/src/ghostpdl/pdf/pdf_int.c
Line | Count | Source |
1 | | /* Copyright (C) 2018-2026 Artifex Software, Inc. |
2 | | All Rights Reserved. |
3 | | |
4 | | This software is provided AS-IS with no warranty, either express or |
5 | | implied. |
6 | | |
7 | | This software is distributed under license and may not be copied, |
8 | | modified or distributed except as expressly authorized under the terms |
9 | | of the license contained in the file LICENSE in this distribution. |
10 | | |
11 | | Refer to licensing information at http://www.artifex.com or contact |
12 | | Artifex Software, Inc., 39 Mesa Street, Suite 108A, San Francisco, |
13 | | CA 94129, USA, for further information. |
14 | | */ |
15 | | |
16 | | /* The PDF interpreter written in C */ |
17 | | |
18 | | #include "pdf_int.h" |
19 | | #include "pdf_file.h" |
20 | | #include "strmio.h" |
21 | | #include "stream.h" |
22 | | #include "pdf_misc.h" |
23 | | #include "pdf_path.h" |
24 | | #include "pdf_colour.h" |
25 | | #include "pdf_image.h" |
26 | | #include "pdf_shading.h" |
27 | | #include "pdf_font.h" |
28 | | #include "pdf_font_types.h" |
29 | | #include "pdf_cmap.h" |
30 | | #include "pdf_text.h" |
31 | | #include "pdf_gstate.h" |
32 | | #include "pdf_stack.h" |
33 | | #include "pdf_xref.h" |
34 | | #include "pdf_dict.h" |
35 | | #include "pdf_array.h" |
36 | | #include "pdf_trans.h" |
37 | | #include "pdf_optcontent.h" |
38 | | #include "pdf_sec.h" |
39 | | #include <stdlib.h> |
40 | | |
41 | | #include "gsstate.h" /* for gs_gstate_free */ |
42 | | |
43 | | /* we use -ve returns for error, 0 for success and +ve for 'take an action' */ |
44 | | /* Defining tis return so we do not need to define a new error */ |
45 | 77.1M | #define REPAIRED_KEYWORD 1 |
46 | | |
47 | | /***********************************************************************************/ |
48 | | /* 'token' reading functions. Tokens in this sense are PDF logical objects and the */ |
49 | | /* related keywords. So that's numbers, booleans, names, strings, dictionaries, */ |
50 | | /* arrays, the null object and indirect references. The keywords are obj/endobj */ |
51 | | /* stream/endstream, xref, startxref and trailer. */ |
52 | | |
53 | | /***********************************************************************************/ |
54 | | /* Some simple functions to find white space, delimiters and hex bytes */ |
55 | | static bool iswhite(char c) |
56 | 4.85G | { |
57 | 4.85G | if (c == 0x00 || c == 0x09 || c == 0x0a || c == 0x0c || c == 0x0d || c == 0x20) |
58 | 755M | return true; |
59 | 4.09G | else |
60 | 4.09G | return false; |
61 | 4.85G | } |
62 | | |
63 | | static bool isdelimiter(char c) |
64 | 2.80G | { |
65 | 2.80G | if (c == '/' || c == '(' || c == ')' || c == '[' || c == ']' || c == '<' || c == '>' || c == '{' || c == '}' || c == '%') |
66 | 66.5M | return true; |
67 | 2.73G | else |
68 | 2.73G | return false; |
69 | 2.80G | } |
70 | | |
71 | | /* The 'read' functions all return the newly created object on the context's stack |
72 | | * which means these objects are created with a reference count of 0, and only when |
73 | | * pushed onto the stack does the reference count become 1, indicating the stack is |
74 | | * the only reference. |
75 | | */ |
76 | | int pdfi_skip_white(pdf_context *ctx, pdf_c_stream *s) |
77 | 1.10G | { |
78 | 1.10G | int c; |
79 | | |
80 | 1.32G | do { |
81 | 1.32G | c = pdfi_read_byte(ctx, s); |
82 | 1.32G | if (c < 0) |
83 | 347k | return 0; |
84 | 1.32G | } while (iswhite(c)); |
85 | | |
86 | 1.10G | pdfi_unread_byte(ctx, s, (byte)c); |
87 | 1.10G | return 0; |
88 | 1.10G | } |
89 | | |
90 | | int pdfi_skip_eol(pdf_context *ctx, pdf_c_stream *s) |
91 | 2.68M | { |
92 | 2.68M | int c; |
93 | | |
94 | 2.86M | do { |
95 | 2.86M | c = pdfi_read_byte(ctx, s); |
96 | 2.86M | if (c < 0 || c == 0x0a) |
97 | 1.21M | return 0; |
98 | 2.86M | } while (c != 0x0d); |
99 | 1.47M | c = pdfi_read_byte(ctx, s); |
100 | 1.47M | if (c == 0x0a) |
101 | 1.46M | return 0; |
102 | 7.94k | if (c >= 0) |
103 | 7.85k | pdfi_unread_byte(ctx, s, (byte)c); |
104 | 7.94k | pdfi_set_warning(ctx, 0, NULL, W_PDF_STREAM_BAD_KEYWORD, "pdfi_skip_eol", NULL); |
105 | 7.94k | return 0; |
106 | 1.47M | } |
107 | | |
108 | | /* Fast(ish) but inaccurate strtof, with Adobe overflow handling, |
109 | | * lifted from MuPDF. */ |
110 | | static float acrobat_compatible_atof(char *s) |
111 | 129M | { |
112 | 129M | int neg = 0; |
113 | 129M | int i = 0; |
114 | | |
115 | 137M | while (*s == '-') { |
116 | 8.05M | neg = 1; |
117 | 8.05M | ++s; |
118 | 8.05M | } |
119 | 129M | while (*s == '+') { |
120 | 7 | ++s; |
121 | 7 | } |
122 | | |
123 | 457M | while (*s >= '0' && *s <= '9') { |
124 | | /* We deliberately ignore overflow here. |
125 | | * Tests show that Acrobat handles * overflows in exactly the same way we do: |
126 | | * 123450000000000000000678 is read as 678. |
127 | | */ |
128 | 327M | i = i * 10 + (*s - '0'); |
129 | 327M | ++s; |
130 | 327M | } |
131 | | |
132 | 129M | if (*s == '.') { |
133 | 129M | float MAX = (MAX_FLOAT-9)/10; |
134 | 129M | float v = (float)i; |
135 | 129M | float n = 0; |
136 | 129M | float d = 1; |
137 | 129M | ++s; |
138 | | /* Bug 705211: Ensure that we don't overflow n here - just ignore any |
139 | | * trailing digits after this. This will be plenty accurate enough. */ |
140 | 550M | while (*s >= '0' && *s <= '9' && n <= MAX) { |
141 | 421M | n = 10 * n + (*s - '0'); |
142 | 421M | d = 10 * d; |
143 | 421M | ++s; |
144 | 421M | } |
145 | 129M | v += n / d; |
146 | 129M | return neg ? -v : v; |
147 | 129M | } else { |
148 | 161k | return (float)(neg ? -i : i); |
149 | 161k | } |
150 | 129M | } |
151 | | |
152 | | int pdfi_read_bare_int(pdf_context *ctx, pdf_c_stream *s, int *parsed_int) |
153 | 105M | { |
154 | 105M | int index = 0; |
155 | 105M | int int_val = 0; |
156 | 105M | int negative = 0; |
157 | 105M | int tenth_max_int = max_int / 10, tenth_max_uint = max_uint / 10; |
158 | 105M | bool overflowed = false; |
159 | 105M | int code = 0; |
160 | | |
161 | 105M | restart: |
162 | 105M | pdfi_skip_white(ctx, s); |
163 | | |
164 | 450M | do { |
165 | 450M | int c = pdfi_read_byte(ctx, s); |
166 | 450M | if (c == EOFC) |
167 | 4.27k | break; |
168 | | |
169 | 450M | if (c < 0) |
170 | 6.30k | return_error(gs_error_ioerror); |
171 | | |
172 | 450M | if (iswhite(c)) { |
173 | 105M | break; |
174 | 345M | } else if (c == '%' && index == 0) { |
175 | 47.1k | pdfi_skip_comment(ctx, s); |
176 | 47.1k | goto restart; |
177 | 345M | } else if (isdelimiter(c)) { |
178 | 14.0k | pdfi_unread_byte(ctx, s, (byte)c); |
179 | 14.0k | break; |
180 | 14.0k | } |
181 | | |
182 | 345M | if (c >= '0' && c <= '9') { |
183 | 345M | if (!overflowed) { |
184 | 345M | if ((negative && int_val <= tenth_max_int) || (!negative && int_val <= tenth_max_uint)) |
185 | 345M | int_val = int_val*10 + c - '0'; |
186 | 1.85k | else { |
187 | 1.85k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_NUMBEROVERFLOW, "pdfi_read_num", NULL)) < 0) { |
188 | 0 | return code; |
189 | 0 | } |
190 | 1.85k | overflowed = true; |
191 | 1.85k | } |
192 | 345M | } |
193 | 345M | } else if (c == '.') { |
194 | 2.54k | goto error; |
195 | 184k | } else if (c == 'e' || c == 'E') { |
196 | 2.50k | pdfi_log_info(ctx, "pdfi_read_bare_int", (char *)"Invalid number format: scientific notation\n"); |
197 | 2.50k | goto error; |
198 | 182k | } else if (c == '-') { |
199 | | /* Any - sign not at the start of the string indicates a malformed number. */ |
200 | 751 | if (index != 0 || negative) { |
201 | 138 | pdfi_log_info(ctx, "pdfi_read_bare_int", (char *)"Invalid number format: sign not at the start\n"); |
202 | 138 | goto error; |
203 | 138 | } |
204 | 613 | negative = 1; |
205 | 181k | } else if (c == '+') { |
206 | 112k | if (index == 0) { |
207 | | /* Just drop the + it's pointless, and it'll get in the way |
208 | | * of our negation handling for floats. */ |
209 | 111k | continue; |
210 | 111k | } else { |
211 | 147 | pdfi_log_info(ctx, "pdfi_read_bare_int", (char *)"Invalid number format: sign not at the start\n"); |
212 | 147 | goto error; |
213 | 147 | } |
214 | 112k | } else { |
215 | 69.2k | if (index > 0) { |
216 | 6.78k | pdfi_log_info(ctx, "pdfi_read_bare_int", (char *)"Invalid number format: Ignoring missing white space while parsing number\n"); |
217 | 6.78k | goto error; |
218 | 6.78k | } |
219 | 62.4k | pdfi_unread_byte(ctx, s, (byte)c); |
220 | 62.4k | goto error; |
221 | 69.2k | } |
222 | 345M | if (++index > 255) |
223 | 210 | goto error; |
224 | 345M | } while(1); |
225 | | |
226 | 105M | *parsed_int = negative ? -int_val : int_val; |
227 | 105M | if (ctx->args.pdfdebug) |
228 | 0 | outprintf(ctx->memory, " %d", *parsed_int); |
229 | 105M | return (index > 0); |
230 | | |
231 | 74.7k | error: |
232 | 74.7k | *parsed_int = 0; |
233 | 74.7k | return_error(gs_error_syntaxerror); |
234 | 105M | } |
235 | | |
236 | | static int pdfi_read_num(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
237 | 299M | { |
238 | 299M | byte Buffer[256]; |
239 | 299M | unsigned short index = 0; |
240 | 299M | bool real = false; |
241 | 299M | bool has_decimal_point = false; |
242 | 299M | bool has_exponent = false; |
243 | 299M | unsigned short exponent_index = 0; |
244 | 299M | pdf_num *num; |
245 | 299M | int code = 0, malformed = false, doubleneg = false, recovered = false, negative = false, overflowed = false; |
246 | 299M | unsigned int int_val = 0; |
247 | 299M | int tenth_max_int = max_int / 10, tenth_max_uint = max_uint / 10; |
248 | | |
249 | 299M | pdfi_skip_white(ctx, s); |
250 | | |
251 | 1.66G | do { |
252 | 1.66G | int c = pdfi_read_byte(ctx, s); |
253 | 1.66G | if (c == EOFC) { |
254 | 11.9k | Buffer[index] = 0x00; |
255 | 11.9k | break; |
256 | 11.9k | } |
257 | | |
258 | 1.66G | if (c < 0) |
259 | 3.89k | return_error(gs_error_ioerror); |
260 | | |
261 | 1.66G | if (iswhite(c)) { |
262 | 266M | Buffer[index] = 0x00; |
263 | 266M | break; |
264 | 1.39G | } else if (isdelimiter(c)) { |
265 | 27.5M | pdfi_unread_byte(ctx, s, (byte)c); |
266 | 27.5M | Buffer[index] = 0x00; |
267 | 27.5M | break; |
268 | 27.5M | } |
269 | 1.37G | Buffer[index] = (byte)c; |
270 | | |
271 | 1.37G | if (c >= '0' && c <= '9') { |
272 | 1.21G | if (!(malformed && recovered) && !overflowed && !real) { |
273 | 746M | if ((negative && int_val <= tenth_max_int) || (!negative && int_val <= tenth_max_uint)) |
274 | 745M | int_val = int_val*10 + c - '0'; |
275 | 1.32M | else { |
276 | 1.32M | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_NUMBEROVERFLOW, "pdfi_read_num", NULL)) < 0) { |
277 | 0 | return code; |
278 | 0 | } |
279 | 1.32M | overflowed = true; |
280 | 1.32M | } |
281 | 746M | } |
282 | 1.21G | } else if (c == '.') { |
283 | 134M | if (has_decimal_point == true) { |
284 | 3.22M | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", NULL)) < 0) { |
285 | 0 | return code; |
286 | 0 | } |
287 | 3.22M | malformed = true; |
288 | 131M | } else { |
289 | 131M | has_decimal_point = true; |
290 | 131M | real = true; |
291 | 131M | } |
292 | 134M | } else if (c == 'e' || c == 'E') { |
293 | | /* TODO: technically scientific notation isn't in PDF spec, |
294 | | * but gs seems to accept it, so we should also? |
295 | | */ |
296 | 460k | if (has_exponent == true) { |
297 | 34.7k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", NULL)) < 0) { |
298 | 0 | return code; |
299 | 0 | } |
300 | 34.7k | malformed = true; |
301 | 425k | } else { |
302 | 425k | pdfi_set_warning(ctx, 0, NULL, W_PDF_NUM_EXPONENT, "pdfi_read_num", NULL); |
303 | 425k | has_exponent = true; |
304 | 425k | exponent_index = index; |
305 | 425k | real = true; |
306 | 425k | } |
307 | 24.4M | } else if (c == '-') { |
308 | | /* Any - sign not at the start of the string, or just after an exponent |
309 | | * indicates a malformed number. */ |
310 | 18.5M | if (!(index == 0 || (has_exponent && index == exponent_index+1))) { |
311 | 4.40M | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", NULL)) < 0) { |
312 | 0 | return code; |
313 | 0 | } |
314 | 4.40M | if (Buffer[index - 1] != '-') { |
315 | | /* We are parsing a number line 123-56. We should continue parsing, but |
316 | | * ignore anything from the second -. */ |
317 | 770k | malformed = true; |
318 | 770k | Buffer[index] = 0; |
319 | 770k | recovered = true; |
320 | 770k | } |
321 | 4.40M | } |
322 | 18.5M | if (!has_exponent && !(malformed && recovered)) { |
323 | 17.8M | doubleneg = negative; |
324 | 17.8M | negative = 1; |
325 | 17.8M | } |
326 | 18.5M | } else if (c == '+') { |
327 | 137k | if (index == 0 || (has_exponent && index == exponent_index+1)) { |
328 | | /* Just drop the + it's pointless, and it'll get in the way |
329 | | * of our negation handling for floats. */ |
330 | 127k | index--; |
331 | 127k | } else { |
332 | 10.3k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", NULL)) < 0) { |
333 | 0 | return code; |
334 | 0 | } |
335 | 10.3k | if (Buffer[index - 1] != '-') { |
336 | | /* We are parsing a number line 123-56. We should continue parsing, but |
337 | | * ignore anything from the second -. */ |
338 | 10.0k | malformed = true; |
339 | 10.0k | Buffer[index] = 0; |
340 | 10.0k | recovered = true; |
341 | 10.0k | } |
342 | 10.3k | } |
343 | 5.78M | } else { |
344 | 5.78M | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MISSINGWHITESPACE, "pdfi_read_num", NULL)) < 0) { |
345 | 0 | return code; |
346 | 0 | } |
347 | 5.78M | pdfi_unread_byte(ctx, s, (byte)c); |
348 | 5.78M | Buffer[index] = 0x00; |
349 | 5.78M | break; |
350 | 5.78M | } |
351 | 1.36G | if (++index > 255) |
352 | 44.8k | return_error(gs_error_syntaxerror); |
353 | 1.36G | } while(1); |
354 | | |
355 | 299M | if (real && (!malformed || (malformed && recovered))) |
356 | 130M | code = pdfi_object_alloc(ctx, PDF_REAL, 0, (pdf_obj **)&num); |
357 | 169M | else |
358 | 169M | code = pdfi_object_alloc(ctx, PDF_INT, 0, (pdf_obj **)&num); |
359 | 299M | if (code < 0) |
360 | 0 | return code; |
361 | | |
362 | 299M | if ((malformed && !recovered) || (!real && doubleneg)) { |
363 | 1.63M | if ((code = pdfi_set_warning_var(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", "Treating malformed number %s as 0", Buffer)) < 0) { |
364 | 0 | goto exit; |
365 | 0 | } |
366 | 1.63M | num->value.i = 0; |
367 | 298M | } else if (has_exponent) { |
368 | 398k | float f, exp; |
369 | 398k | char *p = strstr((const char *)Buffer, "e"); |
370 | | |
371 | 398k | if (p == NULL) |
372 | 37.5k | p = strstr((const char *)Buffer, "E"); |
373 | | |
374 | 398k | if (p == NULL) { |
375 | 2.92k | if ((code = pdfi_set_warning_var(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", "Treating malformed float %s as 0", Buffer)) < 0) { |
376 | 0 | goto exit; |
377 | 0 | } |
378 | 2.92k | num->value.d = 0; |
379 | 396k | } else { |
380 | 396k | p++; |
381 | | |
382 | 396k | if (sscanf((char *)p, "%g", &exp) != 1 || exp > 38) { |
383 | 366k | if ((code = pdfi_set_warning_var(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", "Treating malformed float %s as 0", Buffer)) < 0) { |
384 | 0 | goto exit; |
385 | 0 | } |
386 | 366k | num->value.d = 0; |
387 | 366k | } else { |
388 | 29.5k | if (sscanf((char *)Buffer, "%g", &f) == 1) { |
389 | 29.0k | num->value.d = f; |
390 | 29.0k | } else { |
391 | 574 | if ((code = pdfi_set_warning_var(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MALFORMEDNUMBER, "pdfi_read_num", "Treating malformed float %s as 0", Buffer)) < 0) { |
392 | 0 | goto exit; |
393 | 0 | } |
394 | 574 | num->value.d = 0; |
395 | 574 | } |
396 | 29.5k | } |
397 | 396k | } |
398 | 297M | } else if (real) { |
399 | 129M | num->value.d = acrobat_compatible_atof((char *)Buffer); |
400 | 168M | } else { |
401 | | /* The doubleneg case is taken care of above. */ |
402 | 168M | num->value.i = negative ? (int64_t)int_val * -1 : (int64_t)int_val; |
403 | 168M | } |
404 | 299M | if (ctx->args.pdfdebug) { |
405 | 0 | if (real) |
406 | 0 | outprintf(ctx->memory, " %f", num->value.d); |
407 | 0 | else |
408 | 0 | outprintf(ctx->memory, " %"PRIi64, num->value.i); |
409 | 0 | } |
410 | 299M | num->indirect_num = indirect_num; |
411 | 299M | num->indirect_gen = indirect_gen; |
412 | | |
413 | 299M | code = pdfi_push(ctx, (pdf_obj *)num); |
414 | | |
415 | 299M | exit: |
416 | 299M | if (code < 0) |
417 | 947 | pdfi_free_object((pdf_obj *)num); |
418 | | |
419 | 299M | return code; |
420 | 299M | } |
421 | | |
422 | | static int pdfi_read_name(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
423 | 85.3M | { |
424 | 85.3M | char *Buffer, *NewBuf = NULL; |
425 | 85.3M | unsigned short index = 0; |
426 | 85.3M | short bytes = 0; |
427 | 85.3M | uint32_t size = 256; |
428 | 85.3M | pdf_name *name = NULL; |
429 | 85.3M | int code; |
430 | | |
431 | 85.3M | Buffer = (char *)gs_alloc_bytes(ctx->memory, size, "pdfi_read_name"); |
432 | 85.3M | if (Buffer == NULL) |
433 | 0 | return_error(gs_error_VMerror); |
434 | | |
435 | 620M | do { |
436 | 620M | int c = pdfi_read_byte(ctx, s); |
437 | 620M | if (c < 0) |
438 | 11.0k | break; |
439 | | |
440 | 620M | if (iswhite((char)c)) { |
441 | 56.0M | Buffer[index] = 0x00; |
442 | 56.0M | break; |
443 | 564M | } else if (isdelimiter((char)c)) { |
444 | 29.3M | pdfi_unread_byte(ctx, s, (char)c); |
445 | 29.3M | Buffer[index] = 0x00; |
446 | 29.3M | break; |
447 | 29.3M | } |
448 | 535M | Buffer[index] = (char)c; |
449 | | |
450 | | /* Check for and convert escaped name characters */ |
451 | 535M | if (c == '#') { |
452 | 238k | byte NumBuf[2]; |
453 | | |
454 | 238k | bytes = pdfi_read_bytes(ctx, (byte *)&NumBuf, 1, 2, s); |
455 | 238k | if (bytes < 2 || (!ishex(NumBuf[0]) || !ishex(NumBuf[1]))) { |
456 | 101k | pdfi_set_warning(ctx, 0, NULL, W_PDF_BAD_NAME_ESCAPE, "pdfi_read_name", NULL); |
457 | 101k | pdfi_unread(ctx, s, (byte *)NumBuf, bytes); |
458 | | /* This leaves the name buffer with a # in it, rather than anything sane! */ |
459 | 101k | } |
460 | 136k | else |
461 | 136k | Buffer[index] = (fromhex(NumBuf[0]) << 4) + fromhex(NumBuf[1]); |
462 | 238k | } |
463 | | |
464 | | /* If we ran out of memory, increase the buffer size */ |
465 | 535M | if (index++ >= size - 1) { |
466 | 46.1k | NewBuf = (char *)gs_alloc_bytes(ctx->memory, (size_t)size + 256, "pdfi_read_name"); |
467 | 46.1k | if (NewBuf == NULL) { |
468 | 0 | gs_free_object(ctx->memory, Buffer, "pdfi_read_name error"); |
469 | 0 | return_error(gs_error_VMerror); |
470 | 0 | } |
471 | 46.1k | memcpy(NewBuf, Buffer, size); |
472 | 46.1k | gs_free_object(ctx->memory, Buffer, "pdfi_read_name"); |
473 | 46.1k | Buffer = NewBuf; |
474 | 46.1k | size += 256; |
475 | 46.1k | } |
476 | 535M | } while(1); |
477 | | |
478 | 85.3M | code = pdfi_object_alloc(ctx, PDF_NAME, index, (pdf_obj **)&name); |
479 | 85.3M | if (code < 0) { |
480 | 0 | gs_free_object(ctx->memory, Buffer, "pdfi_read_name error"); |
481 | 0 | return code; |
482 | 0 | } |
483 | 85.3M | memcpy(name->data, Buffer, index); |
484 | 85.3M | name->indirect_num = indirect_num; |
485 | 85.3M | name->indirect_gen = indirect_gen; |
486 | | |
487 | 85.3M | if (ctx->args.pdfdebug) |
488 | 0 | outprintf(ctx->memory, " /%s", Buffer); |
489 | | |
490 | 85.3M | gs_free_object(ctx->memory, Buffer, "pdfi_read_name"); |
491 | | |
492 | 85.3M | code = pdfi_push(ctx, (pdf_obj *)name); |
493 | | |
494 | 85.3M | if (code < 0) |
495 | 0 | pdfi_free_object((pdf_obj *)name); |
496 | | |
497 | 85.3M | return code; |
498 | 85.3M | } |
499 | | |
500 | | static int pdfi_read_hexstring(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
501 | 4.17M | { |
502 | 4.17M | char *Buffer, *NewBuf = NULL; |
503 | 4.17M | unsigned short index = 0; |
504 | 4.17M | uint32_t size = 256; |
505 | 4.17M | pdf_string *string = NULL; |
506 | 4.17M | int code, hex0, hex1; |
507 | | |
508 | 4.17M | Buffer = (char *)gs_alloc_bytes(ctx->memory, size, "pdfi_read_hexstring"); |
509 | 4.17M | if (Buffer == NULL) |
510 | 0 | return_error(gs_error_VMerror); |
511 | | |
512 | 4.17M | if (ctx->args.pdfdebug) |
513 | 0 | outprintf(ctx->memory, " <"); |
514 | | |
515 | 146M | do { |
516 | 146M | do { |
517 | 146M | hex0 = pdfi_read_byte(ctx, s); |
518 | 146M | if (hex0 < 0) |
519 | 808 | break; |
520 | 146M | } while(iswhite(hex0)); |
521 | 146M | if (hex0 < 0) |
522 | 808 | break; |
523 | | |
524 | 146M | if (hex0 == '>') |
525 | 3.85M | break; |
526 | | |
527 | 142M | if (ctx->args.pdfdebug) |
528 | 0 | outprintf(ctx->memory, "%c", (char)hex0); |
529 | | |
530 | 142M | do { |
531 | 142M | hex1 = pdfi_read_byte(ctx, s); |
532 | 142M | if (hex1 < 0) |
533 | 794 | break; |
534 | 142M | } while(iswhite(hex1)); |
535 | 142M | if (hex1 < 0) |
536 | 794 | break; |
537 | | |
538 | 142M | if (hex1 == '>') { |
539 | | /* PDF Reference 1.7 page 56: |
540 | | * "If the final digit of a hexadecimal string is missing that is, |
541 | | * if there is an odd number of digits the final digit is assumed to be 0." |
542 | | */ |
543 | 42.4k | hex1 = 0x30; |
544 | 42.4k | if (!ishex(hex0) || !ishex(hex1)) { |
545 | 2.05k | code = gs_note_error(gs_error_syntaxerror); |
546 | 2.05k | goto exit; |
547 | 2.05k | } |
548 | 40.3k | Buffer[index] = (fromhex(hex0) << 4) + fromhex(hex1); |
549 | 40.3k | if (ctx->args.pdfdebug) |
550 | 0 | outprintf(ctx->memory, "%c", hex1); |
551 | 40.3k | break; |
552 | 42.4k | } |
553 | | |
554 | 142M | if (!ishex(hex0) || !ishex(hex1)) { |
555 | 280k | code = gs_note_error(gs_error_syntaxerror); |
556 | 280k | goto exit; |
557 | 280k | } |
558 | | |
559 | 141M | if (ctx->args.pdfdebug) |
560 | 0 | outprintf(ctx->memory, "%c", (char)hex1); |
561 | | |
562 | 141M | Buffer[index] = (fromhex(hex0) << 4) + fromhex(hex1); |
563 | | |
564 | 141M | if (index++ >= size - 1) { |
565 | 482k | NewBuf = (char *)gs_alloc_bytes(ctx->memory, (size_t)size + 256, "pdfi_read_hexstring"); |
566 | 482k | if (NewBuf == NULL) { |
567 | 0 | code = gs_note_error(gs_error_VMerror); |
568 | 0 | goto exit; |
569 | 0 | } |
570 | 482k | memcpy(NewBuf, Buffer, size); |
571 | 482k | gs_free_object(ctx->memory, Buffer, "pdfi_read_hexstring"); |
572 | 482k | Buffer = NewBuf; |
573 | 482k | size += 256; |
574 | 482k | } |
575 | 141M | } while(1); |
576 | | |
577 | 3.89M | if (ctx->args.pdfdebug) |
578 | 0 | outprintf(ctx->memory, ">"); |
579 | | |
580 | 3.89M | code = pdfi_object_alloc(ctx, PDF_STRING, index, (pdf_obj **)&string); |
581 | 3.89M | if (code < 0) |
582 | 0 | goto exit; |
583 | 3.89M | memcpy(string->data, Buffer, index); |
584 | 3.89M | string->indirect_num = indirect_num; |
585 | 3.89M | string->indirect_gen = indirect_gen; |
586 | | |
587 | 3.89M | if (ctx->encryption.is_encrypted && ctx->encryption.decrypt_strings) { |
588 | 2.28k | code = pdfi_decrypt_string(ctx, string); |
589 | 2.28k | if (code < 0) |
590 | 0 | return code; |
591 | 2.28k | } |
592 | | |
593 | 3.89M | code = pdfi_push(ctx, (pdf_obj *)string); |
594 | 3.89M | if (code < 0) |
595 | 0 | pdfi_free_object((pdf_obj *)string); |
596 | | |
597 | 4.17M | exit: |
598 | 4.17M | gs_free_object(ctx->memory, Buffer, "pdfi_read_hexstring"); |
599 | 4.17M | return code; |
600 | 3.89M | } |
601 | | |
602 | | static int pdfi_read_string(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
603 | 15.4M | { |
604 | 15.4M | char *Buffer, *NewBuf = NULL; |
605 | 15.4M | unsigned short index = 0; |
606 | 15.4M | uint32_t size = 256; |
607 | 15.4M | pdf_string *string = NULL; |
608 | 15.4M | int c, code, nesting = 0; |
609 | 15.4M | bool escape = false, skip_lf = false, exit_loop = false; |
610 | | |
611 | 15.4M | Buffer = (char *)gs_alloc_bytes(ctx->memory, size, "pdfi_read_string"); |
612 | 15.4M | if (Buffer == NULL) |
613 | 0 | return_error(gs_error_VMerror); |
614 | | |
615 | 751M | do { |
616 | 751M | if (index >= size - 1) { |
617 | 1.66M | NewBuf = (char *)gs_alloc_bytes(ctx->memory, (size_t)size + 256, "pdfi_read_string"); |
618 | 1.66M | if (NewBuf == NULL) { |
619 | 0 | gs_free_object(ctx->memory, Buffer, "pdfi_read_string error"); |
620 | 0 | return_error(gs_error_VMerror); |
621 | 0 | } |
622 | 1.66M | memcpy(NewBuf, Buffer, size); |
623 | 1.66M | gs_free_object(ctx->memory, Buffer, "pdfi_read_string"); |
624 | 1.66M | Buffer = NewBuf; |
625 | 1.66M | size += 256; |
626 | 1.66M | } |
627 | | |
628 | 751M | c = pdfi_read_byte(ctx, s); |
629 | | |
630 | 751M | if (c < 0) { |
631 | 22.1k | if (nesting > 0 && (code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_UNESCAPEDSTRING, "pdfi_read_string", NULL)) < 0) { |
632 | 0 | gs_free_object(ctx->memory, Buffer, "pdfi_read_string error"); |
633 | 0 | return code; |
634 | 0 | } |
635 | 22.1k | Buffer[index] = 0x00; |
636 | 22.1k | break; |
637 | 22.1k | } |
638 | | |
639 | 751M | if (skip_lf) { |
640 | 3.27M | skip_lf = false; |
641 | 3.27M | if (c == 0x0a) |
642 | 1.08M | continue; |
643 | 3.27M | } |
644 | 750M | Buffer[index] = (char)c; |
645 | | |
646 | 750M | if (escape) { |
647 | 3.30M | escape = false; |
648 | 3.30M | switch (Buffer[index]) { |
649 | 56.1k | case 0x0d: |
650 | 56.1k | skip_lf = true; |
651 | 67.9k | case 0x0a: |
652 | 67.9k | continue; |
653 | 76.1k | case 'n': |
654 | 76.1k | Buffer[index] = 0x0a; |
655 | 76.1k | break; |
656 | 77.7k | case 'r': |
657 | 77.7k | Buffer[index] = 0x0d; |
658 | 77.7k | break; |
659 | 18.5k | case 't': |
660 | 18.5k | Buffer[index] = 0x09; |
661 | 18.5k | break; |
662 | 22.0k | case 'b': |
663 | 22.0k | Buffer[index] = 0x08; |
664 | 22.0k | break; |
665 | 23.1k | case 'f': |
666 | 23.1k | Buffer[index] = 0x0c; |
667 | 23.1k | break; |
668 | 171k | case '(': |
669 | 345k | case ')': |
670 | 430k | case '\\': |
671 | 430k | break; |
672 | 625k | case '0': |
673 | 643k | case '1': |
674 | 1.01M | case '2': |
675 | 1.29M | case '3': |
676 | 1.30M | case '4': |
677 | 1.31M | case '5': |
678 | 1.32M | case '6': |
679 | 1.33M | case '7': |
680 | 1.33M | { |
681 | | /* Octal chars can be 1, 2 or 3 chars in length, terminated either |
682 | | * by being 3 chars long, EOFC, or a non-octal char. We do not allow |
683 | | * line breaks in the middle of octal chars. */ |
684 | 1.33M | int c1 = pdfi_read_byte(ctx, s); |
685 | 1.33M | c -= '0'; |
686 | 1.33M | if (c1 < 0) { |
687 | | /* Nothing to do, or unread */ |
688 | 1.33M | } else if (c1 < '0' || c1 > '7') { |
689 | 80.9k | pdfi_unread_byte(ctx, s, (char)c1); |
690 | 1.25M | } else { |
691 | 1.25M | c = c*8 + c1 - '0'; |
692 | 1.25M | c1 = pdfi_read_byte(ctx, s); |
693 | 1.25M | if (c1 < 0) { |
694 | | /* Nothing to do, or unread */ |
695 | 1.25M | } else if (c1 < '0' || c1 > '7') { |
696 | 34.1k | pdfi_unread_byte(ctx, s, (char)c1); |
697 | 34.1k | } else |
698 | 1.22M | c = c*8 + c1 - '0'; |
699 | 1.25M | } |
700 | 1.33M | Buffer[index] = c; |
701 | 1.33M | break; |
702 | 1.32M | } |
703 | 1.25M | default: |
704 | | /* PDF Reference, literal strings, if the character following a |
705 | | * escape \ character is not recognised, then it is ignored. |
706 | | */ |
707 | 1.25M | escape = false; |
708 | 1.25M | index++; |
709 | 1.25M | continue; |
710 | 3.30M | } |
711 | 747M | } else { |
712 | 747M | switch(Buffer[index]) { |
713 | 3.22M | case 0x0d: |
714 | 3.22M | Buffer[index] = 0x0a; |
715 | 3.22M | skip_lf = true; |
716 | 3.22M | break; |
717 | 18.2M | case ')': |
718 | 18.2M | if (nesting == 0) { |
719 | 15.3M | Buffer[index] = 0x00; |
720 | 15.3M | exit_loop = true; |
721 | 15.3M | } else |
722 | 2.81M | nesting--; |
723 | 18.2M | break; |
724 | 3.30M | case '\\': |
725 | 3.30M | escape = true; |
726 | 3.30M | continue; |
727 | 3.39M | case '(': |
728 | 3.39M | nesting++; |
729 | 3.39M | break; |
730 | 719M | default: |
731 | 719M | break; |
732 | 747M | } |
733 | 747M | } |
734 | | |
735 | 745M | if (exit_loop) |
736 | 15.3M | break; |
737 | | |
738 | 730M | index++; |
739 | 736M | } while(1); |
740 | | |
741 | 15.4M | code = pdfi_object_alloc(ctx, PDF_STRING, index, (pdf_obj **)&string); |
742 | 15.4M | if (code < 0) { |
743 | 0 | gs_free_object(ctx->memory, Buffer, "pdfi_read_name error"); |
744 | 0 | return code; |
745 | 0 | } |
746 | 15.4M | memcpy(string->data, Buffer, index); |
747 | 15.4M | string->indirect_num = indirect_num; |
748 | 15.4M | string->indirect_gen = indirect_gen; |
749 | | |
750 | 15.4M | gs_free_object(ctx->memory, Buffer, "pdfi_read_string"); |
751 | | |
752 | 15.4M | if (ctx->encryption.is_encrypted && ctx->encryption.decrypt_strings) { |
753 | 13.1k | code = pdfi_decrypt_string(ctx, string); |
754 | 13.1k | if (code < 0) |
755 | 0 | return code; |
756 | 13.1k | } |
757 | | |
758 | 15.4M | if (ctx->args.pdfdebug) { |
759 | 0 | int i; |
760 | 0 | outprintf(ctx->memory, " ("); |
761 | 0 | for (i=0;i<string->length;i++) |
762 | 0 | outprintf(ctx->memory, "%c", string->data[i]); |
763 | 0 | outprintf(ctx->memory, ")"); |
764 | 0 | } |
765 | | |
766 | 15.4M | code = pdfi_push(ctx, (pdf_obj *)string); |
767 | 15.4M | if (code < 0) { |
768 | 0 | pdfi_free_object((pdf_obj *)string); |
769 | 0 | pdfi_set_error(ctx, code, NULL, 0, "pdfi_read_string", NULL); |
770 | 0 | } |
771 | | |
772 | 15.4M | return code; |
773 | 15.4M | } |
774 | | |
775 | | int pdfi_read_dict(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
776 | 9.13k | { |
777 | 9.13k | int code, depth; |
778 | | |
779 | 9.13k | code = pdfi_read_token(ctx, s, indirect_num, indirect_gen); |
780 | 9.13k | if (code < 0) |
781 | 12 | return code; |
782 | 9.12k | if (code == 0) |
783 | 0 | return_error(gs_error_syntaxerror); |
784 | | |
785 | 9.12k | if (pdfi_type_of(ctx->stack_top[-1]) != PDF_DICT_MARK) |
786 | 21 | return_error(gs_error_typecheck); |
787 | 9.10k | depth = pdfi_count_stack(ctx); |
788 | | |
789 | 141k | do { |
790 | 141k | code = pdfi_read_token(ctx, s, indirect_num, indirect_gen); |
791 | 141k | if (code < 0) |
792 | 134 | return code; |
793 | 141k | if (code == 0) |
794 | 17 | return_error(gs_error_syntaxerror); |
795 | 141k | } while(pdfi_count_stack(ctx) > depth); |
796 | 8.95k | return 0; |
797 | 9.10k | } |
798 | | |
799 | | int pdfi_skip_comment(pdf_context *ctx, pdf_c_stream *s) |
800 | 1.87M | { |
801 | 1.87M | int c; |
802 | | |
803 | 1.87M | if (ctx->args.pdfdebug) |
804 | 0 | outprintf (ctx->memory, " %%"); |
805 | | |
806 | 43.3M | do { |
807 | 43.3M | c = pdfi_read_byte(ctx, s); |
808 | 43.3M | if (c < 0) |
809 | 9.97k | break; |
810 | | |
811 | 43.3M | if (ctx->args.pdfdebug) |
812 | 0 | outprintf (ctx->memory, "%c", (char)c); |
813 | | |
814 | 43.3M | } while (c != 0x0a && c != 0x0d); |
815 | | |
816 | 1.87M | return 0; |
817 | 1.87M | } |
818 | | |
819 | | #define PARAM1(A) # A, |
820 | | #define PARAM2(A,B) A, |
821 | | static const char pdf_token_strings[][10] = { |
822 | | #include "pdf_tokens.h" |
823 | | }; |
824 | | |
825 | 123M | #define nelems(A) (sizeof(A)/sizeof(A[0])) |
826 | | |
827 | | typedef int (*bsearch_comparator)(const void *, const void *); |
828 | | |
829 | | int pdfi_read_bare_keyword(pdf_context *ctx, pdf_c_stream *s) |
830 | 3.90M | { |
831 | 3.90M | byte Buffer[256]; |
832 | 3.90M | int code, index = 0; |
833 | 3.90M | int c; |
834 | 3.90M | void *t; |
835 | | |
836 | 3.90M | pdfi_skip_white(ctx, s); |
837 | | |
838 | 24.1M | do { |
839 | 24.1M | c = pdfi_read_byte(ctx, s); |
840 | 24.1M | if (c < 0) |
841 | 26.1k | break; |
842 | | |
843 | 24.1M | if (iswhite(c) || isdelimiter(c)) { |
844 | 3.87M | pdfi_unread_byte(ctx, s, (byte)c); |
845 | 3.87M | break; |
846 | 3.87M | } |
847 | 20.2M | Buffer[index] = (byte)c; |
848 | 20.2M | index++; |
849 | 20.2M | } while (index < 255); |
850 | | |
851 | 3.90M | if (index >= 255 || index == 0) { |
852 | 35.1k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_NOERROR, "pdfi_read_bare_keyword", "")) < 0) { |
853 | 0 | return code; |
854 | 0 | } |
855 | 35.1k | return TOKEN_INVALID_KEY; |
856 | 35.1k | } |
857 | | |
858 | 3.86M | Buffer[index] = 0x00; |
859 | 3.86M | t = bsearch((const void *)Buffer, |
860 | 3.86M | (const void *)pdf_token_strings[TOKEN_INVALID_KEY+1], |
861 | 3.86M | nelems(pdf_token_strings)-(TOKEN_INVALID_KEY+1), |
862 | 3.86M | sizeof(pdf_token_strings[0]), |
863 | 3.86M | (bsearch_comparator)&strcmp); |
864 | 3.86M | if (t == NULL) |
865 | 76.3k | return TOKEN_INVALID_KEY; |
866 | | |
867 | 3.79M | if (ctx->args.pdfdebug) |
868 | 0 | outprintf(ctx->memory, " %s\n", Buffer); |
869 | | |
870 | 3.79M | return (((const char *)t) - pdf_token_strings[0]) / sizeof(pdf_token_strings[0]); |
871 | 3.86M | } |
872 | | |
873 | | static pdf_key lookup_keyword(const byte *Buffer) |
874 | 119M | { |
875 | 119M | void *t = bsearch((const void *)Buffer, |
876 | 119M | (const void *)pdf_token_strings[TOKEN_INVALID_KEY+1], |
877 | 119M | nelems(pdf_token_strings)-(TOKEN_INVALID_KEY+1), |
878 | 119M | sizeof(pdf_token_strings[0]), |
879 | 119M | (bsearch_comparator)&strcmp); |
880 | 119M | if (t == NULL) |
881 | 14.6M | return TOKEN_NOT_A_KEYWORD; |
882 | | |
883 | 104M | return (pdf_key)((((const char *)t) - pdf_token_strings[0]) / |
884 | 104M | sizeof(pdf_token_strings[0])); |
885 | 119M | } |
886 | | |
887 | | /* This function is slightly misnamed. We read 'keywords' from |
888 | | * the stream (including null, true, false and R), and will usually |
889 | | * return them directly as TOKENs cast to be pointers. In the event |
890 | | * that we can't match what we parse to a known keyword, we'll |
891 | | * instead return a PDF_KEYWORD object. In the even that we parse |
892 | | * an 'R', we will return a PDF_INDIRECT object. |
893 | | */ |
894 | | static int pdfi_read_keyword(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
895 | 117M | { |
896 | 117M | byte Buffer[256]; |
897 | 117M | unsigned short index = 0; |
898 | 117M | int c, code; |
899 | 117M | pdf_keyword *keyword; |
900 | 117M | pdf_key key; |
901 | | |
902 | 117M | pdfi_skip_white(ctx, s); |
903 | | |
904 | 464M | do { |
905 | 464M | c = pdfi_read_byte(ctx, s); |
906 | 464M | if (c < 0) |
907 | 228k | break; |
908 | | |
909 | 464M | if (iswhite(c) || isdelimiter(c)) { |
910 | 116M | pdfi_unread_byte(ctx, s, (byte)c); |
911 | 116M | break; |
912 | 116M | } |
913 | 347M | Buffer[index] = (byte)c; |
914 | 347M | index++; |
915 | 347M | } while (index < 255); |
916 | | |
917 | 117M | if (index >= 255 || index == 0) { |
918 | 239k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, 0, "pdfi_read_keyword", NULL)) < 0) { |
919 | 0 | return code; |
920 | 0 | } |
921 | 239k | key = (index >= 255 ? TOKEN_TOO_LONG : TOKEN_INVALID_KEY); |
922 | 239k | index = 0; |
923 | 239k | Buffer[0] = 0; |
924 | 117M | } else { |
925 | 117M | Buffer[index] = 0x00; |
926 | 117M | key = lookup_keyword(Buffer); |
927 | | |
928 | 117M | if (ctx->args.pdfdebug) |
929 | 0 | outprintf(ctx->memory, " %s\n", Buffer); |
930 | | |
931 | 117M | switch (key) { |
932 | 14.5M | case TOKEN_R: |
933 | 14.5M | { |
934 | 14.5M | pdf_indirect_ref *o; |
935 | 14.5M | uint64_t obj_num; |
936 | 14.5M | uint32_t gen_num; |
937 | | |
938 | 14.5M | if(pdfi_count_stack(ctx) < 2) { |
939 | 35.8k | pdfi_clearstack(ctx); |
940 | 35.8k | return_error(gs_error_stackunderflow); |
941 | 35.8k | } |
942 | | |
943 | 14.4M | if(pdfi_type_of(ctx->stack_top[-1]) != PDF_INT || pdfi_type_of(ctx->stack_top[-2]) != PDF_INT) { |
944 | 46.8k | pdfi_clearstack(ctx); |
945 | 46.8k | return_error(gs_error_typecheck); |
946 | 46.8k | } |
947 | | |
948 | 14.4M | gen_num = ((pdf_num *)ctx->stack_top[-1])->value.i; |
949 | 14.4M | pdfi_pop(ctx, 1); |
950 | 14.4M | obj_num = ((pdf_num *)ctx->stack_top[-1])->value.i; |
951 | 14.4M | pdfi_pop(ctx, 1); |
952 | | |
953 | 14.4M | code = pdfi_object_alloc(ctx, PDF_INDIRECT, 0, (pdf_obj **)&o); |
954 | 14.4M | if (code < 0) |
955 | 0 | return code; |
956 | | |
957 | 14.4M | o->ref_generation_num = gen_num; |
958 | 14.4M | o->ref_object_num = obj_num; |
959 | 14.4M | o->indirect_num = indirect_num; |
960 | 14.4M | o->indirect_gen = indirect_gen; |
961 | | |
962 | 14.4M | code = pdfi_push(ctx, (pdf_obj *)o); |
963 | 14.4M | if (code < 0) |
964 | 0 | pdfi_free_object((pdf_obj *)o); |
965 | | |
966 | 14.4M | return code; |
967 | 14.4M | } |
968 | 14.6M | case TOKEN_NOT_A_KEYWORD: |
969 | | /* Unexpected keyword found. We'll allocate an object for the buffer below. */ |
970 | 14.6M | break; |
971 | 2.68M | case TOKEN_STREAM: |
972 | 2.68M | code = pdfi_skip_eol(ctx, s); |
973 | 2.68M | if (code < 0) |
974 | 0 | return code; |
975 | | /* fallthrough */ |
976 | 3.34M | case TOKEN_PDF_TRUE: |
977 | 3.75M | case TOKEN_PDF_FALSE: |
978 | 3.82M | case TOKEN_null: |
979 | 87.8M | default: |
980 | | /* This is the fast, common exit case. We just push the key |
981 | | * onto the stack. No allocation required. No deallocation |
982 | | * in the case of error. */ |
983 | 87.8M | return pdfi_push(ctx, (pdf_obj *)(intptr_t)key); |
984 | 117M | } |
985 | 117M | } |
986 | | |
987 | | /* Unexpected keyword. We can't handle this with the fast no-allocation case. */ |
988 | 14.9M | code = pdfi_object_alloc(ctx, PDF_KEYWORD, index, (pdf_obj **)&keyword); |
989 | 14.9M | if (code < 0) |
990 | 0 | return code; |
991 | | |
992 | 14.9M | if (index) |
993 | 14.6M | memcpy(keyword->data, Buffer, index); |
994 | | |
995 | | /* keyword->length set as part of allocation. */ |
996 | 14.9M | keyword->indirect_num = indirect_num; |
997 | 14.9M | keyword->indirect_gen = indirect_gen; |
998 | | |
999 | 14.9M | code = pdfi_push(ctx, (pdf_obj *)keyword); |
1000 | 14.9M | if (code < 0) |
1001 | 30 | pdfi_free_object((pdf_obj *)keyword); |
1002 | | |
1003 | 14.9M | return code; |
1004 | 14.9M | } |
1005 | | |
1006 | | /* This function reads from the given stream, at the current offset in the stream, |
1007 | | * a single PDF 'token' and returns it on the stack. |
1008 | | */ |
1009 | | int pdfi_read_token(pdf_context *ctx, pdf_c_stream *s, uint32_t indirect_num, uint32_t indirect_gen) |
1010 | 572M | { |
1011 | 572M | int c, code; |
1012 | | |
1013 | 576M | rescan: |
1014 | 576M | pdfi_skip_white(ctx, s); |
1015 | | |
1016 | 576M | c = pdfi_read_byte(ctx, s); |
1017 | 576M | if (c == EOFC) |
1018 | 297k | return 0; |
1019 | 576M | if (c < 0) |
1020 | 13.4k | return_error(gs_error_ioerror); |
1021 | | |
1022 | 576M | switch(c) { |
1023 | 73.7M | case 0x30: |
1024 | 118M | case 0x31: |
1025 | 152M | case 0x32: |
1026 | 190M | case 0x33: |
1027 | 222M | case 0x34: |
1028 | 241M | case 0x35: |
1029 | 259M | case 0x36: |
1030 | 271M | case 0x37: |
1031 | 278M | case 0x38: |
1032 | 283M | case 0x39: |
1033 | 283M | case '+': |
1034 | 298M | case '-': |
1035 | 299M | case '.': |
1036 | 299M | pdfi_unread_byte(ctx, s, (byte)c); |
1037 | 299M | code = pdfi_read_num(ctx, s, indirect_num, indirect_gen); |
1038 | 299M | if (code < 0) |
1039 | 49.6k | return code; |
1040 | 299M | break; |
1041 | 299M | case '/': |
1042 | 85.3M | code = pdfi_read_name(ctx, s, indirect_num, indirect_gen); |
1043 | 85.3M | if (code < 0) |
1044 | 0 | return code; |
1045 | 85.3M | return 1; |
1046 | 0 | break; |
1047 | 16.2M | case '<': |
1048 | 16.2M | c = pdfi_read_byte(ctx, s); |
1049 | 16.2M | if (c < 0) |
1050 | 366 | return (gs_error_ioerror); |
1051 | 16.2M | if (iswhite(c)) { |
1052 | 49.6k | code = pdfi_skip_white(ctx, s); |
1053 | 49.6k | if (code < 0) |
1054 | 0 | return code; |
1055 | 49.6k | c = pdfi_read_byte(ctx, s); |
1056 | 49.6k | } |
1057 | 16.2M | if (c == '<') { |
1058 | 11.5M | if (ctx->args.pdfdebug) |
1059 | 0 | outprintf (ctx->memory, " <<\n"); |
1060 | 11.5M | if (ctx->object_nesting < MAX_NESTING_DEPTH) { |
1061 | 10.9M | ctx->object_nesting++; |
1062 | 10.9M | code = pdfi_mark_stack(ctx, PDF_DICT_MARK); |
1063 | 10.9M | if (code < 0) |
1064 | 1 | return code; |
1065 | 10.9M | } |
1066 | 564k | else if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_limitcheck), NULL, E_PDF_NESTEDTOODEEP, "pdfi_read_token", NULL)) < 0) { |
1067 | 0 | return code; |
1068 | 0 | } |
1069 | 11.5M | return 1; |
1070 | 11.5M | } else if (c == '>') { |
1071 | 29.1k | pdfi_unread_byte(ctx, s, (byte)c); |
1072 | 29.1k | code = pdfi_read_hexstring(ctx, s, indirect_num, indirect_gen); |
1073 | 29.1k | if (code < 0) |
1074 | 0 | return code; |
1075 | 29.1k | return 1; |
1076 | 4.71M | } else if (ishex(c)) { |
1077 | 4.14M | pdfi_unread_byte(ctx, s, (byte)c); |
1078 | 4.14M | code = pdfi_read_hexstring(ctx, s, indirect_num, indirect_gen); |
1079 | 4.14M | if (code < 0) |
1080 | 282k | return code; |
1081 | 4.14M | } |
1082 | 562k | else |
1083 | 562k | return_error(gs_error_syntaxerror); |
1084 | 3.86M | break; |
1085 | 11.9M | case '>': |
1086 | 11.9M | c = pdfi_read_byte(ctx, s); |
1087 | 11.9M | if (c < 0) |
1088 | 712 | return (gs_error_ioerror); |
1089 | 11.9M | if (c == '>') { |
1090 | 11.1M | if (ctx->object_nesting > 0) { |
1091 | 10.5M | ctx->object_nesting--; |
1092 | 10.5M | code = pdfi_dict_from_stack(ctx, indirect_num, indirect_gen, false); |
1093 | 10.5M | if (code < 0) |
1094 | 358k | return code; |
1095 | 10.5M | } else { |
1096 | 663k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_unmatchedmark), NULL, E_PDF_UNMATCHEDMARK, "pdfi_read_token", NULL)) < 0) { |
1097 | 0 | return code; |
1098 | 0 | } |
1099 | 663k | goto rescan; |
1100 | 663k | } |
1101 | 10.1M | return 1; |
1102 | 11.1M | } else { |
1103 | 797k | pdfi_unread_byte(ctx, s, (byte)c); |
1104 | 797k | return_error(gs_error_syntaxerror); |
1105 | 797k | } |
1106 | 0 | break; |
1107 | 15.4M | case '(': |
1108 | 15.4M | code = pdfi_read_string(ctx, s, indirect_num, indirect_gen); |
1109 | 15.4M | if (code < 0) |
1110 | 0 | return code; |
1111 | 15.4M | return 1; |
1112 | 0 | break; |
1113 | 13.9M | case '[': |
1114 | 13.9M | if (ctx->args.pdfdebug) |
1115 | 0 | outprintf (ctx->memory, "["); |
1116 | 13.9M | if (ctx->object_nesting < MAX_NESTING_DEPTH) { |
1117 | 12.6M | ctx->object_nesting++; |
1118 | 12.6M | code = pdfi_mark_stack(ctx, PDF_ARRAY_MARK); |
1119 | 12.6M | if (code < 0) |
1120 | 0 | return code; |
1121 | 12.6M | } else |
1122 | 1.37M | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_NESTEDTOODEEP, "pdfi_read_token", NULL)) < 0) |
1123 | 0 | return code; |
1124 | 13.9M | return 1; |
1125 | 0 | break; |
1126 | 12.7M | case ']': |
1127 | 12.7M | if (ctx->object_nesting > 0) { |
1128 | 12.5M | ctx->object_nesting--; |
1129 | 12.5M | code = pdfi_array_from_stack(ctx, indirect_num, indirect_gen); |
1130 | 12.5M | if (code < 0) |
1131 | 215k | return code; |
1132 | 12.5M | } else { |
1133 | 296k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_unmatchedmark), NULL, E_PDF_UNMATCHEDMARK, "pdfi_read_token", NULL)) < 0) { |
1134 | 0 | return code; |
1135 | 0 | } |
1136 | 296k | goto rescan; |
1137 | 296k | } |
1138 | 12.2M | break; |
1139 | 12.2M | case '{': |
1140 | 486k | if (ctx->args.pdfdebug) |
1141 | 0 | outprintf (ctx->memory, "{"); |
1142 | 486k | code = pdfi_mark_stack(ctx, PDF_PROC_MARK); |
1143 | 486k | if (code < 0) |
1144 | 0 | return code; |
1145 | 486k | return 1; |
1146 | 0 | break; |
1147 | 324k | case '}': |
1148 | 324k | pdfi_clear_to_mark(ctx); |
1149 | 324k | goto rescan; |
1150 | 0 | break; |
1151 | 1.74M | case '%': |
1152 | 1.74M | pdfi_skip_comment(ctx, s); |
1153 | 1.74M | goto rescan; |
1154 | 0 | break; |
1155 | 117M | default: |
1156 | 117M | if (isdelimiter(c)) { |
1157 | 661k | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, 0, "pdfi_read_token", NULL)) < 0) { |
1158 | 0 | return code; |
1159 | 0 | } |
1160 | 661k | goto rescan; |
1161 | 661k | } |
1162 | 117M | pdfi_unread_byte(ctx, s, (byte)c); |
1163 | 117M | code = pdfi_read_keyword(ctx, s, indirect_num, indirect_gen); |
1164 | 117M | if (code < 0) |
1165 | 82.7k | return code; |
1166 | 117M | return 1; |
1167 | 0 | break; |
1168 | 576M | } |
1169 | 315M | return 1; |
1170 | 576M | } |
1171 | | |
1172 | | /* In contrast to the 'read' functions, the 'make' functions create an object with a |
1173 | | * reference count of 1. This indicates that the caller holds the reference. Thus the |
1174 | | * caller need not increment the reference count to the object, but must decrement |
1175 | | * it (pdf_countdown) before exiting. |
1176 | | */ |
1177 | | int pdfi_name_alloc(pdf_context *ctx, byte *n, uint32_t size, pdf_obj **o) |
1178 | 175M | { |
1179 | 175M | int code; |
1180 | 175M | *o = NULL; |
1181 | | |
1182 | 175M | code = pdfi_object_alloc(ctx, PDF_NAME, size, o); |
1183 | 175M | if (code < 0) |
1184 | 0 | return code; |
1185 | | |
1186 | 175M | memcpy(((pdf_name *)*o)->data, n, size); |
1187 | | |
1188 | 175M | return 0; |
1189 | 175M | } |
1190 | | |
1191 | | static char op_table_3[5][3] = { |
1192 | | "BDC", "BMC", "EMC", "SCN", "scn" |
1193 | | }; |
1194 | | |
1195 | | static char op_table_2[39][2] = { |
1196 | | "b*", "BI", "BT", "BX", "cm", "CS", "cs", "EI", "d0", "d1", "Do", "DP", "ET", "EX", "f*", "gs", "ID", "MP", "re", "RG", |
1197 | | "rg", "ri", "SC", "sc", "sh", "T*", "Tc", "Td", "TD", "Tf", "Tj", "TJ", "TL", "Tm", "Tr", "Ts", "Tw", "Tz", "W*", |
1198 | | }; |
1199 | | |
1200 | | static char op_table_1[27][1] = { |
1201 | | "b", "B", "c", "d", "f", "F", "G", "g", "h", "i", "j", "J", "K", "k", "l", "m", "n", "q", "Q", "s", "S", "v", "w", "W", |
1202 | | "y", "'", "\"" |
1203 | | }; |
1204 | | |
1205 | | /* forward definition for the 'split_bogus_operator' function to use */ |
1206 | | static int pdfi_interpret_stream_operator(pdf_context *ctx, pdf_c_stream *source, |
1207 | | pdf_dict *stream_dict, pdf_dict *page_dict); |
1208 | | |
1209 | | static int |
1210 | | make_keyword_obj(pdf_context *ctx, const byte *data, int length, pdf_keyword **pkey) |
1211 | 2.19M | { |
1212 | 2.19M | byte Buffer[256]; |
1213 | 2.19M | pdf_key key; |
1214 | 2.19M | int code; |
1215 | | |
1216 | 2.19M | if (length > 255) |
1217 | 0 | return_error(gs_error_rangecheck); |
1218 | | |
1219 | 2.19M | memcpy(Buffer, data, length); |
1220 | 2.19M | Buffer[length] = 0; |
1221 | 2.19M | key = lookup_keyword(Buffer); |
1222 | 2.19M | if (key != TOKEN_INVALID_KEY) { |
1223 | | /* The common case. We've found a real key, just cast the token to |
1224 | | * a pointer, and return that. */ |
1225 | 2.19M | *pkey = (pdf_keyword *)PDF_TOKEN_AS_OBJ(key); |
1226 | 2.19M | return 1; |
1227 | 2.19M | } |
1228 | | /* We still haven't found a real keyword. Allocate a new object and |
1229 | | * return it. */ |
1230 | 0 | code = pdfi_object_alloc(ctx, PDF_KEYWORD, length, (pdf_obj **)pkey); |
1231 | 0 | if (code < 0) |
1232 | 0 | return code; |
1233 | 0 | if (length) |
1234 | 0 | memcpy((*pkey)->data, Buffer, length); |
1235 | 0 | pdfi_countup(*pkey); |
1236 | |
|
1237 | 0 | return 1; |
1238 | 0 | } |
1239 | | |
1240 | | static int search_table_3(pdf_context *ctx, unsigned char *str, pdf_keyword **key) |
1241 | 2.08M | { |
1242 | 2.08M | int i; |
1243 | | |
1244 | 12.5M | for (i = 0; i < 5; i++) { |
1245 | 10.4M | if (memcmp(str, op_table_3[i], 3) == 0) |
1246 | 7.80k | return make_keyword_obj(ctx, str, 3, key); |
1247 | 10.4M | } |
1248 | 2.07M | return 0; |
1249 | 2.08M | } |
1250 | | |
1251 | | static int search_table_2(pdf_context *ctx, unsigned char *str, pdf_keyword **key) |
1252 | 4.43M | { |
1253 | 4.43M | int i; |
1254 | | |
1255 | 171M | for (i = 0; i < 39; i++) { |
1256 | 167M | if (memcmp(str, op_table_2[i], 2) == 0) |
1257 | 273k | return make_keyword_obj(ctx, str, 2, key); |
1258 | 167M | } |
1259 | 4.16M | return 0; |
1260 | 4.43M | } |
1261 | | |
1262 | | static int search_table_1(pdf_context *ctx, unsigned char *str, pdf_keyword **key) |
1263 | 4.10M | { |
1264 | 4.10M | int i; |
1265 | | |
1266 | 88.2M | for (i = 0; i < 27; i++) { |
1267 | 86.0M | if (memcmp(str, op_table_1[i], 1) == 0) |
1268 | 1.91M | return make_keyword_obj(ctx, str, 1, key); |
1269 | 86.0M | } |
1270 | 2.19M | return 0; |
1271 | 4.10M | } |
1272 | | |
1273 | | static int split_bogus_operator(pdf_context *ctx, pdf_c_stream *source, pdf_dict *stream_dict, pdf_dict *page_dict) |
1274 | 7.94M | { |
1275 | 7.94M | int code = 0; |
1276 | 7.94M | pdf_keyword *keyword = (pdf_keyword *)ctx->stack_top[-1], *key1 = NULL, *key2 = NULL; |
1277 | 7.94M | int length = keyword->length - 6; |
1278 | | |
1279 | 7.94M | if (length > 0) { |
1280 | | /* Longer than 2 3-character operators, we only allow for up to two |
1281 | | * operators. Check to see if it includes an endstream or endobj. |
1282 | | */ |
1283 | 2.74M | if (memcmp(&keyword->data[length], "endobj", 6) == 0) { |
1284 | | /* Keyword is "<something>endobj". So make a keyword just from |
1285 | | * <something>, push that, execute it, then push endobj. */ |
1286 | 24 | code = make_keyword_obj(ctx, keyword->data, length, &key1); |
1287 | 24 | if (code < 0) |
1288 | 0 | goto error_exit; |
1289 | 24 | pdfi_pop(ctx, 1); |
1290 | 24 | pdfi_push(ctx, (pdf_obj *)key1); |
1291 | 24 | pdfi_countdown(key1); /* Drop the reference returned by make_keyword_obj. */ |
1292 | 24 | code = pdfi_interpret_stream_operator(ctx, source, stream_dict, page_dict); |
1293 | 24 | if (code < 0) |
1294 | 15 | goto error_exit; |
1295 | 9 | pdfi_push(ctx, PDF_TOKEN_AS_OBJ(TOKEN_ENDOBJ)); |
1296 | 9 | return 0; |
1297 | 2.74M | } else { |
1298 | 2.74M | length = keyword->length - 9; |
1299 | 2.74M | if (length > 0 && memcmp(&keyword->data[length], "endstream", 9) == 0) { |
1300 | | /* Keyword is "<something>endstream". So make a keyword just from |
1301 | | * <something>, push that, execute it, then push endstream. */ |
1302 | 2 | code = make_keyword_obj(ctx, keyword->data, length, &key1); |
1303 | 2 | if (code < 0) |
1304 | 0 | goto error_exit; |
1305 | 2 | pdfi_pop(ctx, 1); |
1306 | 2 | pdfi_push(ctx, (pdf_obj *)key1); |
1307 | 2 | pdfi_countdown(key1); /* Drop the reference returned by make_keyword_obj. */ |
1308 | 2 | code = pdfi_interpret_stream_operator(ctx, source, stream_dict, page_dict); |
1309 | 2 | if (code < 0) |
1310 | 0 | goto error_exit; |
1311 | 2 | pdfi_push(ctx, PDF_TOKEN_AS_OBJ(TOKEN_ENDSTREAM)); |
1312 | 2 | return 0; |
1313 | 2.74M | } else { |
1314 | 2.74M | pdfi_clearstack(ctx); |
1315 | 2.74M | return 0; |
1316 | 2.74M | } |
1317 | 2.74M | } |
1318 | 2.74M | } |
1319 | | |
1320 | 5.19M | if (keyword->length > 3) { |
1321 | 1.65M | code = search_table_3(ctx, keyword->data, &key1); |
1322 | 1.65M | if (code < 0) |
1323 | 0 | goto error_exit; |
1324 | | |
1325 | 1.65M | if (code > 0) { |
1326 | 6.43k | switch (keyword->length - 3) { |
1327 | 4.95k | case 1: |
1328 | 4.95k | code = search_table_1(ctx, &keyword->data[3], &key2); |
1329 | 4.95k | break; |
1330 | 829 | case 2: |
1331 | 829 | code = search_table_2(ctx, &keyword->data[3], &key2); |
1332 | 829 | break; |
1333 | 656 | case 3: |
1334 | 656 | code = search_table_3(ctx, &keyword->data[3], &key2); |
1335 | 656 | break; |
1336 | 0 | default: |
1337 | 0 | goto error_exit; |
1338 | 6.43k | } |
1339 | 6.43k | } |
1340 | 1.65M | if (code < 0) |
1341 | 0 | goto error_exit; |
1342 | 1.65M | if (code > 0) |
1343 | 1.30k | goto match; |
1344 | 1.65M | } |
1345 | 5.19M | pdfi_countdown(key1); |
1346 | 5.19M | pdfi_countdown(key2); |
1347 | 5.19M | key1 = NULL; |
1348 | 5.19M | key2 = NULL; |
1349 | | |
1350 | 5.19M | if (keyword->length > 5 || keyword->length < 2) |
1351 | 1.31M | goto error_exit; |
1352 | | |
1353 | 3.87M | code = search_table_2(ctx, keyword->data, &key1); |
1354 | 3.87M | if (code < 0) |
1355 | 0 | goto error_exit; |
1356 | | |
1357 | 3.87M | if (code > 0) { |
1358 | 232k | switch(keyword->length - 2) { |
1359 | 88.0k | case 1: |
1360 | 88.0k | code = search_table_1(ctx, &keyword->data[2], &key2); |
1361 | 88.0k | break; |
1362 | 87.1k | case 2: |
1363 | 87.1k | code = search_table_2(ctx, &keyword->data[2], &key2); |
1364 | 87.1k | break; |
1365 | 56.8k | case 3: |
1366 | 56.8k | code = search_table_3(ctx, &keyword->data[2], &key2); |
1367 | 56.8k | break; |
1368 | 0 | default: |
1369 | 0 | goto error_exit; |
1370 | 232k | } |
1371 | 232k | if (code < 0) |
1372 | 0 | goto error_exit; |
1373 | 232k | if (code > 0) |
1374 | 41.6k | goto match; |
1375 | 232k | } |
1376 | 3.83M | pdfi_countdown(key1); |
1377 | 3.83M | pdfi_countdown(key2); |
1378 | 3.83M | key1 = NULL; |
1379 | 3.83M | key2 = NULL; |
1380 | | |
1381 | 3.83M | if (keyword->length > 4) |
1382 | 600k | goto error_exit; |
1383 | | |
1384 | 3.23M | code = search_table_1(ctx, keyword->data, &key1); |
1385 | 3.23M | if (code <= 0) |
1386 | 1.61M | goto error_exit; |
1387 | | |
1388 | 1.62M | switch(keyword->length - 1) { |
1389 | 772k | case 1: |
1390 | 772k | code = search_table_1(ctx, &keyword->data[1], &key2); |
1391 | 772k | break; |
1392 | 470k | case 2: |
1393 | 470k | code = search_table_2(ctx, &keyword->data[1], &key2); |
1394 | 470k | break; |
1395 | 377k | case 3: |
1396 | 377k | code = search_table_3(ctx, &keyword->data[1], &key2); |
1397 | 377k | break; |
1398 | 0 | default: |
1399 | 0 | goto error_exit; |
1400 | 1.62M | } |
1401 | 1.62M | if (code <= 0) |
1402 | 1.33M | goto error_exit; |
1403 | | |
1404 | 333k | match: |
1405 | 333k | pdfi_set_warning(ctx, 0, NULL, W_PDF_MISSING_WHITE_OPS, "split_bogus_operator", NULL); |
1406 | | /* If we get here, we have two PDF_KEYWORD objects. We push them on the stack |
1407 | | * one at a time, and execute them. |
1408 | | */ |
1409 | 333k | pdfi_push(ctx, (pdf_obj *)key1); |
1410 | 333k | code = pdfi_interpret_stream_operator(ctx, source, stream_dict, page_dict); |
1411 | 333k | if (code < 0) |
1412 | 257k | goto error_exit; |
1413 | | |
1414 | 75.5k | pdfi_push(ctx, (pdf_obj *)key2); |
1415 | 75.5k | code = pdfi_interpret_stream_operator(ctx, source, stream_dict, page_dict); |
1416 | | |
1417 | 75.5k | pdfi_countdown(key1); |
1418 | 75.5k | pdfi_countdown(key2); |
1419 | 75.5k | pdfi_clearstack(ctx); |
1420 | 75.5k | return code; |
1421 | | |
1422 | 5.12M | error_exit: |
1423 | 5.12M | pdfi_set_error(ctx, code, NULL, E_PDF_TOKENERROR, "split_bogus_operator", NULL); |
1424 | 5.12M | pdfi_countdown(key1); |
1425 | 5.12M | pdfi_countdown(key2); |
1426 | 5.12M | pdfi_clearstack(ctx); |
1427 | 5.12M | return code; |
1428 | 333k | } |
1429 | | |
1430 | | static int pdfi_interpret_stream_operator(pdf_context *ctx, pdf_c_stream *source, |
1431 | | pdf_dict *stream_dict, pdf_dict *page_dict) |
1432 | 77.5M | { |
1433 | 77.5M | pdf_obj *keyword = ctx->stack_top[-1]; |
1434 | 77.5M | int code = 0; |
1435 | | |
1436 | 77.5M | if (keyword < PDF_TOKEN_AS_OBJ(TOKEN__LAST_KEY)) |
1437 | 69.6M | { |
1438 | 69.6M | switch((uintptr_t)keyword) { |
1439 | 16.0k | case TOKEN_b: /* closepath, fill, stroke */ |
1440 | 16.0k | pdfi_pop(ctx, 1); |
1441 | 16.0k | code = pdfi_b(ctx); |
1442 | 16.0k | break; |
1443 | 40.3k | case TOKEN_B: /* fill, stroke */ |
1444 | 40.3k | pdfi_pop(ctx, 1); |
1445 | 40.3k | code = pdfi_B(ctx); |
1446 | 40.3k | break; |
1447 | 495 | case TOKEN_bstar: /* closepath, eofill, stroke */ |
1448 | 495 | pdfi_pop(ctx, 1); |
1449 | 495 | code = pdfi_b_star(ctx); |
1450 | 495 | break; |
1451 | 1.78k | case TOKEN_Bstar: /* eofill, stroke */ |
1452 | 1.78k | pdfi_pop(ctx, 1); |
1453 | 1.78k | code = pdfi_B_star(ctx); |
1454 | 1.78k | break; |
1455 | 255k | case TOKEN_BI: /* begin inline image */ |
1456 | 255k | pdfi_pop(ctx, 1); |
1457 | 255k | code = pdfi_BI(ctx); |
1458 | 255k | break; |
1459 | 314k | case TOKEN_BDC: /* begin marked content sequence with property list */ |
1460 | 314k | pdfi_pop(ctx, 1); |
1461 | 314k | code = pdfi_op_BDC(ctx, stream_dict, page_dict); |
1462 | 314k | break; |
1463 | 17.5k | case TOKEN_BMC: /* begin marked content sequence */ |
1464 | 17.5k | pdfi_pop(ctx, 1); |
1465 | 17.5k | code = pdfi_op_BMC(ctx); |
1466 | 17.5k | break; |
1467 | 1.78M | case TOKEN_BT: /* begin text */ |
1468 | 1.78M | pdfi_pop(ctx, 1); |
1469 | 1.78M | code = pdfi_BT(ctx); |
1470 | 1.78M | break; |
1471 | 49.8k | case TOKEN_BX: /* begin compatibility section */ |
1472 | 49.8k | pdfi_pop(ctx, 1); |
1473 | 49.8k | break; |
1474 | 10.0M | case TOKEN_c: /* curveto */ |
1475 | 10.0M | pdfi_pop(ctx, 1); |
1476 | 10.0M | code = pdfi_curveto(ctx); |
1477 | 10.0M | break; |
1478 | 2.56M | case TOKEN_cm: /* concat */ |
1479 | 2.56M | pdfi_pop(ctx, 1); |
1480 | 2.56M | code = pdfi_concat(ctx); |
1481 | 2.56M | break; |
1482 | 28.8k | case TOKEN_CS: /* set stroke colour space */ |
1483 | 28.8k | pdfi_pop(ctx, 1); |
1484 | 28.8k | code = pdfi_setstrokecolor_space(ctx, stream_dict, page_dict); |
1485 | 28.8k | break; |
1486 | 236k | case TOKEN_cs: /* set non-stroke colour space */ |
1487 | 236k | pdfi_pop(ctx, 1); |
1488 | 236k | code = pdfi_setfillcolor_space(ctx, stream_dict, page_dict); |
1489 | 236k | break; |
1490 | 726k | case TOKEN_d: /* set dash params */ |
1491 | 726k | pdfi_pop(ctx, 1); |
1492 | 726k | code = pdfi_setdash(ctx); |
1493 | 726k | break; |
1494 | 3.26k | case TOKEN_d0: /* set type 3 font glyph width */ |
1495 | 3.26k | pdfi_pop(ctx, 1); |
1496 | 3.26k | code = pdfi_d0(ctx); |
1497 | 3.26k | break; |
1498 | 39.9k | case TOKEN_d1: /* set type 3 font glyph width and bounding box */ |
1499 | 39.9k | pdfi_pop(ctx, 1); |
1500 | 39.9k | code = pdfi_d1(ctx); |
1501 | 39.9k | break; |
1502 | 260k | case TOKEN_Do: /* invoke named XObject */ |
1503 | 260k | pdfi_pop(ctx, 1); |
1504 | 260k | code = pdfi_Do(ctx, stream_dict, page_dict); |
1505 | 260k | break; |
1506 | 783 | case TOKEN_DP: /* define marked content point with property list */ |
1507 | 783 | pdfi_pop(ctx, 1); |
1508 | 783 | code = pdfi_op_DP(ctx, stream_dict, page_dict); |
1509 | 783 | break; |
1510 | 250k | case TOKEN_EI: /* end inline image */ |
1511 | 250k | pdfi_pop(ctx, 1); |
1512 | 250k | code = pdfi_EI(ctx); |
1513 | 250k | break; |
1514 | 1.75M | case TOKEN_ET: /* end text */ |
1515 | 1.75M | pdfi_pop(ctx, 1); |
1516 | 1.75M | code = pdfi_ET(ctx); |
1517 | 1.75M | break; |
1518 | 324k | case TOKEN_EMC: /* end marked content sequence */ |
1519 | 324k | pdfi_pop(ctx, 1); |
1520 | 324k | code = pdfi_op_EMC(ctx); |
1521 | 324k | break; |
1522 | 47.5k | case TOKEN_EX: /* end compatibility section */ |
1523 | 47.5k | pdfi_pop(ctx, 1); |
1524 | 47.5k | break; |
1525 | 2.17M | case TOKEN_f: /* fill */ |
1526 | 2.17M | pdfi_pop(ctx, 1); |
1527 | 2.17M | code = pdfi_fill(ctx); |
1528 | 2.17M | break; |
1529 | 14.6k | case TOKEN_F: /* fill (obselete operator) */ |
1530 | 14.6k | pdfi_pop(ctx, 1); |
1531 | 14.6k | code = pdfi_fill(ctx); |
1532 | 14.6k | break; |
1533 | 66.3k | case TOKEN_fstar: /* eofill */ |
1534 | 66.3k | pdfi_pop(ctx, 1); |
1535 | 66.3k | code = pdfi_eofill(ctx); |
1536 | 66.3k | break; |
1537 | 479k | case TOKEN_G: /* setgray for stroke */ |
1538 | 479k | pdfi_pop(ctx, 1); |
1539 | 479k | code = pdfi_setgraystroke(ctx); |
1540 | 479k | break; |
1541 | 594k | case TOKEN_g: /* setgray for non-stroke */ |
1542 | 594k | pdfi_pop(ctx, 1); |
1543 | 594k | code = pdfi_setgrayfill(ctx); |
1544 | 594k | break; |
1545 | 355k | case TOKEN_gs: /* set graphics state from dictionary */ |
1546 | 355k | pdfi_pop(ctx, 1); |
1547 | 355k | code = pdfi_setgstate(ctx, stream_dict, page_dict); |
1548 | 355k | break; |
1549 | 487k | case TOKEN_h: /* closepath */ |
1550 | 487k | pdfi_pop(ctx, 1); |
1551 | 487k | code = pdfi_closepath(ctx); |
1552 | 487k | break; |
1553 | 151k | case TOKEN_i: /* setflat */ |
1554 | 151k | pdfi_pop(ctx, 1); |
1555 | 151k | code = pdfi_setflat(ctx); |
1556 | 151k | break; |
1557 | 255k | case TOKEN_ID: /* begin inline image data */ |
1558 | 255k | pdfi_pop(ctx, 1); |
1559 | 255k | code = pdfi_ID(ctx, stream_dict, page_dict, source); |
1560 | 255k | break; |
1561 | 711k | case TOKEN_j: /* setlinejoin */ |
1562 | 711k | pdfi_pop(ctx, 1); |
1563 | 711k | code = pdfi_setlinejoin(ctx); |
1564 | 711k | break; |
1565 | 850k | case TOKEN_J: /* setlinecap */ |
1566 | 850k | pdfi_pop(ctx, 1); |
1567 | 850k | code = pdfi_setlinecap(ctx); |
1568 | 850k | break; |
1569 | 43.1k | case TOKEN_K: /* setcmyk for non-stroke */ |
1570 | 43.1k | pdfi_pop(ctx, 1); |
1571 | 43.1k | code = pdfi_setcmykstroke(ctx); |
1572 | 43.1k | break; |
1573 | 142k | case TOKEN_k: /* setcmyk for non-stroke */ |
1574 | 142k | pdfi_pop(ctx, 1); |
1575 | 142k | code = pdfi_setcmykfill(ctx); |
1576 | 142k | break; |
1577 | 11.3M | case TOKEN_l: /* lineto */ |
1578 | 11.3M | pdfi_pop(ctx, 1); |
1579 | 11.3M | code = pdfi_lineto(ctx); |
1580 | 11.3M | break; |
1581 | 5.85M | case TOKEN_m: /* moveto */ |
1582 | 5.85M | pdfi_pop(ctx, 1); |
1583 | 5.85M | code = pdfi_moveto(ctx); |
1584 | 5.85M | break; |
1585 | 41.5k | case TOKEN_M: /* setmiterlimit */ |
1586 | 41.5k | pdfi_pop(ctx, 1); |
1587 | 41.5k | code = pdfi_setmiterlimit(ctx); |
1588 | 41.5k | break; |
1589 | 2.58k | case TOKEN_MP: /* define marked content point */ |
1590 | 2.58k | pdfi_pop(ctx, 1); |
1591 | 2.58k | code = pdfi_op_MP(ctx); |
1592 | 2.58k | break; |
1593 | 1.11M | case TOKEN_n: /* newpath */ |
1594 | 1.11M | pdfi_pop(ctx, 1); |
1595 | 1.11M | code = pdfi_newpath(ctx); |
1596 | 1.11M | break; |
1597 | 2.89M | case TOKEN_q: /* gsave */ |
1598 | 2.89M | pdfi_pop(ctx, 1); |
1599 | 2.89M | code = pdfi_op_q(ctx); |
1600 | 2.89M | break; |
1601 | 2.81M | case TOKEN_Q: /* grestore */ |
1602 | 2.81M | pdfi_pop(ctx, 1); |
1603 | 2.81M | code = pdfi_op_Q(ctx); |
1604 | 2.81M | break; |
1605 | 47.0k | case TOKEN_r: /* non-standard set rgb colour for non-stroke */ |
1606 | 47.0k | pdfi_pop(ctx, 1); |
1607 | 47.0k | code = pdfi_setrgbfill_array(ctx); |
1608 | 47.0k | break; |
1609 | 2.57M | case TOKEN_re: /* append rectangle */ |
1610 | 2.57M | pdfi_pop(ctx, 1); |
1611 | 2.57M | code = pdfi_rectpath(ctx); |
1612 | 2.57M | break; |
1613 | 829k | case TOKEN_RG: /* set rgb colour for stroke */ |
1614 | 829k | pdfi_pop(ctx, 1); |
1615 | 829k | code = pdfi_setrgbstroke(ctx); |
1616 | 829k | break; |
1617 | 904k | case TOKEN_rg: /* set rgb colour for non-stroke */ |
1618 | 904k | pdfi_pop(ctx, 1); |
1619 | 904k | code = pdfi_setrgbfill(ctx); |
1620 | 904k | break; |
1621 | 87.3k | case TOKEN_ri: /* set rendering intent */ |
1622 | 87.3k | pdfi_pop(ctx, 1); |
1623 | 87.3k | code = pdfi_ri(ctx); |
1624 | 87.3k | break; |
1625 | 76.5k | case TOKEN_s: /* closepath, stroke */ |
1626 | 76.5k | pdfi_pop(ctx, 1); |
1627 | 76.5k | code = pdfi_closepath_stroke(ctx); |
1628 | 76.5k | break; |
1629 | 2.77M | case TOKEN_S: /* stroke */ |
1630 | 2.77M | pdfi_pop(ctx, 1); |
1631 | 2.77M | code = pdfi_stroke(ctx); |
1632 | 2.77M | break; |
1633 | 36.8k | case TOKEN_SC: /* set colour for stroke */ |
1634 | 36.8k | pdfi_pop(ctx, 1); |
1635 | 36.8k | code = pdfi_setstrokecolor(ctx); |
1636 | 36.8k | break; |
1637 | 151k | case TOKEN_sc: /* set colour for non-stroke */ |
1638 | 151k | pdfi_pop(ctx, 1); |
1639 | 151k | code = pdfi_setfillcolor(ctx); |
1640 | 151k | break; |
1641 | 12.5k | case TOKEN_SCN: /* set special colour for stroke */ |
1642 | 12.5k | pdfi_pop(ctx, 1); |
1643 | 12.5k | code = pdfi_setcolorN(ctx, stream_dict, page_dict, false); |
1644 | 12.5k | break; |
1645 | 129k | case TOKEN_scn: /* set special colour for non-stroke */ |
1646 | 129k | pdfi_pop(ctx, 1); |
1647 | 129k | code = pdfi_setcolorN(ctx, stream_dict, page_dict, true); |
1648 | 129k | break; |
1649 | 64.1k | case TOKEN_sh: /* fill with sahding pattern */ |
1650 | 64.1k | pdfi_pop(ctx, 1); |
1651 | 64.1k | code = pdfi_shading(ctx, stream_dict, page_dict); |
1652 | 64.1k | break; |
1653 | 137k | case TOKEN_Tstar: /* Move to start of next text line */ |
1654 | 137k | pdfi_pop(ctx, 1); |
1655 | 137k | code = pdfi_T_star(ctx); |
1656 | 137k | break; |
1657 | 715k | case TOKEN_Tc: /* set character spacing */ |
1658 | 715k | pdfi_pop(ctx, 1); |
1659 | 715k | code = pdfi_Tc(ctx); |
1660 | 715k | break; |
1661 | 1.28M | case TOKEN_Td: /* move text position */ |
1662 | 1.28M | pdfi_pop(ctx, 1); |
1663 | 1.28M | code = pdfi_Td(ctx); |
1664 | 1.28M | break; |
1665 | 394k | case TOKEN_TD: /* Move text position, set leading */ |
1666 | 394k | pdfi_pop(ctx, 1); |
1667 | 394k | code = pdfi_TD(ctx); |
1668 | 394k | break; |
1669 | 1.25M | case TOKEN_Tf: /* set font and size */ |
1670 | 1.25M | pdfi_pop(ctx, 1); |
1671 | 1.25M | code = pdfi_Tf(ctx, stream_dict, page_dict); |
1672 | 1.25M | break; |
1673 | 1.66M | case TOKEN_Tj: /* show text */ |
1674 | 1.66M | pdfi_pop(ctx, 1); |
1675 | 1.66M | code = pdfi_Tj(ctx); |
1676 | 1.66M | break; |
1677 | 1.82M | case TOKEN_TJ: /* show text with individual glyph positioning */ |
1678 | 1.82M | pdfi_pop(ctx, 1); |
1679 | 1.82M | code = pdfi_TJ(ctx); |
1680 | 1.82M | break; |
1681 | 12.1k | case TOKEN_TL: /* set text leading */ |
1682 | 12.1k | pdfi_pop(ctx, 1); |
1683 | 12.1k | code = pdfi_TL(ctx); |
1684 | 12.1k | break; |
1685 | 1.69M | case TOKEN_Tm: /* set text matrix */ |
1686 | 1.69M | pdfi_pop(ctx, 1); |
1687 | 1.69M | code = pdfi_Tm(ctx); |
1688 | 1.69M | break; |
1689 | 581k | case TOKEN_Tr: /* set text rendering mode */ |
1690 | 581k | pdfi_pop(ctx, 1); |
1691 | 581k | code = pdfi_Tr(ctx); |
1692 | 581k | break; |
1693 | 4.03k | case TOKEN_Ts: /* set text rise */ |
1694 | 4.03k | pdfi_pop(ctx, 1); |
1695 | 4.03k | code = pdfi_Ts(ctx); |
1696 | 4.03k | break; |
1697 | 154k | case TOKEN_Tw: /* set word spacing */ |
1698 | 154k | pdfi_pop(ctx, 1); |
1699 | 154k | code = pdfi_Tw(ctx); |
1700 | 154k | break; |
1701 | 102k | case TOKEN_Tz: /* set text matrix */ |
1702 | 102k | pdfi_pop(ctx, 1); |
1703 | 102k | code = pdfi_Tz(ctx); |
1704 | 102k | break; |
1705 | 152k | case TOKEN_v: /* append curve (initial point replicated) */ |
1706 | 152k | pdfi_pop(ctx, 1); |
1707 | 152k | code = pdfi_v_curveto(ctx); |
1708 | 152k | break; |
1709 | 1.94M | case TOKEN_w: /* setlinewidth */ |
1710 | 1.94M | pdfi_pop(ctx, 1); |
1711 | 1.94M | code = pdfi_setlinewidth(ctx); |
1712 | 1.94M | break; |
1713 | 616k | case TOKEN_W: /* clip */ |
1714 | 616k | pdfi_pop(ctx, 1); |
1715 | 616k | ctx->clip_active = true; |
1716 | 616k | ctx->do_eoclip = false; |
1717 | 616k | break; |
1718 | 58.2k | case TOKEN_Wstar: /* eoclip */ |
1719 | 58.2k | pdfi_pop(ctx, 1); |
1720 | 58.2k | ctx->clip_active = true; |
1721 | 58.2k | ctx->do_eoclip = true; |
1722 | 58.2k | break; |
1723 | 170k | case TOKEN_y: /* append curve (final point replicated) */ |
1724 | 170k | pdfi_pop(ctx, 1); |
1725 | 170k | code = pdfi_y_curveto(ctx); |
1726 | 170k | break; |
1727 | 18.2k | case TOKEN_APOSTROPHE: /* move to next line and show text */ |
1728 | 18.2k | pdfi_pop(ctx, 1); |
1729 | 18.2k | code = pdfi_singlequote(ctx); |
1730 | 18.2k | break; |
1731 | 5.70k | case TOKEN_QUOTE: /* set word and character spacing, move to next line, show text */ |
1732 | 5.70k | pdfi_pop(ctx, 1); |
1733 | 5.70k | code = pdfi_doublequote(ctx); |
1734 | 5.70k | break; |
1735 | 4.19k | default: |
1736 | | /* Shouldn't we return an error here? Original code didn't seem to. */ |
1737 | 4.19k | break; |
1738 | 69.6M | } |
1739 | | /* We use a return value of 1 to indicate a repaired keyword (a pair of operators |
1740 | | * was concatenated, and we split them up). We must not return a value > 0 from here |
1741 | | * to avoid tripping that test. |
1742 | | */ |
1743 | 69.6M | if (code > 0) |
1744 | 0 | code = 0; |
1745 | 69.6M | return code; |
1746 | 69.6M | } else { |
1747 | | /* This means we either have a corrupted or illegal operator. The most |
1748 | | * usual corruption is two concatented operators (eg QBT instead of Q BT) |
1749 | | * I plan to tackle this by trying to see if I can make two or more operators |
1750 | | * out of the mangled one. |
1751 | | */ |
1752 | 7.94M | code = split_bogus_operator(ctx, source, stream_dict, page_dict); |
1753 | 7.94M | if (code < 0) |
1754 | 305k | return code; |
1755 | 7.63M | if (pdfi_count_stack(ctx) > 0) { |
1756 | 11 | keyword = ctx->stack_top[-1]; |
1757 | 11 | if (keyword != PDF_TOKEN_AS_OBJ(TOKEN_NOT_A_KEYWORD)) |
1758 | 11 | return REPAIRED_KEYWORD; |
1759 | 11 | } |
1760 | 7.63M | } |
1761 | 7.63M | return 0; |
1762 | 77.5M | } |
1763 | | |
1764 | | void local_save_stream_state(pdf_context *ctx, stream_save *local_save) |
1765 | 638k | { |
1766 | | /* copy the 'save_stream' data from the context to a local structure */ |
1767 | 638k | local_save->stream_offset = ctx->current_stream_save.stream_offset; |
1768 | 638k | local_save->gsave_level = ctx->current_stream_save.gsave_level; |
1769 | 638k | local_save->stack_count = ctx->current_stream_save.stack_count; |
1770 | 638k | local_save->group_depth = ctx->current_stream_save.group_depth; |
1771 | 638k | } |
1772 | | |
1773 | | void cleanup_context_interpretation(pdf_context *ctx, stream_save *local_save) |
1774 | 638k | { |
1775 | 638k | pdfi_seek(ctx, ctx->main_stream, ctx->current_stream_save.stream_offset, SEEK_SET); |
1776 | | /* The transparency group implenetation does a gsave, so the end group does a |
1777 | | * grestore. Therefore we need to do this before we check the saved gstate depth |
1778 | | */ |
1779 | 638k | if (ctx->current_stream_save.group_depth != local_save->group_depth) { |
1780 | 9 | pdfi_set_warning(ctx, 0, NULL, W_PDF_GROUPERROR, "pdfi_cleanup_context_interpretation", NULL); |
1781 | 18 | while (ctx->current_stream_save.group_depth > local_save->group_depth) |
1782 | 9 | pdfi_trans_end_group(ctx); |
1783 | 9 | } |
1784 | 638k | if (ctx->pgs->level > ctx->current_stream_save.gsave_level) |
1785 | 21.6k | pdfi_set_warning(ctx, 0, NULL, W_PDF_TOOMANYq, "pdfi_cleanup_context_interpretation", NULL); |
1786 | 638k | if (pdfi_count_stack(ctx) > ctx->current_stream_save.stack_count) |
1787 | 7.87k | pdfi_set_warning(ctx, 0, NULL, W_PDF_STACKGARBAGE, "pdfi_cleanup_context_interpretation", NULL); |
1788 | 829k | while (ctx->pgs->level > ctx->current_stream_save.gsave_level) |
1789 | 191k | pdfi_grestore(ctx); |
1790 | 638k | pdfi_clearstack(ctx); |
1791 | 638k | } |
1792 | | |
1793 | | void local_restore_stream_state(pdf_context *ctx, stream_save *local_save) |
1794 | 638k | { |
1795 | | /* Put the entries stored in the context back to what they were on entry |
1796 | | * We shouldn't really need to do this, the cleanup above should mean all the |
1797 | | * entries are properly reset. |
1798 | | */ |
1799 | 638k | ctx->current_stream_save.stream_offset = local_save->stream_offset; |
1800 | 638k | ctx->current_stream_save.gsave_level = local_save->gsave_level; |
1801 | 638k | ctx->current_stream_save.stack_count = local_save->stack_count; |
1802 | 638k | ctx->current_stream_save.group_depth = local_save->group_depth; |
1803 | 638k | } |
1804 | | |
1805 | | void initialise_stream_save(pdf_context *ctx) |
1806 | 638k | { |
1807 | | /* Set up the values in the context to the current values */ |
1808 | 638k | ctx->current_stream_save.stream_offset = pdfi_tell(ctx->main_stream); |
1809 | 638k | ctx->current_stream_save.gsave_level = ctx->pgs->level; |
1810 | 638k | ctx->current_stream_save.stack_count = pdfi_count_total_stack(ctx); |
1811 | 638k | } |
1812 | | |
1813 | | /* Run a stream in a sub-context (saves/restores DefaultQState) */ |
1814 | | int pdfi_run_context(pdf_context *ctx, pdf_stream *stream_obj, |
1815 | | pdf_dict *page_dict, bool stoponerror, const char *desc) |
1816 | 300k | { |
1817 | 300k | int code = 0, code1 = 0; |
1818 | 300k | gs_gstate *DefaultQState = NULL; |
1819 | | /* Save any existing Default* colour spaces */ |
1820 | 300k | gs_color_space *PageDefaultGray = ctx->page.DefaultGray_cs; |
1821 | 300k | gs_color_space *PageDefaultRGB = ctx->page.DefaultRGB_cs; |
1822 | 300k | gs_color_space *PageDefaultCMYK = ctx->page.DefaultCMYK_cs; |
1823 | | |
1824 | 300k | ctx->page.DefaultGray_cs = NULL; |
1825 | 300k | ctx->page.DefaultRGB_cs = NULL; |
1826 | 300k | ctx->page.DefaultCMYK_cs = NULL; |
1827 | | |
1828 | | #if DEBUG_CONTEXT |
1829 | | dbgmprintf(ctx->memory, "pdfi_run_context BEGIN\n"); |
1830 | | #endif |
1831 | | /* If the stream has any Default* colour spaces, replace the page level ones. |
1832 | | * This will derement the reference counts to the current spaces if they are replaced. |
1833 | | */ |
1834 | 300k | code = pdfi_setup_DefaultSpaces(ctx, stream_obj->stream_dict); |
1835 | 300k | if (code < 0) |
1836 | 0 | goto exit; |
1837 | | |
1838 | | /* If no Default* space found, try using the Page level ones (if any) */ |
1839 | 300k | if (ctx->page.DefaultGray_cs == NULL) { |
1840 | 300k | ctx->page.DefaultGray_cs = PageDefaultGray; |
1841 | 300k | rc_increment(PageDefaultGray); |
1842 | 300k | } |
1843 | 300k | if (ctx->page.DefaultRGB_cs == NULL) { |
1844 | 300k | ctx->page.DefaultRGB_cs = PageDefaultRGB; |
1845 | 300k | rc_increment(PageDefaultRGB); |
1846 | 300k | } |
1847 | 300k | if (ctx->page.DefaultCMYK_cs == NULL) { |
1848 | 300k | ctx->page.DefaultCMYK_cs = PageDefaultCMYK; |
1849 | 300k | rc_increment(PageDefaultCMYK); |
1850 | 300k | } |
1851 | | |
1852 | 300k | code = pdfi_copy_DefaultQState(ctx, &DefaultQState); |
1853 | 300k | if (code < 0) |
1854 | 0 | goto exit; |
1855 | | |
1856 | 300k | code = pdfi_set_DefaultQState(ctx, ctx->pgs); |
1857 | 300k | if (code < 0) |
1858 | 0 | goto exit; |
1859 | | |
1860 | 300k | code = pdfi_interpret_inner_content_stream(ctx, stream_obj, page_dict, stoponerror, desc); |
1861 | | |
1862 | 300k | code1 = pdfi_restore_DefaultQState(ctx, &DefaultQState); |
1863 | 300k | if (code >= 0) |
1864 | 300k | code = code1; |
1865 | | |
1866 | 300k | exit: |
1867 | 300k | if (DefaultQState != NULL) { |
1868 | 0 | gs_gstate_free(DefaultQState); |
1869 | 0 | DefaultQState = NULL; |
1870 | 0 | } |
1871 | | |
1872 | | /* Count down any Default* colour spaces */ |
1873 | 300k | rc_decrement(ctx->page.DefaultGray_cs, "pdfi_run_context"); |
1874 | 300k | rc_decrement(ctx->page.DefaultRGB_cs, "pdfi_run_context"); |
1875 | 300k | rc_decrement(ctx->page.DefaultCMYK_cs, "pdfi_run_context"); |
1876 | | |
1877 | | /* And restore the page level ones (if any) */ |
1878 | 300k | ctx->page.DefaultGray_cs = PageDefaultGray; |
1879 | 300k | ctx->page.DefaultRGB_cs = PageDefaultRGB; |
1880 | 300k | ctx->page.DefaultCMYK_cs = PageDefaultCMYK; |
1881 | | |
1882 | | #if DEBUG_CONTEXT |
1883 | | dbgmprintf(ctx->memory, "pdfi_run_context END\n"); |
1884 | | #endif |
1885 | 300k | return code; |
1886 | 300k | } |
1887 | | |
1888 | | |
1889 | | /* Interpret a sub-content stream, with some handling of error recovery, clearing stack, etc. |
1890 | | * This temporarily turns on pdfstoponerror if requested. |
1891 | | * It will make sure the stack is cleared and the gstate is matched. |
1892 | | */ |
1893 | | static int |
1894 | | pdfi_interpret_inner_content(pdf_context *ctx, pdf_c_stream *content_stream, pdf_stream *stream_obj, |
1895 | | pdf_dict *page_dict, bool stoponerror, const char *desc) |
1896 | 335k | { |
1897 | 335k | int code = 0; |
1898 | 335k | bool saved_stoponerror = ctx->args.pdfstoponerror; |
1899 | 335k | stream_save local_entry_save; |
1900 | | |
1901 | 335k | local_save_stream_state(ctx, &local_entry_save); |
1902 | 335k | initialise_stream_save(ctx); |
1903 | | |
1904 | | /* This causes several files to render 'incorrectly', even though they are in some sense |
1905 | | * invalid. It doesn't seem to provide any benefits so I have, for now, removed it. If |
1906 | | * there is a good reason for it we can put it back again. |
1907 | | * FIXME - either restore or remove these lines |
1908 | | * /tests_private/pdf/PDF_2.0_FTS/fts_23_2310.pdf |
1909 | | * /tests_private/pdf/PDF_2.0_FTS/fts_23_2311.pdf |
1910 | | * /tests_private/pdf/PDF_2.0_FTS/fts_23_2312.pdf |
1911 | | * /tests_private/pdf/sumatra/recursive_colorspace.pdf |
1912 | | * /tests_private/pdf/uploads/Bug696410.pdf |
1913 | | * /tests_private/pdf/sumatra/1900_-_cairo_transparency_inefficiency.pdf (with pdfwrite) |
1914 | | */ |
1915 | | #if 0 |
1916 | | /* Stop on error in substream, and also be prepared to clean up the stack */ |
1917 | | if (stoponerror) |
1918 | | ctx->args.pdfstoponerror = true; |
1919 | | #endif |
1920 | | |
1921 | | #if DEBUG_CONTEXT |
1922 | | dbgmprintf1(ctx->memory, "BEGIN %s stream\n", desc); |
1923 | | #endif |
1924 | 335k | code = pdfi_interpret_content_stream(ctx, content_stream, stream_obj, page_dict); |
1925 | | #if DEBUG_CONTEXT |
1926 | | dbgmprintf1(ctx->memory, "END %s stream\n", desc); |
1927 | | #endif |
1928 | | |
1929 | 335k | if (code < 0) |
1930 | 335k | dbgmprintf1(ctx->memory, "ERROR: inner_stream: code %d when rendering stream\n", code); |
1931 | | |
1932 | 335k | ctx->args.pdfstoponerror = saved_stoponerror; |
1933 | | |
1934 | | /* Put our state back the way it was on entry */ |
1935 | | #if PROBE_STREAMS |
1936 | | if (ctx->pgs->level > ctx->current_stream_save.gsave_level || |
1937 | | pdfi_count_stack(ctx) > ctx->current_stream_save.stack_count) |
1938 | | code = ((pdf_context *)0)->first_page; |
1939 | | #endif |
1940 | | |
1941 | 335k | cleanup_context_interpretation(ctx, &local_entry_save); |
1942 | 335k | local_restore_stream_state(ctx, &local_entry_save); |
1943 | 335k | if (code < 0) |
1944 | 7.14k | code = pdfi_set_error_stop(ctx, code, NULL, 0, "pdfi_interpret_inner_content", NULL); |
1945 | 335k | return code; |
1946 | 335k | } |
1947 | | |
1948 | | /* Interpret inner content from a buffer |
1949 | | */ |
1950 | | int |
1951 | | pdfi_interpret_inner_content_buffer(pdf_context *ctx, byte *content_data, |
1952 | | uint32_t content_length, |
1953 | | pdf_dict *stream_dict, pdf_dict *page_dict, |
1954 | | bool stoponerror, const char *desc) |
1955 | 34.7k | { |
1956 | 34.7k | int code = 0; |
1957 | 34.7k | pdf_c_stream *stream = NULL; |
1958 | 34.7k | pdf_stream *stream_obj = NULL; |
1959 | | |
1960 | 34.7k | if (content_length == 0) |
1961 | 0 | return 0; |
1962 | | |
1963 | 34.7k | code = pdfi_open_memory_stream_from_memory(ctx, content_length, |
1964 | 34.7k | content_data, &stream, true); |
1965 | 34.7k | if (code < 0) |
1966 | 0 | goto exit; |
1967 | | |
1968 | 34.7k | code = pdfi_obj_dict_to_stream(ctx, stream_dict, &stream_obj, false); |
1969 | 34.7k | if (code < 0) |
1970 | 0 | return code; |
1971 | | |
1972 | | /* NOTE: stream gets closed in here */ |
1973 | 34.7k | code = pdfi_interpret_inner_content(ctx, stream, stream_obj, page_dict, stoponerror, desc); |
1974 | 34.7k | pdfi_countdown(stream_obj); |
1975 | 34.7k | exit: |
1976 | 34.7k | return code; |
1977 | 34.7k | } |
1978 | | |
1979 | | /* Interpret inner content from a C string |
1980 | | */ |
1981 | | int |
1982 | | pdfi_interpret_inner_content_c_string(pdf_context *ctx, char *content_string, |
1983 | | pdf_dict *stream_dict, pdf_dict *page_dict, |
1984 | | bool stoponerror, const char *desc) |
1985 | 20.4k | { |
1986 | 20.4k | uint32_t length = (uint32_t)strlen(content_string); |
1987 | 20.4k | bool decrypt_strings; |
1988 | 20.4k | int code; |
1989 | | |
1990 | 20.4k | if (length == 0) |
1991 | 0 | return 0; |
1992 | | |
1993 | | /* Underlying buffer limit is uint32, so handle the extremely unlikely case that |
1994 | | * our string is too big. |
1995 | | */ |
1996 | 20.4k | if (length != strlen(content_string)) |
1997 | 0 | return_error(gs_error_limitcheck); |
1998 | | |
1999 | | /* Since this is a constructed string content, not part of the file, it can never |
2000 | | * be encrypted. So disable decryption during this call. |
2001 | | */ |
2002 | 20.4k | decrypt_strings = ctx->encryption.decrypt_strings; |
2003 | 20.4k | ctx->encryption.decrypt_strings = false; |
2004 | 20.4k | code = pdfi_interpret_inner_content_buffer(ctx, (byte *)content_string, length, |
2005 | 20.4k | stream_dict, page_dict, stoponerror, desc); |
2006 | 20.4k | ctx->encryption.decrypt_strings = decrypt_strings; |
2007 | | |
2008 | 20.4k | return code; |
2009 | 20.4k | } |
2010 | | |
2011 | | /* Interpret inner content from a string |
2012 | | */ |
2013 | | int |
2014 | | pdfi_interpret_inner_content_string(pdf_context *ctx, pdf_string *content_string, |
2015 | | pdf_dict *stream_dict, pdf_dict *page_dict, |
2016 | | bool stoponerror, const char *desc) |
2017 | 14.2k | { |
2018 | 14.2k | return pdfi_interpret_inner_content_buffer(ctx, content_string->data, content_string->length, |
2019 | 14.2k | stream_dict, page_dict, stoponerror, desc); |
2020 | 14.2k | } |
2021 | | |
2022 | | /* Interpret inner content from a stream_dict |
2023 | | */ |
2024 | | int |
2025 | | pdfi_interpret_inner_content_stream(pdf_context *ctx, pdf_stream *stream_obj, |
2026 | | pdf_dict *page_dict, bool stoponerror, const char *desc) |
2027 | 300k | { |
2028 | 300k | return pdfi_interpret_inner_content(ctx, NULL, stream_obj, page_dict, stoponerror, desc); |
2029 | 300k | } |
2030 | | |
2031 | | /* |
2032 | | * Interpret a content stream. |
2033 | | * content_stream -- content to parse. If NULL, get it from the stream_dict |
2034 | | * stream_dict -- dict containing the stream |
2035 | | */ |
2036 | | int |
2037 | | pdfi_interpret_content_stream(pdf_context *ctx, pdf_c_stream *content_stream, |
2038 | | pdf_stream *stream_obj, pdf_dict *page_dict) |
2039 | 487k | { |
2040 | 487k | int code; |
2041 | 487k | pdf_c_stream *stream = NULL, *SubFile_stream = NULL; |
2042 | 487k | pdf_keyword *keyword; |
2043 | 487k | pdf_stream *s = ctx->current_stream; |
2044 | 487k | pdf_obj_type type; |
2045 | 487k | char EODString[] = "endstream"; |
2046 | | |
2047 | | /* Check this stream, and all the streams currently being executed, to see |
2048 | | * if the stream we've been given is already in train. If it is, then we |
2049 | | * have encountered recursion. This can happen if a non-page stream such |
2050 | | * as a Form or Pattern uses a Resource, but does not declare it in it's |
2051 | | * Resources, and instead inherits it from the parent. We cannot detect that |
2052 | | * before the Resource is used, so all we can do is check here. |
2053 | | */ |
2054 | 677k | while (s != NULL && pdfi_type_of(s) == PDF_STREAM) { |
2055 | 190k | if (s->object_num > 0) { |
2056 | 189k | if (s->object_num == stream_obj->object_num) { |
2057 | 259 | pdf_dict *d = NULL; |
2058 | 259 | bool known = false; |
2059 | | |
2060 | 259 | code = pdfi_dict_from_obj(ctx, (pdf_obj *)stream_obj, &d); |
2061 | 259 | if (code >= 0) { |
2062 | 259 | code = pdfi_dict_known(ctx, d, "Parent", &known); |
2063 | 259 | if (code >= 0 && known) |
2064 | 60 | (void)pdfi_dict_delete(ctx, d, "Parent"); |
2065 | 259 | } |
2066 | 259 | pdfi_set_error(ctx, 0, NULL, E_PDF_CIRCULARREF, "pdfi_interpret_content_stream", "Aborting stream"); |
2067 | 259 | return_error(gs_error_circular_reference); |
2068 | 259 | } |
2069 | 189k | } |
2070 | 189k | s = (pdf_stream *)s->parent_obj; |
2071 | 189k | } |
2072 | | |
2073 | 487k | if (content_stream != NULL) { |
2074 | 78.6k | stream = content_stream; |
2075 | 408k | } else { |
2076 | 408k | code = pdfi_seek(ctx, ctx->main_stream, pdfi_stream_offset(ctx, stream_obj), SEEK_SET); |
2077 | 408k | if (code < 0) |
2078 | 0 | return code; |
2079 | | |
2080 | 408k | if (stream_obj->length_valid) { |
2081 | 401k | if (stream_obj->Length == 0) |
2082 | 2.61k | return 0; |
2083 | 399k | code = pdfi_apply_SubFileDecode_filter(ctx, stream_obj->Length, NULL, ctx->main_stream, &SubFile_stream, false); |
2084 | 399k | } |
2085 | 6.78k | else |
2086 | 6.78k | code = pdfi_apply_SubFileDecode_filter(ctx, 0, EODString, ctx->main_stream, &SubFile_stream, false); |
2087 | 405k | if (code < 0) |
2088 | 0 | return code; |
2089 | | |
2090 | 405k | code = pdfi_filter(ctx, stream_obj, SubFile_stream, &stream, false); |
2091 | 405k | if (code < 0) { |
2092 | 556 | pdfi_close_file(ctx, SubFile_stream); |
2093 | 556 | return code; |
2094 | 556 | } |
2095 | 405k | } |
2096 | | |
2097 | 484k | pdfi_set_stream_parent(ctx, stream_obj, ctx->current_stream); |
2098 | 484k | ctx->current_stream = stream_obj; |
2099 | | |
2100 | 283M | do { |
2101 | 283M | code = pdfi_read_token(ctx, stream, stream_obj->object_num, stream_obj->generation_num); |
2102 | 283M | if (code < 0) { |
2103 | 698k | if (code == gs_error_ioerror || code == gs_error_VMerror || ctx->args.pdfstoponerror) { |
2104 | 14.8k | if (code == gs_error_ioerror) { |
2105 | 14.8k | pdfi_set_error(ctx, code, NULL, E_PDF_BADSTREAM, "pdfi_interpret_content_stream", (char *)"**** Error reading a content stream. The page may be incomplete"); |
2106 | 14.8k | } else if (code == gs_error_VMerror) { |
2107 | 0 | pdfi_set_error(ctx, code, NULL, E_PDF_OUTOFMEMORY, "pdfi_interpret_content_stream", (char *)"**** Error ran out of memory reading a content stream. The page may be incomplete"); |
2108 | 0 | } |
2109 | 14.8k | goto exit; |
2110 | 14.8k | } |
2111 | 683k | continue; |
2112 | 698k | } |
2113 | | |
2114 | 283M | if (pdfi_count_stack(ctx) <= 0) { |
2115 | 251k | if(stream->eof == true) |
2116 | 229k | break; |
2117 | 251k | } |
2118 | | |
2119 | 282M | repaired_keyword: |
2120 | 282M | type = pdfi_type_of(ctx->stack_top[-1]); |
2121 | 282M | if (type == PDF_FAST_KEYWORD) { |
2122 | 69.2M | keyword = (pdf_keyword *)ctx->stack_top[-1]; |
2123 | | |
2124 | 69.2M | switch((uintptr_t)keyword) { |
2125 | 43.8k | case TOKEN_ENDSTREAM: |
2126 | 43.8k | pdfi_pop(ctx,1); |
2127 | 43.8k | goto exit; |
2128 | 0 | break; |
2129 | 164 | case TOKEN_ENDOBJ: |
2130 | 164 | pdfi_clearstack(ctx); |
2131 | 164 | code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MISSINGENDSTREAM, "pdfi_interpret_content_stream", NULL); |
2132 | 164 | goto exit; |
2133 | 0 | break; |
2134 | 0 | case TOKEN_INVALID_KEY: |
2135 | 0 | pdfi_clearstack(ctx); |
2136 | 0 | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_limitcheck), NULL, E_PDF_KEYWORDTOOLONG, "pdfi_interpret_content_stream", NULL)) < 0) |
2137 | 0 | goto exit; |
2138 | 0 | break; |
2139 | 0 | case TOKEN_TOO_LONG: |
2140 | 0 | pdfi_clearstack(ctx); |
2141 | 0 | if ((code = pdfi_set_error_stop(ctx, gs_note_error(gs_error_syntaxerror), NULL, E_PDF_MISSINGENDSTREAM, "pdfi_interpret_content_stream", NULL)) < 0) |
2142 | 0 | goto exit; |
2143 | 0 | break; |
2144 | 69.2M | default: |
2145 | 69.2M | goto execute; |
2146 | 69.2M | } |
2147 | 69.2M | } |
2148 | 213M | else if (type == PDF_KEYWORD) |
2149 | 7.94M | { |
2150 | 77.1M | execute: |
2151 | 77.1M | { |
2152 | 77.1M | pdf_dict *stream_dict = NULL; |
2153 | | |
2154 | 77.1M | code = pdfi_dict_from_obj(ctx, (pdf_obj *)stream_obj, &stream_dict); |
2155 | 77.1M | if (code < 0) |
2156 | 0 | goto exit; |
2157 | | |
2158 | 77.1M | code = pdfi_interpret_stream_operator(ctx, stream, stream_dict, page_dict); |
2159 | 77.1M | if (code == REPAIRED_KEYWORD) |
2160 | 11 | goto repaired_keyword; |
2161 | | |
2162 | 77.1M | if (code < 0) { |
2163 | 3.96M | if ((code = pdfi_set_error_stop(ctx, code, NULL, E_PDF_TOKENERROR, "pdf_interpret_content_stream", NULL)) < 0) { |
2164 | 785 | pdfi_clearstack(ctx); |
2165 | 785 | goto exit; |
2166 | 785 | } |
2167 | 3.96M | } |
2168 | 77.1M | } |
2169 | 77.1M | } |
2170 | 282M | if(stream->eof == true) |
2171 | 194k | break; |
2172 | 283M | }while(1); |
2173 | | |
2174 | 484k | exit: |
2175 | 484k | ctx->current_stream = pdfi_stream_parent(ctx, stream_obj); |
2176 | 484k | pdfi_clear_stream_parent(ctx, stream_obj); |
2177 | 484k | pdfi_close_file(ctx, stream); |
2178 | 484k | if (SubFile_stream != NULL) |
2179 | 405k | pdfi_close_file(ctx, SubFile_stream); |
2180 | 484k | return code; |
2181 | 484k | } |