore); |
5145 | | if (result != XML_ERROR_NONE) |
5146 | | return result; |
5147 | | else if (! next) { |
5148 | | parser->m_processor = ignoreSectionProcessor; |
5149 | | return result; |
5150 | | } |
5151 | | } break; |
5152 | | #endif /* XML_DTD */ |
5153 | 531k | case XML_ROLE_GROUP_OPEN: |
5154 | 531k | if (parser->m_prologState.level >= parser->m_groupSize) { |
5155 | 571 | if (parser->m_groupSize) { |
5156 | 196 | { |
5157 | | /* Detect and prevent integer overflow */ |
5158 | 196 | if (parser->m_groupSize > (unsigned int)(-1) / 2u) { |
5159 | 0 | return XML_ERROR_NO_MEMORY; |
5160 | 0 | } |
5161 | | |
5162 | 196 | char *const new_connector = (char *)REALLOC( |
5163 | 196 | parser, parser->m_groupConnector, parser->m_groupSize *= 2); |
5164 | 196 | if (new_connector == NULL) { |
5165 | 0 | parser->m_groupSize /= 2; |
5166 | 0 | return XML_ERROR_NO_MEMORY; |
5167 | 0 | } |
5168 | 196 | parser->m_groupConnector = new_connector; |
5169 | 196 | } |
5170 | | |
5171 | 196 | if (dtd->scaffIndex) { |
5172 | | /* Detect and prevent integer overflow. |
5173 | | * The preprocessor guard addresses the "always false" warning |
5174 | | * from -Wtype-limits on platforms where |
5175 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
5176 | | #if UINT_MAX >= SIZE_MAX |
5177 | | if (parser->m_groupSize > (size_t)(-1) / sizeof(int)) { |
5178 | | return XML_ERROR_NO_MEMORY; |
5179 | | } |
5180 | | #endif |
5181 | |
|
5182 | 0 | int *const new_scaff_index = (int *)REALLOC( |
5183 | 0 | parser, dtd->scaffIndex, parser->m_groupSize * sizeof(int)); |
5184 | 0 | if (new_scaff_index == NULL) |
5185 | 0 | return XML_ERROR_NO_MEMORY; |
5186 | 0 | dtd->scaffIndex = new_scaff_index; |
5187 | 0 | } |
5188 | 375 | } else { |
5189 | 375 | parser->m_groupConnector |
5190 | 375 | = (char *)MALLOC(parser, parser->m_groupSize = 32); |
5191 | 375 | if (! parser->m_groupConnector) { |
5192 | 0 | parser->m_groupSize = 0; |
5193 | 0 | return XML_ERROR_NO_MEMORY; |
5194 | 0 | } |
5195 | 375 | } |
5196 | 571 | } |
5197 | 531k | parser->m_groupConnector[parser->m_prologState.level] = 0; |
5198 | 531k | if (dtd->in_eldecl) { |
5199 | 0 | int myindex = nextScaffoldPart(parser); |
5200 | 0 | if (myindex < 0) |
5201 | 0 | return XML_ERROR_NO_MEMORY; |
5202 | 0 | assert(dtd->scaffIndex != NULL); |
5203 | 0 | dtd->scaffIndex[dtd->scaffLevel] = myindex; |
5204 | 0 | dtd->scaffLevel++; |
5205 | 0 | dtd->scaffold[myindex].type = XML_CTYPE_SEQ; |
5206 | 0 | if (parser->m_elementDeclHandler) |
5207 | 0 | handleDefault = XML_FALSE; |
5208 | 0 | } |
5209 | 531k | break; |
5210 | 531k | case XML_ROLE_GROUP_SEQUENCE: |
5211 | 1.50k | if (parser->m_groupConnector[parser->m_prologState.level] == ASCII_PIPE) |
5212 | 4 | return XML_ERROR_SYNTAX; |
5213 | 1.49k | parser->m_groupConnector[parser->m_prologState.level] = ASCII_COMMA; |
5214 | 1.49k | if (dtd->in_eldecl && parser->m_elementDeclHandler) |
5215 | 0 | handleDefault = XML_FALSE; |
5216 | 1.49k | break; |
5217 | 3.41k | case XML_ROLE_GROUP_CHOICE: |
5218 | 3.41k | if (parser->m_groupConnector[parser->m_prologState.level] == ASCII_COMMA) |
5219 | 1 | return XML_ERROR_SYNTAX; |
5220 | 3.41k | if (dtd->in_eldecl |
5221 | 3.41k | && ! parser->m_groupConnector[parser->m_prologState.level] |
5222 | 3.41k | && (dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type |
5223 | 0 | != XML_CTYPE_MIXED)) { |
5224 | 0 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type |
5225 | 0 | = XML_CTYPE_CHOICE; |
5226 | 0 | if (parser->m_elementDeclHandler) |
5227 | 0 | handleDefault = XML_FALSE; |
5228 | 0 | } |
5229 | 3.41k | parser->m_groupConnector[parser->m_prologState.level] = ASCII_PIPE; |
5230 | 3.41k | break; |
5231 | 1.52k | case XML_ROLE_PARAM_ENTITY_REF: |
5232 | | #ifdef XML_DTD |
5233 | | case XML_ROLE_INNER_PARAM_ENTITY_REF: |
5234 | | dtd->hasParamEntityRefs = XML_TRUE; |
5235 | | if (! parser->m_paramEntityParsing) |
5236 | | dtd->keepProcessing = dtd->standalone; |
5237 | | else { |
5238 | | const XML_Char *name; |
5239 | | ENTITY *entity; |
5240 | | name = poolStoreString(&dtd->pool, enc, s + enc->minBytesPerChar, |
5241 | | next - enc->minBytesPerChar); |
5242 | | if (! name) |
5243 | | return XML_ERROR_NO_MEMORY; |
5244 | | entity = (ENTITY *)lookup(parser, &dtd->paramEntities, name, 0); |
5245 | | poolDiscard(&dtd->pool); |
5246 | | /* first, determine if a check for an existing declaration is needed; |
5247 | | if yes, check that the entity exists, and that it is internal, |
5248 | | otherwise call the skipped entity handler |
5249 | | */ |
5250 | | if (parser->m_prologState.documentEntity |
5251 | | && (dtd->standalone ? ! parser->m_openInternalEntities |
5252 | | : ! dtd->hasParamEntityRefs)) { |
5253 | | if (! entity) |
5254 | | return XML_ERROR_UNDEFINED_ENTITY; |
5255 | | else if (! entity->is_internal) { |
5256 | | /* It's hard to exhaustively search the code to be sure, |
5257 | | * but there doesn't seem to be a way of executing the |
5258 | | * following line. There are two cases: |
5259 | | * |
5260 | | * If 'standalone' is false, the DTD must have no |
5261 | | * parameter entities or we wouldn't have passed the outer |
5262 | | * 'if' statement. That measn the only entity in the hash |
5263 | | * table is the external subset name "#" which cannot be |
5264 | | * given as a parameter entity name in XML syntax, so the |
5265 | | * lookup must have returned NULL and we don't even reach |
5266 | | * the test for an internal entity. |
5267 | | * |
5268 | | * If 'standalone' is true, it does not seem to be |
5269 | | * possible to create entities taking this code path that |
5270 | | * are not internal entities, so fail the test above. |
5271 | | * |
5272 | | * Because this analysis is very uncertain, the code is |
5273 | | * being left in place and merely removed from the |
5274 | | * coverage test statistics. |
5275 | | */ |
5276 | | return XML_ERROR_ENTITY_DECLARED_IN_PE; /* LCOV_EXCL_LINE */ |
5277 | | } |
5278 | | } else if (! entity) { |
5279 | | dtd->keepProcessing = dtd->standalone; |
5280 | | /* cannot report skipped entities in declarations */ |
5281 | | if ((role == XML_ROLE_PARAM_ENTITY_REF) |
5282 | | && parser->m_skippedEntityHandler) { |
5283 | | parser->m_skippedEntityHandler(parser->m_handlerArg, name, 1); |
5284 | | handleDefault = XML_FALSE; |
5285 | | } |
5286 | | break; |
5287 | | } |
5288 | | if (entity->open) |
5289 | | return XML_ERROR_RECURSIVE_ENTITY_REF; |
5290 | | if (entity->textPtr) { |
5291 | | enum XML_Error result; |
5292 | | XML_Bool betweenDecl |
5293 | | = (role == XML_ROLE_PARAM_ENTITY_REF ? XML_TRUE : XML_FALSE); |
5294 | | result = processInternalEntity(parser, entity, betweenDecl); |
5295 | | if (result != XML_ERROR_NONE) |
5296 | | return result; |
5297 | | handleDefault = XML_FALSE; |
5298 | | break; |
5299 | | } |
5300 | | if (parser->m_externalEntityRefHandler) { |
5301 | | dtd->paramEntityRead = XML_FALSE; |
5302 | | entity->open = XML_TRUE; |
5303 | | entityTrackingOnOpen(parser, entity, __LINE__); |
5304 | | if (! parser->m_externalEntityRefHandler( |
5305 | | parser->m_externalEntityRefHandlerArg, 0, entity->base, |
5306 | | entity->systemId, entity->publicId)) { |
5307 | | entityTrackingOnClose(parser, entity, __LINE__); |
5308 | | entity->open = XML_FALSE; |
5309 | | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
5310 | | } |
5311 | | entityTrackingOnClose(parser, entity, __LINE__); |
5312 | | entity->open = XML_FALSE; |
5313 | | handleDefault = XML_FALSE; |
5314 | | if (! dtd->paramEntityRead) { |
5315 | | dtd->keepProcessing = dtd->standalone; |
5316 | | break; |
5317 | | } |
5318 | | } else { |
5319 | | dtd->keepProcessing = dtd->standalone; |
5320 | | break; |
5321 | | } |
5322 | | } |
5323 | | #endif /* XML_DTD */ |
5324 | 1.52k | if (! dtd->standalone && parser->m_notStandaloneHandler |
5325 | 1.52k | && ! parser->m_notStandaloneHandler(parser->m_handlerArg)) |
5326 | 0 | return XML_ERROR_NOT_STANDALONE; |
5327 | 1.52k | break; |
5328 | | |
5329 | | /* Element declaration stuff */ |
5330 | | |
5331 | 3.66k | case XML_ROLE_ELEMENT_NAME: |
5332 | 3.66k | if (parser->m_elementDeclHandler) { |
5333 | 0 | parser->m_declElementType = getElementType(parser, enc, s, next); |
5334 | 0 | if (! parser->m_declElementType) |
5335 | 0 | return XML_ERROR_NO_MEMORY; |
5336 | 0 | dtd->scaffLevel = 0; |
5337 | 0 | dtd->scaffCount = 0; |
5338 | 0 | dtd->in_eldecl = XML_TRUE; |
5339 | 0 | handleDefault = XML_FALSE; |
5340 | 0 | } |
5341 | 3.66k | break; |
5342 | | |
5343 | 3.66k | case XML_ROLE_CONTENT_ANY: |
5344 | 453 | case XML_ROLE_CONTENT_EMPTY: |
5345 | 453 | if (dtd->in_eldecl) { |
5346 | 0 | if (parser->m_elementDeclHandler) { |
5347 | 0 | XML_Content *content |
5348 | 0 | = (XML_Content *)MALLOC(parser, sizeof(XML_Content)); |
5349 | 0 | if (! content) |
5350 | 0 | return XML_ERROR_NO_MEMORY; |
5351 | 0 | content->quant = XML_CQUANT_NONE; |
5352 | 0 | content->name = NULL; |
5353 | 0 | content->numchildren = 0; |
5354 | 0 | content->children = NULL; |
5355 | 0 | content->type = ((role == XML_ROLE_CONTENT_ANY) ? XML_CTYPE_ANY |
5356 | 0 | : XML_CTYPE_EMPTY); |
5357 | 0 | *eventEndPP = s; |
5358 | 0 | parser->m_elementDeclHandler( |
5359 | 0 | parser->m_handlerArg, parser->m_declElementType->name, content); |
5360 | 0 | handleDefault = XML_FALSE; |
5361 | 0 | } |
5362 | 0 | dtd->in_eldecl = XML_FALSE; |
5363 | 0 | } |
5364 | 453 | break; |
5365 | | |
5366 | 810 | case XML_ROLE_CONTENT_PCDATA: |
5367 | 810 | if (dtd->in_eldecl) { |
5368 | 0 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type |
5369 | 0 | = XML_CTYPE_MIXED; |
5370 | 0 | if (parser->m_elementDeclHandler) |
5371 | 0 | handleDefault = XML_FALSE; |
5372 | 0 | } |
5373 | 810 | break; |
5374 | | |
5375 | 6.87k | case XML_ROLE_CONTENT_ELEMENT: |
5376 | 6.87k | quant = XML_CQUANT_NONE; |
5377 | 6.87k | goto elementContent; |
5378 | 693 | case XML_ROLE_CONTENT_ELEMENT_OPT: |
5379 | 693 | quant = XML_CQUANT_OPT; |
5380 | 693 | goto elementContent; |
5381 | 382 | case XML_ROLE_CONTENT_ELEMENT_REP: |
5382 | 382 | quant = XML_CQUANT_REP; |
5383 | 382 | goto elementContent; |
5384 | 175 | case XML_ROLE_CONTENT_ELEMENT_PLUS: |
5385 | 175 | quant = XML_CQUANT_PLUS; |
5386 | 8.12k | elementContent: |
5387 | 8.12k | if (dtd->in_eldecl) { |
5388 | 0 | ELEMENT_TYPE *el; |
5389 | 0 | const XML_Char *name; |
5390 | 0 | size_t nameLen; |
5391 | 0 | const char *nxt |
5392 | 0 | = (quant == XML_CQUANT_NONE ? next : next - enc->minBytesPerChar); |
5393 | 0 | int myindex = nextScaffoldPart(parser); |
5394 | 0 | if (myindex < 0) |
5395 | 0 | return XML_ERROR_NO_MEMORY; |
5396 | 0 | dtd->scaffold[myindex].type = XML_CTYPE_NAME; |
5397 | 0 | dtd->scaffold[myindex].quant = quant; |
5398 | 0 | el = getElementType(parser, enc, s, nxt); |
5399 | 0 | if (! el) |
5400 | 0 | return XML_ERROR_NO_MEMORY; |
5401 | 0 | name = el->name; |
5402 | 0 | dtd->scaffold[myindex].name = name; |
5403 | 0 | nameLen = 0; |
5404 | 0 | for (; name[nameLen++];) |
5405 | 0 | ; |
5406 | | |
5407 | | /* Detect and prevent integer overflow */ |
5408 | 0 | if (nameLen > UINT_MAX - dtd->contentStringLen) { |
5409 | 0 | return XML_ERROR_NO_MEMORY; |
5410 | 0 | } |
5411 | | |
5412 | 0 | dtd->contentStringLen += (unsigned)nameLen; |
5413 | 0 | if (parser->m_elementDeclHandler) |
5414 | 0 | handleDefault = XML_FALSE; |
5415 | 0 | } |
5416 | 8.12k | break; |
5417 | | |
5418 | 8.12k | case XML_ROLE_GROUP_CLOSE: |
5419 | 2.82k | quant = XML_CQUANT_NONE; |
5420 | 2.82k | goto closeGroup; |
5421 | 1.22k | case XML_ROLE_GROUP_CLOSE_OPT: |
5422 | 1.22k | quant = XML_CQUANT_OPT; |
5423 | 1.22k | goto closeGroup; |
5424 | 1.36k | case XML_ROLE_GROUP_CLOSE_REP: |
5425 | 1.36k | quant = XML_CQUANT_REP; |
5426 | 1.36k | goto closeGroup; |
5427 | 920 | case XML_ROLE_GROUP_CLOSE_PLUS: |
5428 | 920 | quant = XML_CQUANT_PLUS; |
5429 | 6.34k | closeGroup: |
5430 | 6.34k | if (dtd->in_eldecl) { |
5431 | 0 | if (parser->m_elementDeclHandler) |
5432 | 0 | handleDefault = XML_FALSE; |
5433 | 0 | dtd->scaffLevel--; |
5434 | 0 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel]].quant = quant; |
5435 | 0 | if (dtd->scaffLevel == 0) { |
5436 | 0 | if (! handleDefault) { |
5437 | 0 | XML_Content *model = build_model(parser); |
5438 | 0 | if (! model) |
5439 | 0 | return XML_ERROR_NO_MEMORY; |
5440 | 0 | *eventEndPP = s; |
5441 | 0 | parser->m_elementDeclHandler( |
5442 | 0 | parser->m_handlerArg, parser->m_declElementType->name, model); |
5443 | 0 | } |
5444 | 0 | dtd->in_eldecl = XML_FALSE; |
5445 | 0 | dtd->contentStringLen = 0; |
5446 | 0 | } |
5447 | 0 | } |
5448 | 6.34k | break; |
5449 | | /* End element declaration stuff */ |
5450 | | |
5451 | 6.34k | case XML_ROLE_PI: |
5452 | 5.01k | if (! reportProcessingInstruction(parser, enc, s, next)) |
5453 | 0 | return XML_ERROR_NO_MEMORY; |
5454 | 5.01k | handleDefault = XML_FALSE; |
5455 | 5.01k | break; |
5456 | 508 | case XML_ROLE_COMMENT: |
5457 | 508 | if (! reportComment(parser, enc, s, next)) |
5458 | 0 | return XML_ERROR_NO_MEMORY; |
5459 | 508 | handleDefault = XML_FALSE; |
5460 | 508 | break; |
5461 | 3.45k | case XML_ROLE_NONE: |
5462 | 3.45k | switch (tok) { |
5463 | 252 | case XML_TOK_BOM: |
5464 | 252 | handleDefault = XML_FALSE; |
5465 | 252 | break; |
5466 | 3.45k | } |
5467 | 3.45k | break; |
5468 | 9.50k | case XML_ROLE_DOCTYPE_NONE: |
5469 | 9.50k | if (parser->m_startDoctypeDeclHandler) |
5470 | 0 | handleDefault = XML_FALSE; |
5471 | 9.50k | break; |
5472 | 53.8k | case XML_ROLE_ENTITY_NONE: |
5473 | 53.8k | if (dtd->keepProcessing && parser->m_entityDeclHandler) |
5474 | 0 | handleDefault = XML_FALSE; |
5475 | 53.8k | break; |
5476 | 7.90k | case XML_ROLE_NOTATION_NONE: |
5477 | 7.90k | if (parser->m_notationDeclHandler) |
5478 | 0 | handleDefault = XML_FALSE; |
5479 | 7.90k | break; |
5480 | 133k | case XML_ROLE_ATTLIST_NONE: |
5481 | 133k | if (dtd->keepProcessing && parser->m_attlistDeclHandler) |
5482 | 0 | handleDefault = XML_FALSE; |
5483 | 133k | break; |
5484 | 22.5k | case XML_ROLE_ELEMENT_NONE: |
5485 | 22.5k | if (parser->m_elementDeclHandler) |
5486 | 0 | handleDefault = XML_FALSE; |
5487 | 22.5k | break; |
5488 | 923k | } /* end of big switch */ |
5489 | | |
5490 | 919k | if (handleDefault && parser->m_defaultHandler) |
5491 | 0 | reportDefault(parser, enc, s, next); |
5492 | | |
5493 | 919k | switch (parser->m_parsingStatus.parsing) { |
5494 | 0 | case XML_SUSPENDED: |
5495 | 0 | *nextPtr = next; |
5496 | 0 | return XML_ERROR_NONE; |
5497 | 0 | case XML_FINISHED: |
5498 | 0 | return XML_ERROR_ABORTED; |
5499 | 919k | default: |
5500 | 919k | s = next; |
5501 | 919k | tok = XmlPrologTok(enc, s, end, &next); |
5502 | 919k | } |
5503 | 919k | } |
5504 | | /* not reached */ |
5505 | 9.22k | } |
5506 | | |
5507 | | static enum XML_Error PTRCALL |
5508 | | epilogProcessor(XML_Parser parser, const char *s, const char *end, |
5509 | 133 | const char **nextPtr) { |
5510 | 133 | parser->m_processor = epilogProcessor; |
5511 | 133 | parser->m_eventPtr = s; |
5512 | 888 | for (;;) { |
5513 | 888 | const char *next = NULL; |
5514 | 888 | int tok = XmlPrologTok(parser->m_encoding, s, end, &next); |
5515 | | #ifdef XML_DTD |
5516 | | if (! accountingDiffTolerated(parser, tok, s, next, __LINE__, |
5517 | | XML_ACCOUNT_DIRECT)) { |
5518 | | accountingOnAbort(parser); |
5519 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
5520 | | } |
5521 | | #endif |
5522 | 888 | parser->m_eventEndPtr = next; |
5523 | 888 | switch (tok) { |
5524 | | /* report partial linebreak - it might be the last token */ |
5525 | 1 | case -XML_TOK_PROLOG_S: |
5526 | 1 | if (parser->m_defaultHandler) { |
5527 | 0 | reportDefault(parser, parser->m_encoding, s, next); |
5528 | 0 | if (parser->m_parsingStatus.parsing == XML_FINISHED) |
5529 | 0 | return XML_ERROR_ABORTED; |
5530 | 0 | } |
5531 | 1 | *nextPtr = next; |
5532 | 1 | return XML_ERROR_NONE; |
5533 | 97 | case XML_TOK_NONE: |
5534 | 97 | *nextPtr = s; |
5535 | 97 | return XML_ERROR_NONE; |
5536 | 330 | case XML_TOK_PROLOG_S: |
5537 | 330 | if (parser->m_defaultHandler) |
5538 | 0 | reportDefault(parser, parser->m_encoding, s, next); |
5539 | 330 | break; |
5540 | 352 | case XML_TOK_PI: |
5541 | 352 | if (! reportProcessingInstruction(parser, parser->m_encoding, s, next)) |
5542 | 0 | return XML_ERROR_NO_MEMORY; |
5543 | 352 | break; |
5544 | 352 | case XML_TOK_COMMENT: |
5545 | 73 | if (! reportComment(parser, parser->m_encoding, s, next)) |
5546 | 0 | return XML_ERROR_NO_MEMORY; |
5547 | 73 | break; |
5548 | 73 | case XML_TOK_INVALID: |
5549 | 8 | parser->m_eventPtr = next; |
5550 | 8 | return XML_ERROR_INVALID_TOKEN; |
5551 | 8 | case XML_TOK_PARTIAL: |
5552 | 8 | if (! parser->m_parsingStatus.finalBuffer) { |
5553 | 4 | *nextPtr = s; |
5554 | 4 | return XML_ERROR_NONE; |
5555 | 4 | } |
5556 | 4 | return XML_ERROR_UNCLOSED_TOKEN; |
5557 | 2 | case XML_TOK_PARTIAL_CHAR: |
5558 | 2 | if (! parser->m_parsingStatus.finalBuffer) { |
5559 | 1 | *nextPtr = s; |
5560 | 1 | return XML_ERROR_NONE; |
5561 | 1 | } |
5562 | 1 | return XML_ERROR_PARTIAL_CHAR; |
5563 | 17 | default: |
5564 | 17 | return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; |
5565 | 888 | } |
5566 | 755 | parser->m_eventPtr = s = next; |
5567 | 755 | switch (parser->m_parsingStatus.parsing) { |
5568 | 0 | case XML_SUSPENDED: |
5569 | 0 | *nextPtr = next; |
5570 | 0 | return XML_ERROR_NONE; |
5571 | 0 | case XML_FINISHED: |
5572 | 0 | return XML_ERROR_ABORTED; |
5573 | 755 | default:; |
5574 | 755 | } |
5575 | 755 | } |
5576 | 133 | } |
5577 | | |
5578 | | static enum XML_Error |
5579 | 21.0k | processInternalEntity(XML_Parser parser, ENTITY *entity, XML_Bool betweenDecl) { |
5580 | 21.0k | const char *textStart, *textEnd; |
5581 | 21.0k | const char *next; |
5582 | 21.0k | enum XML_Error result; |
5583 | 21.0k | OPEN_INTERNAL_ENTITY *openEntity; |
5584 | | |
5585 | 21.0k | if (parser->m_freeInternalEntities) { |
5586 | 19.5k | openEntity = parser->m_freeInternalEntities; |
5587 | 19.5k | parser->m_freeInternalEntities = openEntity->next; |
5588 | 19.5k | } else { |
5589 | 1.53k | openEntity |
5590 | 1.53k | = (OPEN_INTERNAL_ENTITY *)MALLOC(parser, sizeof(OPEN_INTERNAL_ENTITY)); |
5591 | 1.53k | if (! openEntity) |
5592 | 0 | return XML_ERROR_NO_MEMORY; |
5593 | 1.53k | } |
5594 | 21.0k | entity->open = XML_TRUE; |
5595 | | #ifdef XML_DTD |
5596 | | entityTrackingOnOpen(parser, entity, __LINE__); |
5597 | | #endif |
5598 | 21.0k | entity->processed = 0; |
5599 | 21.0k | openEntity->next = parser->m_openInternalEntities; |
5600 | 21.0k | parser->m_openInternalEntities = openEntity; |
5601 | 21.0k | openEntity->entity = entity; |
5602 | 21.0k | openEntity->startTagLevel = parser->m_tagLevel; |
5603 | 21.0k | openEntity->betweenDecl = betweenDecl; |
5604 | 21.0k | openEntity->internalEventPtr = NULL; |
5605 | 21.0k | openEntity->internalEventEndPtr = NULL; |
5606 | 21.0k | textStart = (const char *)entity->textPtr; |
5607 | 21.0k | textEnd = (const char *)(entity->textPtr + entity->textLen); |
5608 | | /* Set a safe default value in case 'next' does not get set */ |
5609 | 21.0k | next = textStart; |
5610 | | |
5611 | | #ifdef XML_DTD |
5612 | | if (entity->is_param) { |
5613 | | int tok |
5614 | | = XmlPrologTok(parser->m_internalEncoding, textStart, textEnd, &next); |
5615 | | result = doProlog(parser, parser->m_internalEncoding, textStart, textEnd, |
5616 | | tok, next, &next, XML_FALSE, XML_FALSE, |
5617 | | XML_ACCOUNT_ENTITY_EXPANSION); |
5618 | | } else |
5619 | | #endif /* XML_DTD */ |
5620 | 21.0k | result = doContent(parser, parser->m_tagLevel, parser->m_internalEncoding, |
5621 | 21.0k | textStart, textEnd, &next, XML_FALSE, |
5622 | 21.0k | XML_ACCOUNT_ENTITY_EXPANSION); |
5623 | | |
5624 | 21.0k | if (result == XML_ERROR_NONE) { |
5625 | 20.4k | if (textEnd != next && parser->m_parsingStatus.parsing == XML_SUSPENDED) { |
5626 | 0 | entity->processed = (int)(next - textStart); |
5627 | 0 | parser->m_processor = internalEntityProcessor; |
5628 | 20.4k | } else { |
5629 | | #ifdef XML_DTD |
5630 | | entityTrackingOnClose(parser, entity, __LINE__); |
5631 | | #endif /* XML_DTD */ |
5632 | 20.4k | entity->open = XML_FALSE; |
5633 | 20.4k | parser->m_openInternalEntities = openEntity->next; |
5634 | | /* put openEntity back in list of free instances */ |
5635 | 20.4k | openEntity->next = parser->m_freeInternalEntities; |
5636 | 20.4k | parser->m_freeInternalEntities = openEntity; |
5637 | 20.4k | } |
5638 | 20.4k | } |
5639 | 21.0k | return result; |
5640 | 21.0k | } |
5641 | | |
5642 | | static enum XML_Error PTRCALL |
5643 | | internalEntityProcessor(XML_Parser parser, const char *s, const char *end, |
5644 | 0 | const char **nextPtr) { |
5645 | 0 | ENTITY *entity; |
5646 | 0 | const char *textStart, *textEnd; |
5647 | 0 | const char *next; |
5648 | 0 | enum XML_Error result; |
5649 | 0 | OPEN_INTERNAL_ENTITY *openEntity = parser->m_openInternalEntities; |
5650 | 0 | if (! openEntity) |
5651 | 0 | return XML_ERROR_UNEXPECTED_STATE; |
5652 | | |
5653 | 0 | entity = openEntity->entity; |
5654 | 0 | textStart = ((const char *)entity->textPtr) + entity->processed; |
5655 | 0 | textEnd = (const char *)(entity->textPtr + entity->textLen); |
5656 | | /* Set a safe default value in case 'next' does not get set */ |
5657 | 0 | next = textStart; |
5658 | |
|
5659 | | #ifdef XML_DTD |
5660 | | if (entity->is_param) { |
5661 | | int tok |
5662 | | = XmlPrologTok(parser->m_internalEncoding, textStart, textEnd, &next); |
5663 | | result = doProlog(parser, parser->m_internalEncoding, textStart, textEnd, |
5664 | | tok, next, &next, XML_FALSE, XML_TRUE, |
5665 | | XML_ACCOUNT_ENTITY_EXPANSION); |
5666 | | } else |
5667 | | #endif /* XML_DTD */ |
5668 | 0 | result = doContent(parser, openEntity->startTagLevel, |
5669 | 0 | parser->m_internalEncoding, textStart, textEnd, &next, |
5670 | 0 | XML_FALSE, XML_ACCOUNT_ENTITY_EXPANSION); |
5671 | |
|
5672 | 0 | if (result != XML_ERROR_NONE) |
5673 | 0 | return result; |
5674 | 0 | else if (textEnd != next |
5675 | 0 | && parser->m_parsingStatus.parsing == XML_SUSPENDED) { |
5676 | 0 | entity->processed = (int)(next - (const char *)entity->textPtr); |
5677 | 0 | return result; |
5678 | 0 | } else { |
5679 | | #ifdef XML_DTD |
5680 | | entityTrackingOnClose(parser, entity, __LINE__); |
5681 | | #endif |
5682 | 0 | entity->open = XML_FALSE; |
5683 | 0 | parser->m_openInternalEntities = openEntity->next; |
5684 | | /* put openEntity back in list of free instances */ |
5685 | 0 | openEntity->next = parser->m_freeInternalEntities; |
5686 | 0 | parser->m_freeInternalEntities = openEntity; |
5687 | 0 | } |
5688 | | |
5689 | | #ifdef XML_DTD |
5690 | | if (entity->is_param) { |
5691 | | int tok; |
5692 | | parser->m_processor = prologProcessor; |
5693 | | tok = XmlPrologTok(parser->m_encoding, s, end, &next); |
5694 | | return doProlog(parser, parser->m_encoding, s, end, tok, next, nextPtr, |
5695 | | (XML_Bool)! parser->m_parsingStatus.finalBuffer, XML_TRUE, |
5696 | | XML_ACCOUNT_DIRECT); |
5697 | | } else |
5698 | | #endif /* XML_DTD */ |
5699 | 0 | { |
5700 | 0 | parser->m_processor = contentProcessor; |
5701 | | /* see externalEntityContentProcessor vs contentProcessor */ |
5702 | 0 | return doContent(parser, parser->m_parentParser ? 1 : 0, parser->m_encoding, |
5703 | 0 | s, end, nextPtr, |
5704 | 0 | (XML_Bool)! parser->m_parsingStatus.finalBuffer, |
5705 | 0 | XML_ACCOUNT_DIRECT); |
5706 | 0 | } |
5707 | 0 | } |
5708 | | |
5709 | | static enum XML_Error PTRCALL |
5710 | | errorProcessor(XML_Parser parser, const char *s, const char *end, |
5711 | 0 | const char **nextPtr) { |
5712 | 0 | UNUSED_P(s); |
5713 | 0 | UNUSED_P(end); |
5714 | 0 | UNUSED_P(nextPtr); |
5715 | 0 | return parser->m_errorCode; |
5716 | 0 | } |
5717 | | |
5718 | | static enum XML_Error |
5719 | | storeAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, |
5720 | | const char *ptr, const char *end, STRING_POOL *pool, |
5721 | 19.3k | enum XML_Account account) { |
5722 | 19.3k | enum XML_Error result |
5723 | 19.3k | = appendAttributeValue(parser, enc, isCdata, ptr, end, pool, account); |
5724 | 19.3k | if (result) |
5725 | 116 | return result; |
5726 | 19.2k | if (! isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) |
5727 | 628 | poolChop(pool); |
5728 | 19.2k | if (! poolAppendChar(pool, XML_T('\0'))) |
5729 | 0 | return XML_ERROR_NO_MEMORY; |
5730 | 19.2k | return XML_ERROR_NONE; |
5731 | 19.2k | } |
5732 | | |
5733 | | static enum XML_Error |
5734 | | appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, |
5735 | | const char *ptr, const char *end, STRING_POOL *pool, |
5736 | 10.5M | enum XML_Account account) { |
5737 | 10.5M | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
5738 | 10.5M | #ifndef XML_DTD |
5739 | 10.5M | UNUSED_P(account); |
5740 | 10.5M | #endif |
5741 | | |
5742 | 34.9M | for (;;) { |
5743 | 34.9M | const char *next |
5744 | 34.9M | = ptr; /* XmlAttributeValueTok doesn't always set the last arg */ |
5745 | 34.9M | int tok = XmlAttributeValueTok(enc, ptr, end, &next); |
5746 | | #ifdef XML_DTD |
5747 | | if (! accountingDiffTolerated(parser, tok, ptr, next, __LINE__, account)) { |
5748 | | accountingOnAbort(parser); |
5749 | | return XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
5750 | | } |
5751 | | #endif |
5752 | 34.9M | switch (tok) { |
5753 | 10.5M | case XML_TOK_NONE: |
5754 | 10.5M | return XML_ERROR_NONE; |
5755 | 33 | case XML_TOK_INVALID: |
5756 | 33 | if (enc == parser->m_encoding) |
5757 | 21 | parser->m_eventPtr = next; |
5758 | 33 | return XML_ERROR_INVALID_TOKEN; |
5759 | 7 | case XML_TOK_PARTIAL: |
5760 | 7 | if (enc == parser->m_encoding) |
5761 | 5 | parser->m_eventPtr = ptr; |
5762 | 7 | return XML_ERROR_INVALID_TOKEN; |
5763 | 704k | case XML_TOK_CHAR_REF: { |
5764 | 704k | XML_Char buf[XML_ENCODE_MAX]; |
5765 | 704k | int i; |
5766 | 704k | int n = XmlCharRefNumber(enc, ptr); |
5767 | 704k | if (n < 0) { |
5768 | 8 | if (enc == parser->m_encoding) |
5769 | 7 | parser->m_eventPtr = ptr; |
5770 | 8 | return XML_ERROR_BAD_CHAR_REF; |
5771 | 8 | } |
5772 | 704k | if (! isCdata && n == 0x20 /* space */ |
5773 | 704k | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) |
5774 | 364 | break; |
5775 | 704k | n = XmlEncode(n, (ICHAR *)buf); |
5776 | | /* The XmlEncode() functions can never return 0 here. That |
5777 | | * error return happens if the code point passed in is either |
5778 | | * negative or greater than or equal to 0x110000. The |
5779 | | * XmlCharRefNumber() functions will all return a number |
5780 | | * strictly less than 0x110000 or a negative value if an error |
5781 | | * occurred. The negative value is intercepted above, so |
5782 | | * XmlEncode() is never passed a value it might return an |
5783 | | * error for. |
5784 | | */ |
5785 | 2.38M | for (i = 0; i < n; i++) { |
5786 | 1.67M | if (! poolAppendChar(pool, buf[i])) |
5787 | 1 | return XML_ERROR_NO_MEMORY; |
5788 | 1.67M | } |
5789 | 704k | } break; |
5790 | 7.80M | case XML_TOK_DATA_CHARS: |
5791 | 7.80M | if (! poolAppend(pool, enc, ptr, next)) |
5792 | 2 | return XML_ERROR_NO_MEMORY; |
5793 | 7.80M | break; |
5794 | 7.80M | case XML_TOK_TRAILING_CR: |
5795 | 727 | next = ptr + enc->minBytesPerChar; |
5796 | | /* fall through */ |
5797 | 594k | case XML_TOK_ATTRIBUTE_VALUE_S: |
5798 | 5.36M | case XML_TOK_DATA_NEWLINE: |
5799 | 5.36M | if (! isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) |
5800 | 111k | break; |
5801 | 5.25M | if (! poolAppendChar(pool, 0x20)) |
5802 | 0 | return XML_ERROR_NO_MEMORY; |
5803 | 5.25M | break; |
5804 | 10.5M | case XML_TOK_ENTITY_REF: { |
5805 | 10.5M | const XML_Char *name; |
5806 | 10.5M | ENTITY *entity; |
5807 | 10.5M | char checkEntityDecl; |
5808 | 10.5M | XML_Char ch = (XML_Char)XmlPredefinedEntityName( |
5809 | 10.5M | enc, ptr + enc->minBytesPerChar, next - enc->minBytesPerChar); |
5810 | 10.5M | if (ch) { |
5811 | | #ifdef XML_DTD |
5812 | | /* NOTE: We are replacing 4-6 characters original input for 1 character |
5813 | | * so there is no amplification and hence recording without |
5814 | | * protection. */ |
5815 | | accountingDiffTolerated(parser, tok, (char *)&ch, |
5816 | | ((char *)&ch) + sizeof(XML_Char), __LINE__, |
5817 | | XML_ACCOUNT_ENTITY_EXPANSION); |
5818 | | #endif /* XML_DTD */ |
5819 | 221 | if (! poolAppendChar(pool, ch)) |
5820 | 0 | return XML_ERROR_NO_MEMORY; |
5821 | 221 | break; |
5822 | 221 | } |
5823 | 10.5M | name = poolStoreString(&parser->m_temp2Pool, enc, |
5824 | 10.5M | ptr + enc->minBytesPerChar, |
5825 | 10.5M | next - enc->minBytesPerChar); |
5826 | 10.5M | if (! name) |
5827 | 0 | return XML_ERROR_NO_MEMORY; |
5828 | 10.5M | entity = (ENTITY *)lookup(parser, &dtd->generalEntities, name, 0); |
5829 | 10.5M | poolDiscard(&parser->m_temp2Pool); |
5830 | | /* First, determine if a check for an existing declaration is needed; |
5831 | | if yes, check that the entity exists, and that it is internal. |
5832 | | */ |
5833 | 10.5M | if (pool == &dtd->pool) /* are we called from prolog? */ |
5834 | 23.7k | checkEntityDecl = |
5835 | | #ifdef XML_DTD |
5836 | | parser->m_prologState.documentEntity && |
5837 | | #endif /* XML_DTD */ |
5838 | 23.7k | (dtd->standalone ? ! parser->m_openInternalEntities |
5839 | 23.7k | : ! dtd->hasParamEntityRefs); |
5840 | 10.4M | else /* if (pool == &parser->m_tempPool): we are called from content */ |
5841 | 10.4M | checkEntityDecl = ! dtd->hasParamEntityRefs || dtd->standalone; |
5842 | 10.5M | if (checkEntityDecl) { |
5843 | 10.5M | if (! entity) |
5844 | 47 | return XML_ERROR_UNDEFINED_ENTITY; |
5845 | 10.5M | else if (! entity->is_internal) |
5846 | 0 | return XML_ERROR_ENTITY_DECLARED_IN_PE; |
5847 | 10.5M | } else if (! entity) { |
5848 | | /* Cannot report skipped entity here - see comments on |
5849 | | parser->m_skippedEntityHandler. |
5850 | | if (parser->m_skippedEntityHandler) |
5851 | | parser->m_skippedEntityHandler(parser->m_handlerArg, name, 0); |
5852 | | */ |
5853 | | /* Cannot call the default handler because this would be |
5854 | | out of sync with the call to the startElementHandler. |
5855 | | if ((pool == &parser->m_tempPool) && parser->m_defaultHandler) |
5856 | | reportDefault(parser, enc, ptr, next); |
5857 | | */ |
5858 | 2.82k | break; |
5859 | 2.82k | } |
5860 | 10.5M | if (entity->open) { |
5861 | 15 | if (enc == parser->m_encoding) { |
5862 | | /* It does not appear that this line can be executed. |
5863 | | * |
5864 | | * The "if (entity->open)" check catches recursive entity |
5865 | | * definitions. In order to be called with an open |
5866 | | * entity, it must have gone through this code before and |
5867 | | * been through the recursive call to |
5868 | | * appendAttributeValue() some lines below. That call |
5869 | | * sets the local encoding ("enc") to the parser's |
5870 | | * internal encoding (internal_utf8 or internal_utf16), |
5871 | | * which can never be the same as the principle encoding. |
5872 | | * It doesn't appear there is another code path that gets |
5873 | | * here with entity->open being TRUE. |
5874 | | * |
5875 | | * Since it is not certain that this logic is watertight, |
5876 | | * we keep the line and merely exclude it from coverage |
5877 | | * tests. |
5878 | | */ |
5879 | 0 | parser->m_eventPtr = ptr; /* LCOV_EXCL_LINE */ |
5880 | 0 | } |
5881 | 15 | return XML_ERROR_RECURSIVE_ENTITY_REF; |
5882 | 15 | } |
5883 | 10.5M | if (entity->notation) { |
5884 | 1 | if (enc == parser->m_encoding) |
5885 | 1 | parser->m_eventPtr = ptr; |
5886 | 1 | return XML_ERROR_BINARY_ENTITY_REF; |
5887 | 1 | } |
5888 | 10.5M | if (! entity->textPtr) { |
5889 | 2 | if (enc == parser->m_encoding) |
5890 | 1 | parser->m_eventPtr = ptr; |
5891 | 2 | return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; |
5892 | 10.5M | } else { |
5893 | 10.5M | enum XML_Error result; |
5894 | 10.5M | const XML_Char *textEnd = entity->textPtr + entity->textLen; |
5895 | 10.5M | entity->open = XML_TRUE; |
5896 | | #ifdef XML_DTD |
5897 | | entityTrackingOnOpen(parser, entity, __LINE__); |
5898 | | #endif |
5899 | 10.5M | result = appendAttributeValue(parser, parser->m_internalEncoding, |
5900 | 10.5M | isCdata, (const char *)entity->textPtr, |
5901 | 10.5M | (const char *)textEnd, pool, |
5902 | 10.5M | XML_ACCOUNT_ENTITY_EXPANSION); |
5903 | | #ifdef XML_DTD |
5904 | | entityTrackingOnClose(parser, entity, __LINE__); |
5905 | | #endif |
5906 | 10.5M | entity->open = XML_FALSE; |
5907 | 10.5M | if (result) |
5908 | 70 | return result; |
5909 | 10.5M | } |
5910 | 10.5M | } break; |
5911 | 10.5M | default: |
5912 | | /* The only token returned by XmlAttributeValueTok() that does |
5913 | | * not have an explicit case here is XML_TOK_PARTIAL_CHAR. |
5914 | | * Getting that would require an entity name to contain an |
5915 | | * incomplete XML character (e.g. \xE2\x82); however previous |
5916 | | * tokenisers will have already recognised and rejected such |
5917 | | * names before XmlAttributeValueTok() gets a look-in. This |
5918 | | * default case should be retained as a safety net, but the code |
5919 | | * excluded from coverage tests. |
5920 | | * |
5921 | | * LCOV_EXCL_START |
5922 | | */ |
5923 | 0 | if (enc == parser->m_encoding) |
5924 | 0 | parser->m_eventPtr = ptr; |
5925 | 0 | return XML_ERROR_UNEXPECTED_STATE; |
5926 | | /* LCOV_EXCL_STOP */ |
5927 | 34.9M | } |
5928 | 24.3M | ptr = next; |
5929 | 24.3M | } |
5930 | | /* not reached */ |
5931 | 10.5M | } |
5932 | | |
5933 | | static enum XML_Error |
5934 | | storeEntityValue(XML_Parser parser, const ENCODING *enc, |
5935 | | const char *entityTextPtr, const char *entityTextEnd, |
5936 | 5.19k | enum XML_Account account) { |
5937 | 5.19k | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
5938 | 5.19k | STRING_POOL *pool = &(dtd->entityValuePool); |
5939 | 5.19k | enum XML_Error result = XML_ERROR_NONE; |
5940 | | #ifdef XML_DTD |
5941 | | int oldInEntityValue = parser->m_prologState.inEntityValue; |
5942 | | parser->m_prologState.inEntityValue = 1; |
5943 | | #else |
5944 | 5.19k | UNUSED_P(account); |
5945 | 5.19k | #endif /* XML_DTD */ |
5946 | | /* never return Null for the value argument in EntityDeclHandler, |
5947 | | since this would indicate an external entity; therefore we |
5948 | | have to make sure that entityValuePool.start is not null */ |
5949 | 5.19k | if (! pool->blocks) { |
5950 | 1.77k | if (! poolGrow(pool)) |
5951 | 0 | return XML_ERROR_NO_MEMORY; |
5952 | 1.77k | } |
5953 | | |
5954 | 232k | for (;;) { |
5955 | 232k | const char *next |
5956 | 232k | = entityTextPtr; /* XmlEntityValueTok doesn't always set the last arg */ |
5957 | 232k | int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next); |
5958 | | |
5959 | | #ifdef XML_DTD |
5960 | | if (! accountingDiffTolerated(parser, tok, entityTextPtr, next, __LINE__, |
5961 | | account)) { |
5962 | | accountingOnAbort(parser); |
5963 | | result = XML_ERROR_AMPLIFICATION_LIMIT_BREACH; |
5964 | | goto endEntityValue; |
5965 | | } |
5966 | | #endif |
5967 | | |
5968 | 232k | switch (tok) { |
5969 | 1 | case XML_TOK_PARAM_ENTITY_REF: |
5970 | | #ifdef XML_DTD |
5971 | | if (parser->m_isParamEntity || enc != parser->m_encoding) { |
5972 | | const XML_Char *name; |
5973 | | ENTITY *entity; |
5974 | | name = poolStoreString(&parser->m_tempPool, enc, |
5975 | | entityTextPtr + enc->minBytesPerChar, |
5976 | | next - enc->minBytesPerChar); |
5977 | | if (! name) { |
5978 | | result = XML_ERROR_NO_MEMORY; |
5979 | | goto endEntityValue; |
5980 | | } |
5981 | | entity = (ENTITY *)lookup(parser, &dtd->paramEntities, name, 0); |
5982 | | poolDiscard(&parser->m_tempPool); |
5983 | | if (! entity) { |
5984 | | /* not a well-formedness error - see XML 1.0: WFC Entity Declared */ |
5985 | | /* cannot report skipped entity here - see comments on |
5986 | | parser->m_skippedEntityHandler |
5987 | | if (parser->m_skippedEntityHandler) |
5988 | | parser->m_skippedEntityHandler(parser->m_handlerArg, name, 0); |
5989 | | */ |
5990 | | dtd->keepProcessing = dtd->standalone; |
5991 | | goto endEntityValue; |
5992 | | } |
5993 | | if (entity->open) { |
5994 | | if (enc == parser->m_encoding) |
5995 | | parser->m_eventPtr = entityTextPtr; |
5996 | | result = XML_ERROR_RECURSIVE_ENTITY_REF; |
5997 | | goto endEntityValue; |
5998 | | } |
5999 | | if (entity->systemId) { |
6000 | | if (parser->m_externalEntityRefHandler) { |
6001 | | dtd->paramEntityRead = XML_FALSE; |
6002 | | entity->open = XML_TRUE; |
6003 | | entityTrackingOnOpen(parser, entity, __LINE__); |
6004 | | if (! parser->m_externalEntityRefHandler( |
6005 | | parser->m_externalEntityRefHandlerArg, 0, entity->base, |
6006 | | entity->systemId, entity->publicId)) { |
6007 | | entityTrackingOnClose(parser, entity, __LINE__); |
6008 | | entity->open = XML_FALSE; |
6009 | | result = XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
6010 | | goto endEntityValue; |
6011 | | } |
6012 | | entityTrackingOnClose(parser, entity, __LINE__); |
6013 | | entity->open = XML_FALSE; |
6014 | | if (! dtd->paramEntityRead) |
6015 | | dtd->keepProcessing = dtd->standalone; |
6016 | | } else |
6017 | | dtd->keepProcessing = dtd->standalone; |
6018 | | } else { |
6019 | | entity->open = XML_TRUE; |
6020 | | entityTrackingOnOpen(parser, entity, __LINE__); |
6021 | | result = storeEntityValue( |
6022 | | parser, parser->m_internalEncoding, (const char *)entity->textPtr, |
6023 | | (const char *)(entity->textPtr + entity->textLen), |
6024 | | XML_ACCOUNT_ENTITY_EXPANSION); |
6025 | | entityTrackingOnClose(parser, entity, __LINE__); |
6026 | | entity->open = XML_FALSE; |
6027 | | if (result) |
6028 | | goto endEntityValue; |
6029 | | } |
6030 | | break; |
6031 | | } |
6032 | | #endif /* XML_DTD */ |
6033 | | /* In the internal subset, PE references are not legal |
6034 | | within markup declarations, e.g entity values in this case. */ |
6035 | 1 | parser->m_eventPtr = entityTextPtr; |
6036 | 1 | result = XML_ERROR_PARAM_ENTITY_REF; |
6037 | 1 | goto endEntityValue; |
6038 | 5.13k | case XML_TOK_NONE: |
6039 | 5.13k | result = XML_ERROR_NONE; |
6040 | 5.13k | goto endEntityValue; |
6041 | 19.5k | case XML_TOK_ENTITY_REF: |
6042 | 98.9k | case XML_TOK_DATA_CHARS: |
6043 | 98.9k | if (! poolAppend(pool, enc, entityTextPtr, next)) { |
6044 | 0 | result = XML_ERROR_NO_MEMORY; |
6045 | 0 | goto endEntityValue; |
6046 | 0 | } |
6047 | 98.9k | break; |
6048 | 98.9k | case XML_TOK_TRAILING_CR: |
6049 | 236 | next = entityTextPtr + enc->minBytesPerChar; |
6050 | | /* fall through */ |
6051 | 110k | case XML_TOK_DATA_NEWLINE: |
6052 | 110k | if (pool->end == pool->ptr && ! poolGrow(pool)) { |
6053 | 0 | result = XML_ERROR_NO_MEMORY; |
6054 | 0 | goto endEntityValue; |
6055 | 0 | } |
6056 | 110k | *(pool->ptr)++ = 0xA; |
6057 | 110k | break; |
6058 | 17.6k | case XML_TOK_CHAR_REF: { |
6059 | 17.6k | XML_Char buf[XML_ENCODE_MAX]; |
6060 | 17.6k | int i; |
6061 | 17.6k | int n = XmlCharRefNumber(enc, entityTextPtr); |
6062 | 17.6k | if (n < 0) { |
6063 | 2 | if (enc == parser->m_encoding) |
6064 | 2 | parser->m_eventPtr = entityTextPtr; |
6065 | 2 | result = XML_ERROR_BAD_CHAR_REF; |
6066 | 2 | goto endEntityValue; |
6067 | 2 | } |
6068 | 17.6k | n = XmlEncode(n, (ICHAR *)buf); |
6069 | | /* The XmlEncode() functions can never return 0 here. That |
6070 | | * error return happens if the code point passed in is either |
6071 | | * negative or greater than or equal to 0x110000. The |
6072 | | * XmlCharRefNumber() functions will all return a number |
6073 | | * strictly less than 0x110000 or a negative value if an error |
6074 | | * occurred. The negative value is intercepted above, so |
6075 | | * XmlEncode() is never passed a value it might return an |
6076 | | * error for. |
6077 | | */ |
6078 | 70.1k | for (i = 0; i < n; i++) { |
6079 | 52.5k | if (pool->end == pool->ptr && ! poolGrow(pool)) { |
6080 | 0 | result = XML_ERROR_NO_MEMORY; |
6081 | 0 | goto endEntityValue; |
6082 | 0 | } |
6083 | 52.5k | *(pool->ptr)++ = buf[i]; |
6084 | 52.5k | } |
6085 | 17.6k | } break; |
6086 | 17.6k | case XML_TOK_PARTIAL: |
6087 | 23 | if (enc == parser->m_encoding) |
6088 | 23 | parser->m_eventPtr = entityTextPtr; |
6089 | 23 | result = XML_ERROR_INVALID_TOKEN; |
6090 | 23 | goto endEntityValue; |
6091 | 29 | case XML_TOK_INVALID: |
6092 | 29 | if (enc == parser->m_encoding) |
6093 | 29 | parser->m_eventPtr = next; |
6094 | 29 | result = XML_ERROR_INVALID_TOKEN; |
6095 | 29 | goto endEntityValue; |
6096 | 0 | default: |
6097 | | /* This default case should be unnecessary -- all the tokens |
6098 | | * that XmlEntityValueTok() can return have their own explicit |
6099 | | * cases -- but should be retained for safety. We do however |
6100 | | * exclude it from the coverage statistics. |
6101 | | * |
6102 | | * LCOV_EXCL_START |
6103 | | */ |
6104 | 0 | if (enc == parser->m_encoding) |
6105 | 0 | parser->m_eventPtr = entityTextPtr; |
6106 | 0 | result = XML_ERROR_UNEXPECTED_STATE; |
6107 | 0 | goto endEntityValue; |
6108 | | /* LCOV_EXCL_STOP */ |
6109 | 232k | } |
6110 | 226k | entityTextPtr = next; |
6111 | 226k | } |
6112 | 5.19k | endEntityValue: |
6113 | | #ifdef XML_DTD |
6114 | | parser->m_prologState.inEntityValue = oldInEntityValue; |
6115 | | #endif /* XML_DTD */ |
6116 | 5.19k | return result; |
6117 | 5.19k | } |
6118 | | |
6119 | | static void FASTCALL |
6120 | 0 | normalizeLines(XML_Char *s) { |
6121 | 0 | XML_Char *p; |
6122 | 0 | for (;; s++) { |
6123 | 0 | if (*s == XML_T('\0')) |
6124 | 0 | return; |
6125 | 0 | if (*s == 0xD) |
6126 | 0 | break; |
6127 | 0 | } |
6128 | 0 | p = s; |
6129 | 0 | do { |
6130 | 0 | if (*s == 0xD) { |
6131 | 0 | *p++ = 0xA; |
6132 | 0 | if (*++s == 0xA) |
6133 | 0 | s++; |
6134 | 0 | } else |
6135 | 0 | *p++ = *s++; |
6136 | 0 | } while (*s); |
6137 | 0 | *p = XML_T('\0'); |
6138 | 0 | } |
6139 | | |
6140 | | static int |
6141 | | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, |
6142 | 10.6k | const char *start, const char *end) { |
6143 | 10.6k | const XML_Char *target; |
6144 | 10.6k | XML_Char *data; |
6145 | 10.6k | const char *tem; |
6146 | 10.6k | if (! parser->m_processingInstructionHandler) { |
6147 | 10.6k | if (parser->m_defaultHandler) |
6148 | 0 | reportDefault(parser, enc, start, end); |
6149 | 10.6k | return 1; |
6150 | 10.6k | } |
6151 | 0 | start += enc->minBytesPerChar * 2; |
6152 | 0 | tem = start + XmlNameLength(enc, start); |
6153 | 0 | target = poolStoreString(&parser->m_tempPool, enc, start, tem); |
6154 | 0 | if (! target) |
6155 | 0 | return 0; |
6156 | 0 | poolFinish(&parser->m_tempPool); |
6157 | 0 | data = poolStoreString(&parser->m_tempPool, enc, XmlSkipS(enc, tem), |
6158 | 0 | end - enc->minBytesPerChar * 2); |
6159 | 0 | if (! data) |
6160 | 0 | return 0; |
6161 | 0 | normalizeLines(data); |
6162 | 0 | parser->m_processingInstructionHandler(parser->m_handlerArg, target, data); |
6163 | 0 | poolClear(&parser->m_tempPool); |
6164 | 0 | return 1; |
6165 | 0 | } |
6166 | | |
6167 | | static int |
6168 | | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, |
6169 | 1.20k | const char *end) { |
6170 | 1.20k | XML_Char *data; |
6171 | 1.20k | if (! parser->m_commentHandler) { |
6172 | 1.20k | if (parser->m_defaultHandler) |
6173 | 0 | reportDefault(parser, enc, start, end); |
6174 | 1.20k | return 1; |
6175 | 1.20k | } |
6176 | 0 | data = poolStoreString(&parser->m_tempPool, enc, |
6177 | 0 | start + enc->minBytesPerChar * 4, |
6178 | 0 | end - enc->minBytesPerChar * 3); |
6179 | 0 | if (! data) |
6180 | 0 | return 0; |
6181 | 0 | normalizeLines(data); |
6182 | 0 | parser->m_commentHandler(parser->m_handlerArg, data); |
6183 | 0 | poolClear(&parser->m_tempPool); |
6184 | 0 | return 1; |
6185 | 0 | } |
6186 | | |
6187 | | static void |
6188 | | reportDefault(XML_Parser parser, const ENCODING *enc, const char *s, |
6189 | 0 | const char *end) { |
6190 | 0 | if (MUST_CONVERT(enc, s)) { |
6191 | 0 | enum XML_Convert_Result convert_res; |
6192 | 0 | const char **eventPP; |
6193 | 0 | const char **eventEndPP; |
6194 | 0 | if (enc == parser->m_encoding) { |
6195 | 0 | eventPP = &parser->m_eventPtr; |
6196 | 0 | eventEndPP = &parser->m_eventEndPtr; |
6197 | 0 | } else { |
6198 | | /* To get here, two things must be true; the parser must be |
6199 | | * using a character encoding that is not the same as the |
6200 | | * encoding passed in, and the encoding passed in must need |
6201 | | * conversion to the internal format (UTF-8 unless XML_UNICODE |
6202 | | * is defined). The only occasions on which the encoding passed |
6203 | | * in is not the same as the parser's encoding are when it is |
6204 | | * the internal encoding (e.g. a previously defined parameter |
6205 | | * entity, already converted to internal format). This by |
6206 | | * definition doesn't need conversion, so the whole branch never |
6207 | | * gets executed. |
6208 | | * |
6209 | | * For safety's sake we don't delete these lines and merely |
6210 | | * exclude them from coverage statistics. |
6211 | | * |
6212 | | * LCOV_EXCL_START |
6213 | | */ |
6214 | 0 | eventPP = &(parser->m_openInternalEntities->internalEventPtr); |
6215 | 0 | eventEndPP = &(parser->m_openInternalEntities->internalEventEndPtr); |
6216 | | /* LCOV_EXCL_STOP */ |
6217 | 0 | } |
6218 | 0 | do { |
6219 | 0 | ICHAR *dataPtr = (ICHAR *)parser->m_dataBuf; |
6220 | 0 | convert_res |
6221 | 0 | = XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)parser->m_dataBufEnd); |
6222 | 0 | *eventEndPP = s; |
6223 | 0 | parser->m_defaultHandler(parser->m_handlerArg, parser->m_dataBuf, |
6224 | 0 | (int)(dataPtr - (ICHAR *)parser->m_dataBuf)); |
6225 | 0 | *eventPP = s; |
6226 | 0 | } while ((convert_res != XML_CONVERT_COMPLETED) |
6227 | 0 | && (convert_res != XML_CONVERT_INPUT_INCOMPLETE)); |
6228 | 0 | } else |
6229 | 0 | parser->m_defaultHandler(parser->m_handlerArg, (XML_Char *)s, |
6230 | 0 | (int)((XML_Char *)end - (XML_Char *)s)); |
6231 | 0 | } |
6232 | | |
6233 | | static int |
6234 | | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, XML_Bool isCdata, |
6235 | 24.2k | XML_Bool isId, const XML_Char *value, XML_Parser parser) { |
6236 | 24.2k | DEFAULT_ATTRIBUTE *att; |
6237 | 24.2k | if (value || isId) { |
6238 | | /* The handling of default attributes gets messed up if we have |
6239 | | a default which duplicates a non-default. */ |
6240 | 12.9k | int i; |
6241 | 52.3k | for (i = 0; i < type->nDefaultAtts; i++) |
6242 | 48.9k | if (attId == type->defaultAtts[i].id) |
6243 | 9.58k | return 1; |
6244 | 3.35k | if (isId && ! type->idAtt && ! attId->xmlns) |
6245 | 376 | type->idAtt = attId; |
6246 | 3.35k | } |
6247 | 14.6k | if (type->nDefaultAtts == type->allocDefaultAtts) { |
6248 | 2.38k | if (type->allocDefaultAtts == 0) { |
6249 | 1.47k | type->allocDefaultAtts = 8; |
6250 | 1.47k | type->defaultAtts = (DEFAULT_ATTRIBUTE *)MALLOC( |
6251 | 1.47k | parser, type->allocDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); |
6252 | 1.47k | if (! type->defaultAtts) { |
6253 | 0 | type->allocDefaultAtts = 0; |
6254 | 0 | return 0; |
6255 | 0 | } |
6256 | 1.47k | } else { |
6257 | 911 | DEFAULT_ATTRIBUTE *temp; |
6258 | | |
6259 | | /* Detect and prevent integer overflow */ |
6260 | 911 | if (type->allocDefaultAtts > INT_MAX / 2) { |
6261 | 0 | return 0; |
6262 | 0 | } |
6263 | | |
6264 | 911 | int count = type->allocDefaultAtts * 2; |
6265 | | |
6266 | | /* Detect and prevent integer overflow. |
6267 | | * The preprocessor guard addresses the "always false" warning |
6268 | | * from -Wtype-limits on platforms where |
6269 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
6270 | | #if UINT_MAX >= SIZE_MAX |
6271 | | if ((unsigned)count > (size_t)(-1) / sizeof(DEFAULT_ATTRIBUTE)) { |
6272 | | return 0; |
6273 | | } |
6274 | | #endif |
6275 | | |
6276 | 911 | temp = (DEFAULT_ATTRIBUTE *)REALLOC(parser, type->defaultAtts, |
6277 | 911 | (count * sizeof(DEFAULT_ATTRIBUTE))); |
6278 | 911 | if (temp == NULL) |
6279 | 0 | return 0; |
6280 | 911 | type->allocDefaultAtts = count; |
6281 | 911 | type->defaultAtts = temp; |
6282 | 911 | } |
6283 | 2.38k | } |
6284 | 14.6k | att = type->defaultAtts + type->nDefaultAtts; |
6285 | 14.6k | att->id = attId; |
6286 | 14.6k | att->value = value; |
6287 | 14.6k | att->isCdata = isCdata; |
6288 | 14.6k | if (! isCdata) |
6289 | 7.83k | attId->maybeTokenized = XML_TRUE; |
6290 | 14.6k | type->nDefaultAtts += 1; |
6291 | 14.6k | return 1; |
6292 | 14.6k | } |
6293 | | |
6294 | | static int |
6295 | 8.09k | setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) { |
6296 | 8.09k | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
6297 | 8.09k | const XML_Char *name; |
6298 | 673k | for (name = elementType->name; *name; name++) { |
6299 | 670k | if (*name == XML_T(ASCII_COLON)) { |
6300 | 4.86k | PREFIX *prefix; |
6301 | 4.86k | const XML_Char *s; |
6302 | 404k | for (s = elementType->name; s != name; s++) { |
6303 | 400k | if (! poolAppendChar(&dtd->pool, *s)) |
6304 | 0 | return 0; |
6305 | 400k | } |
6306 | 4.86k | if (! poolAppendChar(&dtd->pool, XML_T('\0'))) |
6307 | 0 | return 0; |
6308 | 4.86k | prefix = (PREFIX *)lookup(parser, &dtd->prefixes, poolStart(&dtd->pool), |
6309 | 4.86k | sizeof(PREFIX)); |
6310 | 4.86k | if (! prefix) |
6311 | 0 | return 0; |
6312 | 4.86k | if (prefix->name == poolStart(&dtd->pool)) |
6313 | 4.58k | poolFinish(&dtd->pool); |
6314 | 279 | else |
6315 | 279 | poolDiscard(&dtd->pool); |
6316 | 4.86k | elementType->prefix = prefix; |
6317 | 4.86k | break; |
6318 | 4.86k | } |
6319 | 670k | } |
6320 | 8.09k | return 1; |
6321 | 8.09k | } |
6322 | | |
6323 | | static ATTRIBUTE_ID * |
6324 | | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, |
6325 | 50.0k | const char *end) { |
6326 | 50.0k | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
6327 | 50.0k | ATTRIBUTE_ID *id; |
6328 | 50.0k | const XML_Char *name; |
6329 | 50.0k | if (! poolAppendChar(&dtd->pool, XML_T('\0'))) |
6330 | 0 | return NULL; |
6331 | 50.0k | name = poolStoreString(&dtd->pool, enc, start, end); |
6332 | 50.0k | if (! name) |
6333 | 0 | return NULL; |
6334 | | /* skip quotation mark - its storage will be re-used (like in name[-1]) */ |
6335 | 50.0k | ++name; |
6336 | 50.0k | id = (ATTRIBUTE_ID *)lookup(parser, &dtd->attributeIds, name, |
6337 | 50.0k | sizeof(ATTRIBUTE_ID)); |
6338 | 50.0k | if (! id) |
6339 | 0 | return NULL; |
6340 | 50.0k | if (id->name != name) |
6341 | 45.0k | poolDiscard(&dtd->pool); |
6342 | 5.01k | else { |
6343 | 5.01k | poolFinish(&dtd->pool); |
6344 | 5.01k | if (! parser->m_ns) |
6345 | 5.01k | ; |
6346 | 0 | else if (name[0] == XML_T(ASCII_x) && name[1] == XML_T(ASCII_m) |
6347 | 0 | && name[2] == XML_T(ASCII_l) && name[3] == XML_T(ASCII_n) |
6348 | 0 | && name[4] == XML_T(ASCII_s) |
6349 | 0 | && (name[5] == XML_T('\0') || name[5] == XML_T(ASCII_COLON))) { |
6350 | 0 | if (name[5] == XML_T('\0')) |
6351 | 0 | id->prefix = &dtd->defaultPrefix; |
6352 | 0 | else |
6353 | 0 | id->prefix = (PREFIX *)lookup(parser, &dtd->prefixes, name + 6, |
6354 | 0 | sizeof(PREFIX)); |
6355 | 0 | id->xmlns = XML_TRUE; |
6356 | 0 | } else { |
6357 | 0 | int i; |
6358 | 0 | for (i = 0; name[i]; i++) { |
6359 | | /* attributes without prefix are *not* in the default namespace */ |
6360 | 0 | if (name[i] == XML_T(ASCII_COLON)) { |
6361 | 0 | int j; |
6362 | 0 | for (j = 0; j < i; j++) { |
6363 | 0 | if (! poolAppendChar(&dtd->pool, name[j])) |
6364 | 0 | return NULL; |
6365 | 0 | } |
6366 | 0 | if (! poolAppendChar(&dtd->pool, XML_T('\0'))) |
6367 | 0 | return NULL; |
6368 | 0 | id->prefix = (PREFIX *)lookup(parser, &dtd->prefixes, |
6369 | 0 | poolStart(&dtd->pool), sizeof(PREFIX)); |
6370 | 0 | if (! id->prefix) |
6371 | 0 | return NULL; |
6372 | 0 | if (id->prefix->name == poolStart(&dtd->pool)) |
6373 | 0 | poolFinish(&dtd->pool); |
6374 | 0 | else |
6375 | 0 | poolDiscard(&dtd->pool); |
6376 | 0 | break; |
6377 | 0 | } |
6378 | 0 | } |
6379 | 0 | } |
6380 | 5.01k | } |
6381 | 50.0k | return id; |
6382 | 50.0k | } |
6383 | | |
6384 | 0 | #define CONTEXT_SEP XML_T(ASCII_FF) |
6385 | | |
6386 | | static const XML_Char * |
6387 | 0 | getContext(XML_Parser parser) { |
6388 | 0 | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
6389 | 0 | HASH_TABLE_ITER iter; |
6390 | 0 | XML_Bool needSep = XML_FALSE; |
6391 | |
|
6392 | 0 | if (dtd->defaultPrefix.binding) { |
6393 | 0 | int i; |
6394 | 0 | int len; |
6395 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T(ASCII_EQUALS))) |
6396 | 0 | return NULL; |
6397 | 0 | len = dtd->defaultPrefix.binding->uriLen; |
6398 | 0 | if (parser->m_namespaceSeparator) |
6399 | 0 | len--; |
6400 | 0 | for (i = 0; i < len; i++) { |
6401 | 0 | if (! poolAppendChar(&parser->m_tempPool, |
6402 | 0 | dtd->defaultPrefix.binding->uri[i])) { |
6403 | | /* Because of memory caching, I don't believe this line can be |
6404 | | * executed. |
6405 | | * |
6406 | | * This is part of a loop copying the default prefix binding |
6407 | | * URI into the parser's temporary string pool. Previously, |
6408 | | * that URI was copied into the same string pool, with a |
6409 | | * terminating NUL character, as part of setContext(). When |
6410 | | * the pool was cleared, that leaves a block definitely big |
6411 | | * enough to hold the URI on the free block list of the pool. |
6412 | | * The URI copy in getContext() therefore cannot run out of |
6413 | | * memory. |
6414 | | * |
6415 | | * If the pool is used between the setContext() and |
6416 | | * getContext() calls, the worst it can do is leave a bigger |
6417 | | * block on the front of the free list. Given that this is |
6418 | | * all somewhat inobvious and program logic can be changed, we |
6419 | | * don't delete the line but we do exclude it from the test |
6420 | | * coverage statistics. |
6421 | | */ |
6422 | 0 | return NULL; /* LCOV_EXCL_LINE */ |
6423 | 0 | } |
6424 | 0 | } |
6425 | 0 | needSep = XML_TRUE; |
6426 | 0 | } |
6427 | | |
6428 | 0 | hashTableIterInit(&iter, &(dtd->prefixes)); |
6429 | 0 | for (;;) { |
6430 | 0 | int i; |
6431 | 0 | int len; |
6432 | 0 | const XML_Char *s; |
6433 | 0 | PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); |
6434 | 0 | if (! prefix) |
6435 | 0 | break; |
6436 | 0 | if (! prefix->binding) { |
6437 | | /* This test appears to be (justifiable) paranoia. There does |
6438 | | * not seem to be a way of injecting a prefix without a binding |
6439 | | * that doesn't get errored long before this function is called. |
6440 | | * The test should remain for safety's sake, so we instead |
6441 | | * exclude the following line from the coverage statistics. |
6442 | | */ |
6443 | 0 | continue; /* LCOV_EXCL_LINE */ |
6444 | 0 | } |
6445 | 0 | if (needSep && ! poolAppendChar(&parser->m_tempPool, CONTEXT_SEP)) |
6446 | 0 | return NULL; |
6447 | 0 | for (s = prefix->name; *s; s++) |
6448 | 0 | if (! poolAppendChar(&parser->m_tempPool, *s)) |
6449 | 0 | return NULL; |
6450 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T(ASCII_EQUALS))) |
6451 | 0 | return NULL; |
6452 | 0 | len = prefix->binding->uriLen; |
6453 | 0 | if (parser->m_namespaceSeparator) |
6454 | 0 | len--; |
6455 | 0 | for (i = 0; i < len; i++) |
6456 | 0 | if (! poolAppendChar(&parser->m_tempPool, prefix->binding->uri[i])) |
6457 | 0 | return NULL; |
6458 | 0 | needSep = XML_TRUE; |
6459 | 0 | } |
6460 | | |
6461 | 0 | hashTableIterInit(&iter, &(dtd->generalEntities)); |
6462 | 0 | for (;;) { |
6463 | 0 | const XML_Char *s; |
6464 | 0 | ENTITY *e = (ENTITY *)hashTableIterNext(&iter); |
6465 | 0 | if (! e) |
6466 | 0 | break; |
6467 | 0 | if (! e->open) |
6468 | 0 | continue; |
6469 | 0 | if (needSep && ! poolAppendChar(&parser->m_tempPool, CONTEXT_SEP)) |
6470 | 0 | return NULL; |
6471 | 0 | for (s = e->name; *s; s++) |
6472 | 0 | if (! poolAppendChar(&parser->m_tempPool, *s)) |
6473 | 0 | return 0; |
6474 | 0 | needSep = XML_TRUE; |
6475 | 0 | } |
6476 | | |
6477 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
6478 | 0 | return NULL; |
6479 | 0 | return parser->m_tempPool.start; |
6480 | 0 | } |
6481 | | |
6482 | | static XML_Bool |
6483 | 0 | setContext(XML_Parser parser, const XML_Char *context) { |
6484 | 0 | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
6485 | 0 | const XML_Char *s = context; |
6486 | |
|
6487 | 0 | while (*context != XML_T('\0')) { |
6488 | 0 | if (*s == CONTEXT_SEP || *s == XML_T('\0')) { |
6489 | 0 | ENTITY *e; |
6490 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
6491 | 0 | return XML_FALSE; |
6492 | 0 | e = (ENTITY *)lookup(parser, &dtd->generalEntities, |
6493 | 0 | poolStart(&parser->m_tempPool), 0); |
6494 | 0 | if (e) |
6495 | 0 | e->open = XML_TRUE; |
6496 | 0 | if (*s != XML_T('\0')) |
6497 | 0 | s++; |
6498 | 0 | context = s; |
6499 | 0 | poolDiscard(&parser->m_tempPool); |
6500 | 0 | } else if (*s == XML_T(ASCII_EQUALS)) { |
6501 | 0 | PREFIX *prefix; |
6502 | 0 | if (poolLength(&parser->m_tempPool) == 0) |
6503 | 0 | prefix = &dtd->defaultPrefix; |
6504 | 0 | else { |
6505 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
6506 | 0 | return XML_FALSE; |
6507 | 0 | prefix |
6508 | 0 | = (PREFIX *)lookup(parser, &dtd->prefixes, |
6509 | 0 | poolStart(&parser->m_tempPool), sizeof(PREFIX)); |
6510 | 0 | if (! prefix) |
6511 | 0 | return XML_FALSE; |
6512 | 0 | if (prefix->name == poolStart(&parser->m_tempPool)) { |
6513 | 0 | prefix->name = poolCopyString(&dtd->pool, prefix->name); |
6514 | 0 | if (! prefix->name) |
6515 | 0 | return XML_FALSE; |
6516 | 0 | } |
6517 | 0 | poolDiscard(&parser->m_tempPool); |
6518 | 0 | } |
6519 | 0 | for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); |
6520 | 0 | context++) |
6521 | 0 | if (! poolAppendChar(&parser->m_tempPool, *context)) |
6522 | 0 | return XML_FALSE; |
6523 | 0 | if (! poolAppendChar(&parser->m_tempPool, XML_T('\0'))) |
6524 | 0 | return XML_FALSE; |
6525 | 0 | if (addBinding(parser, prefix, NULL, poolStart(&parser->m_tempPool), |
6526 | 0 | &parser->m_inheritedBindings) |
6527 | 0 | != XML_ERROR_NONE) |
6528 | 0 | return XML_FALSE; |
6529 | 0 | poolDiscard(&parser->m_tempPool); |
6530 | 0 | if (*context != XML_T('\0')) |
6531 | 0 | ++context; |
6532 | 0 | s = context; |
6533 | 0 | } else { |
6534 | 0 | if (! poolAppendChar(&parser->m_tempPool, *s)) |
6535 | 0 | return XML_FALSE; |
6536 | 0 | s++; |
6537 | 0 | } |
6538 | 0 | } |
6539 | 0 | return XML_TRUE; |
6540 | 0 | } |
6541 | | |
6542 | | static void FASTCALL |
6543 | 656 | normalizePublicId(XML_Char *publicId) { |
6544 | 656 | XML_Char *p = publicId; |
6545 | 656 | XML_Char *s; |
6546 | 72.3k | for (s = publicId; *s; s++) { |
6547 | 71.7k | switch (*s) { |
6548 | 670 | case 0x20: |
6549 | 1.48k | case 0xD: |
6550 | 1.70k | case 0xA: |
6551 | 1.70k | if (p != publicId && p[-1] != 0x20) |
6552 | 532 | *p++ = 0x20; |
6553 | 1.70k | break; |
6554 | 70.0k | default: |
6555 | 70.0k | *p++ = *s; |
6556 | 71.7k | } |
6557 | 71.7k | } |
6558 | 656 | if (p != publicId && p[-1] == 0x20) |
6559 | 242 | --p; |
6560 | 656 | *p = XML_T('\0'); |
6561 | 656 | } |
6562 | | |
6563 | | static DTD * |
6564 | 7.48k | dtdCreate(const XML_Memory_Handling_Suite *ms) { |
6565 | 7.48k | DTD *p = ms->malloc_fcn(sizeof(DTD)); |
6566 | 7.48k | if (p == NULL) |
6567 | 0 | return p; |
6568 | 7.48k | poolInit(&(p->pool), ms); |
6569 | 7.48k | poolInit(&(p->entityValuePool), ms); |
6570 | 7.48k | hashTableInit(&(p->generalEntities), ms); |
6571 | 7.48k | hashTableInit(&(p->elementTypes), ms); |
6572 | 7.48k | hashTableInit(&(p->attributeIds), ms); |
6573 | 7.48k | hashTableInit(&(p->prefixes), ms); |
6574 | | #ifdef XML_DTD |
6575 | | p->paramEntityRead = XML_FALSE; |
6576 | | hashTableInit(&(p->paramEntities), ms); |
6577 | | #endif /* XML_DTD */ |
6578 | 7.48k | p->defaultPrefix.name = NULL; |
6579 | 7.48k | p->defaultPrefix.binding = NULL; |
6580 | | |
6581 | 7.48k | p->in_eldecl = XML_FALSE; |
6582 | 7.48k | p->scaffIndex = NULL; |
6583 | 7.48k | p->scaffold = NULL; |
6584 | 7.48k | p->scaffLevel = 0; |
6585 | 7.48k | p->scaffSize = 0; |
6586 | 7.48k | p->scaffCount = 0; |
6587 | 7.48k | p->contentStringLen = 0; |
6588 | | |
6589 | 7.48k | p->keepProcessing = XML_TRUE; |
6590 | 7.48k | p->hasParamEntityRefs = XML_FALSE; |
6591 | 7.48k | p->standalone = XML_FALSE; |
6592 | 7.48k | return p; |
6593 | 7.48k | } |
6594 | | |
6595 | | static void |
6596 | 0 | dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms) { |
6597 | 0 | HASH_TABLE_ITER iter; |
6598 | 0 | hashTableIterInit(&iter, &(p->elementTypes)); |
6599 | 0 | for (;;) { |
6600 | 0 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); |
6601 | 0 | if (! e) |
6602 | 0 | break; |
6603 | 0 | if (e->allocDefaultAtts != 0) |
6604 | 0 | ms->free_fcn(e->defaultAtts); |
6605 | 0 | } |
6606 | 0 | hashTableClear(&(p->generalEntities)); |
6607 | | #ifdef XML_DTD |
6608 | | p->paramEntityRead = XML_FALSE; |
6609 | | hashTableClear(&(p->paramEntities)); |
6610 | | #endif /* XML_DTD */ |
6611 | 0 | hashTableClear(&(p->elementTypes)); |
6612 | 0 | hashTableClear(&(p->attributeIds)); |
6613 | 0 | hashTableClear(&(p->prefixes)); |
6614 | 0 | poolClear(&(p->pool)); |
6615 | 0 | poolClear(&(p->entityValuePool)); |
6616 | 0 | p->defaultPrefix.name = NULL; |
6617 | 0 | p->defaultPrefix.binding = NULL; |
6618 | |
|
6619 | 0 | p->in_eldecl = XML_FALSE; |
6620 | |
|
6621 | 0 | ms->free_fcn(p->scaffIndex); |
6622 | 0 | p->scaffIndex = NULL; |
6623 | 0 | ms->free_fcn(p->scaffold); |
6624 | 0 | p->scaffold = NULL; |
6625 | |
|
6626 | 0 | p->scaffLevel = 0; |
6627 | 0 | p->scaffSize = 0; |
6628 | 0 | p->scaffCount = 0; |
6629 | 0 | p->contentStringLen = 0; |
6630 | |
|
6631 | 0 | p->keepProcessing = XML_TRUE; |
6632 | 0 | p->hasParamEntityRefs = XML_FALSE; |
6633 | 0 | p->standalone = XML_FALSE; |
6634 | 0 | } |
6635 | | |
6636 | | static void |
6637 | 7.48k | dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms) { |
6638 | 7.48k | HASH_TABLE_ITER iter; |
6639 | 7.48k | hashTableIterInit(&iter, &(p->elementTypes)); |
6640 | 24.7k | for (;;) { |
6641 | 24.7k | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); |
6642 | 24.7k | if (! e) |
6643 | 7.48k | break; |
6644 | 17.3k | if (e->allocDefaultAtts != 0) |
6645 | 1.47k | ms->free_fcn(e->defaultAtts); |
6646 | 17.3k | } |
6647 | 7.48k | hashTableDestroy(&(p->generalEntities)); |
6648 | | #ifdef XML_DTD |
6649 | | hashTableDestroy(&(p->paramEntities)); |
6650 | | #endif /* XML_DTD */ |
6651 | 7.48k | hashTableDestroy(&(p->elementTypes)); |
6652 | 7.48k | hashTableDestroy(&(p->attributeIds)); |
6653 | 7.48k | hashTableDestroy(&(p->prefixes)); |
6654 | 7.48k | poolDestroy(&(p->pool)); |
6655 | 7.48k | poolDestroy(&(p->entityValuePool)); |
6656 | 7.48k | if (isDocEntity) { |
6657 | 7.48k | ms->free_fcn(p->scaffIndex); |
6658 | 7.48k | ms->free_fcn(p->scaffold); |
6659 | 7.48k | } |
6660 | 7.48k | ms->free_fcn(p); |
6661 | 7.48k | } |
6662 | | |
6663 | | /* Do a deep copy of the DTD. Return 0 for out of memory, non-zero otherwise. |
6664 | | The new DTD has already been initialized. |
6665 | | */ |
6666 | | static int |
6667 | | dtdCopy(XML_Parser oldParser, DTD *newDtd, const DTD *oldDtd, |
6668 | 0 | const XML_Memory_Handling_Suite *ms) { |
6669 | 0 | HASH_TABLE_ITER iter; |
6670 | | |
6671 | | /* Copy the prefix table. */ |
6672 | |
|
6673 | 0 | hashTableIterInit(&iter, &(oldDtd->prefixes)); |
6674 | 0 | for (;;) { |
6675 | 0 | const XML_Char *name; |
6676 | 0 | const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); |
6677 | 0 | if (! oldP) |
6678 | 0 | break; |
6679 | 0 | name = poolCopyString(&(newDtd->pool), oldP->name); |
6680 | 0 | if (! name) |
6681 | 0 | return 0; |
6682 | 0 | if (! lookup(oldParser, &(newDtd->prefixes), name, sizeof(PREFIX))) |
6683 | 0 | return 0; |
6684 | 0 | } |
6685 | | |
6686 | 0 | hashTableIterInit(&iter, &(oldDtd->attributeIds)); |
6687 | | |
6688 | | /* Copy the attribute id table. */ |
6689 | |
|
6690 | 0 | for (;;) { |
6691 | 0 | ATTRIBUTE_ID *newA; |
6692 | 0 | const XML_Char *name; |
6693 | 0 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); |
6694 | |
|
6695 | 0 | if (! oldA) |
6696 | 0 | break; |
6697 | | /* Remember to allocate the scratch byte before the name. */ |
6698 | 0 | if (! poolAppendChar(&(newDtd->pool), XML_T('\0'))) |
6699 | 0 | return 0; |
6700 | 0 | name = poolCopyString(&(newDtd->pool), oldA->name); |
6701 | 0 | if (! name) |
6702 | 0 | return 0; |
6703 | 0 | ++name; |
6704 | 0 | newA = (ATTRIBUTE_ID *)lookup(oldParser, &(newDtd->attributeIds), name, |
6705 | 0 | sizeof(ATTRIBUTE_ID)); |
6706 | 0 | if (! newA) |
6707 | 0 | return 0; |
6708 | 0 | newA->maybeTokenized = oldA->maybeTokenized; |
6709 | 0 | if (oldA->prefix) { |
6710 | 0 | newA->xmlns = oldA->xmlns; |
6711 | 0 | if (oldA->prefix == &oldDtd->defaultPrefix) |
6712 | 0 | newA->prefix = &newDtd->defaultPrefix; |
6713 | 0 | else |
6714 | 0 | newA->prefix = (PREFIX *)lookup(oldParser, &(newDtd->prefixes), |
6715 | 0 | oldA->prefix->name, 0); |
6716 | 0 | } |
6717 | 0 | } |
6718 | | |
6719 | | /* Copy the element type table. */ |
6720 | | |
6721 | 0 | hashTableIterInit(&iter, &(oldDtd->elementTypes)); |
6722 | |
|
6723 | 0 | for (;;) { |
6724 | 0 | int i; |
6725 | 0 | ELEMENT_TYPE *newE; |
6726 | 0 | const XML_Char *name; |
6727 | 0 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); |
6728 | 0 | if (! oldE) |
6729 | 0 | break; |
6730 | 0 | name = poolCopyString(&(newDtd->pool), oldE->name); |
6731 | 0 | if (! name) |
6732 | 0 | return 0; |
6733 | 0 | newE = (ELEMENT_TYPE *)lookup(oldParser, &(newDtd->elementTypes), name, |
6734 | 0 | sizeof(ELEMENT_TYPE)); |
6735 | 0 | if (! newE) |
6736 | 0 | return 0; |
6737 | 0 | if (oldE->nDefaultAtts) { |
6738 | 0 | newE->defaultAtts |
6739 | 0 | = ms->malloc_fcn(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); |
6740 | 0 | if (! newE->defaultAtts) { |
6741 | 0 | return 0; |
6742 | 0 | } |
6743 | 0 | } |
6744 | 0 | if (oldE->idAtt) |
6745 | 0 | newE->idAtt = (ATTRIBUTE_ID *)lookup(oldParser, &(newDtd->attributeIds), |
6746 | 0 | oldE->idAtt->name, 0); |
6747 | 0 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; |
6748 | 0 | if (oldE->prefix) |
6749 | 0 | newE->prefix = (PREFIX *)lookup(oldParser, &(newDtd->prefixes), |
6750 | 0 | oldE->prefix->name, 0); |
6751 | 0 | for (i = 0; i < newE->nDefaultAtts; i++) { |
6752 | 0 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *)lookup( |
6753 | 0 | oldParser, &(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); |
6754 | 0 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; |
6755 | 0 | if (oldE->defaultAtts[i].value) { |
6756 | 0 | newE->defaultAtts[i].value |
6757 | 0 | = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); |
6758 | 0 | if (! newE->defaultAtts[i].value) |
6759 | 0 | return 0; |
6760 | 0 | } else |
6761 | 0 | newE->defaultAtts[i].value = NULL; |
6762 | 0 | } |
6763 | 0 | } |
6764 | | |
6765 | | /* Copy the entity tables. */ |
6766 | 0 | if (! copyEntityTable(oldParser, &(newDtd->generalEntities), &(newDtd->pool), |
6767 | 0 | &(oldDtd->generalEntities))) |
6768 | 0 | return 0; |
6769 | | |
6770 | | #ifdef XML_DTD |
6771 | | if (! copyEntityTable(oldParser, &(newDtd->paramEntities), &(newDtd->pool), |
6772 | | &(oldDtd->paramEntities))) |
6773 | | return 0; |
6774 | | newDtd->paramEntityRead = oldDtd->paramEntityRead; |
6775 | | #endif /* XML_DTD */ |
6776 | | |
6777 | 0 | newDtd->keepProcessing = oldDtd->keepProcessing; |
6778 | 0 | newDtd->hasParamEntityRefs = oldDtd->hasParamEntityRefs; |
6779 | 0 | newDtd->standalone = oldDtd->standalone; |
6780 | | |
6781 | | /* Don't want deep copying for scaffolding */ |
6782 | 0 | newDtd->in_eldecl = oldDtd->in_eldecl; |
6783 | 0 | newDtd->scaffold = oldDtd->scaffold; |
6784 | 0 | newDtd->contentStringLen = oldDtd->contentStringLen; |
6785 | 0 | newDtd->scaffSize = oldDtd->scaffSize; |
6786 | 0 | newDtd->scaffLevel = oldDtd->scaffLevel; |
6787 | 0 | newDtd->scaffIndex = oldDtd->scaffIndex; |
6788 | |
|
6789 | 0 | return 1; |
6790 | 0 | } /* End dtdCopy */ |
6791 | | |
6792 | | static int |
6793 | | copyEntityTable(XML_Parser oldParser, HASH_TABLE *newTable, |
6794 | 0 | STRING_POOL *newPool, const HASH_TABLE *oldTable) { |
6795 | 0 | HASH_TABLE_ITER iter; |
6796 | 0 | const XML_Char *cachedOldBase = NULL; |
6797 | 0 | const XML_Char *cachedNewBase = NULL; |
6798 | |
|
6799 | 0 | hashTableIterInit(&iter, oldTable); |
6800 | |
|
6801 | 0 | for (;;) { |
6802 | 0 | ENTITY *newE; |
6803 | 0 | const XML_Char *name; |
6804 | 0 | const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); |
6805 | 0 | if (! oldE) |
6806 | 0 | break; |
6807 | 0 | name = poolCopyString(newPool, oldE->name); |
6808 | 0 | if (! name) |
6809 | 0 | return 0; |
6810 | 0 | newE = (ENTITY *)lookup(oldParser, newTable, name, sizeof(ENTITY)); |
6811 | 0 | if (! newE) |
6812 | 0 | return 0; |
6813 | 0 | if (oldE->systemId) { |
6814 | 0 | const XML_Char *tem = poolCopyString(newPool, oldE->systemId); |
6815 | 0 | if (! tem) |
6816 | 0 | return 0; |
6817 | 0 | newE->systemId = tem; |
6818 | 0 | if (oldE->base) { |
6819 | 0 | if (oldE->base == cachedOldBase) |
6820 | 0 | newE->base = cachedNewBase; |
6821 | 0 | else { |
6822 | 0 | cachedOldBase = oldE->base; |
6823 | 0 | tem = poolCopyString(newPool, cachedOldBase); |
6824 | 0 | if (! tem) |
6825 | 0 | return 0; |
6826 | 0 | cachedNewBase = newE->base = tem; |
6827 | 0 | } |
6828 | 0 | } |
6829 | 0 | if (oldE->publicId) { |
6830 | 0 | tem = poolCopyString(newPool, oldE->publicId); |
6831 | 0 | if (! tem) |
6832 | 0 | return 0; |
6833 | 0 | newE->publicId = tem; |
6834 | 0 | } |
6835 | 0 | } else { |
6836 | 0 | const XML_Char *tem |
6837 | 0 | = poolCopyStringN(newPool, oldE->textPtr, oldE->textLen); |
6838 | 0 | if (! tem) |
6839 | 0 | return 0; |
6840 | 0 | newE->textPtr = tem; |
6841 | 0 | newE->textLen = oldE->textLen; |
6842 | 0 | } |
6843 | 0 | if (oldE->notation) { |
6844 | 0 | const XML_Char *tem = poolCopyString(newPool, oldE->notation); |
6845 | 0 | if (! tem) |
6846 | 0 | return 0; |
6847 | 0 | newE->notation = tem; |
6848 | 0 | } |
6849 | 0 | newE->is_param = oldE->is_param; |
6850 | 0 | newE->is_internal = oldE->is_internal; |
6851 | 0 | } |
6852 | 0 | return 1; |
6853 | 0 | } |
6854 | | |
6855 | 15.2k | #define INIT_POWER 6 |
6856 | | |
6857 | | static XML_Bool FASTCALL |
6858 | 12.0M | keyeq(KEY s1, KEY s2) { |
6859 | 78.9M | for (; *s1 == *s2; s1++, s2++) |
6860 | 78.8M | if (*s1 == 0) |
6861 | 11.9M | return XML_TRUE; |
6862 | 94.8k | return XML_FALSE; |
6863 | 12.0M | } |
6864 | | |
6865 | | static size_t |
6866 | 11.9M | keylen(KEY s) { |
6867 | 11.9M | size_t len = 0; |
6868 | 106M | for (; *s; s++, len++) |
6869 | 94.2M | ; |
6870 | 11.9M | return len; |
6871 | 11.9M | } |
6872 | | |
6873 | | static void |
6874 | 11.9M | copy_salt_to_sipkey(XML_Parser parser, struct sipkey *key) { |
6875 | 11.9M | key->k[0] = 0; |
6876 | 11.9M | key->k[1] = get_hash_secret_salt(parser); |
6877 | 11.9M | } |
6878 | | |
6879 | | static unsigned long FASTCALL |
6880 | 11.9M | hash(XML_Parser parser, KEY s) { |
6881 | 11.9M | struct siphash state; |
6882 | 11.9M | struct sipkey key; |
6883 | 11.9M | (void)sip24_valid; |
6884 | 11.9M | copy_salt_to_sipkey(parser, &key); |
6885 | 11.9M | sip24_init(&state, &key); |
6886 | 11.9M | sip24_update(&state, s, keylen(s) * sizeof(XML_Char)); |
6887 | 11.9M | return (unsigned long)sip24_final(&state); |
6888 | 11.9M | } |
6889 | | |
6890 | | static NAMED * |
6891 | 11.9M | lookup(XML_Parser parser, HASH_TABLE *table, KEY name, size_t createSize) { |
6892 | 11.9M | size_t i; |
6893 | 11.9M | if (table->size == 0) { |
6894 | 11.4k | size_t tsize; |
6895 | 11.4k | if (! createSize) |
6896 | 3.85k | return NULL; |
6897 | 7.64k | table->power = INIT_POWER; |
6898 | | /* table->size is a power of 2 */ |
6899 | 7.64k | table->size = (size_t)1 << INIT_POWER; |
6900 | 7.64k | tsize = table->size * sizeof(NAMED *); |
6901 | 7.64k | table->v = table->mem->malloc_fcn(tsize); |
6902 | 7.64k | if (! table->v) { |
6903 | 0 | table->size = 0; |
6904 | 0 | return NULL; |
6905 | 0 | } |
6906 | 7.64k | memset(table->v, 0, tsize); |
6907 | 7.64k | i = hash(parser, name) & ((unsigned long)table->size - 1); |
6908 | 11.9M | } else { |
6909 | 11.9M | unsigned long h = hash(parser, name); |
6910 | 11.9M | unsigned long mask = (unsigned long)table->size - 1; |
6911 | 11.9M | unsigned char step = 0; |
6912 | 11.9M | i = h & mask; |
6913 | 12.0M | while (table->v[i]) { |
6914 | 12.0M | if (keyeq(name, table->v[i]->name)) |
6915 | 11.9M | return table->v[i]; |
6916 | 94.8k | if (! step) |
6917 | 89.1k | step = PROBE_STEP(h, mask, table->power); |
6918 | 94.8k | i < step ? (i += table->size - step) : (i -= step); |
6919 | 94.8k | } |
6920 | 39.6k | if (! createSize) |
6921 | 16.4k | return NULL; |
6922 | | |
6923 | | /* check for overflow (table is half full) */ |
6924 | 23.1k | if (table->used >> (table->power - 1)) { |
6925 | 317 | unsigned char newPower = table->power + 1; |
6926 | | |
6927 | | /* Detect and prevent invalid shift */ |
6928 | 317 | if (newPower >= sizeof(unsigned long) * 8 /* bits per byte */) { |
6929 | 0 | return NULL; |
6930 | 0 | } |
6931 | | |
6932 | 317 | size_t newSize = (size_t)1 << newPower; |
6933 | 317 | unsigned long newMask = (unsigned long)newSize - 1; |
6934 | | |
6935 | | /* Detect and prevent integer overflow */ |
6936 | 317 | if (newSize > (size_t)(-1) / sizeof(NAMED *)) { |
6937 | 0 | return NULL; |
6938 | 0 | } |
6939 | | |
6940 | 317 | size_t tsize = newSize * sizeof(NAMED *); |
6941 | 317 | NAMED **newV = table->mem->malloc_fcn(tsize); |
6942 | 317 | if (! newV) |
6943 | 0 | return NULL; |
6944 | 317 | memset(newV, 0, tsize); |
6945 | 40.3k | for (i = 0; i < table->size; i++) |
6946 | 40.0k | if (table->v[i]) { |
6947 | 20.0k | unsigned long newHash = hash(parser, table->v[i]->name); |
6948 | 20.0k | size_t j = newHash & newMask; |
6949 | 20.0k | step = 0; |
6950 | 22.8k | while (newV[j]) { |
6951 | 2.87k | if (! step) |
6952 | 2.33k | step = PROBE_STEP(newHash, newMask, newPower); |
6953 | 2.87k | j < step ? (j += newSize - step) : (j -= step); |
6954 | 2.87k | } |
6955 | 20.0k | newV[j] = table->v[i]; |
6956 | 20.0k | } |
6957 | 317 | table->mem->free_fcn(table->v); |
6958 | 317 | table->v = newV; |
6959 | 317 | table->power = newPower; |
6960 | 317 | table->size = newSize; |
6961 | 317 | i = h & newMask; |
6962 | 317 | step = 0; |
6963 | 412 | while (table->v[i]) { |
6964 | 95 | if (! step) |
6965 | 68 | step = PROBE_STEP(h, newMask, newPower); |
6966 | 95 | i < step ? (i += newSize - step) : (i -= step); |
6967 | 95 | } |
6968 | 317 | } |
6969 | 23.1k | } |
6970 | 30.8k | table->v[i] = table->mem->malloc_fcn(createSize); |
6971 | 30.8k | if (! table->v[i]) |
6972 | 0 | return NULL; |
6973 | 30.8k | memset(table->v[i], 0, createSize); |
6974 | 30.8k | table->v[i]->name = name; |
6975 | 30.8k | (table->used)++; |
6976 | 30.8k | return table->v[i]; |
6977 | 30.8k | } |
6978 | | |
6979 | | static void FASTCALL |
6980 | 0 | hashTableClear(HASH_TABLE *table) { |
6981 | 0 | size_t i; |
6982 | 0 | for (i = 0; i < table->size; i++) { |
6983 | 0 | table->mem->free_fcn(table->v[i]); |
6984 | 0 | table->v[i] = NULL; |
6985 | 0 | } |
6986 | 0 | table->used = 0; |
6987 | 0 | } |
6988 | | |
6989 | | static void FASTCALL |
6990 | 29.9k | hashTableDestroy(HASH_TABLE *table) { |
6991 | 29.9k | size_t i; |
6992 | 558k | for (i = 0; i < table->size; i++) |
6993 | 529k | table->mem->free_fcn(table->v[i]); |
6994 | 29.9k | table->mem->free_fcn(table->v); |
6995 | 29.9k | } |
6996 | | |
6997 | | static void FASTCALL |
6998 | 29.9k | hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms) { |
6999 | 29.9k | p->power = 0; |
7000 | 29.9k | p->size = 0; |
7001 | 29.9k | p->used = 0; |
7002 | 29.9k | p->v = NULL; |
7003 | 29.9k | p->mem = ms; |
7004 | 29.9k | } |
7005 | | |
7006 | | static void FASTCALL |
7007 | 7.48k | hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table) { |
7008 | 7.48k | iter->p = table->v; |
7009 | 7.48k | iter->end = iter->p ? iter->p + table->size : NULL; |
7010 | 7.48k | } |
7011 | | |
7012 | | static NAMED *FASTCALL |
7013 | 24.7k | hashTableIterNext(HASH_TABLE_ITER *iter) { |
7014 | 291k | while (iter->p != iter->end) { |
7015 | 284k | NAMED *tem = *(iter->p)++; |
7016 | 284k | if (tem) |
7017 | 17.3k | return tem; |
7018 | 284k | } |
7019 | 7.48k | return NULL; |
7020 | 24.7k | } |
7021 | | |
7022 | | static void FASTCALL |
7023 | 29.9k | poolInit(STRING_POOL *pool, const XML_Memory_Handling_Suite *ms) { |
7024 | 29.9k | pool->blocks = NULL; |
7025 | 29.9k | pool->freeBlocks = NULL; |
7026 | 29.9k | pool->start = NULL; |
7027 | 29.9k | pool->ptr = NULL; |
7028 | 29.9k | pool->end = NULL; |
7029 | 29.9k | pool->mem = ms; |
7030 | 29.9k | } |
7031 | | |
7032 | | static void FASTCALL |
7033 | 1.35M | poolClear(STRING_POOL *pool) { |
7034 | 1.35M | if (! pool->freeBlocks) |
7035 | 489k | pool->freeBlocks = pool->blocks; |
7036 | 867k | else { |
7037 | 867k | BLOCK *p = pool->blocks; |
7038 | 872k | while (p) { |
7039 | 5.50k | BLOCK *tem = p->next; |
7040 | 5.50k | p->next = pool->freeBlocks; |
7041 | 5.50k | pool->freeBlocks = p; |
7042 | 5.50k | p = tem; |
7043 | 5.50k | } |
7044 | 867k | } |
7045 | 1.35M | pool->blocks = NULL; |
7046 | 1.35M | pool->start = NULL; |
7047 | 1.35M | pool->ptr = NULL; |
7048 | 1.35M | pool->end = NULL; |
7049 | 1.35M | } |
7050 | | |
7051 | | static void FASTCALL |
7052 | 29.9k | poolDestroy(STRING_POOL *pool) { |
7053 | 29.9k | BLOCK *p = pool->blocks; |
7054 | 43.7k | while (p) { |
7055 | 13.8k | BLOCK *tem = p->next; |
7056 | 13.8k | pool->mem->free_fcn(p); |
7057 | 13.8k | p = tem; |
7058 | 13.8k | } |
7059 | 29.9k | p = pool->freeBlocks; |
7060 | 31.3k | while (p) { |
7061 | 1.41k | BLOCK *tem = p->next; |
7062 | 1.41k | pool->mem->free_fcn(p); |
7063 | 1.41k | p = tem; |
7064 | 1.41k | } |
7065 | 29.9k | } |
7066 | | |
7067 | | static XML_Char * |
7068 | | poolAppend(STRING_POOL *pool, const ENCODING *enc, const char *ptr, |
7069 | 18.5M | const char *end) { |
7070 | 18.5M | if (! pool->ptr && ! poolGrow(pool)) |
7071 | 0 | return NULL; |
7072 | 18.5M | for (;;) { |
7073 | 18.5M | const enum XML_Convert_Result convert_res = XmlConvert( |
7074 | 18.5M | enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); |
7075 | 18.5M | if ((convert_res == XML_CONVERT_COMPLETED) |
7076 | 18.5M | || (convert_res == XML_CONVERT_INPUT_INCOMPLETE)) |
7077 | 18.5M | break; |
7078 | 16.3k | if (! poolGrow(pool)) |
7079 | 2 | return NULL; |
7080 | 16.3k | } |
7081 | 18.5M | return pool->start; |
7082 | 18.5M | } |
7083 | | |
7084 | | static const XML_Char *FASTCALL |
7085 | 9.21k | poolCopyString(STRING_POOL *pool, const XML_Char *s) { |
7086 | 16.3M | do { |
7087 | 16.3M | if (! poolAppendChar(pool, *s)) |
7088 | 0 | return NULL; |
7089 | 16.3M | } while (*s++); |
7090 | 9.21k | s = pool->start; |
7091 | 9.21k | poolFinish(pool); |
7092 | 9.21k | return s; |
7093 | 9.21k | } |
7094 | | |
7095 | | static const XML_Char * |
7096 | 0 | poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) { |
7097 | 0 | if (! pool->ptr && ! poolGrow(pool)) { |
7098 | | /* The following line is unreachable given the current usage of |
7099 | | * poolCopyStringN(). Currently it is called from exactly one |
7100 | | * place to copy the text of a simple general entity. By that |
7101 | | * point, the name of the entity is already stored in the pool, so |
7102 | | * pool->ptr cannot be NULL. |
7103 | | * |
7104 | | * If poolCopyStringN() is used elsewhere as it well might be, |
7105 | | * this line may well become executable again. Regardless, this |
7106 | | * sort of check shouldn't be removed lightly, so we just exclude |
7107 | | * it from the coverage statistics. |
7108 | | */ |
7109 | 0 | return NULL; /* LCOV_EXCL_LINE */ |
7110 | 0 | } |
7111 | 0 | for (; n > 0; --n, s++) { |
7112 | 0 | if (! poolAppendChar(pool, *s)) |
7113 | 0 | return NULL; |
7114 | 0 | } |
7115 | 0 | s = pool->start; |
7116 | 0 | poolFinish(pool); |
7117 | 0 | return s; |
7118 | 0 | } |
7119 | | |
7120 | | static const XML_Char *FASTCALL |
7121 | 0 | poolAppendString(STRING_POOL *pool, const XML_Char *s) { |
7122 | 0 | while (*s) { |
7123 | 0 | if (! poolAppendChar(pool, *s)) |
7124 | 0 | return NULL; |
7125 | 0 | s++; |
7126 | 0 | } |
7127 | 0 | return pool->start; |
7128 | 0 | } |
7129 | | |
7130 | | static XML_Char * |
7131 | | poolStoreString(STRING_POOL *pool, const ENCODING *enc, const char *ptr, |
7132 | 10.6M | const char *end) { |
7133 | 10.6M | if (! poolAppend(pool, enc, ptr, end)) |
7134 | 0 | return NULL; |
7135 | 10.6M | if (pool->ptr == pool->end && ! poolGrow(pool)) |
7136 | 0 | return NULL; |
7137 | 10.6M | *(pool->ptr)++ = 0; |
7138 | 10.6M | return pool->start; |
7139 | 10.6M | } |
7140 | | |
7141 | | static size_t |
7142 | 24.6k | poolBytesToAllocateFor(int blockSize) { |
7143 | | /* Unprotected math would be: |
7144 | | ** return offsetof(BLOCK, s) + blockSize * sizeof(XML_Char); |
7145 | | ** |
7146 | | ** Detect overflow, avoiding _signed_ overflow undefined behavior |
7147 | | ** For a + b * c we check b * c in isolation first, so that addition of a |
7148 | | ** on top has no chance of making us accept a small non-negative number |
7149 | | */ |
7150 | 24.6k | const size_t stretch = sizeof(XML_Char); /* can be 4 bytes */ |
7151 | | |
7152 | 24.6k | if (blockSize <= 0) |
7153 | 0 | return 0; |
7154 | | |
7155 | 24.6k | if (blockSize > (int)(INT_MAX / stretch)) |
7156 | 0 | return 0; |
7157 | | |
7158 | 24.6k | { |
7159 | 24.6k | const int stretchedBlockSize = blockSize * (int)stretch; |
7160 | 24.6k | const int bytesToAllocate |
7161 | 24.6k | = (int)(offsetof(BLOCK, s) + (unsigned)stretchedBlockSize); |
7162 | 24.6k | if (bytesToAllocate < 0) |
7163 | 1 | return 0; |
7164 | | |
7165 | 24.6k | return (size_t)bytesToAllocate; |
7166 | 24.6k | } |
7167 | 24.6k | } |
7168 | | |
7169 | | static XML_Bool FASTCALL |
7170 | 45.9k | poolGrow(STRING_POOL *pool) { |
7171 | 45.9k | if (pool->freeBlocks) { |
7172 | 21.3k | if (pool->start == 0) { |
7173 | 18.5k | pool->blocks = pool->freeBlocks; |
7174 | 18.5k | pool->freeBlocks = pool->freeBlocks->next; |
7175 | 18.5k | pool->blocks->next = NULL; |
7176 | 18.5k | pool->start = pool->blocks->s; |
7177 | 18.5k | pool->end = pool->start + pool->blocks->size; |
7178 | 18.5k | pool->ptr = pool->start; |
7179 | 18.5k | return XML_TRUE; |
7180 | 18.5k | } |
7181 | 2.82k | if (pool->end - pool->start < pool->freeBlocks->size) { |
7182 | 2.65k | BLOCK *tem = pool->freeBlocks->next; |
7183 | 2.65k | pool->freeBlocks->next = pool->blocks; |
7184 | 2.65k | pool->blocks = pool->freeBlocks; |
7185 | 2.65k | pool->freeBlocks = tem; |
7186 | 2.65k | memcpy(pool->blocks->s, pool->start, |
7187 | 2.65k | (pool->end - pool->start) * sizeof(XML_Char)); |
7188 | 2.65k | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); |
7189 | 2.65k | pool->start = pool->blocks->s; |
7190 | 2.65k | pool->end = pool->start + pool->blocks->size; |
7191 | 2.65k | return XML_TRUE; |
7192 | 2.65k | } |
7193 | 2.82k | } |
7194 | 24.6k | if (pool->blocks && pool->start == pool->blocks->s) { |
7195 | 9.45k | BLOCK *temp; |
7196 | 9.45k | int blockSize = (int)((unsigned)(pool->end - pool->start) * 2U); |
7197 | 9.45k | size_t bytesToAllocate; |
7198 | | |
7199 | | /* NOTE: Needs to be calculated prior to calling `realloc` |
7200 | | to avoid dangling pointers: */ |
7201 | 9.45k | const ptrdiff_t offsetInsideBlock = pool->ptr - pool->start; |
7202 | | |
7203 | 9.45k | if (blockSize < 0) { |
7204 | | /* This condition traps a situation where either more than |
7205 | | * INT_MAX/2 bytes have already been allocated. This isn't |
7206 | | * readily testable, since it is unlikely that an average |
7207 | | * machine will have that much memory, so we exclude it from the |
7208 | | * coverage statistics. |
7209 | | */ |
7210 | 2 | return XML_FALSE; /* LCOV_EXCL_LINE */ |
7211 | 2 | } |
7212 | | |
7213 | 9.45k | bytesToAllocate = poolBytesToAllocateFor(blockSize); |
7214 | 9.45k | if (bytesToAllocate == 0) |
7215 | 1 | return XML_FALSE; |
7216 | | |
7217 | 9.45k | temp = (BLOCK *)pool->mem->realloc_fcn(pool->blocks, |
7218 | 9.45k | (unsigned)bytesToAllocate); |
7219 | 9.45k | if (temp == NULL) |
7220 | 0 | return XML_FALSE; |
7221 | 9.45k | pool->blocks = temp; |
7222 | 9.45k | pool->blocks->size = blockSize; |
7223 | 9.45k | pool->ptr = pool->blocks->s + offsetInsideBlock; |
7224 | 9.45k | pool->start = pool->blocks->s; |
7225 | 9.45k | pool->end = pool->start + blockSize; |
7226 | 15.2k | } else { |
7227 | 15.2k | BLOCK *tem; |
7228 | 15.2k | int blockSize = (int)(pool->end - pool->start); |
7229 | 15.2k | size_t bytesToAllocate; |
7230 | | |
7231 | 15.2k | if (blockSize < 0) { |
7232 | | /* This condition traps a situation where either more than |
7233 | | * INT_MAX bytes have already been allocated (which is prevented |
7234 | | * by various pieces of program logic, not least this one, never |
7235 | | * mind the unlikelihood of actually having that much memory) or |
7236 | | * the pool control fields have been corrupted (which could |
7237 | | * conceivably happen in an extremely buggy user handler |
7238 | | * function). Either way it isn't readily testable, so we |
7239 | | * exclude it from the coverage statistics. |
7240 | | */ |
7241 | 0 | return XML_FALSE; /* LCOV_EXCL_LINE */ |
7242 | 0 | } |
7243 | | |
7244 | 15.2k | if (blockSize < INIT_BLOCK_SIZE) |
7245 | 14.7k | blockSize = INIT_BLOCK_SIZE; |
7246 | 530 | else { |
7247 | | /* Detect overflow, avoiding _signed_ overflow undefined behavior */ |
7248 | 530 | if ((int)((unsigned)blockSize * 2U) < 0) { |
7249 | 0 | return XML_FALSE; |
7250 | 0 | } |
7251 | 530 | blockSize *= 2; |
7252 | 530 | } |
7253 | | |
7254 | 15.2k | bytesToAllocate = poolBytesToAllocateFor(blockSize); |
7255 | 15.2k | if (bytesToAllocate == 0) |
7256 | 0 | return XML_FALSE; |
7257 | | |
7258 | 15.2k | tem = pool->mem->malloc_fcn(bytesToAllocate); |
7259 | 15.2k | if (! tem) |
7260 | 0 | return XML_FALSE; |
7261 | 15.2k | tem->size = blockSize; |
7262 | 15.2k | tem->next = pool->blocks; |
7263 | 15.2k | pool->blocks = tem; |
7264 | 15.2k | if (pool->ptr != pool->start) |
7265 | 6.55k | memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char)); |
7266 | 15.2k | pool->ptr = tem->s + (pool->ptr - pool->start); |
7267 | 15.2k | pool->start = tem->s; |
7268 | 15.2k | pool->end = tem->s + blockSize; |
7269 | 15.2k | } |
7270 | 24.6k | return XML_TRUE; |
7271 | 24.6k | } |
7272 | | |
7273 | | static int FASTCALL |
7274 | 0 | nextScaffoldPart(XML_Parser parser) { |
7275 | 0 | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
7276 | 0 | CONTENT_SCAFFOLD *me; |
7277 | 0 | int next; |
7278 | |
|
7279 | 0 | if (! dtd->scaffIndex) { |
7280 | 0 | dtd->scaffIndex = (int *)MALLOC(parser, parser->m_groupSize * sizeof(int)); |
7281 | 0 | if (! dtd->scaffIndex) |
7282 | 0 | return -1; |
7283 | 0 | dtd->scaffIndex[0] = 0; |
7284 | 0 | } |
7285 | | |
7286 | 0 | if (dtd->scaffCount >= dtd->scaffSize) { |
7287 | 0 | CONTENT_SCAFFOLD *temp; |
7288 | 0 | if (dtd->scaffold) { |
7289 | | /* Detect and prevent integer overflow */ |
7290 | 0 | if (dtd->scaffSize > UINT_MAX / 2u) { |
7291 | 0 | return -1; |
7292 | 0 | } |
7293 | | /* Detect and prevent integer overflow. |
7294 | | * The preprocessor guard addresses the "always false" warning |
7295 | | * from -Wtype-limits on platforms where |
7296 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
7297 | | #if UINT_MAX >= SIZE_MAX |
7298 | | if (dtd->scaffSize > (size_t)(-1) / 2u / sizeof(CONTENT_SCAFFOLD)) { |
7299 | | return -1; |
7300 | | } |
7301 | | #endif |
7302 | | |
7303 | 0 | temp = (CONTENT_SCAFFOLD *)REALLOC( |
7304 | 0 | parser, dtd->scaffold, dtd->scaffSize * 2 * sizeof(CONTENT_SCAFFOLD)); |
7305 | 0 | if (temp == NULL) |
7306 | 0 | return -1; |
7307 | 0 | dtd->scaffSize *= 2; |
7308 | 0 | } else { |
7309 | 0 | temp = (CONTENT_SCAFFOLD *)MALLOC(parser, INIT_SCAFFOLD_ELEMENTS |
7310 | 0 | * sizeof(CONTENT_SCAFFOLD)); |
7311 | 0 | if (temp == NULL) |
7312 | 0 | return -1; |
7313 | 0 | dtd->scaffSize = INIT_SCAFFOLD_ELEMENTS; |
7314 | 0 | } |
7315 | 0 | dtd->scaffold = temp; |
7316 | 0 | } |
7317 | 0 | next = dtd->scaffCount++; |
7318 | 0 | me = &dtd->scaffold[next]; |
7319 | 0 | if (dtd->scaffLevel) { |
7320 | 0 | CONTENT_SCAFFOLD *parent |
7321 | 0 | = &dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]]; |
7322 | 0 | if (parent->lastchild) { |
7323 | 0 | dtd->scaffold[parent->lastchild].nextsib = next; |
7324 | 0 | } |
7325 | 0 | if (! parent->childcnt) |
7326 | 0 | parent->firstchild = next; |
7327 | 0 | parent->lastchild = next; |
7328 | 0 | parent->childcnt++; |
7329 | 0 | } |
7330 | 0 | me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0; |
7331 | 0 | return next; |
7332 | 0 | } |
7333 | | |
7334 | | static XML_Content * |
7335 | 0 | build_model(XML_Parser parser) { |
7336 | | /* Function build_model transforms the existing parser->m_dtd->scaffold |
7337 | | * array of CONTENT_SCAFFOLD tree nodes into a new array of |
7338 | | * XML_Content tree nodes followed by a gapless list of zero-terminated |
7339 | | * strings. */ |
7340 | 0 | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
7341 | 0 | XML_Content *ret; |
7342 | 0 | XML_Char *str; /* the current string writing location */ |
7343 | | |
7344 | | /* Detect and prevent integer overflow. |
7345 | | * The preprocessor guard addresses the "always false" warning |
7346 | | * from -Wtype-limits on platforms where |
7347 | | * sizeof(unsigned int) < sizeof(size_t), e.g. on x86_64. */ |
7348 | | #if UINT_MAX >= SIZE_MAX |
7349 | | if (dtd->scaffCount > (size_t)(-1) / sizeof(XML_Content)) { |
7350 | | return NULL; |
7351 | | } |
7352 | | if (dtd->contentStringLen > (size_t)(-1) / sizeof(XML_Char)) { |
7353 | | return NULL; |
7354 | | } |
7355 | | #endif |
7356 | 0 | if (dtd->scaffCount * sizeof(XML_Content) |
7357 | 0 | > (size_t)(-1) - dtd->contentStringLen * sizeof(XML_Char)) { |
7358 | 0 | return NULL; |
7359 | 0 | } |
7360 | | |
7361 | 0 | const size_t allocsize = (dtd->scaffCount * sizeof(XML_Content) |
7362 | 0 | + (dtd->contentStringLen * sizeof(XML_Char))); |
7363 | |
|
7364 | 0 | ret = (XML_Content *)MALLOC(parser, allocsize); |
7365 | 0 | if (! ret) |
7366 | 0 | return NULL; |
7367 | | |
7368 | | /* What follows is an iterative implementation (of what was previously done |
7369 | | * recursively in a dedicated function called "build_node". The old recursive |
7370 | | * build_node could be forced into stack exhaustion from input as small as a |
7371 | | * few megabyte, and so that was a security issue. Hence, a function call |
7372 | | * stack is avoided now by resolving recursion.) |
7373 | | * |
7374 | | * The iterative approach works as follows: |
7375 | | * |
7376 | | * - We have two writing pointers, both walking up the result array; one does |
7377 | | * the work, the other creates "jobs" for its colleague to do, and leads |
7378 | | * the way: |
7379 | | * |
7380 | | * - The faster one, pointer jobDest, always leads and writes "what job |
7381 | | * to do" by the other, once they reach that place in the |
7382 | | * array: leader "jobDest" stores the source node array index (relative |
7383 | | * to array dtd->scaffold) in field "numchildren". |
7384 | | * |
7385 | | * - The slower one, pointer dest, looks at the value stored in the |
7386 | | * "numchildren" field (which actually holds a source node array index |
7387 | | * at that time) and puts the real data from dtd->scaffold in. |
7388 | | * |
7389 | | * - Before the loop starts, jobDest writes source array index 0 |
7390 | | * (where the root node is located) so that dest will have something to do |
7391 | | * when it starts operation. |
7392 | | * |
7393 | | * - Whenever nodes with children are encountered, jobDest appends |
7394 | | * them as new jobs, in order. As a result, tree node siblings are |
7395 | | * adjacent in the resulting array, for example: |
7396 | | * |
7397 | | * [0] root, has two children |
7398 | | * [1] first child of 0, has three children |
7399 | | * [3] first child of 1, does not have children |
7400 | | * [4] second child of 1, does not have children |
7401 | | * [5] third child of 1, does not have children |
7402 | | * [2] second child of 0, does not have children |
7403 | | * |
7404 | | * Or (the same data) presented in flat array view: |
7405 | | * |
7406 | | * [0] root, has two children |
7407 | | * |
7408 | | * [1] first child of 0, has three children |
7409 | | * [2] second child of 0, does not have children |
7410 | | * |
7411 | | * [3] first child of 1, does not have children |
7412 | | * [4] second child of 1, does not have children |
7413 | | * [5] third child of 1, does not have children |
7414 | | * |
7415 | | * - The algorithm repeats until all target array indices have been processed. |
7416 | | */ |
7417 | 0 | XML_Content *dest = ret; /* tree node writing location, moves upwards */ |
7418 | 0 | XML_Content *const destLimit = &ret[dtd->scaffCount]; |
7419 | 0 | XML_Content *jobDest = ret; /* next free writing location in target array */ |
7420 | 0 | str = (XML_Char *)&ret[dtd->scaffCount]; |
7421 | | |
7422 | | /* Add the starting job, the root node (index 0) of the source tree */ |
7423 | 0 | (jobDest++)->numchildren = 0; |
7424 | |
|
7425 | 0 | for (; dest < destLimit; dest++) { |
7426 | | /* Retrieve source tree array index from job storage */ |
7427 | 0 | const int src_node = (int)dest->numchildren; |
7428 | | |
7429 | | /* Convert item */ |
7430 | 0 | dest->type = dtd->scaffold[src_node].type; |
7431 | 0 | dest->quant = dtd->scaffold[src_node].quant; |
7432 | 0 | if (dest->type == XML_CTYPE_NAME) { |
7433 | 0 | const XML_Char *src; |
7434 | 0 | dest->name = str; |
7435 | 0 | src = dtd->scaffold[src_node].name; |
7436 | 0 | for (;;) { |
7437 | 0 | *str++ = *src; |
7438 | 0 | if (! *src) |
7439 | 0 | break; |
7440 | 0 | src++; |
7441 | 0 | } |
7442 | 0 | dest->numchildren = 0; |
7443 | 0 | dest->children = NULL; |
7444 | 0 | } else { |
7445 | 0 | unsigned int i; |
7446 | 0 | int cn; |
7447 | 0 | dest->name = NULL; |
7448 | 0 | dest->numchildren = dtd->scaffold[src_node].childcnt; |
7449 | 0 | dest->children = jobDest; |
7450 | | |
7451 | | /* Append scaffold indices of children to array */ |
7452 | 0 | for (i = 0, cn = dtd->scaffold[src_node].firstchild; |
7453 | 0 | i < dest->numchildren; i++, cn = dtd->scaffold[cn].nextsib) |
7454 | 0 | (jobDest++)->numchildren = (unsigned int)cn; |
7455 | 0 | } |
7456 | 0 | } |
7457 | |
|
7458 | 0 | return ret; |
7459 | 0 | } |
7460 | | |
7461 | | static ELEMENT_TYPE * |
7462 | | getElementType(XML_Parser parser, const ENCODING *enc, const char *ptr, |
7463 | 10.2k | const char *end) { |
7464 | 10.2k | DTD *const dtd = parser->m_dtd; /* save one level of indirection */ |
7465 | 10.2k | const XML_Char *name = poolStoreString(&dtd->pool, enc, ptr, end); |
7466 | 10.2k | ELEMENT_TYPE *ret; |
7467 | | |
7468 | 10.2k | if (! name) |
7469 | 0 | return NULL; |
7470 | 10.2k | ret = (ELEMENT_TYPE *)lookup(parser, &dtd->elementTypes, name, |
7471 | 10.2k | sizeof(ELEMENT_TYPE)); |
7472 | 10.2k | if (! ret) |
7473 | 0 | return NULL; |
7474 | 10.2k | if (ret->name != name) |
7475 | 2.14k | poolDiscard(&dtd->pool); |
7476 | 8.09k | else { |
7477 | 8.09k | poolFinish(&dtd->pool); |
7478 | 8.09k | if (! setElementTypePrefix(parser, ret)) |
7479 | 0 | return NULL; |
7480 | 8.09k | } |
7481 | 10.2k | return ret; |
7482 | 10.2k | } |
7483 | | |
7484 | | static XML_Char * |
7485 | 0 | copyString(const XML_Char *s, const XML_Memory_Handling_Suite *memsuite) { |
7486 | 0 | size_t charsRequired = 0; |
7487 | 0 | XML_Char *result; |
7488 | | |
7489 | | /* First determine how long the string is */ |
7490 | 0 | while (s[charsRequired] != 0) { |
7491 | 0 | charsRequired++; |
7492 | 0 | } |
7493 | | /* Include the terminator */ |
7494 | 0 | charsRequired++; |
7495 | | |
7496 | | /* Now allocate space for the copy */ |
7497 | 0 | result = memsuite->malloc_fcn(charsRequired * sizeof(XML_Char)); |
7498 | 0 | if (result == NULL) |
7499 | 0 | return NULL; |
7500 | | /* Copy the original into place */ |
7501 | 0 | memcpy(result, s, charsRequired * sizeof(XML_Char)); |
7502 | 0 | return result; |
7503 | 0 | } |
7504 | | |
7505 | | #ifdef XML_DTD |
7506 | | |
7507 | | static float |
7508 | | accountingGetCurrentAmplification(XML_Parser rootParser) { |
7509 | | const XmlBigCount countBytesOutput |
7510 | | = rootParser->m_accounting.countBytesDirect |
7511 | | + rootParser->m_accounting.countBytesIndirect; |
7512 | | const float amplificationFactor |
7513 | | = rootParser->m_accounting.countBytesDirect |
7514 | | ? (countBytesOutput |
7515 | | / (float)(rootParser->m_accounting.countBytesDirect)) |
7516 | | : 1.0f; |
7517 | | assert(! rootParser->m_parentParser); |
7518 | | return amplificationFactor; |
7519 | | } |
7520 | | |
7521 | | static void |
7522 | | accountingReportStats(XML_Parser originParser, const char *epilog) { |
7523 | | const XML_Parser rootParser = getRootParserOf(originParser, NULL); |
7524 | | assert(! rootParser->m_parentParser); |
7525 | | |
7526 | | if (rootParser->m_accounting.debugLevel < 1) { |
7527 | | return; |
7528 | | } |
7529 | | |
7530 | | const float amplificationFactor |
7531 | | = accountingGetCurrentAmplification(rootParser); |
7532 | | fprintf(stderr, |
7533 | | "expat: Accounting(%p): Direct " EXPAT_FMT_ULL( |
7534 | | "10") ", indirect " EXPAT_FMT_ULL("10") ", amplification %8.2f%s", |
7535 | | (void *)rootParser, rootParser->m_accounting.countBytesDirect, |
7536 | | rootParser->m_accounting.countBytesIndirect, |
7537 | | (double)amplificationFactor, epilog); |
7538 | | } |
7539 | | |
7540 | | static void |
7541 | | accountingOnAbort(XML_Parser originParser) { |
7542 | | accountingReportStats(originParser, " ABORTING\n"); |
7543 | | } |
7544 | | |
7545 | | static void |
7546 | | accountingReportDiff(XML_Parser rootParser, |
7547 | | unsigned int levelsAwayFromRootParser, const char *before, |
7548 | | const char *after, ptrdiff_t bytesMore, int source_line, |
7549 | | enum XML_Account account) { |
7550 | | assert(! rootParser->m_parentParser); |
7551 | | |
7552 | | fprintf(stderr, |
7553 | | " (+" EXPAT_FMT_PTRDIFF_T("6") " bytes %s|%d, xmlparse.c:%d) %*s\"", |
7554 | | bytesMore, (account == XML_ACCOUNT_DIRECT) ? "DIR" : "EXP", |
7555 | | levelsAwayFromRootParser, source_line, 10, ""); |
7556 | | |
7557 | | const char ellipis[] = "[..]"; |
7558 | | const size_t ellipsisLength = sizeof(ellipis) /* because compile-time */ - 1; |
7559 | | const unsigned int contextLength = 10; |
7560 | | |
7561 | | /* Note: Performance is of no concern here */ |
7562 | | const char *walker = before; |
7563 | | if ((rootParser->m_accounting.debugLevel >= 3) |
7564 | | || (after - before) |
7565 | | <= (ptrdiff_t)(contextLength + ellipsisLength + contextLength)) { |
7566 | | for (; walker < after; walker++) { |
7567 | | fprintf(stderr, "%s", unsignedCharToPrintable(walker[0])); |
7568 | | } |
7569 | | } else { |
7570 | | for (; walker < before + contextLength; walker++) { |
7571 | | fprintf(stderr, "%s", unsignedCharToPrintable(walker[0])); |
7572 | | } |
7573 | | fprintf(stderr, ellipis); |
7574 | | walker = after - contextLength; |
7575 | | for (; walker < after; walker++) { |
7576 | | fprintf(stderr, "%s", unsignedCharToPrintable(walker[0])); |
7577 | | } |
7578 | | } |
7579 | | fprintf(stderr, "\"\n"); |
7580 | | } |
7581 | | |
7582 | | static XML_Bool |
7583 | | accountingDiffTolerated(XML_Parser originParser, int tok, const char *before, |
7584 | | const char *after, int source_line, |
7585 | | enum XML_Account account) { |
7586 | | /* Note: We need to check the token type *first* to be sure that |
7587 | | * we can even access variable <after>, safely. |
7588 | | * E.g. for XML_TOK_NONE <after> may hold an invalid pointer. */ |
7589 | | switch (tok) { |
7590 | | case XML_TOK_INVALID: |
7591 | | case XML_TOK_PARTIAL: |
7592 | | case XML_TOK_PARTIAL_CHAR: |
7593 | | case XML_TOK_NONE: |
7594 | | return XML_TRUE; |
7595 | | } |
7596 | | |
7597 | | if (account == XML_ACCOUNT_NONE) |
7598 | | return XML_TRUE; /* because these bytes have been accounted for, already */ |
7599 | | |
7600 | | unsigned int levelsAwayFromRootParser; |
7601 | | const XML_Parser rootParser |
7602 | | = getRootParserOf(originParser, &levelsAwayFromRootParser); |
7603 | | assert(! rootParser->m_parentParser); |
7604 | | |
7605 | | const int isDirect |
7606 | | = (account == XML_ACCOUNT_DIRECT) && (originParser == rootParser); |
7607 | | const ptrdiff_t bytesMore = after - before; |
7608 | | |
7609 | | XmlBigCount *const additionTarget |
7610 | | = isDirect ? &rootParser->m_accounting.countBytesDirect |
7611 | | : &rootParser->m_accounting.countBytesIndirect; |
7612 | | |
7613 | | /* Detect and avoid integer overflow */ |
7614 | | if (*additionTarget > (XmlBigCount)(-1) - (XmlBigCount)bytesMore) |
7615 | | return XML_FALSE; |
7616 | | *additionTarget += bytesMore; |
7617 | | |
7618 | | const XmlBigCount countBytesOutput |
7619 | | = rootParser->m_accounting.countBytesDirect |
7620 | | + rootParser->m_accounting.countBytesIndirect; |
7621 | | const float amplificationFactor |
7622 | | = accountingGetCurrentAmplification(rootParser); |
7623 | | const XML_Bool tolerated |
7624 | | = (countBytesOutput < rootParser->m_accounting.activationThresholdBytes) |
7625 | | || (amplificationFactor |
7626 | | <= rootParser->m_accounting.maximumAmplificationFactor); |
7627 | | |
7628 | | if (rootParser->m_accounting.debugLevel >= 2) { |
7629 | | accountingReportStats(rootParser, ""); |
7630 | | accountingReportDiff(rootParser, levelsAwayFromRootParser, before, after, |
7631 | | bytesMore, source_line, account); |
7632 | | } |
7633 | | |
7634 | | return tolerated; |
7635 | | } |
7636 | | |
7637 | | unsigned long long |
7638 | | testingAccountingGetCountBytesDirect(XML_Parser parser) { |
7639 | | if (! parser) |
7640 | | return 0; |
7641 | | return parser->m_accounting.countBytesDirect; |
7642 | | } |
7643 | | |
7644 | | unsigned long long |
7645 | | testingAccountingGetCountBytesIndirect(XML_Parser parser) { |
7646 | | if (! parser) |
7647 | | return 0; |
7648 | | return parser->m_accounting.countBytesIndirect; |
7649 | | } |
7650 | | |
7651 | | static void |
7652 | | entityTrackingReportStats(XML_Parser rootParser, ENTITY *entity, |
7653 | | const char *action, int sourceLine) { |
7654 | | assert(! rootParser->m_parentParser); |
7655 | | if (rootParser->m_entity_stats.debugLevel < 1) |
7656 | | return; |
7657 | | |
7658 | | # if defined(XML_UNICODE) |
7659 | | const char *const entityName = "[..]"; |
7660 | | # else |
7661 | | const char *const entityName = entity->name; |
7662 | | # endif |
7663 | | |
7664 | | fprintf( |
7665 | | stderr, |
7666 | | "expat: Entities(%p): Count %9d, depth %2d/%2d %*s%s%s; %s length %d (xmlparse.c:%d)\n", |
7667 | | (void *)rootParser, rootParser->m_entity_stats.countEverOpened, |
7668 | | rootParser->m_entity_stats.currentDepth, |
7669 | | rootParser->m_entity_stats.maximumDepthSeen, |
7670 | | (rootParser->m_entity_stats.currentDepth - 1) * 2, "", |
7671 | | entity->is_param ? "%" : "&", entityName, action, entity->textLen, |
7672 | | sourceLine); |
7673 | | } |
7674 | | |
7675 | | static void |
7676 | | entityTrackingOnOpen(XML_Parser originParser, ENTITY *entity, int sourceLine) { |
7677 | | const XML_Parser rootParser = getRootParserOf(originParser, NULL); |
7678 | | assert(! rootParser->m_parentParser); |
7679 | | |
7680 | | rootParser->m_entity_stats.countEverOpened++; |
7681 | | rootParser->m_entity_stats.currentDepth++; |
7682 | | if (rootParser->m_entity_stats.currentDepth |
7683 | | > rootParser->m_entity_stats.maximumDepthSeen) { |
7684 | | rootParser->m_entity_stats.maximumDepthSeen++; |
7685 | | } |
7686 | | |
7687 | | entityTrackingReportStats(rootParser, entity, "OPEN ", sourceLine); |
7688 | | } |
7689 | | |
7690 | | static void |
7691 | | entityTrackingOnClose(XML_Parser originParser, ENTITY *entity, int sourceLine) { |
7692 | | const XML_Parser rootParser = getRootParserOf(originParser, NULL); |
7693 | | assert(! rootParser->m_parentParser); |
7694 | | |
7695 | | entityTrackingReportStats(rootParser, entity, "CLOSE", sourceLine); |
7696 | | rootParser->m_entity_stats.currentDepth--; |
7697 | | } |
7698 | | |
7699 | | static XML_Parser |
7700 | | getRootParserOf(XML_Parser parser, unsigned int *outLevelDiff) { |
7701 | | XML_Parser rootParser = parser; |
7702 | | unsigned int stepsTakenUpwards = 0; |
7703 | | while (rootParser->m_parentParser) { |
7704 | | rootParser = rootParser->m_parentParser; |
7705 | | stepsTakenUpwards++; |
7706 | | } |
7707 | | assert(! rootParser->m_parentParser); |
7708 | | if (outLevelDiff != NULL) { |
7709 | | *outLevelDiff = stepsTakenUpwards; |
7710 | | } |
7711 | | return rootParser; |
7712 | | } |
7713 | | |
7714 | | const char * |
7715 | | unsignedCharToPrintable(unsigned char c) { |
7716 | | switch (c) { |
7717 | | case 0: |
7718 | | return "\\0"; |
7719 | | case 1: |
7720 | | return "\\x1"; |
7721 | | case 2: |
7722 | | return "\\x2"; |
7723 | | case 3: |
7724 | | return "\\x3"; |
7725 | | case 4: |
7726 | | return "\\x4"; |
7727 | | case 5: |
7728 | | return "\\x5"; |
7729 | | case 6: |
7730 | | return "\\x6"; |
7731 | | case 7: |
7732 | | return "\\x7"; |
7733 | | case 8: |
7734 | | return "\\x8"; |
7735 | | case 9: |
7736 | | return "\\t"; |
7737 | | case 10: |
7738 | | return "\\n"; |
7739 | | case 11: |
7740 | | return "\\xB"; |
7741 | | case 12: |
7742 | | return "\\xC"; |
7743 | | case 13: |
7744 | | return "\\r"; |
7745 | | case 14: |
7746 | | return "\\xE"; |
7747 | | case 15: |
7748 | | return "\\xF"; |
7749 | | case 16: |
7750 | | return "\\x10"; |
7751 | | case 17: |
7752 | | return "\\x11"; |
7753 | | case 18: |
7754 | | return "\\x12"; |
7755 | | case 19: |
7756 | | return "\\x13"; |
7757 | | case 20: |
7758 | | return "\\x14"; |
7759 | | case 21: |
7760 | | return "\\x15"; |
7761 | | case 22: |
7762 | | return "\\x16"; |
7763 | | case 23: |
7764 | | return "\\x17"; |
7765 | | case 24: |
7766 | | return "\\x18"; |
7767 | | case 25: |
7768 | | return "\\x19"; |
7769 | | case 26: |
7770 | | return "\\x1A"; |
7771 | | case 27: |
7772 | | return "\\x1B"; |
7773 | | case 28: |
7774 | | return "\\x1C"; |
7775 | | case 29: |
7776 | | return "\\x1D"; |
7777 | | case 30: |
7778 | | return "\\x1E"; |
7779 | | case 31: |
7780 | | return "\\x1F"; |
7781 | | case 32: |
7782 | | return " "; |
7783 | | case 33: |
7784 | | return "!"; |
7785 | | case 34: |
7786 | | return "\\\""; |
7787 | | case 35: |
7788 | | return "#"; |
7789 | | case 36: |
7790 | | return "$"; |
7791 | | case 37: |
7792 | | return "%"; |
7793 | | case 38: |
7794 | | return "&"; |
7795 | | case 39: |
7796 | | return "'"; |
7797 | | case 40: |
7798 | | return "("; |
7799 | | case 41: |
7800 | | return ")"; |
7801 | | case 42: |
7802 | | return "*"; |
7803 | | case 43: |
7804 | | return "+"; |
7805 | | case 44: |
7806 | | return ","; |
7807 | | case 45: |
7808 | | return "-"; |
7809 | | case 46: |
7810 | | return "."; |
7811 | | case 47: |
7812 | | return "/"; |
7813 | | case 48: |
7814 | | return "0"; |
7815 | | case 49: |
7816 | | return "1"; |
7817 | | case 50: |
7818 | | return "2"; |
7819 | | case 51: |
7820 | | return "3"; |
7821 | | case 52: |
7822 | | return "4"; |
7823 | | case 53: |
7824 | | return "5"; |
7825 | | case 54: |
7826 | | return "6"; |
7827 | | case 55: |
7828 | | return "7"; |
7829 | | case 56: |
7830 | | return "8"; |
7831 | | case 57: |
7832 | | return "9"; |
7833 | | case 58: |
7834 | | return ":"; |
7835 | | case 59: |
7836 | | return ";"; |
7837 | | case 60: |
7838 | | return "<"; |
7839 | | case 61: |
7840 | | return "="; |
7841 | | case 62: |
7842 | | return ">"; |
7843 | | case 63: |
7844 | | return "?"; |
7845 | | case 64: |
7846 | | return "@"; |
7847 | | case 65: |
7848 | | return "A"; |
7849 | | case 66: |
7850 | | return "B"; |
7851 | | case 67: |
7852 | | return "C"; |
7853 | | case 68: |
7854 | | return "D"; |
7855 | | case 69: |
7856 | | return "E"; |
7857 | | case 70: |
7858 | | return "F"; |
7859 | | case 71: |
7860 | | return "G"; |
7861 | | case 72: |
7862 | | return "H"; |
7863 | | case 73: |
7864 | | return "I"; |
7865 | | case 74: |
7866 | | return "J"; |
7867 | | case 75: |
7868 | | return "K"; |
7869 | | case 76: |
7870 | | return "L"; |
7871 | | case 77: |
7872 | | return "M"; |
7873 | | case 78: |
7874 | | return "N"; |
7875 | | case 79: |
7876 | | return "O"; |
7877 | | case 80: |
7878 | | return "P"; |
7879 | | case 81: |
7880 | | return "Q"; |
7881 | | case 82: |
7882 | | return "R"; |
7883 | | case 83: |
7884 | | return "S"; |
7885 | | case 84: |
7886 | | return "T"; |
7887 | | case 85: |
7888 | | return "U"; |
7889 | | case 86: |
7890 | | return "V"; |
7891 | | case 87: |
7892 | | return "W"; |
7893 | | case 88: |
7894 | | return "X"; |
7895 | | case 89: |
7896 | | return "Y"; |
7897 | | case 90: |
7898 | | return "Z"; |
7899 | | case 91: |
7900 | | return "["; |
7901 | | case 92: |
7902 | | return "\\\\"; |
7903 | | case 93: |
7904 | | return "]"; |
7905 | | case 94: |
7906 | | return "^"; |
7907 | | case 95: |
7908 | | return "_"; |
7909 | | case 96: |
7910 | | return "`"; |
7911 | | case 97: |
7912 | | return "a"; |
7913 | | case 98: |
7914 | | return "b"; |
7915 | | case 99: |
7916 | | return "c"; |
7917 | | case 100: |
7918 | | return "d"; |
7919 | | case 101: |
7920 | | return "e"; |
7921 | | case 102: |
7922 | | return "f"; |
7923 | | case 103: |
7924 | | return "g"; |
7925 | | case 104: |
7926 | | return "h"; |
7927 | | case 105: |
7928 | | return "i"; |
7929 | | case 106: |
7930 | | return "j"; |
7931 | | case 107: |
7932 | | return "k"; |
7933 | | case 108: |
7934 | | return "l"; |
7935 | | case 109: |
7936 | | return "m"; |
7937 | | case 110: |
7938 | | return "n"; |
7939 | | case 111: |
7940 | | return "o"; |
7941 | | case 112: |
7942 | | return "p"; |
7943 | | case 113: |
7944 | | return "q"; |
7945 | | case 114: |
7946 | | return "r"; |
7947 | | case 115: |
7948 | | return "s"; |
7949 | | case 116: |
7950 | | return "t"; |
7951 | | case 117: |
7952 | | return "u"; |
7953 | | case 118: |
7954 | | return "v"; |
7955 | | case 119: |
7956 | | return "w"; |
7957 | | case 120: |
7958 | | return "x"; |
7959 | | case 121: |
7960 | | return "y"; |
7961 | | case 122: |
7962 | | return "z"; |
7963 | | case 123: |
7964 | | return "{"; |
7965 | | case 124: |
7966 | | return "|"; |
7967 | | case 125: |
7968 | | return "}"; |
7969 | | case 126: |
7970 | | return "~"; |
7971 | | case 127: |
7972 | | return "\\x7F"; |
7973 | | case 128: |
7974 | | return "\\x80"; |
7975 | | case 129: |
7976 | | return "\\x81"; |
7977 | | case 130: |
7978 | | return "\\x82"; |
7979 | | case 131: |
7980 | | return "\\x83"; |
7981 | | case 132: |
7982 | | return "\\x84"; |
7983 | | case 133: |
7984 | | return "\\x85"; |
7985 | | case 134: |
7986 | | return "\\x86"; |
7987 | | case 135: |
7988 | | return "\\x87"; |
7989 | | case 136: |
7990 | | return "\\x88"; |
7991 | | case 137: |
7992 | | return "\\x89"; |
7993 | | case 138: |
7994 | | return "\\x8A"; |
7995 | | case 139: |
7996 | | return "\\x8B"; |
7997 | | case 140: |
7998 | | return "\\x8C"; |
7999 | | case 141: |
8000 | | return "\\x8D"; |
8001 | | case 142: |
8002 | | return "\\x8E"; |
8003 | | case 143: |
8004 | | return "\\x8F"; |
8005 | | case 144: |
8006 | | return "\\x90"; |
8007 | | case 145: |
8008 | | return "\\x91"; |
8009 | | case 146: |
8010 | | return "\\x92"; |
8011 | | case 147: |
8012 | | return "\\x93"; |
8013 | | case 148: |
8014 | | return "\\x94"; |
8015 | | case 149: |
8016 | | return "\\x95"; |
8017 | | case 150: |
8018 | | return "\\x96"; |
8019 | | case 151: |
8020 | | return "\\x97"; |
8021 | | case 152: |
8022 | | return "\\x98"; |
8023 | | case 153: |
8024 | | return "\\x99"; |
8025 | | case 154: |
8026 | | return "\\x9A"; |
8027 | | case 155: |
8028 | | return "\\x9B"; |
8029 | | case 156: |
8030 | | return "\\x9C"; |
8031 | | case 157: |
8032 | | return "\\x9D"; |
8033 | | case 158: |
8034 | | return "\\x9E"; |
8035 | | case 159: |
8036 | | return "\\x9F"; |
8037 | | case 160: |
8038 | | return "\\xA0"; |
8039 | | case 161: |
8040 | | return "\\xA1"; |
8041 | | case 162: |
8042 | | return "\\xA2"; |
8043 | | case 163: |
8044 | | return "\\xA3"; |
8045 | | case 164: |
8046 | | return "\\xA4"; |
8047 | | case 165: |
8048 | | return "\\xA5"; |
8049 | | case 166: |
8050 | | return "\\xA6"; |
8051 | | case 167: |
8052 | | return "\\xA7"; |
8053 | | case 168: |
8054 | | return "\\xA8"; |
8055 | | case 169: |
8056 | | return "\\xA9"; |
8057 | | case 170: |
8058 | | return "\\xAA"; |
8059 | | case 171: |
8060 | | return "\\xAB"; |
8061 | | case 172: |
8062 | | return "\\xAC"; |
8063 | | case 173: |
8064 | | return "\\xAD"; |
8065 | | case 174: |
8066 | | return "\\xAE"; |
8067 | | case 175: |
8068 | | return "\\xAF"; |
8069 | | case 176: |
8070 | | return "\\xB0"; |
8071 | | case 177: |
8072 | | return "\\xB1"; |
8073 | | case 178: |
8074 | | return "\\xB2"; |
8075 | | case 179: |
8076 | | return "\\xB3"; |
8077 | | case 180: |
8078 | | return "\\xB4"; |
8079 | | case 181: |
8080 | | return "\\xB5"; |
8081 | | case 182: |
8082 | | return "\\xB6"; |
8083 | | case 183: |
8084 | | return "\\xB7"; |
8085 | | case 184: |
8086 | | return "\\xB8"; |
8087 | | case 185: |
8088 | | return "\\xB9"; |
8089 | | case 186: |
8090 | | return "\\xBA"; |
8091 | | case 187: |
8092 | | return "\\xBB"; |
8093 | | case 188: |
8094 | | return "\\xBC"; |
8095 | | case 189: |
8096 | | return "\\xBD"; |
8097 | | case 190: |
8098 | | return "\\xBE"; |
8099 | | case 191: |
8100 | | return "\\xBF"; |
8101 | | case 192: |
8102 | | return "\\xC0"; |
8103 | | case 193: |
8104 | | return "\\xC1"; |
8105 | | case 194: |
8106 | | return "\\xC2"; |
8107 | | case 195: |
8108 | | return "\\xC3"; |
8109 | | case 196: |
8110 | | return "\\xC4"; |
8111 | | case 197: |
8112 | | return "\\xC5"; |
8113 | | case 198: |
8114 | | return "\\xC6"; |
8115 | | case 199: |
8116 | | return "\\xC7"; |
8117 | | case 200: |
8118 | | return "\\xC8"; |
8119 | | case 201: |
8120 | | return "\\xC9"; |
8121 | | case 202: |
8122 | | return "\\xCA"; |
8123 | | case 203: |
8124 | | return "\\xCB"; |
8125 | | case 204: |
8126 | | return "\\xCC"; |
8127 | | case 205: |
8128 | | return "\\xCD"; |
8129 | | case 206: |
8130 | | return "\\xCE"; |
8131 | | case 207: |
8132 | | return "\\xCF"; |
8133 | | case 208: |
8134 | | return "\\xD0"; |
8135 | | case 209: |
8136 | | return "\\xD1"; |
8137 | | case 210: |
8138 | | return "\\xD2"; |
8139 | | case 211: |
8140 | | return "\\xD3"; |
8141 | | case 212: |
8142 | | return "\\xD4"; |
8143 | | case 213: |
8144 | | return "\\xD5"; |
8145 | | case 214: |
8146 | | return "\\xD6"; |
8147 | | case 215: |
8148 | | return "\\xD7"; |
8149 | | case 216: |
8150 | | return "\\xD8"; |
8151 | | case 217: |
8152 | | return "\\xD9"; |
8153 | | case 218: |
8154 | | return "\\xDA"; |
8155 | | case 219: |
8156 | | return "\\xDB"; |
8157 | | case 220: |
8158 | | return "\\xDC"; |
8159 | | case 221: |
8160 | | return "\\xDD"; |
8161 | | case 222: |
8162 | | return "\\xDE"; |
8163 | | case 223: |
8164 | | return "\\xDF"; |
8165 | | case 224: |
8166 | | return "\\xE0"; |
8167 | | case 225: |
8168 | | return "\\xE1"; |
8169 | | case 226: |
8170 | | return "\\xE2"; |
8171 | | case 227: |
8172 | | return "\\xE3"; |
8173 | | case 228: |
8174 | | return "\\xE4"; |
8175 | | case 229: |
8176 | | return "\\xE5"; |
8177 | | case 230: |
8178 | | return "\\xE6"; |
8179 | | case 231: |
8180 | | return "\\xE7"; |
8181 | | case 232: |
8182 | | return "\\xE8"; |
8183 | | case 233: |
8184 | | return "\\xE9"; |
8185 | | case 234: |
8186 | | return "\\xEA"; |
8187 | | case 235: |
8188 | | return "\\xEB"; |
8189 | | case 236: |
8190 | | return "\\xEC"; |
8191 | | case 237: |
8192 | | return "\\xED"; |
8193 | | case 238: |
8194 | | return "\\xEE"; |
8195 | | case 239: |
8196 | | return "\\xEF"; |
8197 | | case 240: |
8198 | | return "\\xF0"; |
8199 | | case 241: |
8200 | | return "\\xF1"; |
8201 | | case 242: |
8202 | | return "\\xF2"; |
8203 | | case 243: |
8204 | | return "\\xF3"; |
8205 | | case 244: |
8206 | | return "\\xF4"; |
8207 | | case 245: |
8208 | | return "\\xF5"; |
8209 | | case 246: |
8210 | | return "\\xF6"; |
8211 | | case 247: |
8212 | | return "\\xF7"; |
8213 | | case 248: |
8214 | | return "\\xF8"; |
8215 | | case 249: |
8216 | | return "\\xF9"; |
8217 | | case 250: |
8218 | | return "\\xFA"; |
8219 | | case 251: |
8220 | | return "\\xFB"; |
8221 | | case 252: |
8222 | | return "\\xFC"; |
8223 | | case 253: |
8224 | | return "\\xFD"; |
8225 | | case 254: |
8226 | | return "\\xFE"; |
8227 | | case 255: |
8228 | | return "\\xFF"; |
8229 | | default: |
8230 | | assert(0); /* never gets here */ |
8231 | | return "dead code"; |
8232 | | } |
8233 | | assert(0); /* never gets here */ |
8234 | | } |
8235 | | |
8236 | | #endif /* XML_DTD */ |
8237 | | |
8238 | | static unsigned long |
8239 | 7.48k | getDebugLevel(const char *variableName, unsigned long defaultDebugLevel) { |
8240 | 7.48k | const char *const valueOrNull = getenv(variableName); |
8241 | 7.48k | if (valueOrNull == NULL) { |
8242 | 7.48k | return defaultDebugLevel; |
8243 | 7.48k | } |
8244 | 0 | const char *const value = valueOrNull; |
8245 | |
|
8246 | 0 | errno = 0; |
8247 | 0 | char *afterValue = (char *)value; |
8248 | 0 | unsigned long debugLevel = strtoul(value, &afterValue, 10); |
8249 | 0 | if ((errno != 0) || (afterValue[0] != '\0')) { |
8250 | 0 | errno = 0; |
8251 | 0 | return defaultDebugLevel; |
8252 | 0 | } |
8253 | | |
8254 | 0 | return debugLevel; |
8255 | 0 | } |