| 1 | #include "mupdf/fitz.h" |
| 2 | #include "html-imp.h" |
| 3 | |
| 4 | #include <string.h> |
| 5 | |
| 6 | struct lexbuf |
| 7 | { |
| 8 | fz_context *ctx; |
| 9 | fz_pool *pool; |
| 10 | const unsigned char *s; |
| 11 | const char *file; |
| 12 | int line; |
| 13 | int lookahead; |
| 14 | int c; |
| 15 | int string_len; |
| 16 | char string[1024]; |
| 17 | }; |
| 18 | |
| 19 | static fz_css_value *parse_expr(struct lexbuf *buf); |
| 20 | static fz_css_selector *parse_selector(struct lexbuf *buf); |
| 21 | |
| 22 | FZ_NORETURN static void fz_css_error(struct lexbuf *buf, const char *msg) |
| 23 | { |
| 24 | fz_throw(buf->ctx, FZ_ERROR_SYNTAX, "css syntax error: %s (%s:%d)" , msg, buf->file, buf->line); |
| 25 | } |
| 26 | |
| 27 | fz_css *fz_new_css(fz_context *ctx) |
| 28 | { |
| 29 | fz_pool *pool = fz_new_pool(ctx); |
| 30 | fz_css *css = NULL; |
| 31 | |
| 32 | fz_try(ctx) |
| 33 | { |
| 34 | css = fz_pool_alloc(ctx, pool, sizeof *css); |
| 35 | css->pool = pool; |
| 36 | css->rule = NULL; |
| 37 | } |
| 38 | fz_catch(ctx) |
| 39 | { |
| 40 | fz_drop_pool(ctx, pool); |
| 41 | fz_rethrow(ctx); |
| 42 | } |
| 43 | |
| 44 | return css; |
| 45 | } |
| 46 | |
| 47 | void fz_drop_css(fz_context *ctx, fz_css *css) |
| 48 | { |
| 49 | if (css) |
| 50 | fz_drop_pool(ctx, css->pool); |
| 51 | } |
| 52 | |
| 53 | static fz_css_rule *fz_new_css_rule(fz_context *ctx, fz_pool *pool, fz_css_selector *selector, fz_css_property *declaration) |
| 54 | { |
| 55 | fz_css_rule *rule = fz_pool_alloc(ctx, pool, sizeof *rule); |
| 56 | rule->selector = selector; |
| 57 | rule->declaration = declaration; |
| 58 | rule->next = NULL; |
| 59 | return rule; |
| 60 | } |
| 61 | |
| 62 | static fz_css_selector *fz_new_css_selector(fz_context *ctx, fz_pool *pool, const char *name) |
| 63 | { |
| 64 | fz_css_selector *sel = fz_pool_alloc(ctx, pool, sizeof *sel); |
| 65 | sel->name = name ? fz_pool_strdup(ctx, pool, name) : NULL; |
| 66 | sel->combine = 0; |
| 67 | sel->cond = NULL; |
| 68 | sel->left = NULL; |
| 69 | sel->right = NULL; |
| 70 | sel->next = NULL; |
| 71 | return sel; |
| 72 | } |
| 73 | |
| 74 | static fz_css_condition *fz_new_css_condition(fz_context *ctx, fz_pool *pool, int type, const char *key, const char *val) |
| 75 | { |
| 76 | fz_css_condition *cond = fz_pool_alloc(ctx, pool, sizeof *cond); |
| 77 | cond->type = type; |
| 78 | cond->key = key ? fz_pool_strdup(ctx, pool, key) : NULL; |
| 79 | cond->val = val ? fz_pool_strdup(ctx, pool, val) : NULL; |
| 80 | cond->next = NULL; |
| 81 | return cond; |
| 82 | } |
| 83 | |
| 84 | static fz_css_property *fz_new_css_property(fz_context *ctx, fz_pool *pool, const char *name, fz_css_value *value, int spec) |
| 85 | { |
| 86 | fz_css_property *prop = fz_pool_alloc(ctx, pool, sizeof *prop); |
| 87 | prop->name = fz_pool_strdup(ctx, pool, name); |
| 88 | prop->value = value; |
| 89 | prop->spec = spec; |
| 90 | prop->important = 0; |
| 91 | prop->next = NULL; |
| 92 | return prop; |
| 93 | } |
| 94 | |
| 95 | static fz_css_value *fz_new_css_value_x(fz_context *ctx, fz_pool *pool, int type) |
| 96 | { |
| 97 | fz_css_value *val = fz_pool_alloc(ctx, pool, sizeof *val); |
| 98 | val->type = type; |
| 99 | val->data = NULL; |
| 100 | val->args = NULL; |
| 101 | val->next = NULL; |
| 102 | return val; |
| 103 | } |
| 104 | |
| 105 | static fz_css_value *fz_new_css_value(fz_context *ctx, fz_pool *pool, int type, const char *data) |
| 106 | { |
| 107 | fz_css_value *val = fz_pool_alloc(ctx, pool, sizeof *val); |
| 108 | val->type = type; |
| 109 | val->data = fz_pool_strdup(ctx, pool, data); |
| 110 | val->args = NULL; |
| 111 | val->next = NULL; |
| 112 | return val; |
| 113 | } |
| 114 | |
| 115 | static void css_lex_next(struct lexbuf *buf) |
| 116 | { |
| 117 | buf->c = *(buf->s++); |
| 118 | if (buf->c == '\n') |
| 119 | ++buf->line; |
| 120 | } |
| 121 | |
| 122 | static void css_lex_init(fz_context *ctx, struct lexbuf *buf, fz_pool *pool, const char *s, const char *file) |
| 123 | { |
| 124 | buf->ctx = ctx; |
| 125 | buf->pool = pool; |
| 126 | buf->s = (const unsigned char *)s; |
| 127 | buf->c = 0; |
| 128 | buf->file = file; |
| 129 | buf->line = 1; |
| 130 | css_lex_next(buf); |
| 131 | |
| 132 | buf->string_len = 0; |
| 133 | } |
| 134 | |
| 135 | static inline int iswhite(int c) |
| 136 | { |
| 137 | return c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == '\f'; |
| 138 | } |
| 139 | |
| 140 | static int isnmstart(int c) |
| 141 | { |
| 142 | return c == '\\' || c == '_' || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || |
| 143 | (c >= 128 && c <= 255); |
| 144 | } |
| 145 | |
| 146 | static int isnmchar(int c) |
| 147 | { |
| 148 | return c == '\\' || c == '_' || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || |
| 149 | (c >= '0' && c <= '9') || c == '-' || (c >= 128 && c <= 255); |
| 150 | } |
| 151 | |
| 152 | static void css_push_char(struct lexbuf *buf, int c) |
| 153 | { |
| 154 | if (buf->string_len + 1 >= nelem(buf->string)) |
| 155 | fz_css_error(buf, "token too long" ); |
| 156 | buf->string[buf->string_len++] = c; |
| 157 | } |
| 158 | |
| 159 | static int css_lex_accept(struct lexbuf *buf, int t) |
| 160 | { |
| 161 | if (buf->c == t) |
| 162 | { |
| 163 | css_lex_next(buf); |
| 164 | return 1; |
| 165 | } |
| 166 | return 0; |
| 167 | } |
| 168 | |
| 169 | static void css_lex_expect(struct lexbuf *buf, int t) |
| 170 | { |
| 171 | if (!css_lex_accept(buf, t)) |
| 172 | fz_css_error(buf, "unexpected character" ); |
| 173 | } |
| 174 | |
| 175 | static int css_lex_number(struct lexbuf *buf) |
| 176 | { |
| 177 | while (buf->c >= '0' && buf->c <= '9') |
| 178 | { |
| 179 | css_push_char(buf, buf->c); |
| 180 | css_lex_next(buf); |
| 181 | } |
| 182 | |
| 183 | if (css_lex_accept(buf, '.')) |
| 184 | { |
| 185 | css_push_char(buf, '.'); |
| 186 | while (buf->c >= '0' && buf->c <= '9') |
| 187 | { |
| 188 | css_push_char(buf, buf->c); |
| 189 | css_lex_next(buf); |
| 190 | } |
| 191 | } |
| 192 | |
| 193 | if (css_lex_accept(buf, '%')) |
| 194 | { |
| 195 | css_push_char(buf, '%'); |
| 196 | css_push_char(buf, 0); |
| 197 | return CSS_PERCENT; |
| 198 | } |
| 199 | |
| 200 | if (isnmstart(buf->c)) |
| 201 | { |
| 202 | css_push_char(buf, buf->c); |
| 203 | css_lex_next(buf); |
| 204 | while (isnmchar(buf->c)) |
| 205 | { |
| 206 | css_push_char(buf, buf->c); |
| 207 | css_lex_next(buf); |
| 208 | } |
| 209 | css_push_char(buf, 0); |
| 210 | return CSS_LENGTH; |
| 211 | } |
| 212 | |
| 213 | css_push_char(buf, 0); |
| 214 | return CSS_NUMBER; |
| 215 | } |
| 216 | |
| 217 | static int css_lex_keyword(struct lexbuf *buf) |
| 218 | { |
| 219 | while (isnmchar(buf->c)) |
| 220 | { |
| 221 | css_push_char(buf, buf->c); |
| 222 | css_lex_next(buf); |
| 223 | } |
| 224 | css_push_char(buf, 0); |
| 225 | return CSS_KEYWORD; |
| 226 | } |
| 227 | |
| 228 | static int css_lex_hash(struct lexbuf *buf) |
| 229 | { |
| 230 | while (isnmchar(buf->c)) |
| 231 | { |
| 232 | css_push_char(buf, buf->c); |
| 233 | css_lex_next(buf); |
| 234 | } |
| 235 | css_push_char(buf, 0); |
| 236 | return CSS_HASH; |
| 237 | } |
| 238 | |
| 239 | static int css_lex_string(struct lexbuf *buf, int q) |
| 240 | { |
| 241 | while (buf->c && buf->c != q) |
| 242 | { |
| 243 | if (css_lex_accept(buf, '\\')) |
| 244 | { |
| 245 | if (css_lex_accept(buf, 'n')) |
| 246 | css_push_char(buf, '\n'); |
| 247 | else if (css_lex_accept(buf, 'r')) |
| 248 | css_push_char(buf, '\r'); |
| 249 | else if (css_lex_accept(buf, 'f')) |
| 250 | css_push_char(buf, '\f'); |
| 251 | else if (css_lex_accept(buf, '\f')) |
| 252 | /* line continuation */ ; |
| 253 | else if (css_lex_accept(buf, '\n')) |
| 254 | /* line continuation */ ; |
| 255 | else if (css_lex_accept(buf, '\r')) |
| 256 | css_lex_accept(buf, '\n'); |
| 257 | else |
| 258 | { |
| 259 | css_push_char(buf, buf->c); |
| 260 | css_lex_next(buf); |
| 261 | } |
| 262 | } |
| 263 | else |
| 264 | { |
| 265 | css_push_char(buf, buf->c); |
| 266 | css_lex_next(buf); |
| 267 | } |
| 268 | } |
| 269 | css_lex_expect(buf, q); |
| 270 | css_push_char(buf, 0); |
| 271 | return CSS_STRING; |
| 272 | } |
| 273 | |
| 274 | static void css_lex_uri(struct lexbuf *buf) |
| 275 | { |
| 276 | while (buf->c && buf->c != ')' && !iswhite(buf->c)) |
| 277 | { |
| 278 | if (css_lex_accept(buf, '\\')) |
| 279 | { |
| 280 | if (css_lex_accept(buf, 'n')) |
| 281 | css_push_char(buf, '\n'); |
| 282 | else if (css_lex_accept(buf, 'r')) |
| 283 | css_push_char(buf, '\r'); |
| 284 | else if (css_lex_accept(buf, 'f')) |
| 285 | css_push_char(buf, '\f'); |
| 286 | else |
| 287 | { |
| 288 | css_push_char(buf, buf->c); |
| 289 | css_lex_next(buf); |
| 290 | } |
| 291 | } |
| 292 | else if (buf->c == '!' || buf->c == '#' || buf->c == '$' || buf->c == '%' || buf->c == '&' || |
| 293 | (buf->c >= '*' && buf->c <= '[') || |
| 294 | (buf->c >= ']' && buf->c <= '~') || |
| 295 | buf->c > 159) |
| 296 | { |
| 297 | css_push_char(buf, buf->c); |
| 298 | css_lex_next(buf); |
| 299 | } |
| 300 | else |
| 301 | fz_css_error(buf, "unexpected character in url" ); |
| 302 | } |
| 303 | css_push_char(buf, 0); |
| 304 | } |
| 305 | |
| 306 | static int css_lex(struct lexbuf *buf) |
| 307 | { |
| 308 | int t; |
| 309 | |
| 310 | // TODO: keyword escape sequences |
| 311 | |
| 312 | buf->string_len = 0; |
| 313 | |
| 314 | restart: |
| 315 | if (buf->c == 0) |
| 316 | return EOF; |
| 317 | |
| 318 | if (iswhite(buf->c)) |
| 319 | { |
| 320 | while (iswhite(buf->c)) |
| 321 | css_lex_next(buf); |
| 322 | return ' '; |
| 323 | } |
| 324 | |
| 325 | if (css_lex_accept(buf, '/')) |
| 326 | { |
| 327 | if (css_lex_accept(buf, '*')) |
| 328 | { |
| 329 | while (buf->c) |
| 330 | { |
| 331 | if (css_lex_accept(buf, '*')) |
| 332 | { |
| 333 | while (buf->c == '*') |
| 334 | css_lex_next(buf); |
| 335 | if (css_lex_accept(buf, '/')) |
| 336 | goto restart; |
| 337 | } |
| 338 | css_lex_next(buf); |
| 339 | } |
| 340 | fz_css_error(buf, "unterminated comment" ); |
| 341 | } |
| 342 | return '/'; |
| 343 | } |
| 344 | |
| 345 | if (css_lex_accept(buf, '<')) |
| 346 | { |
| 347 | if (css_lex_accept(buf, '!')) |
| 348 | { |
| 349 | css_lex_expect(buf, '-'); |
| 350 | css_lex_expect(buf, '-'); |
| 351 | goto restart; /* ignore CDO */ |
| 352 | } |
| 353 | return '<'; |
| 354 | } |
| 355 | |
| 356 | if (css_lex_accept(buf, '-')) |
| 357 | { |
| 358 | if (css_lex_accept(buf, '-')) |
| 359 | { |
| 360 | css_lex_expect(buf, '>'); |
| 361 | goto restart; /* ignore CDC */ |
| 362 | } |
| 363 | if (isnmstart(buf->c)) |
| 364 | { |
| 365 | css_push_char(buf, '-'); |
| 366 | return css_lex_keyword(buf); |
| 367 | } |
| 368 | return '-'; |
| 369 | } |
| 370 | |
| 371 | if (css_lex_accept(buf, '.')) |
| 372 | { |
| 373 | if (buf->c >= '0' && buf->c <= '9') |
| 374 | { |
| 375 | css_push_char(buf, '.'); |
| 376 | return css_lex_number(buf); |
| 377 | } |
| 378 | return '.'; |
| 379 | } |
| 380 | |
| 381 | if (css_lex_accept(buf, '#')) |
| 382 | { |
| 383 | if (isnmchar(buf->c)) |
| 384 | return css_lex_hash(buf); |
| 385 | return '#'; |
| 386 | } |
| 387 | |
| 388 | if (css_lex_accept(buf, '"')) |
| 389 | return css_lex_string(buf, '"'); |
| 390 | if (css_lex_accept(buf, '\'')) |
| 391 | return css_lex_string(buf, '\''); |
| 392 | |
| 393 | if (buf->c >= '0' && buf->c <= '9') |
| 394 | return css_lex_number(buf); |
| 395 | |
| 396 | if (css_lex_accept(buf, 'u')) |
| 397 | { |
| 398 | if (css_lex_accept(buf, 'r')) |
| 399 | { |
| 400 | if (css_lex_accept(buf, 'l')) |
| 401 | { |
| 402 | if (css_lex_accept(buf, '(')) |
| 403 | { |
| 404 | while (iswhite(buf->c)) |
| 405 | css_lex_next(buf); |
| 406 | if (css_lex_accept(buf, '"')) |
| 407 | css_lex_string(buf, '"'); |
| 408 | else if (css_lex_accept(buf, '\'')) |
| 409 | css_lex_string(buf, '\''); |
| 410 | else |
| 411 | css_lex_uri(buf); |
| 412 | while (iswhite(buf->c)) |
| 413 | css_lex_next(buf); |
| 414 | css_lex_expect(buf, ')'); |
| 415 | return CSS_URI; |
| 416 | } |
| 417 | css_push_char(buf, 'u'); |
| 418 | css_push_char(buf, 'r'); |
| 419 | css_push_char(buf, 'l'); |
| 420 | return css_lex_keyword(buf); |
| 421 | } |
| 422 | css_push_char(buf, 'u'); |
| 423 | css_push_char(buf, 'r'); |
| 424 | return css_lex_keyword(buf); |
| 425 | } |
| 426 | css_push_char(buf, 'u'); |
| 427 | return css_lex_keyword(buf); |
| 428 | } |
| 429 | |
| 430 | if (isnmstart(buf->c)) |
| 431 | { |
| 432 | css_push_char(buf, buf->c); |
| 433 | css_lex_next(buf); |
| 434 | return css_lex_keyword(buf); |
| 435 | } |
| 436 | |
| 437 | t = buf->c; |
| 438 | css_lex_next(buf); |
| 439 | return t; |
| 440 | } |
| 441 | |
| 442 | static void next(struct lexbuf *buf) |
| 443 | { |
| 444 | buf->lookahead = css_lex(buf); |
| 445 | } |
| 446 | |
| 447 | static int accept(struct lexbuf *buf, int t) |
| 448 | { |
| 449 | if (buf->lookahead == t) |
| 450 | { |
| 451 | next(buf); |
| 452 | return 1; |
| 453 | } |
| 454 | return 0; |
| 455 | } |
| 456 | |
| 457 | static void expect(struct lexbuf *buf, int t) |
| 458 | { |
| 459 | if (accept(buf, t)) |
| 460 | return; |
| 461 | fz_css_error(buf, "unexpected token" ); |
| 462 | } |
| 463 | |
| 464 | static void white(struct lexbuf *buf) |
| 465 | { |
| 466 | while (buf->lookahead == ' ') |
| 467 | next(buf); |
| 468 | } |
| 469 | |
| 470 | static int iscond(int t) |
| 471 | { |
| 472 | return t == ':' || t == '.' || t == '[' || t == CSS_HASH; |
| 473 | } |
| 474 | |
| 475 | static fz_css_value *parse_term(struct lexbuf *buf) |
| 476 | { |
| 477 | fz_css_value *v; |
| 478 | |
| 479 | if (buf->lookahead == '+' || buf->lookahead == '-') |
| 480 | { |
| 481 | float sign = buf->lookahead == '-' ? -1 : 1; |
| 482 | next(buf); |
| 483 | if (buf->lookahead != CSS_NUMBER && buf->lookahead != CSS_LENGTH && buf->lookahead != CSS_PERCENT) |
| 484 | fz_css_error(buf, "expected number" ); |
| 485 | if (sign < 0) |
| 486 | { |
| 487 | v = fz_new_css_value_x(buf->ctx, buf->pool, buf->lookahead); |
| 488 | v->data = fz_pool_alloc(buf->ctx, buf->pool, strlen(buf->string) + 2); |
| 489 | v->data[0] = '-'; |
| 490 | strcpy(v->data + 1, buf->string); |
| 491 | } |
| 492 | else |
| 493 | { |
| 494 | v = fz_new_css_value(buf->ctx, buf->pool, buf->lookahead, buf->string); |
| 495 | } |
| 496 | next(buf); |
| 497 | white(buf); |
| 498 | return v; |
| 499 | } |
| 500 | |
| 501 | if (buf->lookahead == CSS_KEYWORD) |
| 502 | { |
| 503 | v = fz_new_css_value(buf->ctx, buf->pool, CSS_KEYWORD, buf->string); |
| 504 | next(buf); |
| 505 | if (accept(buf, '(')) |
| 506 | { |
| 507 | white(buf); |
| 508 | v->type = '('; |
| 509 | v->args = parse_expr(buf); |
| 510 | expect(buf, ')'); |
| 511 | } |
| 512 | white(buf); |
| 513 | return v; |
| 514 | } |
| 515 | |
| 516 | switch (buf->lookahead) |
| 517 | { |
| 518 | case CSS_HASH: |
| 519 | case CSS_STRING: |
| 520 | case CSS_URI: |
| 521 | case CSS_NUMBER: |
| 522 | case CSS_LENGTH: |
| 523 | case CSS_PERCENT: |
| 524 | v = fz_new_css_value(buf->ctx, buf->pool, buf->lookahead, buf->string); |
| 525 | next(buf); |
| 526 | white(buf); |
| 527 | return v; |
| 528 | } |
| 529 | |
| 530 | fz_css_error(buf, "expected value" ); |
| 531 | } |
| 532 | |
| 533 | static fz_css_value *parse_expr(struct lexbuf *buf) |
| 534 | { |
| 535 | fz_css_value *head, *tail; |
| 536 | |
| 537 | head = tail = parse_term(buf); |
| 538 | |
| 539 | while (buf->lookahead != '}' && buf->lookahead != ';' && buf->lookahead != '!' && |
| 540 | buf->lookahead != ')' && buf->lookahead != EOF) |
| 541 | { |
| 542 | if (accept(buf, ',')) |
| 543 | { |
| 544 | white(buf); |
| 545 | tail = tail->next = fz_new_css_value(buf->ctx, buf->pool, ',', "," ); |
| 546 | tail = tail->next = parse_term(buf); |
| 547 | } |
| 548 | else if (accept(buf, '/')) |
| 549 | { |
| 550 | white(buf); |
| 551 | tail = tail->next = fz_new_css_value(buf->ctx, buf->pool, '/', "/" ); |
| 552 | tail = tail->next = parse_term(buf); |
| 553 | } |
| 554 | else |
| 555 | { |
| 556 | tail = tail->next = parse_term(buf); |
| 557 | } |
| 558 | } |
| 559 | |
| 560 | return head; |
| 561 | } |
| 562 | |
| 563 | static fz_css_property *parse_declaration(struct lexbuf *buf) |
| 564 | { |
| 565 | fz_css_property *p; |
| 566 | |
| 567 | if (buf->lookahead != CSS_KEYWORD) |
| 568 | fz_css_error(buf, "expected keyword in property" ); |
| 569 | p = fz_new_css_property(buf->ctx, buf->pool, buf->string, NULL, 0); |
| 570 | next(buf); |
| 571 | |
| 572 | white(buf); |
| 573 | expect(buf, ':'); |
| 574 | white(buf); |
| 575 | |
| 576 | p->value = parse_expr(buf); |
| 577 | |
| 578 | /* !important */ |
| 579 | if (accept(buf, '!')) |
| 580 | { |
| 581 | white(buf); |
| 582 | if (buf->lookahead != CSS_KEYWORD || strcmp(buf->string, "important" )) |
| 583 | fz_css_error(buf, "expected keyword 'important' after '!'" ); |
| 584 | p->important = 1; |
| 585 | next(buf); |
| 586 | white(buf); |
| 587 | } |
| 588 | |
| 589 | return p; |
| 590 | } |
| 591 | |
| 592 | static fz_css_property *parse_declaration_list(struct lexbuf *buf) |
| 593 | { |
| 594 | fz_css_property *head, *tail; |
| 595 | |
| 596 | white(buf); |
| 597 | |
| 598 | if (buf->lookahead == '}' || buf->lookahead == EOF) |
| 599 | return NULL; |
| 600 | |
| 601 | head = tail = parse_declaration(buf); |
| 602 | |
| 603 | while (accept(buf, ';')) |
| 604 | { |
| 605 | white(buf); |
| 606 | |
| 607 | if (buf->lookahead != '}' && buf->lookahead != ';' && buf->lookahead != EOF) |
| 608 | { |
| 609 | tail = tail->next = parse_declaration(buf); |
| 610 | } |
| 611 | } |
| 612 | |
| 613 | return head; |
| 614 | } |
| 615 | |
| 616 | static char *parse_attrib_value(struct lexbuf *buf) |
| 617 | { |
| 618 | char *s; |
| 619 | |
| 620 | if (buf->lookahead == CSS_KEYWORD || buf->lookahead == CSS_STRING) |
| 621 | { |
| 622 | s = fz_pool_strdup(buf->ctx, buf->pool, buf->string); |
| 623 | next(buf); |
| 624 | white(buf); |
| 625 | return s; |
| 626 | } |
| 627 | |
| 628 | fz_css_error(buf, "expected attribute value" ); |
| 629 | } |
| 630 | |
| 631 | static fz_css_condition *parse_condition(struct lexbuf *buf) |
| 632 | { |
| 633 | fz_css_condition *c; |
| 634 | |
| 635 | if (accept(buf, ':')) |
| 636 | { |
| 637 | accept(buf, ':'); /* swallow css3 :: syntax and pretend it's a normal pseudo-class */ |
| 638 | if (buf->lookahead != CSS_KEYWORD) |
| 639 | fz_css_error(buf, "expected keyword after ':'" ); |
| 640 | c = fz_new_css_condition(buf->ctx, buf->pool, ':', "pseudo" , buf->string); |
| 641 | next(buf); |
| 642 | if (accept(buf, '(')) |
| 643 | { |
| 644 | white(buf); |
| 645 | if (accept(buf, CSS_KEYWORD)) |
| 646 | white(buf); |
| 647 | expect(buf, ')'); |
| 648 | } |
| 649 | return c; |
| 650 | } |
| 651 | |
| 652 | if (accept(buf, '.')) |
| 653 | { |
| 654 | if (buf->lookahead != CSS_KEYWORD) |
| 655 | fz_css_error(buf, "expected keyword after '.'" ); |
| 656 | c = fz_new_css_condition(buf->ctx, buf->pool, '.', "class" , buf->string); |
| 657 | next(buf); |
| 658 | return c; |
| 659 | } |
| 660 | |
| 661 | if (accept(buf, '[')) |
| 662 | { |
| 663 | white(buf); |
| 664 | |
| 665 | if (buf->lookahead != CSS_KEYWORD) |
| 666 | fz_css_error(buf, "expected keyword after '['" ); |
| 667 | c = fz_new_css_condition(buf->ctx, buf->pool, '[', buf->string, NULL); |
| 668 | next(buf); |
| 669 | |
| 670 | white(buf); |
| 671 | |
| 672 | if (accept(buf, '=')) |
| 673 | { |
| 674 | c->type = '='; |
| 675 | c->val = parse_attrib_value(buf); |
| 676 | } |
| 677 | else if (accept(buf, '|')) |
| 678 | { |
| 679 | expect(buf, '='); |
| 680 | c->type = '|'; |
| 681 | c->val = parse_attrib_value(buf); |
| 682 | } |
| 683 | else if (accept(buf, '~')) |
| 684 | { |
| 685 | expect(buf, '='); |
| 686 | c->type = '~'; |
| 687 | c->val = parse_attrib_value(buf); |
| 688 | } |
| 689 | |
| 690 | expect(buf, ']'); |
| 691 | |
| 692 | return c; |
| 693 | } |
| 694 | |
| 695 | if (buf->lookahead == CSS_HASH) |
| 696 | { |
| 697 | c = fz_new_css_condition(buf->ctx, buf->pool, '#', "id" , buf->string); |
| 698 | next(buf); |
| 699 | return c; |
| 700 | } |
| 701 | |
| 702 | fz_css_error(buf, "expected condition" ); |
| 703 | } |
| 704 | |
| 705 | static fz_css_condition *parse_condition_list(struct lexbuf *buf) |
| 706 | { |
| 707 | fz_css_condition *head, *tail; |
| 708 | |
| 709 | head = tail = parse_condition(buf); |
| 710 | while (iscond(buf->lookahead)) |
| 711 | { |
| 712 | tail = tail->next = parse_condition(buf); |
| 713 | } |
| 714 | return head; |
| 715 | } |
| 716 | |
| 717 | static fz_css_selector *parse_simple_selector(struct lexbuf *buf) |
| 718 | { |
| 719 | fz_css_selector *s; |
| 720 | |
| 721 | if (accept(buf, '*')) |
| 722 | { |
| 723 | s = fz_new_css_selector(buf->ctx, buf->pool, NULL); |
| 724 | if (iscond(buf->lookahead)) |
| 725 | s->cond = parse_condition_list(buf); |
| 726 | return s; |
| 727 | } |
| 728 | else if (buf->lookahead == CSS_KEYWORD) |
| 729 | { |
| 730 | s = fz_new_css_selector(buf->ctx, buf->pool, buf->string); |
| 731 | next(buf); |
| 732 | if (iscond(buf->lookahead)) |
| 733 | s->cond = parse_condition_list(buf); |
| 734 | return s; |
| 735 | } |
| 736 | else if (iscond(buf->lookahead)) |
| 737 | { |
| 738 | s = fz_new_css_selector(buf->ctx, buf->pool, NULL); |
| 739 | s->cond = parse_condition_list(buf); |
| 740 | return s; |
| 741 | } |
| 742 | |
| 743 | fz_css_error(buf, "expected selector" ); |
| 744 | } |
| 745 | |
| 746 | static fz_css_selector *parse_combinator(struct lexbuf *buf, int c, fz_css_selector *a) |
| 747 | { |
| 748 | fz_css_selector *sel, *b; |
| 749 | white(buf); |
| 750 | b = parse_simple_selector(buf); |
| 751 | sel = fz_new_css_selector(buf->ctx, buf->pool, NULL); |
| 752 | sel->combine = c; |
| 753 | sel->left = a; |
| 754 | sel->right = b; |
| 755 | return sel; |
| 756 | } |
| 757 | |
| 758 | static fz_css_selector *parse_selector(struct lexbuf *buf) |
| 759 | { |
| 760 | fz_css_selector *sel = parse_simple_selector(buf); |
| 761 | for (;;) |
| 762 | { |
| 763 | if (accept(buf, ' ')) |
| 764 | { |
| 765 | if (accept(buf, '+')) |
| 766 | sel = parse_combinator(buf, '+', sel); |
| 767 | else if (accept(buf, '>')) |
| 768 | sel = parse_combinator(buf, '>', sel); |
| 769 | else if (buf->lookahead != ',' && buf->lookahead != '{' && buf->lookahead != EOF) |
| 770 | sel = parse_combinator(buf, ' ', sel); |
| 771 | else |
| 772 | break; |
| 773 | } |
| 774 | else if (accept(buf, '+')) |
| 775 | sel = parse_combinator(buf, '+', sel); |
| 776 | else if (accept(buf, '>')) |
| 777 | sel = parse_combinator(buf, '>', sel); |
| 778 | else |
| 779 | break; |
| 780 | } |
| 781 | return sel; |
| 782 | } |
| 783 | |
| 784 | static fz_css_selector *parse_selector_list(struct lexbuf *buf) |
| 785 | { |
| 786 | fz_css_selector *head, *tail; |
| 787 | |
| 788 | head = tail = parse_selector(buf); |
| 789 | while (accept(buf, ',')) |
| 790 | { |
| 791 | white(buf); |
| 792 | tail = tail->next = parse_selector(buf); |
| 793 | } |
| 794 | return head; |
| 795 | } |
| 796 | |
| 797 | static fz_css_rule *parse_ruleset(struct lexbuf *buf) |
| 798 | { |
| 799 | fz_css_selector *s = NULL; |
| 800 | fz_css_property *p = NULL; |
| 801 | |
| 802 | fz_try(buf->ctx) |
| 803 | { |
| 804 | s = parse_selector_list(buf); |
| 805 | expect(buf, '{'); |
| 806 | p = parse_declaration_list(buf); |
| 807 | expect(buf, '}'); |
| 808 | white(buf); |
| 809 | } |
| 810 | fz_catch(buf->ctx) |
| 811 | { |
| 812 | if (fz_caught(buf->ctx) != FZ_ERROR_SYNTAX) |
| 813 | fz_rethrow(buf->ctx); |
| 814 | while (buf->lookahead != EOF) |
| 815 | { |
| 816 | if (accept(buf, '}')) |
| 817 | { |
| 818 | white(buf); |
| 819 | break; |
| 820 | } |
| 821 | next(buf); |
| 822 | } |
| 823 | return NULL; |
| 824 | } |
| 825 | |
| 826 | return fz_new_css_rule(buf->ctx, buf->pool, s, p); |
| 827 | } |
| 828 | |
| 829 | static fz_css_rule *parse_at_page(struct lexbuf *buf) |
| 830 | { |
| 831 | fz_css_selector *s = NULL; |
| 832 | fz_css_property *p = NULL; |
| 833 | |
| 834 | white(buf); |
| 835 | if (accept(buf, ':')) |
| 836 | { |
| 837 | expect(buf, CSS_KEYWORD); |
| 838 | white(buf); |
| 839 | } |
| 840 | expect(buf, '{'); |
| 841 | p = parse_declaration_list(buf); |
| 842 | expect(buf, '}'); |
| 843 | white(buf); |
| 844 | |
| 845 | s = fz_new_css_selector(buf->ctx, buf->pool, "@page" ); |
| 846 | return fz_new_css_rule(buf->ctx, buf->pool, s, p); |
| 847 | } |
| 848 | |
| 849 | static fz_css_rule *parse_at_font_face(struct lexbuf *buf) |
| 850 | { |
| 851 | fz_css_selector *s = NULL; |
| 852 | fz_css_property *p = NULL; |
| 853 | |
| 854 | white(buf); |
| 855 | expect(buf, '{'); |
| 856 | p = parse_declaration_list(buf); |
| 857 | expect(buf, '}'); |
| 858 | white(buf); |
| 859 | |
| 860 | s = fz_new_css_selector(buf->ctx, buf->pool, "@font-face" ); |
| 861 | return fz_new_css_rule(buf->ctx, buf->pool, s, p); |
| 862 | } |
| 863 | |
| 864 | static void parse_at_rule(struct lexbuf *buf) |
| 865 | { |
| 866 | expect(buf, CSS_KEYWORD); |
| 867 | |
| 868 | /* skip until '{' or ';' */ |
| 869 | while (buf->lookahead != EOF) |
| 870 | { |
| 871 | if (accept(buf, ';')) |
| 872 | { |
| 873 | white(buf); |
| 874 | return; |
| 875 | } |
| 876 | if (accept(buf, '{')) |
| 877 | { |
| 878 | int depth = 1; |
| 879 | while (buf->lookahead != EOF && depth > 0) |
| 880 | { |
| 881 | if (accept(buf, '{')) |
| 882 | ++depth; |
| 883 | else if (accept(buf, '}')) |
| 884 | --depth; |
| 885 | else |
| 886 | next(buf); |
| 887 | } |
| 888 | white(buf); |
| 889 | return; |
| 890 | } |
| 891 | next(buf); |
| 892 | } |
| 893 | } |
| 894 | |
| 895 | static fz_css_rule *parse_stylesheet(struct lexbuf *buf, fz_css_rule *chain) |
| 896 | { |
| 897 | fz_css_rule *rule, **nextp, *tail; |
| 898 | |
| 899 | tail = chain; |
| 900 | if (tail) |
| 901 | { |
| 902 | while (tail->next) |
| 903 | tail = tail->next; |
| 904 | nextp = &tail->next; |
| 905 | } |
| 906 | else |
| 907 | { |
| 908 | nextp = &tail; |
| 909 | } |
| 910 | |
| 911 | white(buf); |
| 912 | |
| 913 | while (buf->lookahead != EOF) |
| 914 | { |
| 915 | if (accept(buf, '@')) |
| 916 | { |
| 917 | if (buf->lookahead == CSS_KEYWORD && !strcmp(buf->string, "page" )) |
| 918 | { |
| 919 | next(buf); |
| 920 | rule = *nextp = parse_at_page(buf); |
| 921 | nextp = &rule->next; |
| 922 | } |
| 923 | else if (buf->lookahead == CSS_KEYWORD && !strcmp(buf->string, "font-face" )) |
| 924 | { |
| 925 | next(buf); |
| 926 | rule = *nextp = parse_at_font_face(buf); |
| 927 | nextp = &rule->next; |
| 928 | } |
| 929 | else |
| 930 | { |
| 931 | parse_at_rule(buf); |
| 932 | } |
| 933 | } |
| 934 | else |
| 935 | { |
| 936 | fz_css_rule *x = parse_ruleset(buf); |
| 937 | if (x) |
| 938 | { |
| 939 | rule = *nextp = x; |
| 940 | nextp = &rule->next; |
| 941 | } |
| 942 | } |
| 943 | white(buf); |
| 944 | } |
| 945 | |
| 946 | return chain ? chain : tail; |
| 947 | } |
| 948 | |
| 949 | fz_css_property *fz_parse_css_properties(fz_context *ctx, fz_pool *pool, const char *source) |
| 950 | { |
| 951 | struct lexbuf buf; |
| 952 | css_lex_init(ctx, &buf, pool, source, "<inline>" ); |
| 953 | next(&buf); |
| 954 | return parse_declaration_list(&buf); |
| 955 | } |
| 956 | |
| 957 | void fz_parse_css(fz_context *ctx, fz_css *css, const char *source, const char *file) |
| 958 | { |
| 959 | struct lexbuf buf; |
| 960 | css_lex_init(ctx, &buf, css->pool, source, file); |
| 961 | next(&buf); |
| 962 | css->rule = parse_stylesheet(&buf, css->rule); |
| 963 | } |
| 964 | |