/src/tesseract/src/ccutil/scanutils.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // Copyright 2006 Google Inc. |
2 | | // All Rights Reserved. |
3 | | // Author: renn |
4 | | // |
5 | | // Licensed under the Apache License, Version 2.0 (the "License"); |
6 | | // you may not use this file except in compliance with the License. |
7 | | // You may obtain a copy of the License at |
8 | | // http://www.apache.org/licenses/LICENSE-2.0 |
9 | | // Unless required by applicable law or agreed to in writing, software |
10 | | // distributed under the License is distributed on an "AS IS" BASIS, |
11 | | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | | // See the License for the specific language governing permissions and |
13 | | // limitations under the License. |
14 | | |
15 | | #ifdef HAVE_CONFIG_H |
16 | | # include "config_auto.h" |
17 | | #endif |
18 | | |
19 | | #include <cctype> |
20 | | #include <climits> // for CHAR_BIT |
21 | | #include <cmath> |
22 | | #include <cstdarg> |
23 | | #include <cstddef> |
24 | | #include <cstdint> |
25 | | #include <cstdio> |
26 | | #include <cstring> |
27 | | #include <limits> // for std::numeric_limits |
28 | | |
29 | | #include "scanutils.h" |
30 | | |
31 | | enum Flags { |
32 | | FL_SPLAT = 0x01, // Drop the value, do not assign |
33 | | FL_INV = 0x02, // Character-set with inverse |
34 | | FL_WIDTH = 0x04, // Field width specified |
35 | | FL_MINUS = 0x08, // Negative number |
36 | | }; |
37 | | |
38 | | enum Ranks { |
39 | | RANK_CHAR = -2, |
40 | | RANK_SHORT = -1, |
41 | | RANK_INT = 0, |
42 | | RANK_LONG = 1, |
43 | | RANK_LONGLONG = 2, |
44 | | RANK_PTR = std::numeric_limits<int>::max() // Special value used for pointers |
45 | | }; |
46 | | |
47 | | const enum Ranks kMinRank = RANK_CHAR; |
48 | | const enum Ranks kMaxRank = RANK_LONGLONG; |
49 | | |
50 | | const enum Ranks kIntMaxRank = RANK_LONGLONG; |
51 | | const enum Ranks kSizeTRank = RANK_LONG; |
52 | | const enum Ranks kPtrDiffRank = RANK_LONG; |
53 | | |
54 | | enum Bail { |
55 | | BAIL_NONE = 0, // No error condition |
56 | | BAIL_EOF, // Hit EOF |
57 | | BAIL_ERR // Conversion mismatch |
58 | | }; |
59 | | |
60 | | // Helper functions ------------------------------------------------------------ |
61 | 0 | inline size_t LongBit() { |
62 | 0 | return CHAR_BIT * sizeof(long); |
63 | 0 | } |
64 | | |
65 | 0 | static inline int SkipSpace(FILE *s) { |
66 | 0 | int p; |
67 | 0 | while (isascii(p = fgetc(s)) && isspace(p)) { |
68 | 0 | ; |
69 | 0 | } |
70 | 0 | ungetc(p, s); // Make sure next char is available for reading |
71 | 0 | return p; |
72 | 0 | } |
73 | | |
74 | 0 | static inline void SetBit(unsigned long *bitmap, unsigned int bit) { |
75 | 0 | bitmap[bit / LongBit()] |= 1UL << (bit % LongBit()); |
76 | 0 | } |
77 | | |
78 | 0 | static inline int TestBit(unsigned long *bitmap, unsigned int bit) { |
79 | 0 | return static_cast<int>(bitmap[bit / LongBit()] >> (bit % LongBit())) & 1; |
80 | 0 | } |
81 | | |
82 | 0 | static inline int DigitValue(int ch, int base) { |
83 | 0 | if (ch >= '0' && ch <= '9') { |
84 | 0 | if (base >= 10 || ch <= '7') { |
85 | 0 | return ch - '0'; |
86 | 0 | } |
87 | 0 | } else if (ch >= 'A' && ch <= 'Z' && base == 16) { |
88 | 0 | return ch - 'A' + 10; |
89 | 0 | } else if (ch >= 'a' && ch <= 'z' && base == 16) { |
90 | 0 | return ch - 'a' + 10; |
91 | 0 | } |
92 | 0 | return -1; |
93 | 0 | } |
94 | | |
95 | | // IO (re-)implementations ----------------------------------------------------- |
96 | 0 | static uintmax_t streamtoumax(FILE *s, int base) { |
97 | 0 | int minus = 0; |
98 | 0 | uintmax_t v = 0; |
99 | 0 | int d, c = 0; |
100 | |
|
101 | 0 | for (c = fgetc(s); isascii(c) && isspace(c); c = fgetc(s)) { |
102 | 0 | ; |
103 | 0 | } |
104 | | |
105 | | // Single optional + or - |
106 | 0 | if (c == '-' || c == '+') { |
107 | 0 | minus = (c == '-'); |
108 | 0 | c = fgetc(s); |
109 | 0 | } |
110 | | |
111 | | // Assign correct base |
112 | 0 | if (base == 0) { |
113 | 0 | if (c == '0') { |
114 | 0 | c = fgetc(s); |
115 | 0 | if (c == 'x' || c == 'X') { |
116 | 0 | base = 16; |
117 | 0 | c = fgetc(s); |
118 | 0 | } else { |
119 | 0 | base = 8; |
120 | 0 | } |
121 | 0 | } |
122 | 0 | } else if (base == 16) { |
123 | 0 | if (c == '0') { |
124 | 0 | c = fgetc(s); |
125 | 0 | if (c == 'x' || c == 'X') { |
126 | 0 | c = fgetc(s); |
127 | 0 | } |
128 | 0 | } |
129 | 0 | } |
130 | | |
131 | | // Actual number parsing |
132 | 0 | for (; (c != EOF) && (d = DigitValue(c, base)) >= 0; c = fgetc(s)) { |
133 | 0 | v = v * base + d; |
134 | 0 | } |
135 | |
|
136 | 0 | ungetc(c, s); |
137 | 0 | return minus ? -v : v; |
138 | 0 | } |
139 | | |
140 | 0 | static double streamtofloat(FILE *s) { |
141 | 0 | bool minus = false; |
142 | 0 | uint64_t v = 0; |
143 | 0 | int d, c; |
144 | 0 | uint64_t k = 1; |
145 | 0 | uint64_t w = 0; |
146 | |
|
147 | 0 | for (c = fgetc(s); isascii(c) && isspace(c); c = fgetc(s)) { |
148 | 0 | ; |
149 | 0 | } |
150 | | |
151 | | // Single optional + or - |
152 | 0 | if (c == '-' || c == '+') { |
153 | 0 | minus = (c == '-'); |
154 | 0 | c = fgetc(s); |
155 | 0 | } |
156 | | |
157 | | // Actual number parsing |
158 | 0 | for (; c != EOF && (d = DigitValue(c, 10)) >= 0; c = fgetc(s)) { |
159 | 0 | v = v * 10 + d; |
160 | 0 | } |
161 | 0 | if (c == '.') { |
162 | 0 | for (c = fgetc(s); c != EOF && (d = DigitValue(c, 10)) >= 0; c = fgetc(s)) { |
163 | 0 | w = w * 10 + d; |
164 | 0 | k *= 10; |
165 | 0 | } |
166 | 0 | } |
167 | 0 | double f = v + static_cast<double>(w) / k; |
168 | 0 | if (c == 'e' || c == 'E') { |
169 | 0 | c = fgetc(s); |
170 | 0 | int expsign = 1; |
171 | 0 | if (c == '-' || c == '+') { |
172 | 0 | expsign = (c == '-') ? -1 : 1; |
173 | 0 | c = fgetc(s); |
174 | 0 | } |
175 | 0 | int exponent = 0; |
176 | 0 | for (; (c != EOF) && (d = DigitValue(c, 10)) >= 0; c = fgetc(s)) { |
177 | 0 | exponent = exponent * 10 + d; |
178 | 0 | } |
179 | 0 | exponent *= expsign; |
180 | 0 | f *= pow(10.0, static_cast<double>(exponent)); |
181 | 0 | } |
182 | 0 | ungetc(c, s); |
183 | |
|
184 | 0 | return minus ? -f : f; |
185 | 0 | } |
186 | | |
187 | | static int tvfscanf(FILE *stream, const char *format, va_list ap); |
188 | | |
189 | 0 | int tfscanf(FILE *stream, const char *format, ...) { |
190 | 0 | va_list ap; |
191 | 0 | int rv; |
192 | |
|
193 | 0 | va_start(ap, format); |
194 | 0 | rv = tvfscanf(stream, format, ap); |
195 | 0 | va_end(ap); |
196 | |
|
197 | 0 | return rv; |
198 | 0 | } |
199 | | |
200 | 0 | static int tvfscanf(FILE *stream, const char *format, va_list ap) { |
201 | 0 | const char *p = format; |
202 | 0 | char ch; |
203 | 0 | int q = 0; |
204 | 0 | uintmax_t val = 0; |
205 | 0 | int rank = RANK_INT; // Default rank |
206 | 0 | unsigned int width = UINT_MAX; |
207 | 0 | int base; |
208 | 0 | int flags = 0; |
209 | 0 | enum { |
210 | 0 | ST_NORMAL, // Ground state |
211 | 0 | ST_FLAGS, // Special flags |
212 | 0 | ST_WIDTH, // Field width |
213 | 0 | ST_MODIFIERS, // Length or conversion modifiers |
214 | 0 | ST_MATCH_INIT, // Initial state of %[ sequence |
215 | 0 | ST_MATCH, // Main state of %[ sequence |
216 | 0 | ST_MATCH_RANGE, // After - in a %[ sequence |
217 | 0 | } state = ST_NORMAL; |
218 | 0 | char *sarg = nullptr; // %s %c or %[ string argument |
219 | 0 | enum Bail bail = BAIL_NONE; |
220 | 0 | int converted = 0; // Successful conversions |
221 | 0 | unsigned long |
222 | 0 | matchmap[((1 << CHAR_BIT) + (CHAR_BIT * sizeof(long) - 1)) / (CHAR_BIT * sizeof(long))]; |
223 | 0 | int matchinv = 0; // Is match map inverted? |
224 | 0 | unsigned char range_start = 0; |
225 | 0 | auto start_off = std::ftell(stream); |
226 | | |
227 | | // Skip leading spaces |
228 | 0 | SkipSpace(stream); |
229 | |
|
230 | 0 | while ((ch = *p++) && !bail) { |
231 | 0 | switch (state) { |
232 | 0 | case ST_NORMAL: |
233 | 0 | if (ch == '%') { |
234 | 0 | state = ST_FLAGS; |
235 | 0 | flags = 0; |
236 | 0 | rank = RANK_INT; |
237 | 0 | width = UINT_MAX; |
238 | 0 | } else if (isascii(ch) && isspace(ch)) { |
239 | 0 | SkipSpace(stream); |
240 | 0 | } else { |
241 | 0 | if (fgetc(stream) != ch) { |
242 | 0 | bail = BAIL_ERR; // Match failure |
243 | 0 | } |
244 | 0 | } |
245 | 0 | break; |
246 | | |
247 | 0 | case ST_FLAGS: |
248 | 0 | if (ch == '*') { |
249 | 0 | flags |= FL_SPLAT; |
250 | 0 | } else if ('0' <= ch && ch <= '9') { |
251 | 0 | width = (ch - '0'); |
252 | 0 | state = ST_WIDTH; |
253 | 0 | flags |= FL_WIDTH; |
254 | 0 | } else { |
255 | 0 | state = ST_MODIFIERS; |
256 | 0 | p--; // Process this character again |
257 | 0 | } |
258 | 0 | break; |
259 | | |
260 | 0 | case ST_WIDTH: |
261 | 0 | if (ch >= '0' && ch <= '9') { |
262 | 0 | width = width * 10 + (ch - '0'); |
263 | 0 | } else { |
264 | 0 | state = ST_MODIFIERS; |
265 | 0 | p--; // Process this character again |
266 | 0 | } |
267 | 0 | break; |
268 | | |
269 | 0 | case ST_MODIFIERS: |
270 | 0 | switch (ch) { |
271 | | // Length modifiers - nonterminal sequences |
272 | 0 | case 'h': |
273 | 0 | rank--; // Shorter rank |
274 | 0 | break; |
275 | 0 | case 'l': |
276 | 0 | rank++; // Longer rank |
277 | 0 | break; |
278 | 0 | case 'j': |
279 | 0 | rank = kIntMaxRank; |
280 | 0 | break; |
281 | 0 | case 'z': |
282 | 0 | rank = kSizeTRank; |
283 | 0 | break; |
284 | 0 | case 't': |
285 | 0 | rank = kPtrDiffRank; |
286 | 0 | break; |
287 | 0 | case 'L': |
288 | 0 | case 'q': |
289 | 0 | rank = RANK_LONGLONG; // long double/long long |
290 | 0 | break; |
291 | | |
292 | 0 | default: |
293 | | // Output modifiers - terminal sequences |
294 | 0 | state = ST_NORMAL; // Next state will be normal |
295 | 0 | if (rank < kMinRank) { // Canonicalize rank |
296 | 0 | rank = kMinRank; |
297 | 0 | } else if (rank > kMaxRank) { |
298 | 0 | rank = kMaxRank; |
299 | 0 | } |
300 | |
|
301 | 0 | switch (ch) { |
302 | 0 | case 'P': // Upper case pointer |
303 | 0 | case 'p': // Pointer |
304 | 0 | rank = RANK_PTR; |
305 | 0 | base = 0; |
306 | 0 | goto scan_int; |
307 | | |
308 | 0 | case 'i': // Base-independent integer |
309 | 0 | base = 0; |
310 | 0 | goto scan_int; |
311 | | |
312 | 0 | case 'd': // Decimal integer |
313 | 0 | base = 10; |
314 | 0 | goto scan_int; |
315 | | |
316 | 0 | case 'o': // Octal integer |
317 | 0 | base = 8; |
318 | 0 | goto scan_int; |
319 | | |
320 | 0 | case 'u': // Unsigned decimal integer |
321 | 0 | base = 10; |
322 | 0 | goto scan_int; |
323 | | |
324 | 0 | case 'x': // Hexadecimal integer |
325 | 0 | case 'X': |
326 | 0 | base = 16; |
327 | 0 | goto scan_int; |
328 | | |
329 | 0 | case 'n': // Number of characters consumed |
330 | 0 | val = std::ftell(stream) - start_off; |
331 | 0 | goto set_integer; |
332 | | |
333 | 0 | scan_int: |
334 | 0 | q = SkipSpace(stream); |
335 | 0 | if (q <= 0) { |
336 | 0 | bail = BAIL_EOF; |
337 | 0 | break; |
338 | 0 | } |
339 | 0 | val = streamtoumax(stream, base); |
340 | | // fall through |
341 | |
|
342 | 0 | set_integer: |
343 | 0 | if (!(flags & FL_SPLAT)) { |
344 | 0 | converted++; |
345 | 0 | switch (rank) { |
346 | 0 | case RANK_CHAR: |
347 | 0 | *va_arg(ap, unsigned char *) = static_cast<unsigned char>(val); |
348 | 0 | break; |
349 | 0 | case RANK_SHORT: |
350 | 0 | *va_arg(ap, unsigned short *) = static_cast<unsigned short>(val); |
351 | 0 | break; |
352 | 0 | case RANK_INT: |
353 | 0 | *va_arg(ap, unsigned int *) = static_cast<unsigned int>(val); |
354 | 0 | break; |
355 | 0 | case RANK_LONG: |
356 | 0 | *va_arg(ap, unsigned long *) = static_cast<unsigned long>(val); |
357 | 0 | break; |
358 | 0 | case RANK_LONGLONG: |
359 | 0 | *va_arg(ap, unsigned long long *) = static_cast<unsigned long long>(val); |
360 | 0 | break; |
361 | 0 | case RANK_PTR: |
362 | 0 | *va_arg(ap, void **) = reinterpret_cast<void *>(static_cast<uintptr_t>(val)); |
363 | 0 | break; |
364 | 0 | } |
365 | 0 | } |
366 | 0 | break; |
367 | | |
368 | 0 | case 'f': // Preliminary float value parsing |
369 | 0 | case 'g': |
370 | 0 | case 'G': |
371 | 0 | case 'e': |
372 | 0 | case 'E': |
373 | 0 | q = SkipSpace(stream); |
374 | 0 | if (q <= 0) { |
375 | 0 | bail = BAIL_EOF; |
376 | 0 | break; |
377 | 0 | } |
378 | | |
379 | 0 | { |
380 | 0 | double fval = streamtofloat(stream); |
381 | 0 | if (!(flags & FL_SPLAT)) { |
382 | 0 | if (rank == RANK_INT) { |
383 | 0 | *va_arg(ap, float *) = static_cast<float>(fval); |
384 | 0 | } else if (rank == RANK_LONG) { |
385 | 0 | *va_arg(ap, double *) = static_cast<double>(fval); |
386 | 0 | } |
387 | 0 | converted++; |
388 | 0 | } |
389 | 0 | } |
390 | 0 | break; |
391 | | |
392 | 0 | case 'c': // Character |
393 | 0 | width = (flags & FL_WIDTH) ? width : 1; // Default width == 1 |
394 | 0 | sarg = va_arg(ap, char *); |
395 | 0 | while (width--) { |
396 | 0 | if ((q = fgetc(stream)) <= 0) { |
397 | 0 | bail = BAIL_EOF; |
398 | 0 | break; |
399 | 0 | } |
400 | 0 | if (!(flags & FL_SPLAT)) { |
401 | 0 | *sarg++ = q; |
402 | 0 | converted++; |
403 | 0 | } |
404 | 0 | } |
405 | 0 | break; |
406 | | |
407 | 0 | case 's': // String |
408 | 0 | { |
409 | 0 | if (!(flags & FL_SPLAT)) { |
410 | 0 | sarg = va_arg(ap, char *); |
411 | 0 | } |
412 | 0 | unsigned length = 0; |
413 | 0 | while (width--) { |
414 | 0 | q = fgetc(stream); |
415 | 0 | if ((isascii(q) && isspace(q)) || (q <= 0)) { |
416 | 0 | ungetc(q, stream); |
417 | 0 | break; |
418 | 0 | } |
419 | 0 | if (!(flags & FL_SPLAT)) { |
420 | 0 | sarg[length] = q; |
421 | 0 | } |
422 | 0 | length++; |
423 | 0 | } |
424 | 0 | if (length == 0) { |
425 | 0 | bail = BAIL_EOF; |
426 | 0 | } else if (!(flags & FL_SPLAT)) { |
427 | 0 | sarg[length] = '\0'; // Terminate output |
428 | 0 | converted++; |
429 | 0 | } |
430 | 0 | } break; |
431 | | |
432 | 0 | case '[': // Character range |
433 | 0 | sarg = va_arg(ap, char *); |
434 | 0 | state = ST_MATCH_INIT; |
435 | 0 | matchinv = 0; |
436 | 0 | memset(matchmap, 0, sizeof matchmap); |
437 | 0 | break; |
438 | | |
439 | 0 | case '%': // %% sequence |
440 | 0 | if (fgetc(stream) != '%') { |
441 | 0 | bail = BAIL_ERR; |
442 | 0 | } |
443 | 0 | break; |
444 | | |
445 | 0 | default: // Anything else |
446 | 0 | bail = BAIL_ERR; // Unknown sequence |
447 | 0 | break; |
448 | 0 | } |
449 | 0 | } |
450 | 0 | break; |
451 | | |
452 | 0 | case ST_MATCH_INIT: // Initial state for %[ match |
453 | 0 | if (ch == '^' && !(flags & FL_INV)) { |
454 | 0 | matchinv = 1; |
455 | 0 | } else { |
456 | 0 | SetBit(matchmap, static_cast<unsigned char>(ch)); |
457 | 0 | state = ST_MATCH; |
458 | 0 | } |
459 | 0 | break; |
460 | | |
461 | 0 | case ST_MATCH: // Main state for %[ match |
462 | 0 | if (ch == ']') { |
463 | 0 | goto match_run; |
464 | 0 | } else if (ch == '-') { |
465 | 0 | range_start = static_cast<unsigned char>(ch); |
466 | 0 | state = ST_MATCH_RANGE; |
467 | 0 | } else { |
468 | 0 | SetBit(matchmap, static_cast<unsigned char>(ch)); |
469 | 0 | } |
470 | 0 | break; |
471 | | |
472 | 0 | case ST_MATCH_RANGE: // %[ match after - |
473 | 0 | if (ch == ']') { |
474 | 0 | SetBit(matchmap, static_cast<unsigned char>('-')); |
475 | 0 | goto match_run; |
476 | 0 | } else { |
477 | 0 | int i; |
478 | 0 | for (i = range_start; i < (static_cast<unsigned char>(ch)); i++) { |
479 | 0 | SetBit(matchmap, i); |
480 | 0 | } |
481 | 0 | state = ST_MATCH; |
482 | 0 | } |
483 | 0 | break; |
484 | | |
485 | 0 | match_run: // Match expression finished |
486 | 0 | char *oarg = sarg; |
487 | 0 | while (width) { |
488 | 0 | q = fgetc(stream); |
489 | 0 | auto qc = static_cast<unsigned char>(q); |
490 | 0 | if (q <= 0 || !(TestBit(matchmap, qc) ^ matchinv)) { |
491 | 0 | ungetc(q, stream); |
492 | 0 | break; |
493 | 0 | } |
494 | 0 | if (!(flags & FL_SPLAT)) { |
495 | 0 | *sarg = q; |
496 | 0 | } |
497 | 0 | sarg++; |
498 | 0 | } |
499 | 0 | if (oarg == sarg) { |
500 | 0 | bail = (q <= 0) ? BAIL_EOF : BAIL_ERR; |
501 | 0 | } else if (!(flags & FL_SPLAT)) { |
502 | 0 | *sarg = '\0'; |
503 | 0 | converted++; |
504 | 0 | } |
505 | 0 | break; |
506 | 0 | } |
507 | 0 | } |
508 | | |
509 | 0 | if (bail == BAIL_EOF && !converted) { |
510 | 0 | converted = -1; // Return EOF (-1) |
511 | 0 | } |
512 | |
|
513 | 0 | return converted; |
514 | 0 | } |