Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * wparser_def.c
4 : * Default text search parser
5 : *
6 : * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
7 : *
8 : *
9 : * IDENTIFICATION
10 : * src/backend/tsearch/wparser_def.c
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 :
15 : #include "postgres.h"
16 :
17 : #include <limits.h>
18 : #include <wctype.h>
19 :
20 : #include "commands/defrem.h"
21 : #include "mb/pg_wchar.h"
22 : #include "miscadmin.h"
23 : #include "tsearch/ts_public.h"
24 : #include "tsearch/ts_type.h"
25 : #include "tsearch/ts_utils.h"
26 : #include "utils/builtins.h"
27 : #include "utils/pg_locale.h"
28 :
29 :
30 : /* Define me to enable tracing of parser behavior */
31 : /* #define WPARSER_TRACE */
32 :
33 :
34 : /* Output token categories */
35 :
36 : #define ASCIIWORD 1
37 : #define WORD_T 2
38 : #define NUMWORD 3
39 : #define EMAIL 4
40 : #define URL_T 5
41 : #define HOST 6
42 : #define SCIENTIFIC 7
43 : #define VERSIONNUMBER 8
44 : #define NUMPARTHWORD 9
45 : #define PARTHWORD 10
46 : #define ASCIIPARTHWORD 11
47 : #define SPACE 12
48 : #define TAG_T 13
49 : #define PROTOCOL 14
50 : #define NUMHWORD 15
51 : #define ASCIIHWORD 16
52 : #define HWORD 17
53 : #define URLPATH 18
54 : #define FILEPATH 19
55 : #define DECIMAL_T 20
56 : #define SIGNEDINT 21
57 : #define UNSIGNEDINT 22
58 : #define XMLENTITY 23
59 :
60 : #define LASTNUM 23
61 :
62 : static const char *const tok_alias[] = {
63 : "",
64 : "asciiword",
65 : "word",
66 : "numword",
67 : "email",
68 : "url",
69 : "host",
70 : "sfloat",
71 : "version",
72 : "hword_numpart",
73 : "hword_part",
74 : "hword_asciipart",
75 : "blank",
76 : "tag",
77 : "protocol",
78 : "numhword",
79 : "asciihword",
80 : "hword",
81 : "url_path",
82 : "file",
83 : "float",
84 : "int",
85 : "uint",
86 : "entity"
87 : };
88 :
89 : static const char *const lex_descr[] = {
90 : "",
91 : "Word, all ASCII",
92 : "Word, all letters",
93 : "Word, letters and digits",
94 : "Email address",
95 : "URL",
96 : "Host",
97 : "Scientific notation",
98 : "Version number",
99 : "Hyphenated word part, letters and digits",
100 : "Hyphenated word part, all letters",
101 : "Hyphenated word part, all ASCII",
102 : "Space symbols",
103 : "XML tag",
104 : "Protocol head",
105 : "Hyphenated word, letters and digits",
106 : "Hyphenated word, all ASCII",
107 : "Hyphenated word, all letters",
108 : "URL path",
109 : "File or path name",
110 : "Decimal notation",
111 : "Signed integer",
112 : "Unsigned integer",
113 : "XML entity"
114 : };
115 :
116 :
117 : /* Parser states */
118 :
119 : typedef enum
120 : {
121 : TPS_Base = 0,
122 : TPS_InNumWord,
123 : TPS_InAsciiWord,
124 : TPS_InWord,
125 : TPS_InUnsignedInt,
126 : TPS_InSignedIntFirst,
127 : TPS_InSignedInt,
128 : TPS_InSpace,
129 : TPS_InUDecimalFirst,
130 : TPS_InUDecimal,
131 : TPS_InDecimalFirst,
132 : TPS_InDecimal,
133 : TPS_InVerVersion,
134 : TPS_InSVerVersion,
135 : TPS_InVersionFirst,
136 : TPS_InVersion,
137 : TPS_InMantissaFirst,
138 : TPS_InMantissaSign,
139 : TPS_InMantissa,
140 : TPS_InXMLEntityFirst,
141 : TPS_InXMLEntity,
142 : TPS_InXMLEntityNumFirst,
143 : TPS_InXMLEntityNum,
144 : TPS_InXMLEntityHexNumFirst,
145 : TPS_InXMLEntityHexNum,
146 : TPS_InXMLEntityEnd,
147 : TPS_InTagFirst,
148 : TPS_InXMLBegin,
149 : TPS_InTagCloseFirst,
150 : TPS_InTagName,
151 : TPS_InTagBeginEnd,
152 : TPS_InTag,
153 : TPS_InTagEscapeK,
154 : TPS_InTagEscapeKK,
155 : TPS_InTagBackSleshed,
156 : TPS_InTagEnd,
157 : TPS_InCommentFirst,
158 : TPS_InCommentLast,
159 : TPS_InComment,
160 : TPS_InCloseCommentFirst,
161 : TPS_InCloseCommentLast,
162 : TPS_InCommentEnd,
163 : TPS_InHostFirstDomain,
164 : TPS_InHostDomainSecond,
165 : TPS_InHostDomain,
166 : TPS_InPortFirst,
167 : TPS_InPort,
168 : TPS_InHostFirstAN,
169 : TPS_InHost,
170 : TPS_InEmail,
171 : TPS_InFileFirst,
172 : TPS_InFileTwiddle,
173 : TPS_InPathFirst,
174 : TPS_InPathFirstFirst,
175 : TPS_InPathSecond,
176 : TPS_InFile,
177 : TPS_InFileNext,
178 : TPS_InURLPathFirst,
179 : TPS_InURLPathStart,
180 : TPS_InURLPath,
181 : TPS_InFURL,
182 : TPS_InProtocolFirst,
183 : TPS_InProtocolSecond,
184 : TPS_InProtocolEnd,
185 : TPS_InHyphenAsciiWordFirst,
186 : TPS_InHyphenAsciiWord,
187 : TPS_InHyphenWordFirst,
188 : TPS_InHyphenWord,
189 : TPS_InHyphenNumWordFirst,
190 : TPS_InHyphenNumWord,
191 : TPS_InHyphenDigitLookahead,
192 : TPS_InParseHyphen,
193 : TPS_InParseHyphenHyphen,
194 : TPS_InHyphenWordPart,
195 : TPS_InHyphenAsciiWordPart,
196 : TPS_InHyphenNumWordPart,
197 : TPS_InHyphenUnsignedInt,
198 : TPS_Null /* last state (fake value) */
199 : } TParserState;
200 :
201 : /* forward declaration */
202 : struct TParser;
203 :
204 : typedef int (*TParserCharTest) (struct TParser *); /* any p_is* functions
205 : * except p_iseq */
206 : typedef void (*TParserSpecial) (struct TParser *); /* special handler for
207 : * special cases... */
208 :
209 : typedef struct
210 : {
211 : TParserCharTest isclass;
212 : char c;
213 : uint16 flags;
214 : TParserState tostate;
215 : int type;
216 : TParserSpecial special;
217 : } TParserStateActionItem;
218 :
219 : /* Flag bits in TParserStateActionItem.flags */
220 : #define A_NEXT 0x0000
221 : #define A_BINGO 0x0001
222 : #define A_POP 0x0002
223 : #define A_PUSH 0x0004
224 : #define A_RERUN 0x0008
225 : #define A_CLEAR 0x0010
226 : #define A_MERGE 0x0020
227 : #define A_CLRALL 0x0040
228 :
229 : typedef struct TParserPosition
230 : {
231 : int posbyte; /* position of parser in bytes */
232 : int poschar; /* position of parser in characters */
233 : int charlen; /* length of current char */
234 : int lenbytetoken; /* length of token-so-far in bytes */
235 : int lenchartoken; /* and in chars */
236 : TParserState state;
237 : struct TParserPosition *prev;
238 : const TParserStateActionItem *pushedAtAction;
239 : } TParserPosition;
240 :
241 : typedef struct TParser
242 : {
243 : /* string and position information */
244 : char *str; /* multibyte string */
245 : int lenstr; /* length of mbstring */
246 : wchar_t *wstr; /* wide character string */
247 : pg_wchar *pgwstr; /* wide character string for C-locale */
248 : bool usewide;
249 :
250 : /* State of parse */
251 : int charmaxlen;
252 : TParserPosition *state;
253 : bool ignore;
254 : bool wanthost;
255 :
256 : /* silly char */
257 : char c;
258 :
259 : /* out */
260 : char *token;
261 : int lenbytetoken;
262 : int lenchartoken;
263 : int type;
264 : } TParser;
265 :
266 :
267 : /* forward decls here */
268 : static bool TParserGet(TParser *prs);
269 :
270 :
271 : static TParserPosition *
272 10232 : newTParserPosition(TParserPosition *prev)
273 : {
274 10232 : TParserPosition *res = (TParserPosition *) palloc(sizeof(TParserPosition));
275 :
276 10232 : if (prev)
277 5238 : memcpy(res, prev, sizeof(TParserPosition));
278 : else
279 4994 : memset(res, 0, sizeof(TParserPosition));
280 :
281 10232 : res->prev = prev;
282 :
283 10232 : res->pushedAtAction = NULL;
284 :
285 10232 : return res;
286 : }
287 :
288 : static TParser *
289 4754 : TParserInit(char *str, int len)
290 : {
291 4754 : TParser *prs = (TParser *) palloc0(sizeof(TParser));
292 :
293 4754 : prs->charmaxlen = pg_database_encoding_max_length();
294 4754 : prs->str = str;
295 4754 : prs->lenstr = len;
296 :
297 : /*
298 : * Use wide char code only when max encoding length > 1.
299 : */
300 4754 : if (prs->charmaxlen > 1)
301 : {
302 4754 : pg_locale_t mylocale = 0; /* TODO */
303 :
304 4754 : prs->usewide = true;
305 4754 : if (database_ctype_is_c)
306 : {
307 : /*
308 : * char2wchar doesn't work for C-locale and sizeof(pg_wchar) could
309 : * be different from sizeof(wchar_t)
310 : */
311 1582 : prs->pgwstr = (pg_wchar *) palloc(sizeof(pg_wchar) * (prs->lenstr + 1));
312 1582 : pg_mb2wchar_with_len(prs->str, prs->pgwstr, prs->lenstr);
313 : }
314 : else
315 : {
316 3172 : prs->wstr = (wchar_t *) palloc(sizeof(wchar_t) * (prs->lenstr + 1));
317 3172 : char2wchar(prs->wstr, prs->lenstr + 1, prs->str, prs->lenstr,
318 : mylocale);
319 : }
320 : }
321 : else
322 0 : prs->usewide = false;
323 :
324 4754 : prs->state = newTParserPosition(NULL);
325 4754 : prs->state->state = TPS_Base;
326 :
327 : #ifdef WPARSER_TRACE
328 : fprintf(stderr, "parsing \"%.*s\"\n", len, str);
329 : #endif
330 :
331 4754 : return prs;
332 : }
333 :
334 : /*
335 : * As an alternative to a full TParserInit one can create a
336 : * TParserCopy which basically is a regular TParser without a private
337 : * copy of the string - instead it uses the one from another TParser.
338 : * This is useful because at some places TParsers are created
339 : * recursively and the repeated copying around of the strings can
340 : * cause major inefficiency if the source string is long.
341 : * The new parser starts parsing at the original's current position.
342 : *
343 : * Obviously one must not close the original TParser before the copy.
344 : */
345 : static TParser *
346 240 : TParserCopyInit(const TParser *orig)
347 : {
348 240 : TParser *prs = (TParser *) palloc0(sizeof(TParser));
349 :
350 240 : prs->charmaxlen = orig->charmaxlen;
351 240 : prs->str = orig->str + orig->state->posbyte;
352 240 : prs->lenstr = orig->lenstr - orig->state->posbyte;
353 240 : prs->usewide = orig->usewide;
354 :
355 240 : if (orig->pgwstr)
356 80 : prs->pgwstr = orig->pgwstr + orig->state->poschar;
357 240 : if (orig->wstr)
358 160 : prs->wstr = orig->wstr + orig->state->poschar;
359 :
360 240 : prs->state = newTParserPosition(NULL);
361 240 : prs->state->state = TPS_Base;
362 :
363 : #ifdef WPARSER_TRACE
364 : fprintf(stderr, "parsing copy of \"%.*s\"\n", prs->lenstr, prs->str);
365 : #endif
366 :
367 240 : return prs;
368 : }
369 :
370 :
371 : static void
372 4754 : TParserClose(TParser *prs)
373 : {
374 9508 : while (prs->state)
375 : {
376 4754 : TParserPosition *ptr = prs->state->prev;
377 :
378 4754 : pfree(prs->state);
379 4754 : prs->state = ptr;
380 : }
381 :
382 4754 : if (prs->wstr)
383 3172 : pfree(prs->wstr);
384 4754 : if (prs->pgwstr)
385 1582 : pfree(prs->pgwstr);
386 :
387 : #ifdef WPARSER_TRACE
388 : fprintf(stderr, "closing parser\n");
389 : #endif
390 4754 : pfree(prs);
391 4754 : }
392 :
393 : /*
394 : * Close a parser created with TParserCopyInit
395 : */
396 : static void
397 240 : TParserCopyClose(TParser *prs)
398 : {
399 612 : while (prs->state)
400 : {
401 372 : TParserPosition *ptr = prs->state->prev;
402 :
403 372 : pfree(prs->state);
404 372 : prs->state = ptr;
405 : }
406 :
407 : #ifdef WPARSER_TRACE
408 : fprintf(stderr, "closing parser copy\n");
409 : #endif
410 240 : pfree(prs);
411 240 : }
412 :
413 :
414 : /*
415 : * Character-type support functions, equivalent to is* macros, but
416 : * working with any possible encodings and locales. Notes:
417 : * - with multibyte encoding and C-locale isw* function may fail
418 : * or give wrong result.
419 : * - multibyte encoding and C-locale often are used for
420 : * Asian languages.
421 : * - if locale is C then we use pgwstr instead of wstr.
422 : */
423 :
424 : #define p_iswhat(type, nonascii) \
425 : \
426 : static int \
427 : p_is##type(TParser *prs) \
428 : { \
429 : Assert(prs->state); \
430 : if (prs->usewide) \
431 : { \
432 : if (prs->pgwstr) \
433 : { \
434 : unsigned int c = *(prs->pgwstr + prs->state->poschar); \
435 : if (c > 0x7f) \
436 : return nonascii; \
437 : return is##type(c); \
438 : } \
439 : return isw##type(*(prs->wstr + prs->state->poschar)); \
440 : } \
441 : return is##type(*(unsigned char *) (prs->str + prs->state->posbyte)); \
442 : } \
443 : \
444 : static int \
445 : p_isnot##type(TParser *prs) \
446 : { \
447 : return !p_is##type(prs); \
448 : }
449 :
450 : /*
451 : * In C locale with a multibyte encoding, any non-ASCII symbol is considered
452 : * an alpha character, but not a member of other char classes.
453 : */
454 25122 : p_iswhat(alnum, 1)
455 93856 : p_iswhat(alpha, 1)
456 37132 : p_iswhat(digit, 0)
457 0 : p_iswhat(lower, 0)
458 0 : p_iswhat(print, 0)
459 0 : p_iswhat(punct, 0)
460 678 : p_iswhat(space, 0)
461 0 : p_iswhat(upper, 0)
462 18 : p_iswhat(xdigit, 0)
463 :
464 : /* p_iseq should be used only for ascii symbols */
465 :
466 : static int
467 231392 : p_iseq(TParser *prs, char c)
468 : {
469 : Assert(prs->state);
470 231392 : return ((prs->state->charlen == 1 && *(prs->str + prs->state->posbyte) == c)) ? 1 : 0;
471 : }
472 :
473 : static int
474 100158 : p_isEOF(TParser *prs)
475 : {
476 : Assert(prs->state);
477 100158 : return (prs->state->posbyte == prs->lenstr || prs->state->charlen == 0) ? 1 : 0;
478 : }
479 :
480 : static int
481 231392 : p_iseqC(TParser *prs)
482 : {
483 231392 : return p_iseq(prs, prs->c);
484 : }
485 :
486 : static int
487 0 : p_isneC(TParser *prs)
488 : {
489 0 : return !p_iseq(prs, prs->c);
490 : }
491 :
492 : static int
493 73544 : p_isascii(TParser *prs)
494 : {
495 73544 : return (prs->state->charlen == 1 && isascii((unsigned char) *(prs->str + prs->state->posbyte))) ? 1 : 0;
496 : }
497 :
498 : static int
499 73544 : p_isasclet(TParser *prs)
500 : {
501 73544 : return (p_isascii(prs) && p_isalpha(prs)) ? 1 : 0;
502 : }
503 :
504 : static int
505 2658 : p_isurlchar(TParser *prs)
506 : {
507 : char ch;
508 :
509 : /* no non-ASCII need apply */
510 2658 : if (prs->state->charlen != 1)
511 0 : return 0;
512 2658 : ch = *(prs->str + prs->state->posbyte);
513 : /* no spaces or control characters */
514 2658 : if (ch <= 0x20 || ch >= 0x7F)
515 234 : return 0;
516 : /* reject characters disallowed by RFC 3986 */
517 2424 : switch (ch)
518 : {
519 24 : case '"':
520 : case '<':
521 : case '>':
522 : case '\\':
523 : case '^':
524 : case '`':
525 : case '{':
526 : case '|':
527 : case '}':
528 24 : return 0;
529 : }
530 2400 : return 1;
531 : }
532 :
533 :
534 : /* deliberately suppress unused-function complaints for the above */
535 : void _make_compiler_happy(void);
536 : void
537 0 : _make_compiler_happy(void)
538 : {
539 0 : p_isalnum(NULL);
540 0 : p_isnotalnum(NULL);
541 0 : p_isalpha(NULL);
542 0 : p_isnotalpha(NULL);
543 0 : p_isdigit(NULL);
544 0 : p_isnotdigit(NULL);
545 0 : p_islower(NULL);
546 0 : p_isnotlower(NULL);
547 0 : p_isprint(NULL);
548 0 : p_isnotprint(NULL);
549 0 : p_ispunct(NULL);
550 0 : p_isnotpunct(NULL);
551 0 : p_isspace(NULL);
552 0 : p_isnotspace(NULL);
553 0 : p_isupper(NULL);
554 0 : p_isnotupper(NULL);
555 0 : p_isxdigit(NULL);
556 0 : p_isnotxdigit(NULL);
557 0 : p_isEOF(NULL);
558 0 : p_iseqC(NULL);
559 0 : p_isneC(NULL);
560 0 : }
561 :
562 :
563 : static void
564 252 : SpecialTags(TParser *prs)
565 : {
566 252 : switch (prs->state->lenchartoken)
567 : {
568 6 : case 8: /* </script */
569 6 : if (pg_strncasecmp(prs->token, "</script", 8) == 0)
570 6 : prs->ignore = false;
571 6 : break;
572 24 : case 7: /* <script || </style */
573 24 : if (pg_strncasecmp(prs->token, "</style", 7) == 0)
574 0 : prs->ignore = false;
575 24 : else if (pg_strncasecmp(prs->token, "<script", 7) == 0)
576 6 : prs->ignore = true;
577 24 : break;
578 18 : case 6: /* <style */
579 18 : if (pg_strncasecmp(prs->token, "<style", 6) == 0)
580 0 : prs->ignore = true;
581 18 : break;
582 204 : default:
583 204 : break;
584 : }
585 252 : }
586 :
587 : static void
588 132 : SpecialFURL(TParser *prs)
589 : {
590 132 : prs->wanthost = true;
591 132 : prs->state->posbyte -= prs->state->lenbytetoken;
592 132 : prs->state->poschar -= prs->state->lenchartoken;
593 132 : }
594 :
595 : static void
596 36 : SpecialHyphen(TParser *prs)
597 : {
598 36 : prs->state->posbyte -= prs->state->lenbytetoken;
599 36 : prs->state->poschar -= prs->state->lenchartoken;
600 36 : }
601 :
602 : static void
603 0 : SpecialVerVersion(TParser *prs)
604 : {
605 0 : prs->state->posbyte -= prs->state->lenbytetoken;
606 0 : prs->state->poschar -= prs->state->lenchartoken;
607 0 : prs->state->lenbytetoken = 0;
608 0 : prs->state->lenchartoken = 0;
609 0 : }
610 :
611 : static int
612 480 : p_isstophost(TParser *prs)
613 : {
614 480 : if (prs->wanthost)
615 : {
616 204 : prs->wanthost = false;
617 204 : return 1;
618 : }
619 276 : return 0;
620 : }
621 :
622 : static int
623 36086 : p_isignore(TParser *prs)
624 : {
625 36086 : return (prs->ignore) ? 1 : 0;
626 : }
627 :
628 : static int
629 90 : p_ishost(TParser *prs)
630 : {
631 90 : TParser *tmpprs = TParserCopyInit(prs);
632 90 : int res = 0;
633 :
634 90 : tmpprs->wanthost = true;
635 :
636 : /*
637 : * Check stack depth before recursing. (Since TParserGet() doesn't
638 : * normally recurse, we put the cost of checking here not there.)
639 : */
640 90 : check_stack_depth();
641 :
642 90 : if (TParserGet(tmpprs) && tmpprs->type == HOST)
643 : {
644 72 : prs->state->posbyte += tmpprs->lenbytetoken;
645 72 : prs->state->poschar += tmpprs->lenchartoken;
646 72 : prs->state->lenbytetoken += tmpprs->lenbytetoken;
647 72 : prs->state->lenchartoken += tmpprs->lenchartoken;
648 72 : prs->state->charlen = tmpprs->state->charlen;
649 72 : res = 1;
650 : }
651 90 : TParserCopyClose(tmpprs);
652 :
653 90 : return res;
654 : }
655 :
656 : static int
657 150 : p_isURLPath(TParser *prs)
658 : {
659 150 : TParser *tmpprs = TParserCopyInit(prs);
660 150 : int res = 0;
661 :
662 150 : tmpprs->state = newTParserPosition(tmpprs->state);
663 150 : tmpprs->state->state = TPS_InURLPathFirst;
664 :
665 : /*
666 : * Check stack depth before recursing. (Since TParserGet() doesn't
667 : * normally recurse, we put the cost of checking here not there.)
668 : */
669 150 : check_stack_depth();
670 :
671 150 : if (TParserGet(tmpprs) && tmpprs->type == URLPATH)
672 : {
673 132 : prs->state->posbyte += tmpprs->lenbytetoken;
674 132 : prs->state->poschar += tmpprs->lenchartoken;
675 132 : prs->state->lenbytetoken += tmpprs->lenbytetoken;
676 132 : prs->state->lenchartoken += tmpprs->lenchartoken;
677 132 : prs->state->charlen = tmpprs->state->charlen;
678 132 : res = 1;
679 : }
680 150 : TParserCopyClose(tmpprs);
681 :
682 150 : return res;
683 : }
684 :
685 : /*
686 : * returns true if current character has zero display length or
687 : * it's a special sign in several languages. Such characters
688 : * aren't a word-breaker although they aren't an isalpha.
689 : * In beginning of word they aren't a part of it.
690 : */
691 : static int
692 8724 : p_isspecial(TParser *prs)
693 : {
694 : /*
695 : * pg_dsplen could return -1 which means error or control character
696 : */
697 8724 : if (pg_dsplen(prs->str + prs->state->posbyte) == 0)
698 0 : return 1;
699 :
700 : /*
701 : * Unicode Characters in the 'Mark, Spacing Combining' Category That
702 : * characters are not alpha although they are not breakers of word too.
703 : * Check that only in utf encoding, because other encodings aren't
704 : * supported by postgres or even exists.
705 : */
706 8724 : if (GetDatabaseEncoding() == PG_UTF8 && prs->usewide)
707 : {
708 : static const pg_wchar strange_letter[] = {
709 : /*
710 : * use binary search, so elements should be ordered
711 : */
712 : 0x0903, /* DEVANAGARI SIGN VISARGA */
713 : 0x093E, /* DEVANAGARI VOWEL SIGN AA */
714 : 0x093F, /* DEVANAGARI VOWEL SIGN I */
715 : 0x0940, /* DEVANAGARI VOWEL SIGN II */
716 : 0x0949, /* DEVANAGARI VOWEL SIGN CANDRA O */
717 : 0x094A, /* DEVANAGARI VOWEL SIGN SHORT O */
718 : 0x094B, /* DEVANAGARI VOWEL SIGN O */
719 : 0x094C, /* DEVANAGARI VOWEL SIGN AU */
720 : 0x0982, /* BENGALI SIGN ANUSVARA */
721 : 0x0983, /* BENGALI SIGN VISARGA */
722 : 0x09BE, /* BENGALI VOWEL SIGN AA */
723 : 0x09BF, /* BENGALI VOWEL SIGN I */
724 : 0x09C0, /* BENGALI VOWEL SIGN II */
725 : 0x09C7, /* BENGALI VOWEL SIGN E */
726 : 0x09C8, /* BENGALI VOWEL SIGN AI */
727 : 0x09CB, /* BENGALI VOWEL SIGN O */
728 : 0x09CC, /* BENGALI VOWEL SIGN AU */
729 : 0x09D7, /* BENGALI AU LENGTH MARK */
730 : 0x0A03, /* GURMUKHI SIGN VISARGA */
731 : 0x0A3E, /* GURMUKHI VOWEL SIGN AA */
732 : 0x0A3F, /* GURMUKHI VOWEL SIGN I */
733 : 0x0A40, /* GURMUKHI VOWEL SIGN II */
734 : 0x0A83, /* GUJARATI SIGN VISARGA */
735 : 0x0ABE, /* GUJARATI VOWEL SIGN AA */
736 : 0x0ABF, /* GUJARATI VOWEL SIGN I */
737 : 0x0AC0, /* GUJARATI VOWEL SIGN II */
738 : 0x0AC9, /* GUJARATI VOWEL SIGN CANDRA O */
739 : 0x0ACB, /* GUJARATI VOWEL SIGN O */
740 : 0x0ACC, /* GUJARATI VOWEL SIGN AU */
741 : 0x0B02, /* ORIYA SIGN ANUSVARA */
742 : 0x0B03, /* ORIYA SIGN VISARGA */
743 : 0x0B3E, /* ORIYA VOWEL SIGN AA */
744 : 0x0B40, /* ORIYA VOWEL SIGN II */
745 : 0x0B47, /* ORIYA VOWEL SIGN E */
746 : 0x0B48, /* ORIYA VOWEL SIGN AI */
747 : 0x0B4B, /* ORIYA VOWEL SIGN O */
748 : 0x0B4C, /* ORIYA VOWEL SIGN AU */
749 : 0x0B57, /* ORIYA AU LENGTH MARK */
750 : 0x0BBE, /* TAMIL VOWEL SIGN AA */
751 : 0x0BBF, /* TAMIL VOWEL SIGN I */
752 : 0x0BC1, /* TAMIL VOWEL SIGN U */
753 : 0x0BC2, /* TAMIL VOWEL SIGN UU */
754 : 0x0BC6, /* TAMIL VOWEL SIGN E */
755 : 0x0BC7, /* TAMIL VOWEL SIGN EE */
756 : 0x0BC8, /* TAMIL VOWEL SIGN AI */
757 : 0x0BCA, /* TAMIL VOWEL SIGN O */
758 : 0x0BCB, /* TAMIL VOWEL SIGN OO */
759 : 0x0BCC, /* TAMIL VOWEL SIGN AU */
760 : 0x0BD7, /* TAMIL AU LENGTH MARK */
761 : 0x0C01, /* TELUGU SIGN CANDRABINDU */
762 : 0x0C02, /* TELUGU SIGN ANUSVARA */
763 : 0x0C03, /* TELUGU SIGN VISARGA */
764 : 0x0C41, /* TELUGU VOWEL SIGN U */
765 : 0x0C42, /* TELUGU VOWEL SIGN UU */
766 : 0x0C43, /* TELUGU VOWEL SIGN VOCALIC R */
767 : 0x0C44, /* TELUGU VOWEL SIGN VOCALIC RR */
768 : 0x0C82, /* KANNADA SIGN ANUSVARA */
769 : 0x0C83, /* KANNADA SIGN VISARGA */
770 : 0x0CBE, /* KANNADA VOWEL SIGN AA */
771 : 0x0CC0, /* KANNADA VOWEL SIGN II */
772 : 0x0CC1, /* KANNADA VOWEL SIGN U */
773 : 0x0CC2, /* KANNADA VOWEL SIGN UU */
774 : 0x0CC3, /* KANNADA VOWEL SIGN VOCALIC R */
775 : 0x0CC4, /* KANNADA VOWEL SIGN VOCALIC RR */
776 : 0x0CC7, /* KANNADA VOWEL SIGN EE */
777 : 0x0CC8, /* KANNADA VOWEL SIGN AI */
778 : 0x0CCA, /* KANNADA VOWEL SIGN O */
779 : 0x0CCB, /* KANNADA VOWEL SIGN OO */
780 : 0x0CD5, /* KANNADA LENGTH MARK */
781 : 0x0CD6, /* KANNADA AI LENGTH MARK */
782 : 0x0D02, /* MALAYALAM SIGN ANUSVARA */
783 : 0x0D03, /* MALAYALAM SIGN VISARGA */
784 : 0x0D3E, /* MALAYALAM VOWEL SIGN AA */
785 : 0x0D3F, /* MALAYALAM VOWEL SIGN I */
786 : 0x0D40, /* MALAYALAM VOWEL SIGN II */
787 : 0x0D46, /* MALAYALAM VOWEL SIGN E */
788 : 0x0D47, /* MALAYALAM VOWEL SIGN EE */
789 : 0x0D48, /* MALAYALAM VOWEL SIGN AI */
790 : 0x0D4A, /* MALAYALAM VOWEL SIGN O */
791 : 0x0D4B, /* MALAYALAM VOWEL SIGN OO */
792 : 0x0D4C, /* MALAYALAM VOWEL SIGN AU */
793 : 0x0D57, /* MALAYALAM AU LENGTH MARK */
794 : 0x0D82, /* SINHALA SIGN ANUSVARAYA */
795 : 0x0D83, /* SINHALA SIGN VISARGAYA */
796 : 0x0DCF, /* SINHALA VOWEL SIGN AELA-PILLA */
797 : 0x0DD0, /* SINHALA VOWEL SIGN KETTI AEDA-PILLA */
798 : 0x0DD1, /* SINHALA VOWEL SIGN DIGA AEDA-PILLA */
799 : 0x0DD8, /* SINHALA VOWEL SIGN GAETTA-PILLA */
800 : 0x0DD9, /* SINHALA VOWEL SIGN KOMBUVA */
801 : 0x0DDA, /* SINHALA VOWEL SIGN DIGA KOMBUVA */
802 : 0x0DDB, /* SINHALA VOWEL SIGN KOMBU DEKA */
803 : 0x0DDC, /* SINHALA VOWEL SIGN KOMBUVA HAA AELA-PILLA */
804 : 0x0DDD, /* SINHALA VOWEL SIGN KOMBUVA HAA DIGA
805 : * AELA-PILLA */
806 : 0x0DDE, /* SINHALA VOWEL SIGN KOMBUVA HAA GAYANUKITTA */
807 : 0x0DDF, /* SINHALA VOWEL SIGN GAYANUKITTA */
808 : 0x0DF2, /* SINHALA VOWEL SIGN DIGA GAETTA-PILLA */
809 : 0x0DF3, /* SINHALA VOWEL SIGN DIGA GAYANUKITTA */
810 : 0x0F3E, /* TIBETAN SIGN YAR TSHES */
811 : 0x0F3F, /* TIBETAN SIGN MAR TSHES */
812 : 0x0F7F, /* TIBETAN SIGN RNAM BCAD */
813 : 0x102B, /* MYANMAR VOWEL SIGN TALL AA */
814 : 0x102C, /* MYANMAR VOWEL SIGN AA */
815 : 0x1031, /* MYANMAR VOWEL SIGN E */
816 : 0x1038, /* MYANMAR SIGN VISARGA */
817 : 0x103B, /* MYANMAR CONSONANT SIGN MEDIAL YA */
818 : 0x103C, /* MYANMAR CONSONANT SIGN MEDIAL RA */
819 : 0x1056, /* MYANMAR VOWEL SIGN VOCALIC R */
820 : 0x1057, /* MYANMAR VOWEL SIGN VOCALIC RR */
821 : 0x1062, /* MYANMAR VOWEL SIGN SGAW KAREN EU */
822 : 0x1063, /* MYANMAR TONE MARK SGAW KAREN HATHI */
823 : 0x1064, /* MYANMAR TONE MARK SGAW KAREN KE PHO */
824 : 0x1067, /* MYANMAR VOWEL SIGN WESTERN PWO KAREN EU */
825 : 0x1068, /* MYANMAR VOWEL SIGN WESTERN PWO KAREN UE */
826 : 0x1069, /* MYANMAR SIGN WESTERN PWO KAREN TONE-1 */
827 : 0x106A, /* MYANMAR SIGN WESTERN PWO KAREN TONE-2 */
828 : 0x106B, /* MYANMAR SIGN WESTERN PWO KAREN TONE-3 */
829 : 0x106C, /* MYANMAR SIGN WESTERN PWO KAREN TONE-4 */
830 : 0x106D, /* MYANMAR SIGN WESTERN PWO KAREN TONE-5 */
831 : 0x1083, /* MYANMAR VOWEL SIGN SHAN AA */
832 : 0x1084, /* MYANMAR VOWEL SIGN SHAN E */
833 : 0x1087, /* MYANMAR SIGN SHAN TONE-2 */
834 : 0x1088, /* MYANMAR SIGN SHAN TONE-3 */
835 : 0x1089, /* MYANMAR SIGN SHAN TONE-5 */
836 : 0x108A, /* MYANMAR SIGN SHAN TONE-6 */
837 : 0x108B, /* MYANMAR SIGN SHAN COUNCIL TONE-2 */
838 : 0x108C, /* MYANMAR SIGN SHAN COUNCIL TONE-3 */
839 : 0x108F, /* MYANMAR SIGN RUMAI PALAUNG TONE-5 */
840 : 0x17B6, /* KHMER VOWEL SIGN AA */
841 : 0x17BE, /* KHMER VOWEL SIGN OE */
842 : 0x17BF, /* KHMER VOWEL SIGN YA */
843 : 0x17C0, /* KHMER VOWEL SIGN IE */
844 : 0x17C1, /* KHMER VOWEL SIGN E */
845 : 0x17C2, /* KHMER VOWEL SIGN AE */
846 : 0x17C3, /* KHMER VOWEL SIGN AI */
847 : 0x17C4, /* KHMER VOWEL SIGN OO */
848 : 0x17C5, /* KHMER VOWEL SIGN AU */
849 : 0x17C7, /* KHMER SIGN REAHMUK */
850 : 0x17C8, /* KHMER SIGN YUUKALEAPINTU */
851 : 0x1923, /* LIMBU VOWEL SIGN EE */
852 : 0x1924, /* LIMBU VOWEL SIGN AI */
853 : 0x1925, /* LIMBU VOWEL SIGN OO */
854 : 0x1926, /* LIMBU VOWEL SIGN AU */
855 : 0x1929, /* LIMBU SUBJOINED LETTER YA */
856 : 0x192A, /* LIMBU SUBJOINED LETTER RA */
857 : 0x192B, /* LIMBU SUBJOINED LETTER WA */
858 : 0x1930, /* LIMBU SMALL LETTER KA */
859 : 0x1931, /* LIMBU SMALL LETTER NGA */
860 : 0x1933, /* LIMBU SMALL LETTER TA */
861 : 0x1934, /* LIMBU SMALL LETTER NA */
862 : 0x1935, /* LIMBU SMALL LETTER PA */
863 : 0x1936, /* LIMBU SMALL LETTER MA */
864 : 0x1937, /* LIMBU SMALL LETTER RA */
865 : 0x1938, /* LIMBU SMALL LETTER LA */
866 : 0x19B0, /* NEW TAI LUE VOWEL SIGN VOWEL SHORTENER */
867 : 0x19B1, /* NEW TAI LUE VOWEL SIGN AA */
868 : 0x19B2, /* NEW TAI LUE VOWEL SIGN II */
869 : 0x19B3, /* NEW TAI LUE VOWEL SIGN U */
870 : 0x19B4, /* NEW TAI LUE VOWEL SIGN UU */
871 : 0x19B5, /* NEW TAI LUE VOWEL SIGN E */
872 : 0x19B6, /* NEW TAI LUE VOWEL SIGN AE */
873 : 0x19B7, /* NEW TAI LUE VOWEL SIGN O */
874 : 0x19B8, /* NEW TAI LUE VOWEL SIGN OA */
875 : 0x19B9, /* NEW TAI LUE VOWEL SIGN UE */
876 : 0x19BA, /* NEW TAI LUE VOWEL SIGN AY */
877 : 0x19BB, /* NEW TAI LUE VOWEL SIGN AAY */
878 : 0x19BC, /* NEW TAI LUE VOWEL SIGN UY */
879 : 0x19BD, /* NEW TAI LUE VOWEL SIGN OY */
880 : 0x19BE, /* NEW TAI LUE VOWEL SIGN OAY */
881 : 0x19BF, /* NEW TAI LUE VOWEL SIGN UEY */
882 : 0x19C0, /* NEW TAI LUE VOWEL SIGN IY */
883 : 0x19C8, /* NEW TAI LUE TONE MARK-1 */
884 : 0x19C9, /* NEW TAI LUE TONE MARK-2 */
885 : 0x1A19, /* BUGINESE VOWEL SIGN E */
886 : 0x1A1A, /* BUGINESE VOWEL SIGN O */
887 : 0x1A1B, /* BUGINESE VOWEL SIGN AE */
888 : 0x1B04, /* BALINESE SIGN BISAH */
889 : 0x1B35, /* BALINESE VOWEL SIGN TEDUNG */
890 : 0x1B3B, /* BALINESE VOWEL SIGN RA REPA TEDUNG */
891 : 0x1B3D, /* BALINESE VOWEL SIGN LA LENGA TEDUNG */
892 : 0x1B3E, /* BALINESE VOWEL SIGN TALING */
893 : 0x1B3F, /* BALINESE VOWEL SIGN TALING REPA */
894 : 0x1B40, /* BALINESE VOWEL SIGN TALING TEDUNG */
895 : 0x1B41, /* BALINESE VOWEL SIGN TALING REPA TEDUNG */
896 : 0x1B43, /* BALINESE VOWEL SIGN PEPET TEDUNG */
897 : 0x1B44, /* BALINESE ADEG ADEG */
898 : 0x1B82, /* SUNDANESE SIGN PANGWISAD */
899 : 0x1BA1, /* SUNDANESE CONSONANT SIGN PAMINGKAL */
900 : 0x1BA6, /* SUNDANESE VOWEL SIGN PANAELAENG */
901 : 0x1BA7, /* SUNDANESE VOWEL SIGN PANOLONG */
902 : 0x1BAA, /* SUNDANESE SIGN PAMAAEH */
903 : 0x1C24, /* LEPCHA SUBJOINED LETTER YA */
904 : 0x1C25, /* LEPCHA SUBJOINED LETTER RA */
905 : 0x1C26, /* LEPCHA VOWEL SIGN AA */
906 : 0x1C27, /* LEPCHA VOWEL SIGN I */
907 : 0x1C28, /* LEPCHA VOWEL SIGN O */
908 : 0x1C29, /* LEPCHA VOWEL SIGN OO */
909 : 0x1C2A, /* LEPCHA VOWEL SIGN U */
910 : 0x1C2B, /* LEPCHA VOWEL SIGN UU */
911 : 0x1C34, /* LEPCHA CONSONANT SIGN NYIN-DO */
912 : 0x1C35, /* LEPCHA CONSONANT SIGN KANG */
913 : 0xA823, /* SYLOTI NAGRI VOWEL SIGN A */
914 : 0xA824, /* SYLOTI NAGRI VOWEL SIGN I */
915 : 0xA827, /* SYLOTI NAGRI VOWEL SIGN OO */
916 : 0xA880, /* SAURASHTRA SIGN ANUSVARA */
917 : 0xA881, /* SAURASHTRA SIGN VISARGA */
918 : 0xA8B4, /* SAURASHTRA CONSONANT SIGN HAARU */
919 : 0xA8B5, /* SAURASHTRA VOWEL SIGN AA */
920 : 0xA8B6, /* SAURASHTRA VOWEL SIGN I */
921 : 0xA8B7, /* SAURASHTRA VOWEL SIGN II */
922 : 0xA8B8, /* SAURASHTRA VOWEL SIGN U */
923 : 0xA8B9, /* SAURASHTRA VOWEL SIGN UU */
924 : 0xA8BA, /* SAURASHTRA VOWEL SIGN VOCALIC R */
925 : 0xA8BB, /* SAURASHTRA VOWEL SIGN VOCALIC RR */
926 : 0xA8BC, /* SAURASHTRA VOWEL SIGN VOCALIC L */
927 : 0xA8BD, /* SAURASHTRA VOWEL SIGN VOCALIC LL */
928 : 0xA8BE, /* SAURASHTRA VOWEL SIGN E */
929 : 0xA8BF, /* SAURASHTRA VOWEL SIGN EE */
930 : 0xA8C0, /* SAURASHTRA VOWEL SIGN AI */
931 : 0xA8C1, /* SAURASHTRA VOWEL SIGN O */
932 : 0xA8C2, /* SAURASHTRA VOWEL SIGN OO */
933 : 0xA8C3, /* SAURASHTRA VOWEL SIGN AU */
934 : 0xA952, /* REJANG CONSONANT SIGN H */
935 : 0xA953, /* REJANG VIRAMA */
936 : 0xAA2F, /* CHAM VOWEL SIGN O */
937 : 0xAA30, /* CHAM VOWEL SIGN AI */
938 : 0xAA33, /* CHAM CONSONANT SIGN YA */
939 : 0xAA34, /* CHAM CONSONANT SIGN RA */
940 : 0xAA4D /* CHAM CONSONANT SIGN FINAL H */
941 : };
942 8724 : const pg_wchar *StopLow = strange_letter,
943 8724 : *StopHigh = strange_letter + lengthof(strange_letter),
944 : *StopMiddle;
945 : pg_wchar c;
946 :
947 8724 : if (prs->pgwstr)
948 2908 : c = *(prs->pgwstr + prs->state->poschar);
949 : else
950 5816 : c = (pg_wchar) *(prs->wstr + prs->state->poschar);
951 :
952 78516 : while (StopLow < StopHigh)
953 : {
954 69792 : StopMiddle = StopLow + ((StopHigh - StopLow) >> 1);
955 69792 : if (*StopMiddle == c)
956 0 : return 1;
957 69792 : else if (*StopMiddle < c)
958 0 : StopLow = StopMiddle + 1;
959 : else
960 69792 : StopHigh = StopMiddle;
961 : }
962 : }
963 :
964 8724 : return 0;
965 : }
966 :
967 : /*
968 : * Table of state/action of parser
969 : */
970 :
971 : static const TParserStateActionItem actionTPS_Base[] = {
972 : {p_isEOF, 0, A_NEXT, TPS_Null, 0, NULL},
973 : {p_iseqC, '<', A_PUSH, TPS_InTagFirst, 0, NULL},
974 : {p_isignore, 0, A_NEXT, TPS_InSpace, 0, NULL},
975 : {p_isasclet, 0, A_NEXT, TPS_InAsciiWord, 0, NULL},
976 : {p_isalpha, 0, A_NEXT, TPS_InWord, 0, NULL},
977 : {p_isdigit, 0, A_NEXT, TPS_InUnsignedInt, 0, NULL},
978 : {p_iseqC, '-', A_PUSH, TPS_InSignedIntFirst, 0, NULL},
979 : {p_iseqC, '+', A_PUSH, TPS_InSignedIntFirst, 0, NULL},
980 : {p_iseqC, '&', A_PUSH, TPS_InXMLEntityFirst, 0, NULL},
981 : {p_iseqC, '~', A_PUSH, TPS_InFileTwiddle, 0, NULL},
982 : {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL},
983 : {p_iseqC, '.', A_PUSH, TPS_InPathFirstFirst, 0, NULL},
984 : {NULL, 0, A_NEXT, TPS_InSpace, 0, NULL}
985 : };
986 :
987 :
988 : static const TParserStateActionItem actionTPS_InNumWord[] = {
989 : {p_isEOF, 0, A_BINGO, TPS_Base, NUMWORD, NULL},
990 : {p_isalnum, 0, A_NEXT, TPS_InNumWord, 0, NULL},
991 : {p_isspecial, 0, A_NEXT, TPS_InNumWord, 0, NULL},
992 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
993 : {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL},
994 : {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL},
995 : {p_iseqC, '-', A_PUSH, TPS_InHyphenNumWordFirst, 0, NULL},
996 : {NULL, 0, A_BINGO, TPS_Base, NUMWORD, NULL}
997 : };
998 :
999 : static const TParserStateActionItem actionTPS_InAsciiWord[] = {
1000 : {p_isEOF, 0, A_BINGO, TPS_Base, ASCIIWORD, NULL},
1001 : {p_isasclet, 0, A_NEXT, TPS_Null, 0, NULL},
1002 : {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL},
1003 : {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL},
1004 : {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1005 : {p_iseqC, '-', A_PUSH, TPS_InHyphenAsciiWordFirst, 0, NULL},
1006 : {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1007 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
1008 : {p_iseqC, ':', A_PUSH, TPS_InProtocolFirst, 0, NULL},
1009 : {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL},
1010 : {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL},
1011 : {p_isdigit, 0, A_NEXT, TPS_InNumWord, 0, NULL},
1012 : {p_isalpha, 0, A_NEXT, TPS_InWord, 0, NULL},
1013 : {p_isspecial, 0, A_NEXT, TPS_InWord, 0, NULL},
1014 : {NULL, 0, A_BINGO, TPS_Base, ASCIIWORD, NULL}
1015 : };
1016 :
1017 : static const TParserStateActionItem actionTPS_InWord[] = {
1018 : {p_isEOF, 0, A_BINGO, TPS_Base, WORD_T, NULL},
1019 : {p_isalpha, 0, A_NEXT, TPS_Null, 0, NULL},
1020 : {p_isspecial, 0, A_NEXT, TPS_Null, 0, NULL},
1021 : {p_isdigit, 0, A_NEXT, TPS_InNumWord, 0, NULL},
1022 : {p_iseqC, '-', A_PUSH, TPS_InHyphenWordFirst, 0, NULL},
1023 : {NULL, 0, A_BINGO, TPS_Base, WORD_T, NULL}
1024 : };
1025 :
1026 : static const TParserStateActionItem actionTPS_InUnsignedInt[] = {
1027 : {p_isEOF, 0, A_BINGO, TPS_Base, UNSIGNEDINT, NULL},
1028 : {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL},
1029 : {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL},
1030 : {p_iseqC, '.', A_PUSH, TPS_InUDecimalFirst, 0, NULL},
1031 : {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1032 : {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1033 : {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1034 : {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1035 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
1036 : {p_isasclet, 0, A_PUSH, TPS_InHost, 0, NULL},
1037 : {p_isalpha, 0, A_NEXT, TPS_InNumWord, 0, NULL},
1038 : {p_isspecial, 0, A_NEXT, TPS_InNumWord, 0, NULL},
1039 : {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL},
1040 : {NULL, 0, A_BINGO, TPS_Base, UNSIGNEDINT, NULL}
1041 : };
1042 :
1043 : static const TParserStateActionItem actionTPS_InSignedIntFirst[] = {
1044 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1045 : {p_isdigit, 0, A_NEXT | A_CLEAR, TPS_InSignedInt, 0, NULL},
1046 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1047 : };
1048 :
1049 : static const TParserStateActionItem actionTPS_InSignedInt[] = {
1050 : {p_isEOF, 0, A_BINGO, TPS_Base, SIGNEDINT, NULL},
1051 : {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL},
1052 : {p_iseqC, '.', A_PUSH, TPS_InDecimalFirst, 0, NULL},
1053 : {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1054 : {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1055 : {NULL, 0, A_BINGO, TPS_Base, SIGNEDINT, NULL}
1056 : };
1057 :
1058 : static const TParserStateActionItem actionTPS_InSpace[] = {
1059 : {p_isEOF, 0, A_BINGO, TPS_Base, SPACE, NULL},
1060 : {p_iseqC, '<', A_BINGO, TPS_Base, SPACE, NULL},
1061 : {p_isignore, 0, A_NEXT, TPS_Null, 0, NULL},
1062 : {p_iseqC, '-', A_BINGO, TPS_Base, SPACE, NULL},
1063 : {p_iseqC, '+', A_BINGO, TPS_Base, SPACE, NULL},
1064 : {p_iseqC, '&', A_BINGO, TPS_Base, SPACE, NULL},
1065 : {p_iseqC, '/', A_BINGO, TPS_Base, SPACE, NULL},
1066 : {p_isnotalnum, 0, A_NEXT, TPS_InSpace, 0, NULL},
1067 : {NULL, 0, A_BINGO, TPS_Base, SPACE, NULL}
1068 : };
1069 :
1070 : static const TParserStateActionItem actionTPS_InUDecimalFirst[] = {
1071 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1072 : {p_isdigit, 0, A_CLEAR, TPS_InUDecimal, 0, NULL},
1073 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1074 : };
1075 :
1076 : static const TParserStateActionItem actionTPS_InUDecimal[] = {
1077 : {p_isEOF, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL},
1078 : {p_isdigit, 0, A_NEXT, TPS_InUDecimal, 0, NULL},
1079 : {p_iseqC, '.', A_PUSH, TPS_InVersionFirst, 0, NULL},
1080 : {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1081 : {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1082 : {NULL, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL}
1083 : };
1084 :
1085 : static const TParserStateActionItem actionTPS_InDecimalFirst[] = {
1086 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1087 : {p_isdigit, 0, A_CLEAR, TPS_InDecimal, 0, NULL},
1088 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1089 : };
1090 :
1091 : static const TParserStateActionItem actionTPS_InDecimal[] = {
1092 : {p_isEOF, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL},
1093 : {p_isdigit, 0, A_NEXT, TPS_InDecimal, 0, NULL},
1094 : {p_iseqC, '.', A_PUSH, TPS_InVerVersion, 0, NULL},
1095 : {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1096 : {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL},
1097 : {NULL, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL}
1098 : };
1099 :
1100 : static const TParserStateActionItem actionTPS_InVerVersion[] = {
1101 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1102 : {p_isdigit, 0, A_RERUN, TPS_InSVerVersion, 0, SpecialVerVersion},
1103 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1104 : };
1105 :
1106 : static const TParserStateActionItem actionTPS_InSVerVersion[] = {
1107 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1108 : {p_isdigit, 0, A_BINGO | A_CLRALL, TPS_InUnsignedInt, SPACE, NULL},
1109 : {NULL, 0, A_NEXT, TPS_Null, 0, NULL}
1110 : };
1111 :
1112 :
1113 : static const TParserStateActionItem actionTPS_InVersionFirst[] = {
1114 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1115 : {p_isdigit, 0, A_CLEAR, TPS_InVersion, 0, NULL},
1116 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1117 : };
1118 :
1119 : static const TParserStateActionItem actionTPS_InVersion[] = {
1120 : {p_isEOF, 0, A_BINGO, TPS_Base, VERSIONNUMBER, NULL},
1121 : {p_isdigit, 0, A_NEXT, TPS_InVersion, 0, NULL},
1122 : {p_iseqC, '.', A_PUSH, TPS_InVersionFirst, 0, NULL},
1123 : {NULL, 0, A_BINGO, TPS_Base, VERSIONNUMBER, NULL}
1124 : };
1125 :
1126 : static const TParserStateActionItem actionTPS_InMantissaFirst[] = {
1127 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1128 : {p_isdigit, 0, A_CLEAR, TPS_InMantissa, 0, NULL},
1129 : {p_iseqC, '+', A_NEXT, TPS_InMantissaSign, 0, NULL},
1130 : {p_iseqC, '-', A_NEXT, TPS_InMantissaSign, 0, NULL},
1131 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1132 : };
1133 :
1134 : static const TParserStateActionItem actionTPS_InMantissaSign[] = {
1135 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1136 : {p_isdigit, 0, A_CLEAR, TPS_InMantissa, 0, NULL},
1137 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1138 : };
1139 :
1140 : static const TParserStateActionItem actionTPS_InMantissa[] = {
1141 : {p_isEOF, 0, A_BINGO, TPS_Base, SCIENTIFIC, NULL},
1142 : {p_isdigit, 0, A_NEXT, TPS_InMantissa, 0, NULL},
1143 : {NULL, 0, A_BINGO, TPS_Base, SCIENTIFIC, NULL}
1144 : };
1145 :
1146 : static const TParserStateActionItem actionTPS_InXMLEntityFirst[] = {
1147 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1148 : {p_iseqC, '#', A_NEXT, TPS_InXMLEntityNumFirst, 0, NULL},
1149 : {p_isasclet, 0, A_NEXT, TPS_InXMLEntity, 0, NULL},
1150 : {p_iseqC, ':', A_NEXT, TPS_InXMLEntity, 0, NULL},
1151 : {p_iseqC, '_', A_NEXT, TPS_InXMLEntity, 0, NULL},
1152 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1153 : };
1154 :
1155 : static const TParserStateActionItem actionTPS_InXMLEntity[] = {
1156 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1157 : {p_isalnum, 0, A_NEXT, TPS_InXMLEntity, 0, NULL},
1158 : {p_iseqC, ':', A_NEXT, TPS_InXMLEntity, 0, NULL},
1159 : {p_iseqC, '_', A_NEXT, TPS_InXMLEntity, 0, NULL},
1160 : {p_iseqC, '.', A_NEXT, TPS_InXMLEntity, 0, NULL},
1161 : {p_iseqC, '-', A_NEXT, TPS_InXMLEntity, 0, NULL},
1162 : {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL},
1163 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1164 : };
1165 :
1166 : static const TParserStateActionItem actionTPS_InXMLEntityNumFirst[] = {
1167 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1168 : {p_iseqC, 'x', A_NEXT, TPS_InXMLEntityHexNumFirst, 0, NULL},
1169 : {p_iseqC, 'X', A_NEXT, TPS_InXMLEntityHexNumFirst, 0, NULL},
1170 : {p_isdigit, 0, A_NEXT, TPS_InXMLEntityNum, 0, NULL},
1171 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1172 : };
1173 :
1174 : static const TParserStateActionItem actionTPS_InXMLEntityHexNumFirst[] = {
1175 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1176 : {p_isxdigit, 0, A_NEXT, TPS_InXMLEntityHexNum, 0, NULL},
1177 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1178 : };
1179 :
1180 : static const TParserStateActionItem actionTPS_InXMLEntityNum[] = {
1181 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1182 : {p_isdigit, 0, A_NEXT, TPS_InXMLEntityNum, 0, NULL},
1183 : {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL},
1184 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1185 : };
1186 :
1187 : static const TParserStateActionItem actionTPS_InXMLEntityHexNum[] = {
1188 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1189 : {p_isxdigit, 0, A_NEXT, TPS_InXMLEntityHexNum, 0, NULL},
1190 : {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL},
1191 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1192 : };
1193 :
1194 : static const TParserStateActionItem actionTPS_InXMLEntityEnd[] = {
1195 : {NULL, 0, A_BINGO | A_CLEAR, TPS_Base, XMLENTITY, NULL}
1196 : };
1197 :
1198 : static const TParserStateActionItem actionTPS_InTagFirst[] = {
1199 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1200 : {p_iseqC, '/', A_PUSH, TPS_InTagCloseFirst, 0, NULL},
1201 : {p_iseqC, '!', A_PUSH, TPS_InCommentFirst, 0, NULL},
1202 : {p_iseqC, '?', A_PUSH, TPS_InXMLBegin, 0, NULL},
1203 : {p_isasclet, 0, A_PUSH, TPS_InTagName, 0, NULL},
1204 : {p_iseqC, ':', A_PUSH, TPS_InTagName, 0, NULL},
1205 : {p_iseqC, '_', A_PUSH, TPS_InTagName, 0, NULL},
1206 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1207 : };
1208 :
1209 : static const TParserStateActionItem actionTPS_InXMLBegin[] = {
1210 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1211 : /* <?xml ... */
1212 : /* XXX do we wants states for the m and l ? Right now this accepts <?xZ */
1213 : {p_iseqC, 'x', A_NEXT, TPS_InTag, 0, NULL},
1214 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1215 : };
1216 :
1217 : static const TParserStateActionItem actionTPS_InTagCloseFirst[] = {
1218 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1219 : {p_isasclet, 0, A_NEXT, TPS_InTagName, 0, NULL},
1220 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1221 : };
1222 :
1223 : static const TParserStateActionItem actionTPS_InTagName[] = {
1224 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1225 : /* <br/> case */
1226 : {p_iseqC, '/', A_NEXT, TPS_InTagBeginEnd, 0, NULL},
1227 : {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, SpecialTags},
1228 : {p_isspace, 0, A_NEXT, TPS_InTag, 0, SpecialTags},
1229 : {p_isalnum, 0, A_NEXT, TPS_Null, 0, NULL},
1230 : {p_iseqC, ':', A_NEXT, TPS_Null, 0, NULL},
1231 : {p_iseqC, '_', A_NEXT, TPS_Null, 0, NULL},
1232 : {p_iseqC, '.', A_NEXT, TPS_Null, 0, NULL},
1233 : {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL},
1234 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1235 : };
1236 :
1237 : static const TParserStateActionItem actionTPS_InTagBeginEnd[] = {
1238 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1239 : {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, NULL},
1240 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1241 : };
1242 :
1243 : static const TParserStateActionItem actionTPS_InTag[] = {
1244 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1245 : {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, SpecialTags},
1246 : {p_iseqC, '\'', A_NEXT, TPS_InTagEscapeK, 0, NULL},
1247 : {p_iseqC, '"', A_NEXT, TPS_InTagEscapeKK, 0, NULL},
1248 : {p_isasclet, 0, A_NEXT, TPS_Null, 0, NULL},
1249 : {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL},
1250 : {p_iseqC, '=', A_NEXT, TPS_Null, 0, NULL},
1251 : {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL},
1252 : {p_iseqC, '_', A_NEXT, TPS_Null, 0, NULL},
1253 : {p_iseqC, '#', A_NEXT, TPS_Null, 0, NULL},
1254 : {p_iseqC, '/', A_NEXT, TPS_Null, 0, NULL},
1255 : {p_iseqC, ':', A_NEXT, TPS_Null, 0, NULL},
1256 : {p_iseqC, '.', A_NEXT, TPS_Null, 0, NULL},
1257 : {p_iseqC, '&', A_NEXT, TPS_Null, 0, NULL},
1258 : {p_iseqC, '?', A_NEXT, TPS_Null, 0, NULL},
1259 : {p_iseqC, '%', A_NEXT, TPS_Null, 0, NULL},
1260 : {p_iseqC, '~', A_NEXT, TPS_Null, 0, NULL},
1261 : {p_isspace, 0, A_NEXT, TPS_Null, 0, SpecialTags},
1262 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1263 : };
1264 :
1265 : static const TParserStateActionItem actionTPS_InTagEscapeK[] = {
1266 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1267 : {p_iseqC, '\\', A_PUSH, TPS_InTagBackSleshed, 0, NULL},
1268 : {p_iseqC, '\'', A_NEXT, TPS_InTag, 0, NULL},
1269 : {NULL, 0, A_NEXT, TPS_InTagEscapeK, 0, NULL}
1270 : };
1271 :
1272 : static const TParserStateActionItem actionTPS_InTagEscapeKK[] = {
1273 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1274 : {p_iseqC, '\\', A_PUSH, TPS_InTagBackSleshed, 0, NULL},
1275 : {p_iseqC, '"', A_NEXT, TPS_InTag, 0, NULL},
1276 : {NULL, 0, A_NEXT, TPS_InTagEscapeKK, 0, NULL}
1277 : };
1278 :
1279 : static const TParserStateActionItem actionTPS_InTagBackSleshed[] = {
1280 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1281 : {NULL, 0, A_MERGE, TPS_Null, 0, NULL}
1282 : };
1283 :
1284 : static const TParserStateActionItem actionTPS_InTagEnd[] = {
1285 : {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, TAG_T, NULL}
1286 : };
1287 :
1288 : static const TParserStateActionItem actionTPS_InCommentFirst[] = {
1289 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1290 : {p_iseqC, '-', A_NEXT, TPS_InCommentLast, 0, NULL},
1291 : /* <!DOCTYPE ...> */
1292 : {p_iseqC, 'D', A_NEXT, TPS_InTag, 0, NULL},
1293 : {p_iseqC, 'd', A_NEXT, TPS_InTag, 0, NULL},
1294 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1295 : };
1296 :
1297 : static const TParserStateActionItem actionTPS_InCommentLast[] = {
1298 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1299 : {p_iseqC, '-', A_NEXT, TPS_InComment, 0, NULL},
1300 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1301 : };
1302 :
1303 : static const TParserStateActionItem actionTPS_InComment[] = {
1304 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1305 : {p_iseqC, '-', A_NEXT, TPS_InCloseCommentFirst, 0, NULL},
1306 : {NULL, 0, A_NEXT, TPS_Null, 0, NULL}
1307 : };
1308 :
1309 : static const TParserStateActionItem actionTPS_InCloseCommentFirst[] = {
1310 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1311 : {p_iseqC, '-', A_NEXT, TPS_InCloseCommentLast, 0, NULL},
1312 : {NULL, 0, A_NEXT, TPS_InComment, 0, NULL}
1313 : };
1314 :
1315 : static const TParserStateActionItem actionTPS_InCloseCommentLast[] = {
1316 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1317 : {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL},
1318 : {p_iseqC, '>', A_NEXT, TPS_InCommentEnd, 0, NULL},
1319 : {NULL, 0, A_NEXT, TPS_InComment, 0, NULL}
1320 : };
1321 :
1322 : static const TParserStateActionItem actionTPS_InCommentEnd[] = {
1323 : {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, TAG_T, NULL}
1324 : };
1325 :
1326 : static const TParserStateActionItem actionTPS_InHostFirstDomain[] = {
1327 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1328 : {p_isasclet, 0, A_NEXT, TPS_InHostDomainSecond, 0, NULL},
1329 : {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL},
1330 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1331 : };
1332 :
1333 : static const TParserStateActionItem actionTPS_InHostDomainSecond[] = {
1334 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1335 : {p_isasclet, 0, A_NEXT, TPS_InHostDomain, 0, NULL},
1336 : {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL},
1337 : {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1338 : {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1339 : {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL},
1340 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
1341 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1342 : };
1343 :
1344 : static const TParserStateActionItem actionTPS_InHostDomain[] = {
1345 : {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL},
1346 : {p_isasclet, 0, A_NEXT, TPS_InHostDomain, 0, NULL},
1347 : {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL},
1348 : {p_iseqC, ':', A_PUSH, TPS_InPortFirst, 0, NULL},
1349 : {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1350 : {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1351 : {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL},
1352 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
1353 : {p_isdigit, 0, A_POP, TPS_Null, 0, NULL},
1354 : {p_isstophost, 0, A_BINGO | A_CLRALL, TPS_InURLPathStart, HOST, NULL},
1355 : {p_iseqC, '/', A_PUSH, TPS_InFURL, 0, NULL},
1356 : {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL}
1357 : };
1358 :
1359 : static const TParserStateActionItem actionTPS_InPortFirst[] = {
1360 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1361 : {p_isdigit, 0, A_NEXT, TPS_InPort, 0, NULL},
1362 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1363 : };
1364 :
1365 : static const TParserStateActionItem actionTPS_InPort[] = {
1366 : {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL},
1367 : {p_isdigit, 0, A_NEXT, TPS_InPort, 0, NULL},
1368 : {p_isstophost, 0, A_BINGO | A_CLRALL, TPS_InURLPathStart, HOST, NULL},
1369 : {p_iseqC, '/', A_PUSH, TPS_InFURL, 0, NULL},
1370 : {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL}
1371 : };
1372 :
1373 : static const TParserStateActionItem actionTPS_InHostFirstAN[] = {
1374 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1375 : {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL},
1376 : {p_isasclet, 0, A_NEXT, TPS_InHost, 0, NULL},
1377 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1378 : };
1379 :
1380 : static const TParserStateActionItem actionTPS_InHost[] = {
1381 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1382 : {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL},
1383 : {p_isasclet, 0, A_NEXT, TPS_InHost, 0, NULL},
1384 : {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL},
1385 : {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL},
1386 : {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1387 : {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL},
1388 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1389 : };
1390 :
1391 : static const TParserStateActionItem actionTPS_InEmail[] = {
1392 : {p_isstophost, 0, A_POP, TPS_Null, 0, NULL},
1393 : {p_ishost, 0, A_BINGO | A_CLRALL, TPS_Base, EMAIL, NULL},
1394 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1395 : };
1396 :
1397 : static const TParserStateActionItem actionTPS_InFileFirst[] = {
1398 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1399 : {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL},
1400 : {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL},
1401 : {p_iseqC, '.', A_NEXT, TPS_InPathFirst, 0, NULL},
1402 : {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL},
1403 : {p_iseqC, '~', A_PUSH, TPS_InFileTwiddle, 0, NULL},
1404 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1405 : };
1406 :
1407 : static const TParserStateActionItem actionTPS_InFileTwiddle[] = {
1408 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1409 : {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL},
1410 : {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL},
1411 : {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL},
1412 : {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL},
1413 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1414 : };
1415 :
1416 : static const TParserStateActionItem actionTPS_InPathFirst[] = {
1417 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1418 : {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL},
1419 : {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL},
1420 : {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL},
1421 : {p_iseqC, '.', A_NEXT, TPS_InPathSecond, 0, NULL},
1422 : {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL},
1423 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1424 : };
1425 :
1426 : static const TParserStateActionItem actionTPS_InPathFirstFirst[] = {
1427 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1428 : {p_iseqC, '.', A_NEXT, TPS_InPathSecond, 0, NULL},
1429 : {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL},
1430 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1431 : };
1432 :
1433 : static const TParserStateActionItem actionTPS_InPathSecond[] = {
1434 : {p_isEOF, 0, A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL},
1435 : {p_iseqC, '/', A_NEXT | A_PUSH, TPS_InFileFirst, 0, NULL},
1436 : {p_iseqC, '/', A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL},
1437 : {p_isspace, 0, A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL},
1438 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1439 : };
1440 :
1441 : static const TParserStateActionItem actionTPS_InFile[] = {
1442 : {p_isEOF, 0, A_BINGO, TPS_Base, FILEPATH, NULL},
1443 : {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL},
1444 : {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL},
1445 : {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL},
1446 : {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL},
1447 : {p_iseqC, '-', A_NEXT, TPS_InFile, 0, NULL},
1448 : {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL},
1449 : {NULL, 0, A_BINGO, TPS_Base, FILEPATH, NULL}
1450 : };
1451 :
1452 : static const TParserStateActionItem actionTPS_InFileNext[] = {
1453 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1454 : {p_isasclet, 0, A_CLEAR, TPS_InFile, 0, NULL},
1455 : {p_isdigit, 0, A_CLEAR, TPS_InFile, 0, NULL},
1456 : {p_iseqC, '_', A_CLEAR, TPS_InFile, 0, NULL},
1457 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1458 : };
1459 :
1460 : static const TParserStateActionItem actionTPS_InURLPathFirst[] = {
1461 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1462 : {p_isurlchar, 0, A_NEXT, TPS_InURLPath, 0, NULL},
1463 : {NULL, 0, A_POP, TPS_Null, 0, NULL},
1464 : };
1465 :
1466 : static const TParserStateActionItem actionTPS_InURLPathStart[] = {
1467 : {NULL, 0, A_NEXT, TPS_InURLPath, 0, NULL}
1468 : };
1469 :
1470 : static const TParserStateActionItem actionTPS_InURLPath[] = {
1471 : {p_isEOF, 0, A_BINGO, TPS_Base, URLPATH, NULL},
1472 : {p_isurlchar, 0, A_NEXT, TPS_InURLPath, 0, NULL},
1473 : {NULL, 0, A_BINGO, TPS_Base, URLPATH, NULL}
1474 : };
1475 :
1476 : static const TParserStateActionItem actionTPS_InFURL[] = {
1477 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1478 : {p_isURLPath, 0, A_BINGO | A_CLRALL, TPS_Base, URL_T, SpecialFURL},
1479 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1480 : };
1481 :
1482 : static const TParserStateActionItem actionTPS_InProtocolFirst[] = {
1483 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1484 : {p_iseqC, '/', A_NEXT, TPS_InProtocolSecond, 0, NULL},
1485 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1486 : };
1487 :
1488 : static const TParserStateActionItem actionTPS_InProtocolSecond[] = {
1489 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1490 : {p_iseqC, '/', A_NEXT, TPS_InProtocolEnd, 0, NULL},
1491 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1492 : };
1493 :
1494 : static const TParserStateActionItem actionTPS_InProtocolEnd[] = {
1495 : {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, PROTOCOL, NULL}
1496 : };
1497 :
1498 : static const TParserStateActionItem actionTPS_InHyphenAsciiWordFirst[] = {
1499 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1500 : {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWord, 0, NULL},
1501 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1502 : {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL},
1503 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1504 : };
1505 :
1506 : static const TParserStateActionItem actionTPS_InHyphenAsciiWord[] = {
1507 : {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, ASCIIHWORD, SpecialHyphen},
1508 : {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWord, 0, NULL},
1509 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1510 : {p_isspecial, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1511 : {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1512 : {p_iseqC, '-', A_PUSH, TPS_InHyphenAsciiWordFirst, 0, NULL},
1513 : {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, ASCIIHWORD, SpecialHyphen}
1514 : };
1515 :
1516 : static const TParserStateActionItem actionTPS_InHyphenWordFirst[] = {
1517 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1518 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1519 : {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL},
1520 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1521 : };
1522 :
1523 : static const TParserStateActionItem actionTPS_InHyphenWord[] = {
1524 : {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, HWORD, SpecialHyphen},
1525 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1526 : {p_isspecial, 0, A_NEXT, TPS_InHyphenWord, 0, NULL},
1527 : {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1528 : {p_iseqC, '-', A_PUSH, TPS_InHyphenWordFirst, 0, NULL},
1529 : {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, HWORD, SpecialHyphen}
1530 : };
1531 :
1532 : static const TParserStateActionItem actionTPS_InHyphenNumWordFirst[] = {
1533 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1534 : {p_isalpha, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1535 : {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL},
1536 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1537 : };
1538 :
1539 : static const TParserStateActionItem actionTPS_InHyphenNumWord[] = {
1540 : {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, NUMHWORD, SpecialHyphen},
1541 : {p_isalnum, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1542 : {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1543 : {p_iseqC, '-', A_PUSH, TPS_InHyphenNumWordFirst, 0, NULL},
1544 : {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, NUMHWORD, SpecialHyphen}
1545 : };
1546 :
1547 : static const TParserStateActionItem actionTPS_InHyphenDigitLookahead[] = {
1548 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1549 : {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL},
1550 : {p_isalpha, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1551 : {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL},
1552 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1553 : };
1554 :
1555 : static const TParserStateActionItem actionTPS_InParseHyphen[] = {
1556 : {p_isEOF, 0, A_RERUN, TPS_Base, 0, NULL},
1557 : {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWordPart, 0, NULL},
1558 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL},
1559 : {p_isdigit, 0, A_PUSH, TPS_InHyphenUnsignedInt, 0, NULL},
1560 : {p_iseqC, '-', A_PUSH, TPS_InParseHyphenHyphen, 0, NULL},
1561 : {NULL, 0, A_RERUN, TPS_Base, 0, NULL}
1562 : };
1563 :
1564 : static const TParserStateActionItem actionTPS_InParseHyphenHyphen[] = {
1565 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1566 : {p_isalnum, 0, A_BINGO | A_CLEAR, TPS_InParseHyphen, SPACE, NULL},
1567 : {p_isspecial, 0, A_BINGO | A_CLEAR, TPS_InParseHyphen, SPACE, NULL},
1568 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1569 : };
1570 :
1571 : static const TParserStateActionItem actionTPS_InHyphenWordPart[] = {
1572 : {p_isEOF, 0, A_BINGO, TPS_Base, PARTHWORD, NULL},
1573 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL},
1574 : {p_isspecial, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL},
1575 : {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL},
1576 : {NULL, 0, A_BINGO, TPS_InParseHyphen, PARTHWORD, NULL}
1577 : };
1578 :
1579 : static const TParserStateActionItem actionTPS_InHyphenAsciiWordPart[] = {
1580 : {p_isEOF, 0, A_BINGO, TPS_Base, ASCIIPARTHWORD, NULL},
1581 : {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWordPart, 0, NULL},
1582 : {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL},
1583 : {p_isspecial, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL},
1584 : {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL},
1585 : {NULL, 0, A_BINGO, TPS_InParseHyphen, ASCIIPARTHWORD, NULL}
1586 : };
1587 :
1588 : static const TParserStateActionItem actionTPS_InHyphenNumWordPart[] = {
1589 : {p_isEOF, 0, A_BINGO, TPS_Base, NUMPARTHWORD, NULL},
1590 : {p_isalnum, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL},
1591 : {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL},
1592 : {NULL, 0, A_BINGO, TPS_InParseHyphen, NUMPARTHWORD, NULL}
1593 : };
1594 :
1595 : static const TParserStateActionItem actionTPS_InHyphenUnsignedInt[] = {
1596 : {p_isEOF, 0, A_POP, TPS_Null, 0, NULL},
1597 : {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL},
1598 : {p_isalpha, 0, A_CLEAR, TPS_InHyphenNumWordPart, 0, NULL},
1599 : {p_isspecial, 0, A_CLEAR, TPS_InHyphenNumWordPart, 0, NULL},
1600 : {NULL, 0, A_POP, TPS_Null, 0, NULL}
1601 : };
1602 :
1603 :
1604 : /*
1605 : * main table of per-state parser actions
1606 : */
1607 : typedef struct
1608 : {
1609 : const TParserStateActionItem *action; /* the actual state info */
1610 : TParserState state; /* only for Assert crosscheck */
1611 : #ifdef WPARSER_TRACE
1612 : const char *state_name; /* only for debug printout */
1613 : #endif
1614 : } TParserStateAction;
1615 :
1616 : #ifdef WPARSER_TRACE
1617 : #define TPARSERSTATEACTION(state) \
1618 : { CppConcat(action,state), state, CppAsString(state) }
1619 : #else
1620 : #define TPARSERSTATEACTION(state) \
1621 : { CppConcat(action,state), state }
1622 : #endif
1623 :
1624 : /*
1625 : * order must be the same as in typedef enum {} TParserState!!
1626 : */
1627 :
1628 : static const TParserStateAction Actions[] = {
1629 : TPARSERSTATEACTION(TPS_Base),
1630 : TPARSERSTATEACTION(TPS_InNumWord),
1631 : TPARSERSTATEACTION(TPS_InAsciiWord),
1632 : TPARSERSTATEACTION(TPS_InWord),
1633 : TPARSERSTATEACTION(TPS_InUnsignedInt),
1634 : TPARSERSTATEACTION(TPS_InSignedIntFirst),
1635 : TPARSERSTATEACTION(TPS_InSignedInt),
1636 : TPARSERSTATEACTION(TPS_InSpace),
1637 : TPARSERSTATEACTION(TPS_InUDecimalFirst),
1638 : TPARSERSTATEACTION(TPS_InUDecimal),
1639 : TPARSERSTATEACTION(TPS_InDecimalFirst),
1640 : TPARSERSTATEACTION(TPS_InDecimal),
1641 : TPARSERSTATEACTION(TPS_InVerVersion),
1642 : TPARSERSTATEACTION(TPS_InSVerVersion),
1643 : TPARSERSTATEACTION(TPS_InVersionFirst),
1644 : TPARSERSTATEACTION(TPS_InVersion),
1645 : TPARSERSTATEACTION(TPS_InMantissaFirst),
1646 : TPARSERSTATEACTION(TPS_InMantissaSign),
1647 : TPARSERSTATEACTION(TPS_InMantissa),
1648 : TPARSERSTATEACTION(TPS_InXMLEntityFirst),
1649 : TPARSERSTATEACTION(TPS_InXMLEntity),
1650 : TPARSERSTATEACTION(TPS_InXMLEntityNumFirst),
1651 : TPARSERSTATEACTION(TPS_InXMLEntityNum),
1652 : TPARSERSTATEACTION(TPS_InXMLEntityHexNumFirst),
1653 : TPARSERSTATEACTION(TPS_InXMLEntityHexNum),
1654 : TPARSERSTATEACTION(TPS_InXMLEntityEnd),
1655 : TPARSERSTATEACTION(TPS_InTagFirst),
1656 : TPARSERSTATEACTION(TPS_InXMLBegin),
1657 : TPARSERSTATEACTION(TPS_InTagCloseFirst),
1658 : TPARSERSTATEACTION(TPS_InTagName),
1659 : TPARSERSTATEACTION(TPS_InTagBeginEnd),
1660 : TPARSERSTATEACTION(TPS_InTag),
1661 : TPARSERSTATEACTION(TPS_InTagEscapeK),
1662 : TPARSERSTATEACTION(TPS_InTagEscapeKK),
1663 : TPARSERSTATEACTION(TPS_InTagBackSleshed),
1664 : TPARSERSTATEACTION(TPS_InTagEnd),
1665 : TPARSERSTATEACTION(TPS_InCommentFirst),
1666 : TPARSERSTATEACTION(TPS_InCommentLast),
1667 : TPARSERSTATEACTION(TPS_InComment),
1668 : TPARSERSTATEACTION(TPS_InCloseCommentFirst),
1669 : TPARSERSTATEACTION(TPS_InCloseCommentLast),
1670 : TPARSERSTATEACTION(TPS_InCommentEnd),
1671 : TPARSERSTATEACTION(TPS_InHostFirstDomain),
1672 : TPARSERSTATEACTION(TPS_InHostDomainSecond),
1673 : TPARSERSTATEACTION(TPS_InHostDomain),
1674 : TPARSERSTATEACTION(TPS_InPortFirst),
1675 : TPARSERSTATEACTION(TPS_InPort),
1676 : TPARSERSTATEACTION(TPS_InHostFirstAN),
1677 : TPARSERSTATEACTION(TPS_InHost),
1678 : TPARSERSTATEACTION(TPS_InEmail),
1679 : TPARSERSTATEACTION(TPS_InFileFirst),
1680 : TPARSERSTATEACTION(TPS_InFileTwiddle),
1681 : TPARSERSTATEACTION(TPS_InPathFirst),
1682 : TPARSERSTATEACTION(TPS_InPathFirstFirst),
1683 : TPARSERSTATEACTION(TPS_InPathSecond),
1684 : TPARSERSTATEACTION(TPS_InFile),
1685 : TPARSERSTATEACTION(TPS_InFileNext),
1686 : TPARSERSTATEACTION(TPS_InURLPathFirst),
1687 : TPARSERSTATEACTION(TPS_InURLPathStart),
1688 : TPARSERSTATEACTION(TPS_InURLPath),
1689 : TPARSERSTATEACTION(TPS_InFURL),
1690 : TPARSERSTATEACTION(TPS_InProtocolFirst),
1691 : TPARSERSTATEACTION(TPS_InProtocolSecond),
1692 : TPARSERSTATEACTION(TPS_InProtocolEnd),
1693 : TPARSERSTATEACTION(TPS_InHyphenAsciiWordFirst),
1694 : TPARSERSTATEACTION(TPS_InHyphenAsciiWord),
1695 : TPARSERSTATEACTION(TPS_InHyphenWordFirst),
1696 : TPARSERSTATEACTION(TPS_InHyphenWord),
1697 : TPARSERSTATEACTION(TPS_InHyphenNumWordFirst),
1698 : TPARSERSTATEACTION(TPS_InHyphenNumWord),
1699 : TPARSERSTATEACTION(TPS_InHyphenDigitLookahead),
1700 : TPARSERSTATEACTION(TPS_InParseHyphen),
1701 : TPARSERSTATEACTION(TPS_InParseHyphenHyphen),
1702 : TPARSERSTATEACTION(TPS_InHyphenWordPart),
1703 : TPARSERSTATEACTION(TPS_InHyphenAsciiWordPart),
1704 : TPARSERSTATEACTION(TPS_InHyphenNumWordPart),
1705 : TPARSERSTATEACTION(TPS_InHyphenUnsignedInt)
1706 : };
1707 :
1708 :
1709 : static bool
1710 28924 : TParserGet(TParser *prs)
1711 : {
1712 28924 : const TParserStateActionItem *item = NULL;
1713 :
1714 28924 : CHECK_FOR_INTERRUPTS();
1715 :
1716 : Assert(prs->state);
1717 :
1718 28924 : if (prs->state->posbyte >= prs->lenstr)
1719 4754 : return false;
1720 :
1721 24170 : prs->token = prs->str + prs->state->posbyte;
1722 24170 : prs->state->pushedAtAction = NULL;
1723 :
1724 : /* look at string */
1725 103278 : while (prs->state->posbyte <= prs->lenstr)
1726 : {
1727 103278 : if (prs->state->posbyte == prs->lenstr)
1728 4904 : prs->state->charlen = 0;
1729 : else
1730 196748 : prs->state->charlen = (prs->charmaxlen == 1) ? prs->charmaxlen :
1731 98374 : pg_mblen(prs->str + prs->state->posbyte);
1732 :
1733 : Assert(prs->state->posbyte + prs->state->charlen <= prs->lenstr);
1734 : Assert(prs->state->state >= TPS_Base && prs->state->state < TPS_Null);
1735 : Assert(Actions[prs->state->state].state == prs->state->state);
1736 :
1737 103278 : if (prs->state->pushedAtAction)
1738 : {
1739 : /* After a POP, pick up at the next test */
1740 2592 : item = prs->state->pushedAtAction + 1;
1741 2592 : prs->state->pushedAtAction = NULL;
1742 : }
1743 : else
1744 : {
1745 100686 : item = Actions[prs->state->state].action;
1746 : Assert(item != NULL);
1747 : }
1748 :
1749 : /* find action by character class */
1750 555708 : while (item->isclass)
1751 : {
1752 524364 : prs->c = item->c;
1753 524364 : if (item->isclass(prs) != 0)
1754 71934 : break;
1755 452430 : item++;
1756 : }
1757 :
1758 : #ifdef WPARSER_TRACE
1759 : {
1760 : TParserPosition *ptr;
1761 :
1762 : fprintf(stderr, "state ");
1763 : /* indent according to stack depth */
1764 : for (ptr = prs->state->prev; ptr; ptr = ptr->prev)
1765 : fprintf(stderr, " ");
1766 : fprintf(stderr, "%s ", Actions[prs->state->state].state_name);
1767 : if (prs->state->posbyte < prs->lenstr)
1768 : fprintf(stderr, "at %c", *(prs->str + prs->state->posbyte));
1769 : else
1770 : fprintf(stderr, "at EOF");
1771 : fprintf(stderr, " matched rule %d flags%s%s%s%s%s%s%s%s%s%s%s\n",
1772 : (int) (item - Actions[prs->state->state].action),
1773 : (item->flags & A_BINGO) ? " BINGO" : "",
1774 : (item->flags & A_POP) ? " POP" : "",
1775 : (item->flags & A_PUSH) ? " PUSH" : "",
1776 : (item->flags & A_RERUN) ? " RERUN" : "",
1777 : (item->flags & A_CLEAR) ? " CLEAR" : "",
1778 : (item->flags & A_MERGE) ? " MERGE" : "",
1779 : (item->flags & A_CLRALL) ? " CLRALL" : "",
1780 : (item->tostate != TPS_Null) ? " tostate " : "",
1781 : (item->tostate != TPS_Null) ? Actions[item->tostate].state_name : "",
1782 : (item->type > 0) ? " type " : "",
1783 : tok_alias[item->type]);
1784 : }
1785 : #endif
1786 :
1787 : /* call special handler if exists */
1788 103278 : if (item->special)
1789 420 : item->special(prs);
1790 :
1791 : /* BINGO, token is found */
1792 103278 : if (item->flags & A_BINGO)
1793 : {
1794 : Assert(item->type > 0);
1795 24170 : prs->lenbytetoken = prs->state->lenbytetoken;
1796 24170 : prs->lenchartoken = prs->state->lenchartoken;
1797 24170 : prs->state->lenbytetoken = prs->state->lenchartoken = 0;
1798 24170 : prs->type = item->type;
1799 : }
1800 :
1801 : /* do various actions by flags */
1802 103278 : if (item->flags & A_POP)
1803 : { /* pop stored state in stack */
1804 2610 : TParserPosition *ptr = prs->state->prev;
1805 :
1806 2610 : pfree(prs->state);
1807 2610 : prs->state = ptr;
1808 : Assert(prs->state);
1809 : }
1810 100668 : else if (item->flags & A_PUSH)
1811 : { /* push (store) state in stack */
1812 5088 : prs->state->pushedAtAction = item; /* remember where we push */
1813 5088 : prs->state = newTParserPosition(prs->state);
1814 : }
1815 95580 : else if (item->flags & A_CLEAR)
1816 : { /* clear previous pushed state */
1817 : TParserPosition *ptr;
1818 :
1819 : Assert(prs->state->prev);
1820 498 : ptr = prs->state->prev->prev;
1821 498 : pfree(prs->state->prev);
1822 498 : prs->state->prev = ptr;
1823 : }
1824 95082 : else if (item->flags & A_CLRALL)
1825 : { /* clear all previous pushed state */
1826 : TParserPosition *ptr;
1827 :
1828 2778 : while (prs->state->prev)
1829 : {
1830 1998 : ptr = prs->state->prev->prev;
1831 1998 : pfree(prs->state->prev);
1832 1998 : prs->state->prev = ptr;
1833 : }
1834 : }
1835 94302 : else if (item->flags & A_MERGE)
1836 : { /* merge posinfo with current and pushed state */
1837 0 : TParserPosition *ptr = prs->state;
1838 :
1839 : Assert(prs->state->prev);
1840 0 : prs->state = prs->state->prev;
1841 :
1842 0 : prs->state->posbyte = ptr->posbyte;
1843 0 : prs->state->poschar = ptr->poschar;
1844 0 : prs->state->charlen = ptr->charlen;
1845 0 : prs->state->lenbytetoken = ptr->lenbytetoken;
1846 0 : prs->state->lenchartoken = ptr->lenchartoken;
1847 0 : pfree(ptr);
1848 : }
1849 :
1850 : /* set new state if pointed */
1851 103278 : if (item->tostate != TPS_Null)
1852 66202 : prs->state->state = item->tostate;
1853 :
1854 : /* check for go away */
1855 103278 : if ((item->flags & A_BINGO) ||
1856 79108 : (prs->state->posbyte >= prs->lenstr &&
1857 0 : (item->flags & A_RERUN) == 0))
1858 : break;
1859 :
1860 : /* go to beginning of loop if we should rerun or we just restore state */
1861 79108 : if (item->flags & (A_RERUN | A_POP))
1862 2634 : continue;
1863 :
1864 : /* move forward */
1865 76474 : if (prs->state->charlen)
1866 : {
1867 76474 : prs->state->posbyte += prs->state->charlen;
1868 76474 : prs->state->lenbytetoken += prs->state->charlen;
1869 76474 : prs->state->poschar++;
1870 76474 : prs->state->lenchartoken++;
1871 : }
1872 : }
1873 :
1874 24170 : return (item && (item->flags & A_BINGO));
1875 : }
1876 :
1877 : Datum
1878 7948 : prsd_lextype(PG_FUNCTION_ARGS)
1879 : {
1880 7948 : LexDescr *descr = (LexDescr *) palloc(sizeof(LexDescr) * (LASTNUM + 1));
1881 : int i;
1882 :
1883 190752 : for (i = 1; i <= LASTNUM; i++)
1884 : {
1885 182804 : descr[i - 1].lexid = i;
1886 182804 : descr[i - 1].alias = pstrdup(tok_alias[i]);
1887 182804 : descr[i - 1].descr = pstrdup(lex_descr[i]);
1888 : }
1889 :
1890 7948 : descr[LASTNUM].lexid = 0;
1891 :
1892 7948 : PG_RETURN_POINTER(descr);
1893 : }
1894 :
1895 : Datum
1896 4754 : prsd_start(PG_FUNCTION_ARGS)
1897 : {
1898 4754 : PG_RETURN_POINTER(TParserInit((char *) PG_GETARG_POINTER(0), PG_GETARG_INT32(1)));
1899 : }
1900 :
1901 : Datum
1902 28684 : prsd_nexttoken(PG_FUNCTION_ARGS)
1903 : {
1904 28684 : TParser *p = (TParser *) PG_GETARG_POINTER(0);
1905 28684 : char **t = (char **) PG_GETARG_POINTER(1);
1906 28684 : int *tlen = (int *) PG_GETARG_POINTER(2);
1907 :
1908 28684 : if (!TParserGet(p))
1909 4754 : PG_RETURN_INT32(0);
1910 :
1911 23930 : *t = p->token;
1912 23930 : *tlen = p->lenbytetoken;
1913 :
1914 23930 : PG_RETURN_INT32(p->type);
1915 : }
1916 :
1917 : Datum
1918 4754 : prsd_end(PG_FUNCTION_ARGS)
1919 : {
1920 4754 : TParser *p = (TParser *) PG_GETARG_POINTER(0);
1921 :
1922 4754 : TParserClose(p);
1923 4754 : PG_RETURN_VOID();
1924 : }
1925 :
1926 :
1927 : /*
1928 : * ts_headline support begins here
1929 : */
1930 :
1931 : /* token type classification macros */
1932 : #define TS_IDIGNORE(x) ( (x)==TAG_T || (x)==PROTOCOL || (x)==SPACE || (x)==XMLENTITY )
1933 : #define HLIDREPLACE(x) ( (x)==TAG_T )
1934 : #define HLIDSKIP(x) ( (x)==URL_T || (x)==NUMHWORD || (x)==ASCIIHWORD || (x)==HWORD )
1935 : #define XMLHLIDSKIP(x) ( (x)==URL_T || (x)==NUMHWORD || (x)==ASCIIHWORD || (x)==HWORD )
1936 : #define NONWORDTOKEN(x) ( (x)==SPACE || HLIDREPLACE(x) || HLIDSKIP(x) )
1937 : #define NOENDTOKEN(x) ( NONWORDTOKEN(x) || (x)==SCIENTIFIC || (x)==VERSIONNUMBER || (x)==DECIMAL_T || (x)==SIGNEDINT || (x)==UNSIGNEDINT || TS_IDIGNORE(x) )
1938 :
1939 : /*
1940 : * Macros useful in headline selection. These rely on availability of
1941 : * "HeadlineParsedText *prs" describing some text, and "int shortword"
1942 : * describing the "short word" length parameter.
1943 : */
1944 :
1945 : /* Interesting words are non-repeated search terms */
1946 : #define INTERESTINGWORD(j) \
1947 : (prs->words[j].item && !prs->words[j].repeated)
1948 :
1949 : /* Don't want to end at a non-word or a short word, unless interesting */
1950 : #define BADENDPOINT(j) \
1951 : ((NOENDTOKEN(prs->words[j].type) || prs->words[j].len <= shortword) && \
1952 : !INTERESTINGWORD(j))
1953 :
1954 : typedef struct
1955 : {
1956 : /* one cover (well, really one fragment) for mark_hl_fragments */
1957 : int32 startpos; /* fragment's starting word index */
1958 : int32 endpos; /* ending word index (inclusive) */
1959 : int32 poslen; /* number of interesting words */
1960 : int32 curlen; /* total number of words */
1961 : bool chosen; /* chosen? */
1962 : bool excluded; /* excluded? */
1963 : } CoverPos;
1964 :
1965 : typedef struct
1966 : {
1967 : /* callback data for checkcondition_HL */
1968 : HeadlineWordEntry *words;
1969 : int len;
1970 : } hlCheck;
1971 :
1972 :
1973 : /*
1974 : * TS_execute callback for matching a tsquery operand to headline words
1975 : *
1976 : * Note: it's tempting to report words[] indexes as pos values to save
1977 : * searching in hlCover; but that would screw up phrase matching, which
1978 : * expects to measure distances in lexemes not tokens.
1979 : */
1980 : static TSTernaryValue
1981 1000 : checkcondition_HL(void *opaque, QueryOperand *val, ExecPhraseData *data)
1982 : {
1983 1000 : hlCheck *checkval = (hlCheck *) opaque;
1984 : int i;
1985 :
1986 : /* scan words array for matching items */
1987 25450 : for (i = 0; i < checkval->len; i++)
1988 : {
1989 24650 : if (checkval->words[i].item == val)
1990 : {
1991 : /* if data == NULL, don't need to report positions */
1992 874 : if (!data)
1993 200 : return TS_YES;
1994 :
1995 674 : if (!data->pos)
1996 : {
1997 476 : data->pos = palloc(sizeof(WordEntryPos) * checkval->len);
1998 476 : data->allocated = true;
1999 476 : data->npos = 1;
2000 476 : data->pos[0] = checkval->words[i].pos;
2001 : }
2002 198 : else if (data->pos[data->npos - 1] < checkval->words[i].pos)
2003 : {
2004 198 : data->pos[data->npos++] = checkval->words[i].pos;
2005 : }
2006 : }
2007 : }
2008 :
2009 800 : if (data && data->npos > 0)
2010 476 : return TS_YES;
2011 :
2012 324 : return TS_NO;
2013 : }
2014 :
2015 : /*
2016 : * hlCover: try to find a substring of prs' word list that satisfies query
2017 : *
2018 : * locations is the result of TS_execute_locations() for the query.
2019 : * We use this to identify plausible subranges of the query.
2020 : *
2021 : * *nextpos is the lexeme position (NOT word index) to start the search
2022 : * at. Caller should initialize this to zero. If successful, we'll
2023 : * advance it to the next place to search at.
2024 : *
2025 : * On success, sets *p to first word index and *q to last word index of the
2026 : * cover substring, and returns true.
2027 : *
2028 : * The result is a minimal cover, in the sense that both *p and *q will be
2029 : * words used in the query.
2030 : */
2031 : static bool
2032 562 : hlCover(HeadlineParsedText *prs, TSQuery query, List *locations,
2033 : int *nextpos, int *p, int *q)
2034 : {
2035 562 : int pos = *nextpos;
2036 :
2037 : /* This loop repeats when our selected word-range fails the query */
2038 : for (;;)
2039 60 : {
2040 : int posb,
2041 : pose;
2042 : ListCell *lc;
2043 :
2044 : /*
2045 : * For each AND'ed query term or phrase, find its first occurrence at
2046 : * or after pos; set pose to the maximum of those positions.
2047 : *
2048 : * We need not consider ORs or NOTs here; see the comments for
2049 : * TS_execute_locations(). Rechecking the match with TS_execute(),
2050 : * below, will deal with any ensuing imprecision.
2051 : */
2052 622 : pose = -1;
2053 966 : foreach(lc, locations)
2054 : {
2055 466 : ExecPhraseData *pdata = (ExecPhraseData *) lfirst(lc);
2056 466 : int first = -1;
2057 :
2058 792 : for (int i = 0; i < pdata->npos; i++)
2059 : {
2060 : /* For phrase matches, use the ending lexeme */
2061 670 : int endp = pdata->pos[i];
2062 :
2063 670 : if (endp >= pos)
2064 : {
2065 344 : first = endp;
2066 344 : break;
2067 : }
2068 : }
2069 466 : if (first < 0)
2070 122 : return false; /* no more matches for this term */
2071 344 : if (first > pose)
2072 326 : pose = first;
2073 : }
2074 :
2075 500 : if (pose < 0)
2076 246 : return false; /* we only get here if empty list */
2077 :
2078 : /*
2079 : * Now, for each AND'ed query term or phrase, find its last occurrence
2080 : * at or before pose; set posb to the minimum of those positions.
2081 : *
2082 : * We start posb at INT_MAX - 1 to guarantee no overflow if we compute
2083 : * posb + 1 below.
2084 : */
2085 254 : posb = INT_MAX - 1;
2086 586 : foreach(lc, locations)
2087 : {
2088 332 : ExecPhraseData *pdata = (ExecPhraseData *) lfirst(lc);
2089 332 : int last = -1;
2090 :
2091 494 : for (int i = pdata->npos - 1; i >= 0; i--)
2092 : {
2093 : /* For phrase matches, use the starting lexeme */
2094 494 : int startp = pdata->pos[i] - pdata->width;
2095 :
2096 494 : if (startp <= pose)
2097 : {
2098 332 : last = startp;
2099 332 : break;
2100 : }
2101 : }
2102 332 : if (last < posb)
2103 272 : posb = last;
2104 : }
2105 :
2106 : /*
2107 : * We could end up with posb to the left of pos, in case some phrase
2108 : * match crosses pos. Try the match starting at pos anyway, since the
2109 : * result of TS_execute_locations is imprecise for phrase matches OR'd
2110 : * with plain matches; that is, if the query is "(A <-> B) | C" then C
2111 : * could match at pos even though the phrase match would have to
2112 : * extend to the left of pos.
2113 : */
2114 254 : posb = Max(posb, pos);
2115 :
2116 : /* This test probably always succeeds, but be paranoid */
2117 254 : if (posb <= pose)
2118 : {
2119 : /*
2120 : * posb .. pose is now the shortest, earliest-after-pos range of
2121 : * lexeme positions containing all the query terms. It will
2122 : * contain all phrase matches, too, except in the corner case
2123 : * described just above.
2124 : *
2125 : * Now convert these lexeme positions to indexes in prs->words[].
2126 : */
2127 254 : int idxb = -1;
2128 254 : int idxe = -1;
2129 :
2130 11624 : for (int i = 0; i < prs->curwords; i++)
2131 : {
2132 11496 : if (prs->words[i].item == NULL)
2133 10612 : continue;
2134 884 : if (idxb < 0 && prs->words[i].pos >= posb)
2135 254 : idxb = i;
2136 884 : if (prs->words[i].pos <= pose)
2137 758 : idxe = i;
2138 : else
2139 126 : break;
2140 : }
2141 :
2142 : /* This test probably always succeeds, but be paranoid */
2143 254 : if (idxb >= 0 && idxe >= idxb)
2144 : {
2145 : /*
2146 : * Finally, check that the selected range satisfies the query.
2147 : * This should succeed in all simple cases; but odd cases
2148 : * involving non-top-level NOT conditions or phrase matches
2149 : * OR'd with other things could fail, since the result of
2150 : * TS_execute_locations doesn't fully represent such things.
2151 : */
2152 : hlCheck ch;
2153 :
2154 254 : ch.words = &(prs->words[idxb]);
2155 254 : ch.len = idxe - idxb + 1;
2156 254 : if (TS_execute(GETQUERY(query), &ch,
2157 : TS_EXEC_EMPTY, checkcondition_HL))
2158 : {
2159 : /* Match! Advance *nextpos and return the word range. */
2160 194 : *nextpos = posb + 1;
2161 194 : *p = idxb;
2162 194 : *q = idxe;
2163 194 : return true;
2164 : }
2165 : }
2166 : }
2167 :
2168 : /*
2169 : * Advance pos and try again. Any later workable match must start
2170 : * beyond posb.
2171 : */
2172 60 : pos = posb + 1;
2173 : }
2174 : /* Can't get here, but stupider compilers complain if we leave it off */
2175 : return false;
2176 : }
2177 :
2178 : /*
2179 : * Apply suitable highlight marking to words selected by headline selector
2180 : *
2181 : * The words from startpos to endpos inclusive are marked per highlightall
2182 : */
2183 : static void
2184 386 : mark_fragment(HeadlineParsedText *prs, bool highlightall,
2185 : int startpos, int endpos)
2186 : {
2187 : int i;
2188 :
2189 5654 : for (i = startpos; i <= endpos; i++)
2190 : {
2191 5268 : if (prs->words[i].item)
2192 500 : prs->words[i].selected = 1;
2193 5268 : if (!highlightall)
2194 : {
2195 5022 : if (HLIDREPLACE(prs->words[i].type))
2196 0 : prs->words[i].replace = 1;
2197 5022 : else if (HLIDSKIP(prs->words[i].type))
2198 0 : prs->words[i].skip = 1;
2199 : }
2200 : else
2201 : {
2202 246 : if (XMLHLIDSKIP(prs->words[i].type))
2203 6 : prs->words[i].skip = 1;
2204 : }
2205 :
2206 5268 : prs->words[i].in = (prs->words[i].repeated) ? 0 : 1;
2207 : }
2208 386 : }
2209 :
2210 : /*
2211 : * split a cover substring into fragments not longer than max_words
2212 : *
2213 : * At entry, *startpos and *endpos are the (remaining) bounds of the cover
2214 : * substring. They are updated to hold the bounds of the next fragment.
2215 : *
2216 : * *curlen and *poslen are set to the fragment's length, in words and
2217 : * interesting words respectively.
2218 : */
2219 : static void
2220 36 : get_next_fragment(HeadlineParsedText *prs, int *startpos, int *endpos,
2221 : int *curlen, int *poslen, int max_words)
2222 : {
2223 : int i;
2224 :
2225 : /*
2226 : * Objective: select a fragment of words between startpos and endpos such
2227 : * that it has at most max_words and both ends have query words. If the
2228 : * startpos and endpos are the endpoints of the cover and the cover has
2229 : * fewer words than max_words, then this function should just return the
2230 : * cover
2231 : */
2232 : /* first move startpos to an item */
2233 888 : for (i = *startpos; i <= *endpos; i++)
2234 : {
2235 888 : *startpos = i;
2236 888 : if (INTERESTINGWORD(i))
2237 36 : break;
2238 : }
2239 : /* cut endpos to have only max_words */
2240 36 : *curlen = 0;
2241 36 : *poslen = 0;
2242 960 : for (i = *startpos; i <= *endpos && *curlen < max_words; i++)
2243 : {
2244 924 : if (!NONWORDTOKEN(prs->words[i].type))
2245 480 : *curlen += 1;
2246 924 : if (INTERESTINGWORD(i))
2247 54 : *poslen += 1;
2248 : }
2249 : /* if the cover was cut then move back endpos to a query item */
2250 36 : if (*endpos > i)
2251 : {
2252 12 : *endpos = i;
2253 840 : for (i = *endpos; i >= *startpos; i--)
2254 : {
2255 840 : *endpos = i;
2256 840 : if (INTERESTINGWORD(i))
2257 12 : break;
2258 828 : if (!NONWORDTOKEN(prs->words[i].type))
2259 408 : *curlen -= 1;
2260 : }
2261 : }
2262 36 : }
2263 :
2264 : /*
2265 : * Headline selector used when MaxFragments > 0
2266 : *
2267 : * Note: in this mode, highlightall is disregarded for phrase selection;
2268 : * it only controls presentation details.
2269 : */
2270 : static void
2271 30 : mark_hl_fragments(HeadlineParsedText *prs, TSQuery query, List *locations,
2272 : bool highlightall,
2273 : int shortword, int min_words,
2274 : int max_words, int max_fragments)
2275 : {
2276 : int32 poslen,
2277 : curlen,
2278 : i,
2279 : f,
2280 30 : num_f = 0;
2281 : int32 stretch,
2282 : maxstretch,
2283 : posmarker;
2284 :
2285 30 : int32 startpos = 0,
2286 30 : endpos = 0,
2287 30 : nextpos = 0,
2288 30 : p = 0,
2289 30 : q = 0;
2290 :
2291 30 : int32 numcovers = 0,
2292 30 : maxcovers = 32;
2293 :
2294 : int32 minI,
2295 : minwords,
2296 : maxitems;
2297 : CoverPos *covers;
2298 :
2299 30 : covers = palloc(maxcovers * sizeof(CoverPos));
2300 :
2301 : /* get all covers */
2302 54 : while (hlCover(prs, query, locations, &nextpos, &p, &q))
2303 : {
2304 24 : startpos = p;
2305 24 : endpos = q;
2306 :
2307 : /*
2308 : * Break the cover into smaller fragments such that each fragment has
2309 : * at most max_words. Also ensure that each end of each fragment is a
2310 : * query word. This will allow us to stretch the fragment in either
2311 : * direction
2312 : */
2313 :
2314 60 : while (startpos <= endpos)
2315 : {
2316 36 : get_next_fragment(prs, &startpos, &endpos, &curlen, &poslen, max_words);
2317 36 : if (numcovers >= maxcovers)
2318 : {
2319 0 : maxcovers *= 2;
2320 0 : covers = repalloc(covers, sizeof(CoverPos) * maxcovers);
2321 : }
2322 36 : covers[numcovers].startpos = startpos;
2323 36 : covers[numcovers].endpos = endpos;
2324 36 : covers[numcovers].curlen = curlen;
2325 36 : covers[numcovers].poslen = poslen;
2326 36 : covers[numcovers].chosen = false;
2327 36 : covers[numcovers].excluded = false;
2328 36 : numcovers++;
2329 36 : startpos = endpos + 1;
2330 36 : endpos = q;
2331 : }
2332 : }
2333 :
2334 : /* choose best covers */
2335 66 : for (f = 0; f < max_fragments; f++)
2336 : {
2337 48 : maxitems = 0;
2338 48 : minwords = PG_INT32_MAX;
2339 48 : minI = -1;
2340 :
2341 : /*
2342 : * Choose the cover that contains max items. In case of tie choose the
2343 : * one with smaller number of words.
2344 : */
2345 114 : for (i = 0; i < numcovers; i++)
2346 : {
2347 66 : if (!covers[i].chosen && !covers[i].excluded &&
2348 48 : (maxitems < covers[i].poslen ||
2349 12 : (maxitems == covers[i].poslen &&
2350 12 : minwords > covers[i].curlen)))
2351 : {
2352 36 : maxitems = covers[i].poslen;
2353 36 : minwords = covers[i].curlen;
2354 36 : minI = i;
2355 : }
2356 : }
2357 : /* if a cover was found mark it */
2358 48 : if (minI >= 0)
2359 : {
2360 36 : covers[minI].chosen = true;
2361 : /* adjust the size of cover */
2362 36 : startpos = covers[minI].startpos;
2363 36 : endpos = covers[minI].endpos;
2364 36 : curlen = covers[minI].curlen;
2365 : /* stretch the cover if cover size is lower than max_words */
2366 36 : if (curlen < max_words)
2367 : {
2368 : /* divide the stretch on both sides of cover */
2369 36 : maxstretch = (max_words - curlen) / 2;
2370 :
2371 : /*
2372 : * first stretch the startpos stop stretching if 1. we hit the
2373 : * beginning of document 2. exceed maxstretch 3. we hit an
2374 : * already marked fragment
2375 : */
2376 36 : stretch = 0;
2377 36 : posmarker = startpos;
2378 600 : for (i = startpos - 1; i >= 0 && stretch < maxstretch && !prs->words[i].in; i--)
2379 : {
2380 564 : if (!NONWORDTOKEN(prs->words[i].type))
2381 : {
2382 270 : curlen++;
2383 270 : stretch++;
2384 : }
2385 564 : posmarker = i;
2386 : }
2387 : /* cut back startpos till we find a good endpoint */
2388 132 : for (i = posmarker; i < startpos && BADENDPOINT(i); i++)
2389 : {
2390 96 : if (!NONWORDTOKEN(prs->words[i].type))
2391 36 : curlen--;
2392 : }
2393 36 : startpos = i;
2394 : /* now stretch the endpos as much as possible */
2395 36 : posmarker = endpos;
2396 966 : for (i = endpos + 1; i < prs->curwords && curlen < max_words && !prs->words[i].in; i++)
2397 : {
2398 930 : if (!NONWORDTOKEN(prs->words[i].type))
2399 462 : curlen++;
2400 930 : posmarker = i;
2401 : }
2402 : /* cut back endpos till we find a good endpoint */
2403 90 : for (i = posmarker; i > endpos && BADENDPOINT(i); i--)
2404 : {
2405 54 : if (!NONWORDTOKEN(prs->words[i].type))
2406 24 : curlen--;
2407 : }
2408 36 : endpos = i;
2409 : }
2410 36 : covers[minI].startpos = startpos;
2411 36 : covers[minI].endpos = endpos;
2412 36 : covers[minI].curlen = curlen;
2413 : /* Mark the chosen fragments (covers) */
2414 36 : mark_fragment(prs, highlightall, startpos, endpos);
2415 36 : num_f++;
2416 : /* Exclude covers overlapping this one from future consideration */
2417 96 : for (i = 0; i < numcovers; i++)
2418 : {
2419 60 : if (i != minI &&
2420 24 : ((covers[i].startpos >= startpos &&
2421 12 : covers[i].startpos <= endpos) ||
2422 24 : (covers[i].endpos >= startpos &&
2423 12 : covers[i].endpos <= endpos) ||
2424 24 : (covers[i].startpos < startpos &&
2425 12 : covers[i].endpos > endpos)))
2426 0 : covers[i].excluded = true;
2427 : }
2428 : }
2429 : else
2430 12 : break; /* no selectable covers remain */
2431 : }
2432 :
2433 : /* show the first min_words words if we have not marked anything */
2434 30 : if (num_f <= 0)
2435 : {
2436 6 : startpos = curlen = 0;
2437 6 : endpos = -1;
2438 186 : for (i = 0; i < prs->curwords && curlen < min_words; i++)
2439 : {
2440 180 : if (!NONWORDTOKEN(prs->words[i].type))
2441 90 : curlen++;
2442 180 : endpos = i;
2443 : }
2444 6 : mark_fragment(prs, highlightall, startpos, endpos);
2445 : }
2446 :
2447 30 : pfree(covers);
2448 30 : }
2449 :
2450 : /*
2451 : * Headline selector used when MaxFragments == 0
2452 : */
2453 : static void
2454 344 : mark_hl_words(HeadlineParsedText *prs, TSQuery query, List *locations,
2455 : bool highlightall,
2456 : int shortword, int min_words, int max_words)
2457 : {
2458 344 : int nextpos = 0,
2459 344 : p = 0,
2460 344 : q = 0;
2461 344 : int bestb = -1,
2462 344 : beste = -1;
2463 344 : int bestlen = -1;
2464 344 : bool bestcover = false;
2465 : int pose,
2466 : posb,
2467 : poslen,
2468 : curlen;
2469 : bool poscover;
2470 : int i;
2471 :
2472 344 : if (!highlightall)
2473 : {
2474 : /* examine all covers, select a headline using the best one */
2475 508 : while (hlCover(prs, query, locations, &nextpos, &p, &q))
2476 : {
2477 : /*
2478 : * Count words (curlen) and interesting words (poslen) within
2479 : * cover, but stop once we reach max_words. This step doesn't
2480 : * consider whether that's a good stopping point. posb and pose
2481 : * are set to the start and end indexes of the possible headline.
2482 : */
2483 170 : curlen = 0;
2484 170 : poslen = 0;
2485 170 : posb = pose = p;
2486 1456 : for (i = p; i <= q && curlen < max_words; i++)
2487 : {
2488 1286 : if (!NONWORDTOKEN(prs->words[i].type))
2489 728 : curlen++;
2490 1286 : if (INTERESTINGWORD(i))
2491 290 : poslen++;
2492 1286 : pose = i;
2493 : }
2494 :
2495 170 : if (curlen < max_words)
2496 : {
2497 : /*
2498 : * We have room to lengthen the headline, so search forward
2499 : * until it's full or we find a good stopping point. We'll
2500 : * reconsider the word at "q", then move forward.
2501 : */
2502 2938 : for (i = i - 1; i < prs->curwords && curlen < max_words; i++)
2503 : {
2504 2912 : if (i > q)
2505 : {
2506 2754 : if (!NONWORDTOKEN(prs->words[i].type))
2507 1374 : curlen++;
2508 2754 : if (INTERESTINGWORD(i))
2509 120 : poslen++;
2510 : }
2511 2912 : pose = i;
2512 2912 : if (BADENDPOINT(i))
2513 1944 : continue;
2514 968 : if (curlen >= min_words)
2515 132 : break;
2516 : }
2517 158 : if (curlen < min_words)
2518 : {
2519 : /*
2520 : * Reached end of text and our headline is still shorter
2521 : * than min_words, so try to extend it to the left.
2522 : */
2523 366 : for (i = p - 1; i >= 0; i--)
2524 : {
2525 364 : if (!NONWORDTOKEN(prs->words[i].type))
2526 182 : curlen++;
2527 364 : if (INTERESTINGWORD(i))
2528 6 : poslen++;
2529 364 : if (curlen >= max_words)
2530 0 : break;
2531 364 : if (BADENDPOINT(i))
2532 236 : continue;
2533 128 : if (curlen >= min_words)
2534 24 : break;
2535 : }
2536 26 : posb = (i >= 0) ? i : 0;
2537 : }
2538 : }
2539 : else
2540 : {
2541 : /*
2542 : * Can't make headline longer, so consider making it shorter
2543 : * if needed to avoid a bad endpoint.
2544 : */
2545 12 : if (i > q)
2546 6 : i = q;
2547 30 : for (; curlen > min_words; i--)
2548 : {
2549 30 : if (!BADENDPOINT(i))
2550 : break;
2551 18 : if (!NONWORDTOKEN(prs->words[i].type))
2552 6 : curlen--;
2553 18 : if (INTERESTINGWORD(i))
2554 0 : poslen--;
2555 18 : pose = i - 1;
2556 : }
2557 : }
2558 :
2559 : /*
2560 : * Check whether the proposed headline includes the original
2561 : * cover; it might not if we trimmed it due to max_words.
2562 : */
2563 170 : poscover = (posb <= p && pose >= q);
2564 :
2565 : /*
2566 : * Adopt this headline if it's better than the last one, giving
2567 : * highest priority to headlines including the cover, then to
2568 : * headlines with more interesting words, then to headlines with
2569 : * good stopping points. (Since bestlen is initially -1, we will
2570 : * certainly adopt the first headline.)
2571 : */
2572 170 : if (poscover > bestcover ||
2573 78 : (poscover == bestcover && poslen > bestlen) ||
2574 72 : (poscover == bestcover && poslen == bestlen &&
2575 12 : !BADENDPOINT(pose) && BADENDPOINT(beste)))
2576 : {
2577 98 : bestb = posb;
2578 98 : beste = pose;
2579 98 : bestlen = poslen;
2580 98 : bestcover = poscover;
2581 : }
2582 : }
2583 :
2584 : /*
2585 : * If we found nothing acceptable, select min_words words starting at
2586 : * the beginning.
2587 : */
2588 338 : if (bestlen < 0)
2589 : {
2590 240 : curlen = 0;
2591 240 : pose = -1;
2592 1038 : for (i = 0; i < prs->curwords && curlen < min_words; i++)
2593 : {
2594 798 : if (!NONWORDTOKEN(prs->words[i].type))
2595 516 : curlen++;
2596 798 : pose = i;
2597 : }
2598 240 : bestb = 0;
2599 240 : beste = pose;
2600 : }
2601 : }
2602 : else
2603 : {
2604 : /* highlightall mode: headline is whole document */
2605 6 : bestb = 0;
2606 6 : beste = prs->curwords - 1;
2607 : }
2608 :
2609 344 : mark_fragment(prs, highlightall, bestb, beste);
2610 344 : }
2611 :
2612 : /*
2613 : * Default parser's prsheadline function
2614 : */
2615 : Datum
2616 374 : prsd_headline(PG_FUNCTION_ARGS)
2617 : {
2618 374 : HeadlineParsedText *prs = (HeadlineParsedText *) PG_GETARG_POINTER(0);
2619 374 : List *prsoptions = (List *) PG_GETARG_POINTER(1);
2620 374 : TSQuery query = PG_GETARG_TSQUERY(2);
2621 : List *locations;
2622 :
2623 : /* default option values: */
2624 374 : int min_words = 15;
2625 374 : int max_words = 35;
2626 374 : int shortword = 3;
2627 374 : int max_fragments = 0;
2628 374 : bool highlightall = false;
2629 : ListCell *l;
2630 :
2631 : /* Extract configuration option values */
2632 374 : prs->startsel = NULL;
2633 374 : prs->stopsel = NULL;
2634 374 : prs->fragdelim = NULL;
2635 728 : foreach(l, prsoptions)
2636 : {
2637 354 : DefElem *defel = (DefElem *) lfirst(l);
2638 354 : char *val = defGetString(defel);
2639 :
2640 354 : if (pg_strcasecmp(defel->defname, "MaxWords") == 0)
2641 36 : max_words = pg_strtoint32(val);
2642 318 : else if (pg_strcasecmp(defel->defname, "MinWords") == 0)
2643 36 : min_words = pg_strtoint32(val);
2644 282 : else if (pg_strcasecmp(defel->defname, "ShortWord") == 0)
2645 0 : shortword = pg_strtoint32(val);
2646 282 : else if (pg_strcasecmp(defel->defname, "MaxFragments") == 0)
2647 30 : max_fragments = pg_strtoint32(val);
2648 252 : else if (pg_strcasecmp(defel->defname, "StartSel") == 0)
2649 120 : prs->startsel = pstrdup(val);
2650 132 : else if (pg_strcasecmp(defel->defname, "StopSel") == 0)
2651 120 : prs->stopsel = pstrdup(val);
2652 12 : else if (pg_strcasecmp(defel->defname, "FragmentDelimiter") == 0)
2653 6 : prs->fragdelim = pstrdup(val);
2654 6 : else if (pg_strcasecmp(defel->defname, "HighlightAll") == 0)
2655 18 : highlightall = (pg_strcasecmp(val, "1") == 0 ||
2656 12 : pg_strcasecmp(val, "on") == 0 ||
2657 6 : pg_strcasecmp(val, "true") == 0 ||
2658 0 : pg_strcasecmp(val, "t") == 0 ||
2659 12 : pg_strcasecmp(val, "y") == 0 ||
2660 0 : pg_strcasecmp(val, "yes") == 0);
2661 : else
2662 0 : ereport(ERROR,
2663 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
2664 : errmsg("unrecognized headline parameter: \"%s\"",
2665 : defel->defname)));
2666 : }
2667 :
2668 : /* in HighlightAll mode these parameters are ignored */
2669 374 : if (!highlightall)
2670 : {
2671 368 : if (min_words >= max_words)
2672 0 : ereport(ERROR,
2673 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
2674 : errmsg("MinWords should be less than MaxWords")));
2675 368 : if (min_words <= 0)
2676 0 : ereport(ERROR,
2677 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
2678 : errmsg("MinWords should be positive")));
2679 368 : if (shortword < 0)
2680 0 : ereport(ERROR,
2681 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
2682 : errmsg("ShortWord should be >= 0")));
2683 368 : if (max_fragments < 0)
2684 0 : ereport(ERROR,
2685 : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
2686 : errmsg("MaxFragments should be >= 0")));
2687 : }
2688 :
2689 : /* Locate words and phrases matching the query */
2690 374 : if (query->size > 0)
2691 : {
2692 : hlCheck ch;
2693 :
2694 362 : ch.words = prs->words;
2695 362 : ch.len = prs->curwords;
2696 362 : locations = TS_execute_locations(GETQUERY(query), &ch, TS_EXEC_EMPTY,
2697 : checkcondition_HL);
2698 : }
2699 : else
2700 12 : locations = NIL; /* empty query matches nothing */
2701 :
2702 : /* Apply appropriate headline selector */
2703 374 : if (max_fragments == 0)
2704 344 : mark_hl_words(prs, query, locations, highlightall, shortword,
2705 : min_words, max_words);
2706 : else
2707 30 : mark_hl_fragments(prs, query, locations, highlightall, shortword,
2708 : min_words, max_words, max_fragments);
2709 :
2710 : /* Fill in default values for string options */
2711 374 : if (!prs->startsel)
2712 254 : prs->startsel = pstrdup("<b>");
2713 374 : if (!prs->stopsel)
2714 254 : prs->stopsel = pstrdup("</b>");
2715 374 : if (!prs->fragdelim)
2716 368 : prs->fragdelim = pstrdup(" ... ");
2717 :
2718 : /* Caller will need these lengths, too */
2719 374 : prs->startsellen = strlen(prs->startsel);
2720 374 : prs->stopsellen = strlen(prs->stopsel);
2721 374 : prs->fragdelimlen = strlen(prs->fragdelim);
2722 :
2723 374 : PG_RETURN_POINTER(prs);
2724 : }
|