Rev 4680 | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 4680 | right-hear | 1 | #include "fitz.h" |
| 2 | #include "mupdf.h" |
||
| 3 | |||
| 4 | #define IS_NUMBER \ |
||
| 5 | '+':case'-':case'.':case'0':case'1':case'2':case'3':\ |
||
| 6 | case'4':case'5':case'6':case'7':case'8':case'9' |
||
| 7 | #define IS_WHITE \ |
||
| 8 | '\000':case'\011':case'\012':case'\014':case'\015':case'\040' |
||
| 9 | #define IS_HEX \ |
||
| 10 | '0':case'1':case'2':case'3':case'4':case'5':case'6':\ |
||
| 11 | case'7':case'8':case'9':case'A':case'B':case'C':\ |
||
| 12 | case'D':case'E':case'F':case'a':case'b':case'c':\ |
||
| 13 | case'd':case'e':case'f' |
||
| 14 | #define IS_DELIM \ |
||
| 15 | '(':case')':case'<':case'>':case'[':case']':case'{':\ |
||
| 16 | case'}':case'/':case'%' |
||
| 17 | |||
| 18 | #define RANGE_0_9 \ |
||
| 19 | '0':case'1':case'2':case'3':case'4':case'5':\ |
||
| 20 | case'6':case'7':case'8':case'9' |
||
| 21 | #define RANGE_a_f \ |
||
| 22 | 'a':case'b':case'c':case'd':case'e':case'f' |
||
| 23 | #define RANGE_A_F \ |
||
| 24 | 'A':case'B':case'C':case'D':case'E':case'F' |
||
| 25 | |||
| 26 | static inline int iswhite(int ch) |
||
| 27 | { |
||
| 28 | return |
||
| 29 | ch == '\000' || |
||
| 30 | ch == '\011' || |
||
| 31 | ch == '\012' || |
||
| 32 | ch == '\014' || |
||
| 33 | ch == '\015' || |
||
| 34 | ch == '\040'; |
||
| 35 | } |
||
| 36 | |||
| 37 | static inline int unhex(int ch) |
||
| 38 | { |
||
| 39 | if (ch >= '0' && ch <= '9') return ch - '0'; |
||
| 40 | if (ch >= 'A' && ch <= 'F') return ch - 'A' + 0xA; |
||
| 41 | if (ch >= 'a' && ch <= 'f') return ch - 'a' + 0xA; |
||
| 42 | return 0; |
||
| 43 | } |
||
| 44 | |||
| 45 | static void |
||
| 46 | lex_white(fz_stream *f) |
||
| 47 | { |
||
| 48 | int c; |
||
| 49 | do { |
||
| 50 | c = fz_read_byte(f); |
||
| 51 | } while ((c <= 32) && (iswhite(c))); |
||
| 52 | if (c != EOF) |
||
| 53 | fz_unread_byte(f); |
||
| 54 | } |
||
| 55 | |||
| 56 | static void |
||
| 57 | lex_comment(fz_stream *f) |
||
| 58 | { |
||
| 59 | int c; |
||
| 60 | do { |
||
| 61 | c = fz_read_byte(f); |
||
| 62 | } while ((c != '\012') && (c != '\015') && (c != EOF)); |
||
| 63 | } |
||
| 64 | |||
| 65 | static int |
||
| 66 | lex_number(fz_stream *f, char *s, int n, int *tok) |
||
| 67 | { |
||
| 68 | char *buf = s; |
||
| 69 | *tok = PDF_TOK_INT; |
||
| 70 | |||
| 71 | /* Initially we might have +, -, . or a digit */ |
||
| 72 | if (n > 1) |
||
| 73 | { |
||
| 74 | int c = fz_read_byte(f); |
||
| 75 | switch (c) |
||
| 76 | { |
||
| 77 | case '.': |
||
| 78 | *tok = PDF_TOK_REAL; |
||
| 79 | *s++ = c; |
||
| 80 | n--; |
||
| 81 | goto loop_after_dot; |
||
| 82 | case '+': |
||
| 83 | case '-': |
||
| 84 | case RANGE_0_9: |
||
| 85 | *s++ = c; |
||
| 86 | n--; |
||
| 87 | goto loop_after_sign; |
||
| 88 | default: |
||
| 89 | fz_unread_byte(f); |
||
| 90 | goto end; |
||
| 91 | case EOF: |
||
| 92 | goto end; |
||
| 93 | } |
||
| 94 | } |
||
| 95 | |||
| 96 | /* We can't accept a sign from here on in, just . or a digit */ |
||
| 97 | loop_after_sign: |
||
| 98 | while (n > 1) |
||
| 99 | { |
||
| 100 | int c = fz_read_byte(f); |
||
| 101 | switch (c) |
||
| 102 | { |
||
| 103 | case '.': |
||
| 104 | *tok = PDF_TOK_REAL; |
||
| 105 | *s++ = c; |
||
| 106 | n--; |
||
| 107 | goto loop_after_dot; |
||
| 108 | case RANGE_0_9: |
||
| 109 | *s++ = c; |
||
| 110 | break; |
||
| 111 | default: |
||
| 112 | fz_unread_byte(f); |
||
| 113 | goto end; |
||
| 114 | case EOF: |
||
| 115 | goto end; |
||
| 116 | } |
||
| 117 | n--; |
||
| 118 | } |
||
| 119 | |||
| 120 | /* In here, we've seen a dot, so can accept just digits */ |
||
| 121 | loop_after_dot: |
||
| 122 | while (n > 1) |
||
| 123 | { |
||
| 124 | int c = fz_read_byte(f); |
||
| 125 | switch (c) |
||
| 126 | { |
||
| 127 | case RANGE_0_9: |
||
| 128 | *s++ = c; |
||
| 129 | break; |
||
| 130 | default: |
||
| 131 | fz_unread_byte(f); |
||
| 132 | goto end; |
||
| 133 | case EOF: |
||
| 134 | goto end; |
||
| 135 | } |
||
| 136 | n--; |
||
| 137 | } |
||
| 138 | |||
| 139 | end: |
||
| 140 | *s = '\0'; |
||
| 141 | return s-buf; |
||
| 142 | } |
||
| 143 | |||
| 144 | static void |
||
| 145 | lex_name(fz_stream *f, char *s, int n) |
||
| 146 | { |
||
| 147 | while (n > 1) |
||
| 148 | { |
||
| 149 | int c = fz_read_byte(f); |
||
| 150 | switch (c) |
||
| 151 | { |
||
| 152 | case IS_WHITE: |
||
| 153 | case IS_DELIM: |
||
| 154 | fz_unread_byte(f); |
||
| 155 | goto end; |
||
| 156 | case EOF: |
||
| 157 | goto end; |
||
| 158 | case '#': |
||
| 159 | { |
||
| 160 | int d; |
||
| 161 | c = fz_read_byte(f); |
||
| 162 | switch (c) |
||
| 163 | { |
||
| 164 | case RANGE_0_9: |
||
| 165 | d = (c - '0') << 4; |
||
| 166 | break; |
||
| 167 | case RANGE_a_f: |
||
| 168 | d = (c - 'a' + 10) << 4; |
||
| 169 | break; |
||
| 170 | case RANGE_A_F: |
||
| 171 | d = (c - 'A' + 10) << 4; |
||
| 172 | break; |
||
| 173 | default: |
||
| 174 | fz_unread_byte(f); |
||
| 175 | /* fallthrough */ |
||
| 176 | case EOF: |
||
| 177 | goto end; |
||
| 178 | } |
||
| 179 | c = fz_read_byte(f); |
||
| 180 | switch (c) |
||
| 181 | { |
||
| 182 | case RANGE_0_9: |
||
| 183 | c -= '0'; |
||
| 184 | break; |
||
| 185 | case RANGE_a_f: |
||
| 186 | c -= 'a' - 10; |
||
| 187 | break; |
||
| 188 | case RANGE_A_F: |
||
| 189 | c -= 'A' - 10; |
||
| 190 | break; |
||
| 191 | default: |
||
| 192 | fz_unread_byte(f); |
||
| 193 | /* fallthrough */ |
||
| 194 | case EOF: |
||
| 195 | *s++ = d; |
||
| 196 | n--; |
||
| 197 | goto end; |
||
| 198 | } |
||
| 199 | *s++ = d + c; |
||
| 200 | n--; |
||
| 201 | break; |
||
| 202 | } |
||
| 203 | default: |
||
| 204 | *s++ = c; |
||
| 205 | n--; |
||
| 206 | break; |
||
| 207 | } |
||
| 208 | } |
||
| 209 | end: |
||
| 210 | *s = '\0'; |
||
| 211 | } |
||
| 212 | |||
| 213 | static int |
||
| 214 | lex_string(fz_stream *f, char *buf, int n) |
||
| 215 | { |
||
| 216 | char *s = buf; |
||
| 217 | char *e = buf + n; |
||
| 218 | int bal = 1; |
||
| 219 | int oct; |
||
| 220 | int c; |
||
| 221 | |||
| 222 | while (s < e) |
||
| 223 | { |
||
| 224 | c = fz_read_byte(f); |
||
| 225 | switch (c) |
||
| 226 | { |
||
| 227 | case EOF: |
||
| 228 | goto end; |
||
| 229 | case '(': |
||
| 230 | bal++; |
||
| 231 | *s++ = c; |
||
| 232 | break; |
||
| 233 | case ')': |
||
| 234 | bal --; |
||
| 235 | if (bal == 0) |
||
| 236 | goto end; |
||
| 237 | *s++ = c; |
||
| 238 | break; |
||
| 239 | case '\\': |
||
| 240 | c = fz_read_byte(f); |
||
| 241 | switch (c) |
||
| 242 | { |
||
| 243 | case EOF: |
||
| 244 | goto end; |
||
| 245 | case 'n': |
||
| 246 | *s++ = '\n'; |
||
| 247 | break; |
||
| 248 | case 'r': |
||
| 249 | *s++ = '\r'; |
||
| 250 | break; |
||
| 251 | case 't': |
||
| 252 | *s++ = '\t'; |
||
| 253 | break; |
||
| 254 | case 'b': |
||
| 255 | *s++ = '\b'; |
||
| 256 | break; |
||
| 257 | case 'f': |
||
| 258 | *s++ = '\f'; |
||
| 259 | break; |
||
| 260 | case '(': |
||
| 261 | *s++ = '('; |
||
| 262 | break; |
||
| 263 | case ')': |
||
| 264 | *s++ = ')'; |
||
| 265 | break; |
||
| 266 | case '\\': |
||
| 267 | *s++ = '\\'; |
||
| 268 | break; |
||
| 269 | case RANGE_0_9: |
||
| 270 | oct = c - '0'; |
||
| 271 | c = fz_read_byte(f); |
||
| 272 | if (c >= '0' && c <= '9') |
||
| 273 | { |
||
| 274 | oct = oct * 8 + (c - '0'); |
||
| 275 | c = fz_read_byte(f); |
||
| 276 | if (c >= '0' && c <= '9') |
||
| 277 | oct = oct * 8 + (c - '0'); |
||
| 278 | else if (c != EOF) |
||
| 279 | fz_unread_byte(f); |
||
| 280 | } |
||
| 281 | else if (c != EOF) |
||
| 282 | fz_unread_byte(f); |
||
| 283 | *s++ = oct; |
||
| 284 | break; |
||
| 285 | case '\n': |
||
| 286 | break; |
||
| 287 | case '\r': |
||
| 288 | c = fz_read_byte(f); |
||
| 289 | if ((c != '\n') && (c != EOF)) |
||
| 290 | fz_unread_byte(f); |
||
| 291 | break; |
||
| 292 | default: |
||
| 293 | *s++ = c; |
||
| 294 | } |
||
| 295 | break; |
||
| 296 | default: |
||
| 297 | *s++ = c; |
||
| 298 | break; |
||
| 299 | } |
||
| 300 | } |
||
| 301 | end: |
||
| 302 | return s - buf; |
||
| 303 | } |
||
| 304 | |||
| 305 | static int |
||
| 306 | lex_hex_string(fz_stream *f, char *buf, int n) |
||
| 307 | { |
||
| 308 | char *s = buf; |
||
| 309 | char *e = buf + n; |
||
| 310 | int a = 0, x = 0; |
||
| 311 | int c; |
||
| 312 | |||
| 313 | while (s < e) |
||
| 314 | { |
||
| 315 | c = fz_read_byte(f); |
||
| 316 | switch (c) |
||
| 317 | { |
||
| 318 | case IS_WHITE: |
||
| 319 | break; |
||
| 320 | case IS_HEX: |
||
| 321 | if (x) |
||
| 322 | { |
||
| 323 | *s++ = a * 16 + unhex(c); |
||
| 324 | x = !x; |
||
| 325 | } |
||
| 326 | else |
||
| 327 | { |
||
| 328 | a = unhex(c); |
||
| 329 | x = !x; |
||
| 330 | } |
||
| 331 | break; |
||
| 332 | case '>': |
||
| 333 | case EOF: |
||
| 334 | goto end; |
||
| 335 | default: |
||
| 336 | fz_warn("ignoring invalid character in hex string: '%c'", c); |
||
| 337 | } |
||
| 338 | } |
||
| 339 | end: |
||
| 340 | return s - buf; |
||
| 341 | } |
||
| 342 | |||
| 343 | static int |
||
| 344 | pdf_token_from_keyword(char *key) |
||
| 345 | { |
||
| 346 | switch (*key) |
||
| 347 | { |
||
| 348 | case 'R': |
||
| 349 | if (!strcmp(key, "R")) return PDF_TOK_R; |
||
| 350 | break; |
||
| 351 | case 't': |
||
| 352 | if (!strcmp(key, "true")) return PDF_TOK_TRUE; |
||
| 353 | if (!strcmp(key, "trailer")) return PDF_TOK_TRAILER; |
||
| 354 | break; |
||
| 355 | case 'f': |
||
| 356 | if (!strcmp(key, "false")) return PDF_TOK_FALSE; |
||
| 357 | break; |
||
| 358 | case 'n': |
||
| 359 | if (!strcmp(key, "null")) return PDF_TOK_NULL; |
||
| 360 | break; |
||
| 361 | case 'o': |
||
| 362 | if (!strcmp(key, "obj")) return PDF_TOK_OBJ; |
||
| 363 | break; |
||
| 364 | case 'e': |
||
| 365 | if (!strcmp(key, "endobj")) return PDF_TOK_ENDOBJ; |
||
| 366 | if (!strcmp(key, "endstream")) return PDF_TOK_ENDSTREAM; |
||
| 367 | break; |
||
| 368 | case 's': |
||
| 369 | if (!strcmp(key, "stream")) return PDF_TOK_STREAM; |
||
| 370 | if (!strcmp(key, "startxref")) return PDF_TOK_STARTXREF; |
||
| 371 | break; |
||
| 372 | case 'x': |
||
| 373 | if (!strcmp(key, "xref")) return PDF_TOK_XREF; |
||
| 374 | break; |
||
| 375 | default: |
||
| 376 | break; |
||
| 377 | } |
||
| 378 | |||
| 379 | return PDF_TOK_KEYWORD; |
||
| 380 | } |
||
| 381 | |||
| 382 | fz_error |
||
| 383 | pdf_lex(int *tok, fz_stream *f, char *buf, int n, int *sl) |
||
| 384 | { |
||
| 385 | while (1) |
||
| 386 | { |
||
| 387 | int c = fz_read_byte(f); |
||
| 388 | switch (c) |
||
| 389 | { |
||
| 390 | case EOF: |
||
| 391 | *tok = PDF_TOK_EOF; |
||
| 392 | return fz_okay; |
||
| 393 | case IS_WHITE: |
||
| 394 | lex_white(f); |
||
| 395 | break; |
||
| 396 | case '%': |
||
| 397 | lex_comment(f); |
||
| 398 | break; |
||
| 399 | case '/': |
||
| 400 | lex_name(f, buf, n); |
||
| 401 | *sl = strlen(buf); |
||
| 402 | *tok = PDF_TOK_NAME; |
||
| 403 | return fz_okay; |
||
| 404 | case '(': |
||
| 405 | *sl = lex_string(f, buf, n); |
||
| 406 | *tok = PDF_TOK_STRING; |
||
| 407 | return fz_okay; |
||
| 408 | case ')': |
||
| 409 | *tok = PDF_TOK_ERROR; |
||
| 410 | goto cleanuperror; |
||
| 411 | case '<': |
||
| 412 | c = fz_read_byte(f); |
||
| 413 | if (c == '<') |
||
| 414 | { |
||
| 415 | *tok = PDF_TOK_OPEN_DICT; |
||
| 416 | } |
||
| 417 | else |
||
| 418 | { |
||
| 419 | fz_unread_byte(f); |
||
| 420 | *sl = lex_hex_string(f, buf, n); |
||
| 421 | *tok = PDF_TOK_STRING; |
||
| 422 | } |
||
| 423 | return fz_okay; |
||
| 424 | case '>': |
||
| 425 | c = fz_read_byte(f); |
||
| 426 | if (c == '>') |
||
| 427 | { |
||
| 428 | *tok = PDF_TOK_CLOSE_DICT; |
||
| 429 | return fz_okay; |
||
| 430 | } |
||
| 431 | *tok = PDF_TOK_ERROR; |
||
| 432 | goto cleanuperror; |
||
| 433 | case '[': |
||
| 434 | *tok = PDF_TOK_OPEN_ARRAY; |
||
| 435 | return fz_okay; |
||
| 436 | case ']': |
||
| 437 | *tok = PDF_TOK_CLOSE_ARRAY; |
||
| 438 | return fz_okay; |
||
| 439 | case '{': |
||
| 440 | *tok = PDF_TOK_OPEN_BRACE; |
||
| 441 | return fz_okay; |
||
| 442 | case '}': |
||
| 443 | *tok = PDF_TOK_CLOSE_BRACE; |
||
| 444 | return fz_okay; |
||
| 445 | case IS_NUMBER: |
||
| 446 | fz_unread_byte(f); |
||
| 447 | *sl = lex_number(f, buf, n, tok); |
||
| 448 | return fz_okay; |
||
| 449 | default: /* isregular: !isdelim && !iswhite && c != EOF */ |
||
| 450 | fz_unread_byte(f); |
||
| 451 | lex_name(f, buf, n); |
||
| 452 | *sl = strlen(buf); |
||
| 453 | *tok = pdf_token_from_keyword(buf); |
||
| 454 | return fz_okay; |
||
| 455 | } |
||
| 456 | } |
||
| 457 | |||
| 458 | cleanuperror: |
||
| 459 | *tok = PDF_TOK_ERROR; |
||
| 460 | return fz_throw("lexical error"); |
||
| 461 | } |