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 | /* Scan file for objects and reconstruct xref table */ |
||
| 5 | |||
| 6 | struct entry |
||
| 7 | { |
||
| 8 | int num; |
||
| 9 | int gen; |
||
| 10 | int ofs; |
||
| 11 | int stm_ofs; |
||
| 12 | int stm_len; |
||
| 13 | }; |
||
| 14 | |||
| 15 | static fz_error |
||
| 16 | pdf_repair_obj(fz_stream *file, char *buf, int cap, int *stmofsp, int *stmlenp, fz_obj **encrypt, fz_obj **id) |
||
| 17 | { |
||
| 18 | fz_error error; |
||
| 19 | int tok; |
||
| 20 | int stm_len; |
||
| 21 | int len; |
||
| 22 | int n; |
||
| 23 | |||
| 24 | *stmofsp = 0; |
||
| 25 | *stmlenp = -1; |
||
| 26 | |||
| 27 | stm_len = 0; |
||
| 28 | |||
| 29 | error = pdf_lex(&tok, file, buf, cap, &len); |
||
| 30 | if (error) |
||
| 31 | return fz_rethrow(error, "cannot parse object"); |
||
| 32 | if (tok == PDF_TOK_OPEN_DICT) |
||
| 33 | { |
||
| 34 | fz_obj *dict, *obj; |
||
| 35 | |||
| 36 | /* Send NULL xref so we don't try to resolve references */ |
||
| 37 | error = pdf_parse_dict(&dict, NULL, file, buf, cap); |
||
| 38 | if (error) |
||
| 39 | return fz_rethrow(error, "cannot parse object"); |
||
| 40 | |||
| 41 | obj = fz_dict_gets(dict, "Type"); |
||
| 42 | if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "XRef")) |
||
| 43 | { |
||
| 44 | obj = fz_dict_gets(dict, "Encrypt"); |
||
| 45 | if (obj) |
||
| 46 | { |
||
| 47 | if (*encrypt) |
||
| 48 | fz_drop_obj(*encrypt); |
||
| 49 | *encrypt = fz_keep_obj(obj); |
||
| 50 | } |
||
| 51 | |||
| 52 | obj = fz_dict_gets(dict, "ID"); |
||
| 53 | if (obj) |
||
| 54 | { |
||
| 55 | if (*id) |
||
| 56 | fz_drop_obj(*id); |
||
| 57 | *id = fz_keep_obj(obj); |
||
| 58 | } |
||
| 59 | } |
||
| 60 | |||
| 61 | obj = fz_dict_gets(dict, "Length"); |
||
| 62 | if (fz_is_int(obj)) |
||
| 63 | stm_len = fz_to_int(obj); |
||
| 64 | |||
| 65 | fz_drop_obj(dict); |
||
| 66 | } |
||
| 67 | |||
| 68 | while ( tok != PDF_TOK_STREAM && |
||
| 69 | tok != PDF_TOK_ENDOBJ && |
||
| 70 | tok != PDF_TOK_ERROR && |
||
| 71 | tok != PDF_TOK_EOF ) |
||
| 72 | { |
||
| 73 | error = pdf_lex(&tok, file, buf, cap, &len); |
||
| 74 | if (error) |
||
| 75 | return fz_rethrow(error, "cannot scan for endobj or stream token"); |
||
| 76 | } |
||
| 77 | |||
| 78 | if (tok == PDF_TOK_STREAM) |
||
| 79 | { |
||
| 80 | int c = fz_read_byte(file); |
||
| 81 | if (c == '\r') { |
||
| 82 | c = fz_peek_byte(file); |
||
| 83 | if (c == '\n') |
||
| 84 | fz_read_byte(file); |
||
| 85 | } |
||
| 86 | |||
| 87 | *stmofsp = fz_tell(file); |
||
| 88 | if (*stmofsp < 0) |
||
| 89 | return fz_throw("cannot seek in file"); |
||
| 90 | |||
| 91 | if (stm_len > 0) |
||
| 92 | { |
||
| 93 | fz_seek(file, *stmofsp + stm_len, 0); |
||
| 94 | error = pdf_lex(&tok, file, buf, cap, &len); |
||
| 95 | if (error) |
||
| 96 | fz_catch(error, "cannot find endstream token, falling back to scanning"); |
||
| 97 | if (tok == PDF_TOK_ENDSTREAM) |
||
| 98 | goto atobjend; |
||
| 99 | fz_seek(file, *stmofsp, 0); |
||
| 100 | } |
||
| 101 | |||
| 102 | n = fz_read(file, (unsigned char *) buf, 9); |
||
| 103 | if (n < 0) |
||
| 104 | return fz_rethrow(n, "cannot read from file"); |
||
| 105 | |||
| 106 | while (memcmp(buf, "endstream", 9) != 0) |
||
| 107 | { |
||
| 108 | c = fz_read_byte(file); |
||
| 109 | if (c == EOF) |
||
| 110 | break; |
||
| 111 | memmove(buf, buf + 1, 8); |
||
| 112 | buf[8] = c; |
||
| 113 | } |
||
| 114 | |||
| 115 | *stmlenp = fz_tell(file) - *stmofsp - 9; |
||
| 116 | |||
| 117 | atobjend: |
||
| 118 | error = pdf_lex(&tok, file, buf, cap, &len); |
||
| 119 | if (error) |
||
| 120 | return fz_rethrow(error, "cannot scan for endobj token"); |
||
| 121 | if (tok != PDF_TOK_ENDOBJ) |
||
| 122 | fz_warn("object missing 'endobj' token"); |
||
| 123 | } |
||
| 124 | |||
| 125 | return fz_okay; |
||
| 126 | } |
||
| 127 | |||
| 128 | static fz_error |
||
| 129 | pdf_repair_obj_stm(pdf_xref *xref, int num, int gen) |
||
| 130 | { |
||
| 131 | fz_error error; |
||
| 132 | fz_obj *obj; |
||
| 133 | fz_stream *stm; |
||
| 134 | int tok; |
||
| 135 | int i, n, count; |
||
| 136 | char buf[256]; |
||
| 137 | |||
| 138 | error = pdf_load_object(&obj, xref, num, gen); |
||
| 139 | if (error) |
||
| 140 | return fz_rethrow(error, "cannot load object stream object (%d %d R)", num, gen); |
||
| 141 | |||
| 142 | count = fz_to_int(fz_dict_gets(obj, "N")); |
||
| 143 | |||
| 144 | fz_drop_obj(obj); |
||
| 145 | |||
| 146 | error = pdf_open_stream(&stm, xref, num, gen); |
||
| 147 | if (error) |
||
| 148 | return fz_rethrow(error, "cannot open object stream object (%d %d R)", num, gen); |
||
| 149 | |||
| 150 | for (i = 0; i < count; i++) |
||
| 151 | { |
||
| 152 | error = pdf_lex(&tok, stm, buf, sizeof buf, &n); |
||
| 153 | if (error || tok != PDF_TOK_INT) |
||
| 154 | { |
||
| 155 | fz_close(stm); |
||
| 156 | return fz_rethrow(error, "corrupt object stream (%d %d R)", num, gen); |
||
| 157 | } |
||
| 158 | |||
| 159 | n = atoi(buf); |
||
| 160 | if (n >= xref->len) |
||
| 161 | pdf_resize_xref(xref, n + 1); |
||
| 162 | |||
| 163 | xref->table[n].ofs = num; |
||
| 164 | xref->table[n].gen = i; |
||
| 165 | xref->table[n].stm_ofs = 0; |
||
| 166 | xref->table[n].obj = NULL; |
||
| 167 | xref->table[n].type = 'o'; |
||
| 168 | |||
| 169 | error = pdf_lex(&tok, stm, buf, sizeof buf, &n); |
||
| 170 | if (error || tok != PDF_TOK_INT) |
||
| 171 | { |
||
| 172 | fz_close(stm); |
||
| 173 | return fz_rethrow(error, "corrupt object stream (%d %d R)", num, gen); |
||
| 174 | } |
||
| 175 | } |
||
| 176 | |||
| 177 | fz_close(stm); |
||
| 178 | return fz_okay; |
||
| 179 | } |
||
| 180 | |||
| 181 | fz_error |
||
| 182 | pdf_repair_xref(pdf_xref *xref, char *buf, int bufsize) |
||
| 183 | { |
||
| 184 | fz_error error; |
||
| 185 | fz_obj *dict, *obj; |
||
| 186 | fz_obj *length; |
||
| 187 | |||
| 188 | fz_obj *encrypt = NULL; |
||
| 189 | fz_obj *id = NULL; |
||
| 190 | fz_obj *root = NULL; |
||
| 191 | fz_obj *info = NULL; |
||
| 192 | |||
| 193 | struct entry *list = NULL; |
||
| 194 | int listlen; |
||
| 195 | int listcap; |
||
| 196 | int maxnum = 0; |
||
| 197 | |||
| 198 | int num = 0; |
||
| 199 | int gen = 0; |
||
| 200 | int tmpofs, numofs = 0, genofs = 0; |
||
| 201 | int stm_len, stm_ofs = 0; |
||
| 202 | int tok; |
||
| 203 | int next; |
||
| 204 | int i, n, c; |
||
| 205 | |||
| 206 | fz_seek(xref->file, 0, 0); |
||
| 207 | |||
| 208 | listlen = 0; |
||
| 209 | listcap = 1024; |
||
| 210 | list = fz_calloc(listcap, sizeof(struct entry)); |
||
| 211 | |||
| 212 | /* look for '%PDF' version marker within first kilobyte of file */ |
||
| 213 | n = fz_read(xref->file, (unsigned char *)buf, MAX(bufsize, 1024)); |
||
| 214 | if (n < 0) |
||
| 215 | { |
||
| 216 | error = fz_rethrow(n, "cannot read from file"); |
||
| 217 | goto cleanup; |
||
| 218 | } |
||
| 219 | |||
| 220 | fz_seek(xref->file, 0, 0); |
||
| 221 | for (i = 0; i < n - 4; i++) |
||
| 222 | { |
||
| 223 | if (memcmp(buf + i, "%PDF", 4) == 0) |
||
| 224 | { |
||
| 225 | fz_seek(xref->file, i + 8, 0); /* skip "%PDF-X.Y" */ |
||
| 226 | break; |
||
| 227 | } |
||
| 228 | } |
||
| 229 | |||
| 230 | /* skip comment line after version marker since some generators |
||
| 231 | * forget to terminate the comment with a newline */ |
||
| 232 | c = fz_read_byte(xref->file); |
||
| 233 | while (c >= 0 && (c == ' ' || c == '%')) |
||
| 234 | c = fz_read_byte(xref->file); |
||
| 235 | fz_unread_byte(xref->file); |
||
| 236 | |||
| 237 | while (1) |
||
| 238 | { |
||
| 239 | tmpofs = fz_tell(xref->file); |
||
| 240 | if (tmpofs < 0) |
||
| 241 | { |
||
| 242 | error = fz_throw("cannot tell in file"); |
||
| 243 | goto cleanup; |
||
| 244 | } |
||
| 245 | |||
| 246 | error = pdf_lex(&tok, xref->file, buf, bufsize, &n); |
||
| 247 | if (error) |
||
| 248 | { |
||
| 249 | fz_catch(error, "ignoring the rest of the file"); |
||
| 250 | break; |
||
| 251 | } |
||
| 252 | |||
| 253 | if (tok == PDF_TOK_INT) |
||
| 254 | { |
||
| 255 | numofs = genofs; |
||
| 256 | num = gen; |
||
| 257 | genofs = tmpofs; |
||
| 258 | gen = atoi(buf); |
||
| 259 | } |
||
| 260 | |||
| 261 | else if (tok == PDF_TOK_OBJ) |
||
| 262 | { |
||
| 263 | error = pdf_repair_obj(xref->file, buf, bufsize, &stm_ofs, &stm_len, &encrypt, &id); |
||
| 264 | if (error) |
||
| 265 | { |
||
| 266 | error = fz_rethrow(error, "cannot parse object (%d %d R)", num, gen); |
||
| 267 | goto cleanup; |
||
| 268 | } |
||
| 269 | |||
| 270 | if (listlen + 1 == listcap) |
||
| 271 | { |
||
| 272 | listcap = (listcap * 3) / 2; |
||
| 273 | list = fz_realloc(list, listcap, sizeof(struct entry)); |
||
| 274 | } |
||
| 275 | |||
| 276 | list[listlen].num = num; |
||
| 277 | list[listlen].gen = gen; |
||
| 278 | list[listlen].ofs = numofs; |
||
| 279 | list[listlen].stm_ofs = stm_ofs; |
||
| 280 | list[listlen].stm_len = stm_len; |
||
| 281 | listlen ++; |
||
| 282 | |||
| 283 | if (num > maxnum) |
||
| 284 | maxnum = num; |
||
| 285 | } |
||
| 286 | |||
| 287 | /* trailer dictionary */ |
||
| 288 | else if (tok == PDF_TOK_OPEN_DICT) |
||
| 289 | { |
||
| 290 | error = pdf_parse_dict(&dict, xref, xref->file, buf, bufsize); |
||
| 291 | if (error) |
||
| 292 | { |
||
| 293 | error = fz_rethrow(error, "cannot parse object"); |
||
| 294 | goto cleanup; |
||
| 295 | } |
||
| 296 | |||
| 297 | obj = fz_dict_gets(dict, "Encrypt"); |
||
| 298 | if (obj) |
||
| 299 | { |
||
| 300 | if (encrypt) |
||
| 301 | fz_drop_obj(encrypt); |
||
| 302 | encrypt = fz_keep_obj(obj); |
||
| 303 | } |
||
| 304 | |||
| 305 | obj = fz_dict_gets(dict, "ID"); |
||
| 306 | if (obj) |
||
| 307 | { |
||
| 308 | if (id) |
||
| 309 | fz_drop_obj(id); |
||
| 310 | id = fz_keep_obj(obj); |
||
| 311 | } |
||
| 312 | |||
| 313 | obj = fz_dict_gets(dict, "Root"); |
||
| 314 | if (obj) |
||
| 315 | { |
||
| 316 | if (root) |
||
| 317 | fz_drop_obj(root); |
||
| 318 | root = fz_keep_obj(obj); |
||
| 319 | } |
||
| 320 | |||
| 321 | obj = fz_dict_gets(dict, "Info"); |
||
| 322 | if (obj) |
||
| 323 | { |
||
| 324 | if (info) |
||
| 325 | fz_drop_obj(info); |
||
| 326 | info = fz_keep_obj(obj); |
||
| 327 | } |
||
| 328 | |||
| 329 | fz_drop_obj(dict); |
||
| 330 | } |
||
| 331 | |||
| 332 | else if (tok == PDF_TOK_ERROR) |
||
| 333 | fz_read_byte(xref->file); |
||
| 334 | |||
| 335 | else if (tok == PDF_TOK_EOF) |
||
| 336 | break; |
||
| 337 | } |
||
| 338 | |||
| 339 | /* make xref reasonable */ |
||
| 340 | |||
| 341 | pdf_resize_xref(xref, maxnum + 1); |
||
| 342 | |||
| 343 | for (i = 0; i < listlen; i++) |
||
| 344 | { |
||
| 345 | xref->table[list[i].num].type = 'n'; |
||
| 346 | xref->table[list[i].num].ofs = list[i].ofs; |
||
| 347 | xref->table[list[i].num].gen = list[i].gen; |
||
| 348 | |||
| 349 | xref->table[list[i].num].stm_ofs = list[i].stm_ofs; |
||
| 350 | |||
| 351 | /* corrected stream length */ |
||
| 352 | if (list[i].stm_len >= 0) |
||
| 353 | { |
||
| 354 | error = pdf_load_object(&dict, xref, list[i].num, list[i].gen); |
||
| 355 | if (error) |
||
| 356 | { |
||
| 357 | error = fz_rethrow(error, "cannot load stream object (%d %d R)", list[i].num, list[i].gen); |
||
| 358 | goto cleanup; |
||
| 359 | } |
||
| 360 | |||
| 361 | length = fz_new_int(list[i].stm_len); |
||
| 362 | fz_dict_puts(dict, "Length", length); |
||
| 363 | fz_drop_obj(length); |
||
| 364 | |||
| 365 | fz_drop_obj(dict); |
||
| 366 | } |
||
| 367 | |||
| 368 | } |
||
| 369 | |||
| 370 | xref->table[0].type = 'f'; |
||
| 371 | xref->table[0].ofs = 0; |
||
| 372 | xref->table[0].gen = 65535; |
||
| 373 | xref->table[0].stm_ofs = 0; |
||
| 374 | xref->table[0].obj = NULL; |
||
| 375 | |||
| 376 | next = 0; |
||
| 377 | for (i = xref->len - 1; i >= 0; i--) |
||
| 378 | { |
||
| 379 | if (xref->table[i].type == 'f') |
||
| 380 | { |
||
| 381 | xref->table[i].ofs = next; |
||
| 382 | if (xref->table[i].gen < 65535) |
||
| 383 | xref->table[i].gen ++; |
||
| 384 | next = i; |
||
| 385 | } |
||
| 386 | } |
||
| 387 | |||
| 388 | /* create a repaired trailer, Root will be added later */ |
||
| 389 | |||
| 390 | xref->trailer = fz_new_dict(5); |
||
| 391 | |||
| 392 | obj = fz_new_int(maxnum + 1); |
||
| 393 | fz_dict_puts(xref->trailer, "Size", obj); |
||
| 394 | fz_drop_obj(obj); |
||
| 395 | |||
| 396 | if (root) |
||
| 397 | { |
||
| 398 | fz_dict_puts(xref->trailer, "Root", root); |
||
| 399 | fz_drop_obj(root); |
||
| 400 | } |
||
| 401 | if (info) |
||
| 402 | { |
||
| 403 | fz_dict_puts(xref->trailer, "Info", info); |
||
| 404 | fz_drop_obj(info); |
||
| 405 | } |
||
| 406 | |||
| 407 | if (encrypt) |
||
| 408 | { |
||
| 409 | if (fz_is_indirect(encrypt)) |
||
| 410 | { |
||
| 411 | /* create new reference with non-NULL xref pointer */ |
||
| 412 | obj = fz_new_indirect(fz_to_num(encrypt), fz_to_gen(encrypt), xref); |
||
| 413 | fz_drop_obj(encrypt); |
||
| 414 | encrypt = obj; |
||
| 415 | } |
||
| 416 | fz_dict_puts(xref->trailer, "Encrypt", encrypt); |
||
| 417 | fz_drop_obj(encrypt); |
||
| 418 | } |
||
| 419 | |||
| 420 | if (id) |
||
| 421 | { |
||
| 422 | if (fz_is_indirect(id)) |
||
| 423 | { |
||
| 424 | /* create new reference with non-NULL xref pointer */ |
||
| 425 | obj = fz_new_indirect(fz_to_num(id), fz_to_gen(id), xref); |
||
| 426 | fz_drop_obj(id); |
||
| 427 | id = obj; |
||
| 428 | } |
||
| 429 | fz_dict_puts(xref->trailer, "ID", id); |
||
| 430 | fz_drop_obj(id); |
||
| 431 | } |
||
| 432 | |||
| 433 | fz_free(list); |
||
| 434 | return fz_okay; |
||
| 435 | |||
| 436 | cleanup: |
||
| 437 | if (encrypt) fz_drop_obj(encrypt); |
||
| 438 | if (id) fz_drop_obj(id); |
||
| 439 | if (root) fz_drop_obj(root); |
||
| 440 | if (info) fz_drop_obj(info); |
||
| 441 | fz_free(list); |
||
| 442 | return error; /* already rethrown */ |
||
| 443 | } |
||
| 444 | |||
| 445 | fz_error |
||
| 446 | pdf_repair_obj_stms(pdf_xref *xref) |
||
| 447 | { |
||
| 448 | fz_obj *dict; |
||
| 449 | int i; |
||
| 450 | |||
| 451 | for (i = 0; i < xref->len; i++) |
||
| 452 | { |
||
| 453 | if (xref->table[i].stm_ofs) |
||
| 454 | { |
||
| 455 | pdf_load_object(&dict, xref, i, 0); |
||
| 456 | if (!strcmp(fz_to_name(fz_dict_gets(dict, "Type")), "ObjStm")) |
||
| 457 | pdf_repair_obj_stm(xref, i, 0); |
||
| 458 | fz_drop_obj(dict); |
||
| 459 | } |
||
| 460 | } |
||
| 461 | |||
| 462 | return fz_okay; |
||
| 463 | } |