Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 4680 | right-hear | 1 | #include "fitz.h" |
| 2 | #include "mupdf.h" |
||
| 3 | |||
| 4 | /* |
||
| 5 | * CMap parser |
||
| 6 | */ |
||
| 7 | |||
| 8 | enum |
||
| 9 | { |
||
| 10 | TOK_USECMAP = PDF_NUM_TOKENS, |
||
| 11 | TOK_BEGIN_CODESPACE_RANGE, |
||
| 12 | TOK_END_CODESPACE_RANGE, |
||
| 13 | TOK_BEGIN_BF_CHAR, |
||
| 14 | TOK_END_BF_CHAR, |
||
| 15 | TOK_BEGIN_BF_RANGE, |
||
| 16 | TOK_END_BF_RANGE, |
||
| 17 | TOK_BEGIN_CID_CHAR, |
||
| 18 | TOK_END_CID_CHAR, |
||
| 19 | TOK_BEGIN_CID_RANGE, |
||
| 20 | TOK_END_CID_RANGE, |
||
| 21 | TOK_END_CMAP |
||
| 22 | }; |
||
| 23 | |||
| 24 | static int |
||
| 25 | pdf_cmap_token_from_keyword(char *key) |
||
| 26 | { |
||
| 27 | if (!strcmp(key, "usecmap")) return TOK_USECMAP; |
||
| 28 | if (!strcmp(key, "begincodespacerange")) return TOK_BEGIN_CODESPACE_RANGE; |
||
| 29 | if (!strcmp(key, "endcodespacerange")) return TOK_END_CODESPACE_RANGE; |
||
| 30 | if (!strcmp(key, "beginbfchar")) return TOK_BEGIN_BF_CHAR; |
||
| 31 | if (!strcmp(key, "endbfchar")) return TOK_END_BF_CHAR; |
||
| 32 | if (!strcmp(key, "beginbfrange")) return TOK_BEGIN_BF_RANGE; |
||
| 33 | if (!strcmp(key, "endbfrange")) return TOK_END_BF_RANGE; |
||
| 34 | if (!strcmp(key, "begincidchar")) return TOK_BEGIN_CID_CHAR; |
||
| 35 | if (!strcmp(key, "endcidchar")) return TOK_END_CID_CHAR; |
||
| 36 | if (!strcmp(key, "begincidrange")) return TOK_BEGIN_CID_RANGE; |
||
| 37 | if (!strcmp(key, "endcidrange")) return TOK_END_CID_RANGE; |
||
| 38 | if (!strcmp(key, "endcmap")) return TOK_END_CMAP; |
||
| 39 | return PDF_TOK_KEYWORD; |
||
| 40 | } |
||
| 41 | |||
| 42 | static int |
||
| 43 | pdf_code_from_string(char *buf, int len) |
||
| 44 | { |
||
| 45 | int a = 0; |
||
| 46 | while (len--) |
||
| 47 | a = (a << 8) | *(unsigned char *)buf++; |
||
| 48 | return a; |
||
| 49 | } |
||
| 50 | |||
| 51 | static fz_error |
||
| 52 | pdf_lex_cmap(int *tok, fz_stream *file, char *buf, int n, int *sl) |
||
| 53 | { |
||
| 54 | fz_error error; |
||
| 55 | |||
| 56 | error = pdf_lex(tok, file, buf, n, sl); |
||
| 57 | if (error) |
||
| 58 | return fz_rethrow(error, "cannot parse cmap token"); |
||
| 59 | |||
| 60 | if (*tok == PDF_TOK_KEYWORD) |
||
| 61 | *tok = pdf_cmap_token_from_keyword(buf); |
||
| 62 | |||
| 63 | return fz_okay; |
||
| 64 | } |
||
| 65 | |||
| 66 | static fz_error |
||
| 67 | pdf_parse_cmap_name(pdf_cmap *cmap, fz_stream *file) |
||
| 68 | { |
||
| 69 | fz_error error; |
||
| 70 | char buf[256]; |
||
| 71 | int tok; |
||
| 72 | int len; |
||
| 73 | |||
| 74 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 75 | if (error) |
||
| 76 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 77 | |||
| 78 | if (tok == PDF_TOK_NAME) |
||
| 79 | fz_strlcpy(cmap->cmap_name, buf, sizeof(cmap->cmap_name)); |
||
| 80 | else |
||
| 81 | fz_warn("expected name after CMapName in cmap"); |
||
| 82 | |||
| 83 | return fz_okay; |
||
| 84 | } |
||
| 85 | |||
| 86 | static fz_error |
||
| 87 | pdf_parse_wmode(pdf_cmap *cmap, fz_stream *file) |
||
| 88 | { |
||
| 89 | fz_error error; |
||
| 90 | char buf[256]; |
||
| 91 | int tok; |
||
| 92 | int len; |
||
| 93 | |||
| 94 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 95 | if (error) |
||
| 96 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 97 | |||
| 98 | if (tok == PDF_TOK_INT) |
||
| 99 | pdf_set_wmode(cmap, atoi(buf)); |
||
| 100 | else |
||
| 101 | fz_warn("expected integer after WMode in cmap"); |
||
| 102 | |||
| 103 | return fz_okay; |
||
| 104 | } |
||
| 105 | |||
| 106 | static fz_error |
||
| 107 | pdf_parse_codespace_range(pdf_cmap *cmap, fz_stream *file) |
||
| 108 | { |
||
| 109 | fz_error error; |
||
| 110 | char buf[256]; |
||
| 111 | int tok; |
||
| 112 | int len; |
||
| 113 | int lo, hi; |
||
| 114 | |||
| 115 | while (1) |
||
| 116 | { |
||
| 117 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 118 | if (error) |
||
| 119 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 120 | |||
| 121 | if (tok == TOK_END_CODESPACE_RANGE) |
||
| 122 | return fz_okay; |
||
| 123 | |||
| 124 | else if (tok == PDF_TOK_STRING) |
||
| 125 | { |
||
| 126 | lo = pdf_code_from_string(buf, len); |
||
| 127 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 128 | if (error) |
||
| 129 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 130 | if (tok == PDF_TOK_STRING) |
||
| 131 | { |
||
| 132 | hi = pdf_code_from_string(buf, len); |
||
| 133 | pdf_add_codespace(cmap, lo, hi, len); |
||
| 134 | } |
||
| 135 | else break; |
||
| 136 | } |
||
| 137 | |||
| 138 | else break; |
||
| 139 | } |
||
| 140 | |||
| 141 | return fz_throw("expected string or endcodespacerange"); |
||
| 142 | } |
||
| 143 | |||
| 144 | static fz_error |
||
| 145 | pdf_parse_cid_range(pdf_cmap *cmap, fz_stream *file) |
||
| 146 | { |
||
| 147 | fz_error error; |
||
| 148 | char buf[256]; |
||
| 149 | int tok; |
||
| 150 | int len; |
||
| 151 | int lo, hi, dst; |
||
| 152 | |||
| 153 | while (1) |
||
| 154 | { |
||
| 155 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 156 | if (error) |
||
| 157 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 158 | |||
| 159 | if (tok == TOK_END_CID_RANGE) |
||
| 160 | return fz_okay; |
||
| 161 | |||
| 162 | else if (tok != PDF_TOK_STRING) |
||
| 163 | return fz_throw("expected string or endcidrange"); |
||
| 164 | |||
| 165 | lo = pdf_code_from_string(buf, len); |
||
| 166 | |||
| 167 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 168 | if (error) |
||
| 169 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 170 | if (tok != PDF_TOK_STRING) |
||
| 171 | return fz_throw("expected string"); |
||
| 172 | |||
| 173 | hi = pdf_code_from_string(buf, len); |
||
| 174 | |||
| 175 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 176 | if (error) |
||
| 177 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 178 | if (tok != PDF_TOK_INT) |
||
| 179 | return fz_throw("expected integer"); |
||
| 180 | |||
| 181 | dst = atoi(buf); |
||
| 182 | |||
| 183 | pdf_map_range_to_range(cmap, lo, hi, dst); |
||
| 184 | } |
||
| 185 | } |
||
| 186 | |||
| 187 | static fz_error |
||
| 188 | pdf_parse_cid_char(pdf_cmap *cmap, fz_stream *file) |
||
| 189 | { |
||
| 190 | fz_error error; |
||
| 191 | char buf[256]; |
||
| 192 | int tok; |
||
| 193 | int len; |
||
| 194 | int src, dst; |
||
| 195 | |||
| 196 | while (1) |
||
| 197 | { |
||
| 198 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 199 | if (error) |
||
| 200 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 201 | |||
| 202 | if (tok == TOK_END_CID_CHAR) |
||
| 203 | return fz_okay; |
||
| 204 | |||
| 205 | else if (tok != PDF_TOK_STRING) |
||
| 206 | return fz_throw("expected string or endcidchar"); |
||
| 207 | |||
| 208 | src = pdf_code_from_string(buf, len); |
||
| 209 | |||
| 210 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 211 | if (error) |
||
| 212 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 213 | if (tok != PDF_TOK_INT) |
||
| 214 | return fz_throw("expected integer"); |
||
| 215 | |||
| 216 | dst = atoi(buf); |
||
| 217 | |||
| 218 | pdf_map_range_to_range(cmap, src, src, dst); |
||
| 219 | } |
||
| 220 | } |
||
| 221 | |||
| 222 | static fz_error |
||
| 223 | pdf_parse_bf_range_array(pdf_cmap *cmap, fz_stream *file, int lo, int hi) |
||
| 224 | { |
||
| 225 | fz_error error; |
||
| 226 | char buf[256]; |
||
| 227 | int tok; |
||
| 228 | int len; |
||
| 229 | int dst[256]; |
||
| 230 | int i; |
||
| 231 | |||
| 232 | while (1) |
||
| 233 | { |
||
| 234 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 235 | if (error) |
||
| 236 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 237 | |||
| 238 | if (tok == PDF_TOK_CLOSE_ARRAY) |
||
| 239 | return fz_okay; |
||
| 240 | |||
| 241 | /* Note: does not handle [ /Name /Name ... ] */ |
||
| 242 | else if (tok != PDF_TOK_STRING) |
||
| 243 | return fz_throw("expected string or ]"); |
||
| 244 | |||
| 245 | if (len / 2) |
||
| 246 | { |
||
| 247 | for (i = 0; i < len / 2; i++) |
||
| 248 | dst[i] = pdf_code_from_string(buf + i * 2, 2); |
||
| 249 | |||
| 250 | pdf_map_one_to_many(cmap, lo, dst, len / 2); |
||
| 251 | } |
||
| 252 | |||
| 253 | lo ++; |
||
| 254 | } |
||
| 255 | } |
||
| 256 | |||
| 257 | static fz_error |
||
| 258 | pdf_parse_bf_range(pdf_cmap *cmap, fz_stream *file) |
||
| 259 | { |
||
| 260 | fz_error error; |
||
| 261 | char buf[256]; |
||
| 262 | int tok; |
||
| 263 | int len; |
||
| 264 | int lo, hi, dst; |
||
| 265 | |||
| 266 | while (1) |
||
| 267 | { |
||
| 268 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 269 | if (error) |
||
| 270 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 271 | |||
| 272 | if (tok == TOK_END_BF_RANGE) |
||
| 273 | return fz_okay; |
||
| 274 | |||
| 275 | else if (tok != PDF_TOK_STRING) |
||
| 276 | return fz_throw("expected string or endbfrange"); |
||
| 277 | |||
| 278 | lo = pdf_code_from_string(buf, len); |
||
| 279 | |||
| 280 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 281 | if (error) |
||
| 282 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 283 | if (tok != PDF_TOK_STRING) |
||
| 284 | return fz_throw("expected string"); |
||
| 285 | |||
| 286 | hi = pdf_code_from_string(buf, len); |
||
| 287 | |||
| 288 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 289 | if (error) |
||
| 290 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 291 | |||
| 292 | if (tok == PDF_TOK_STRING) |
||
| 293 | { |
||
| 294 | if (len == 2) |
||
| 295 | { |
||
| 296 | dst = pdf_code_from_string(buf, len); |
||
| 297 | pdf_map_range_to_range(cmap, lo, hi, dst); |
||
| 298 | } |
||
| 299 | else |
||
| 300 | { |
||
| 301 | int dststr[256]; |
||
| 302 | int i; |
||
| 303 | |||
| 304 | if (len / 2) |
||
| 305 | { |
||
| 306 | for (i = 0; i < len / 2; i++) |
||
| 307 | dststr[i] = pdf_code_from_string(buf + i * 2, 2); |
||
| 308 | |||
| 309 | while (lo <= hi) |
||
| 310 | { |
||
| 311 | dststr[i-1] ++; |
||
| 312 | pdf_map_one_to_many(cmap, lo, dststr, i); |
||
| 313 | lo ++; |
||
| 314 | } |
||
| 315 | } |
||
| 316 | } |
||
| 317 | } |
||
| 318 | |||
| 319 | else if (tok == PDF_TOK_OPEN_ARRAY) |
||
| 320 | { |
||
| 321 | error = pdf_parse_bf_range_array(cmap, file, lo, hi); |
||
| 322 | if (error) |
||
| 323 | return fz_rethrow(error, "cannot map bfrange"); |
||
| 324 | } |
||
| 325 | |||
| 326 | else |
||
| 327 | { |
||
| 328 | return fz_throw("expected string or array or endbfrange"); |
||
| 329 | } |
||
| 330 | } |
||
| 331 | } |
||
| 332 | |||
| 333 | static fz_error |
||
| 334 | pdf_parse_bf_char(pdf_cmap *cmap, fz_stream *file) |
||
| 335 | { |
||
| 336 | fz_error error; |
||
| 337 | char buf[256]; |
||
| 338 | int tok; |
||
| 339 | int len; |
||
| 340 | int dst[256]; |
||
| 341 | int src; |
||
| 342 | int i; |
||
| 343 | |||
| 344 | while (1) |
||
| 345 | { |
||
| 346 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 347 | if (error) |
||
| 348 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 349 | |||
| 350 | if (tok == TOK_END_BF_CHAR) |
||
| 351 | return fz_okay; |
||
| 352 | |||
| 353 | else if (tok != PDF_TOK_STRING) |
||
| 354 | return fz_throw("expected string or endbfchar"); |
||
| 355 | |||
| 356 | src = pdf_code_from_string(buf, len); |
||
| 357 | |||
| 358 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 359 | if (error) |
||
| 360 | return fz_rethrow(error, "syntaxerror in cmap"); |
||
| 361 | /* Note: does not handle /dstName */ |
||
| 362 | if (tok != PDF_TOK_STRING) |
||
| 363 | return fz_throw("expected string"); |
||
| 364 | |||
| 365 | if (len / 2) |
||
| 366 | { |
||
| 367 | for (i = 0; i < len / 2; i++) |
||
| 368 | dst[i] = pdf_code_from_string(buf + i * 2, 2); |
||
| 369 | pdf_map_one_to_many(cmap, src, dst, i); |
||
| 370 | } |
||
| 371 | } |
||
| 372 | } |
||
| 373 | |||
| 374 | fz_error |
||
| 375 | pdf_parse_cmap(pdf_cmap **cmapp, fz_stream *file) |
||
| 376 | { |
||
| 377 | fz_error error; |
||
| 378 | pdf_cmap *cmap; |
||
| 379 | char key[64]; |
||
| 380 | char buf[256]; |
||
| 381 | int tok; |
||
| 382 | int len; |
||
| 383 | |||
| 384 | cmap = pdf_new_cmap(); |
||
| 385 | |||
| 386 | strcpy(key, ".notdef"); |
||
| 387 | |||
| 388 | while (1) |
||
| 389 | { |
||
| 390 | error = pdf_lex_cmap(&tok, file, buf, sizeof buf, &len); |
||
| 391 | if (error) |
||
| 392 | { |
||
| 393 | error = fz_rethrow(error, "syntaxerror in cmap"); |
||
| 394 | goto cleanup; |
||
| 395 | } |
||
| 396 | |||
| 397 | if (tok == PDF_TOK_EOF || tok == TOK_END_CMAP) |
||
| 398 | break; |
||
| 399 | |||
| 400 | else if (tok == PDF_TOK_NAME) |
||
| 401 | { |
||
| 402 | if (!strcmp(buf, "CMapName")) |
||
| 403 | { |
||
| 404 | error = pdf_parse_cmap_name(cmap, file); |
||
| 405 | if (error) |
||
| 406 | { |
||
| 407 | error = fz_rethrow(error, "syntaxerror in cmap after CMapName"); |
||
| 408 | goto cleanup; |
||
| 409 | } |
||
| 410 | } |
||
| 411 | else if (!strcmp(buf, "WMode")) |
||
| 412 | { |
||
| 413 | error = pdf_parse_wmode(cmap, file); |
||
| 414 | if (error) |
||
| 415 | { |
||
| 416 | error = fz_rethrow(error, "syntaxerror in cmap after WMode"); |
||
| 417 | goto cleanup; |
||
| 418 | } |
||
| 419 | } |
||
| 420 | else |
||
| 421 | fz_strlcpy(key, buf, sizeof key); |
||
| 422 | } |
||
| 423 | |||
| 424 | else if (tok == TOK_USECMAP) |
||
| 425 | { |
||
| 426 | fz_strlcpy(cmap->usecmap_name, key, sizeof(cmap->usecmap_name)); |
||
| 427 | } |
||
| 428 | |||
| 429 | else if (tok == TOK_BEGIN_CODESPACE_RANGE) |
||
| 430 | { |
||
| 431 | error = pdf_parse_codespace_range(cmap, file); |
||
| 432 | if (error) |
||
| 433 | { |
||
| 434 | error = fz_rethrow(error, "syntaxerror in cmap codespacerange"); |
||
| 435 | goto cleanup; |
||
| 436 | } |
||
| 437 | } |
||
| 438 | |||
| 439 | else if (tok == TOK_BEGIN_BF_CHAR) |
||
| 440 | { |
||
| 441 | error = pdf_parse_bf_char(cmap, file); |
||
| 442 | if (error) |
||
| 443 | { |
||
| 444 | error = fz_rethrow(error, "syntaxerror in cmap bfchar"); |
||
| 445 | goto cleanup; |
||
| 446 | } |
||
| 447 | } |
||
| 448 | |||
| 449 | else if (tok == TOK_BEGIN_CID_CHAR) |
||
| 450 | { |
||
| 451 | error = pdf_parse_cid_char(cmap, file); |
||
| 452 | if (error) |
||
| 453 | { |
||
| 454 | error = fz_rethrow(error, "syntaxerror in cmap cidchar"); |
||
| 455 | goto cleanup; |
||
| 456 | } |
||
| 457 | } |
||
| 458 | |||
| 459 | else if (tok == TOK_BEGIN_BF_RANGE) |
||
| 460 | { |
||
| 461 | error = pdf_parse_bf_range(cmap, file); |
||
| 462 | if (error) |
||
| 463 | { |
||
| 464 | error = fz_rethrow(error, "syntaxerror in cmap bfrange"); |
||
| 465 | goto cleanup; |
||
| 466 | } |
||
| 467 | } |
||
| 468 | |||
| 469 | else if (tok == TOK_BEGIN_CID_RANGE) |
||
| 470 | { |
||
| 471 | error = pdf_parse_cid_range(cmap, file); |
||
| 472 | if (error) |
||
| 473 | { |
||
| 474 | error = fz_rethrow(error, "syntaxerror in cmap cidrange"); |
||
| 475 | goto cleanup; |
||
| 476 | } |
||
| 477 | } |
||
| 478 | |||
| 479 | /* ignore everything else */ |
||
| 480 | } |
||
| 481 | |||
| 482 | pdf_sort_cmap(cmap); |
||
| 483 | |||
| 484 | *cmapp = cmap; |
||
| 485 | return fz_okay; |
||
| 486 | |||
| 487 | cleanup: |
||
| 488 | pdf_drop_cmap(cmap); |
||
| 489 | return error; /* already rethrown */ |
||
| 490 | } |