Header And Logo
|
00001 /*------------------------------------------------------------------------- 00002 * 00003 * wparser_def.c 00004 * Default text search parser 00005 * 00006 * Portions Copyright (c) 1996-2013, PostgreSQL Global Development Group 00007 * 00008 * 00009 * IDENTIFICATION 00010 * src/backend/tsearch/wparser_def.c 00011 * 00012 *------------------------------------------------------------------------- 00013 */ 00014 00015 #include "postgres.h" 00016 00017 #include "catalog/pg_collation.h" 00018 #include "commands/defrem.h" 00019 #include "tsearch/ts_locale.h" 00020 #include "tsearch/ts_public.h" 00021 #include "tsearch/ts_type.h" 00022 #include "tsearch/ts_utils.h" 00023 #include "utils/builtins.h" 00024 00025 00026 /* Define me to enable tracing of parser behavior */ 00027 /* #define WPARSER_TRACE */ 00028 00029 00030 /* Output token categories */ 00031 00032 #define ASCIIWORD 1 00033 #define WORD_T 2 00034 #define NUMWORD 3 00035 #define EMAIL 4 00036 #define URL_T 5 00037 #define HOST 6 00038 #define SCIENTIFIC 7 00039 #define VERSIONNUMBER 8 00040 #define NUMPARTHWORD 9 00041 #define PARTHWORD 10 00042 #define ASCIIPARTHWORD 11 00043 #define SPACE 12 00044 #define TAG_T 13 00045 #define PROTOCOL 14 00046 #define NUMHWORD 15 00047 #define ASCIIHWORD 16 00048 #define HWORD 17 00049 #define URLPATH 18 00050 #define FILEPATH 19 00051 #define DECIMAL_T 20 00052 #define SIGNEDINT 21 00053 #define UNSIGNEDINT 22 00054 #define XMLENTITY 23 00055 00056 #define LASTNUM 23 00057 00058 static const char *const tok_alias[] = { 00059 "", 00060 "asciiword", 00061 "word", 00062 "numword", 00063 "email", 00064 "url", 00065 "host", 00066 "sfloat", 00067 "version", 00068 "hword_numpart", 00069 "hword_part", 00070 "hword_asciipart", 00071 "blank", 00072 "tag", 00073 "protocol", 00074 "numhword", 00075 "asciihword", 00076 "hword", 00077 "url_path", 00078 "file", 00079 "float", 00080 "int", 00081 "uint", 00082 "entity" 00083 }; 00084 00085 static const char *const lex_descr[] = { 00086 "", 00087 "Word, all ASCII", 00088 "Word, all letters", 00089 "Word, letters and digits", 00090 "Email address", 00091 "URL", 00092 "Host", 00093 "Scientific notation", 00094 "Version number", 00095 "Hyphenated word part, letters and digits", 00096 "Hyphenated word part, all letters", 00097 "Hyphenated word part, all ASCII", 00098 "Space symbols", 00099 "XML tag", 00100 "Protocol head", 00101 "Hyphenated word, letters and digits", 00102 "Hyphenated word, all ASCII", 00103 "Hyphenated word, all letters", 00104 "URL path", 00105 "File or path name", 00106 "Decimal notation", 00107 "Signed integer", 00108 "Unsigned integer", 00109 "XML entity" 00110 }; 00111 00112 00113 /* Parser states */ 00114 00115 typedef enum 00116 { 00117 TPS_Base = 0, 00118 TPS_InNumWord, 00119 TPS_InAsciiWord, 00120 TPS_InWord, 00121 TPS_InUnsignedInt, 00122 TPS_InSignedIntFirst, 00123 TPS_InSignedInt, 00124 TPS_InSpace, 00125 TPS_InUDecimalFirst, 00126 TPS_InUDecimal, 00127 TPS_InDecimalFirst, 00128 TPS_InDecimal, 00129 TPS_InVerVersion, 00130 TPS_InSVerVersion, 00131 TPS_InVersionFirst, 00132 TPS_InVersion, 00133 TPS_InMantissaFirst, 00134 TPS_InMantissaSign, 00135 TPS_InMantissa, 00136 TPS_InXMLEntityFirst, 00137 TPS_InXMLEntity, 00138 TPS_InXMLEntityNumFirst, 00139 TPS_InXMLEntityNum, 00140 TPS_InXMLEntityHexNumFirst, 00141 TPS_InXMLEntityHexNum, 00142 TPS_InXMLEntityEnd, 00143 TPS_InTagFirst, 00144 TPS_InXMLBegin, 00145 TPS_InTagCloseFirst, 00146 TPS_InTagName, 00147 TPS_InTagBeginEnd, 00148 TPS_InTag, 00149 TPS_InTagEscapeK, 00150 TPS_InTagEscapeKK, 00151 TPS_InTagBackSleshed, 00152 TPS_InTagEnd, 00153 TPS_InCommentFirst, 00154 TPS_InCommentLast, 00155 TPS_InComment, 00156 TPS_InCloseCommentFirst, 00157 TPS_InCloseCommentLast, 00158 TPS_InCommentEnd, 00159 TPS_InHostFirstDomain, 00160 TPS_InHostDomainSecond, 00161 TPS_InHostDomain, 00162 TPS_InPortFirst, 00163 TPS_InPort, 00164 TPS_InHostFirstAN, 00165 TPS_InHost, 00166 TPS_InEmail, 00167 TPS_InFileFirst, 00168 TPS_InFileTwiddle, 00169 TPS_InPathFirst, 00170 TPS_InPathFirstFirst, 00171 TPS_InPathSecond, 00172 TPS_InFile, 00173 TPS_InFileNext, 00174 TPS_InURLPathFirst, 00175 TPS_InURLPathStart, 00176 TPS_InURLPath, 00177 TPS_InFURL, 00178 TPS_InProtocolFirst, 00179 TPS_InProtocolSecond, 00180 TPS_InProtocolEnd, 00181 TPS_InHyphenAsciiWordFirst, 00182 TPS_InHyphenAsciiWord, 00183 TPS_InHyphenWordFirst, 00184 TPS_InHyphenWord, 00185 TPS_InHyphenNumWordFirst, 00186 TPS_InHyphenNumWord, 00187 TPS_InHyphenDigitLookahead, 00188 TPS_InParseHyphen, 00189 TPS_InParseHyphenHyphen, 00190 TPS_InHyphenWordPart, 00191 TPS_InHyphenAsciiWordPart, 00192 TPS_InHyphenNumWordPart, 00193 TPS_InHyphenUnsignedInt, 00194 TPS_Null /* last state (fake value) */ 00195 } TParserState; 00196 00197 /* forward declaration */ 00198 struct TParser; 00199 00200 typedef int (*TParserCharTest) (struct TParser *); /* any p_is* functions 00201 * except p_iseq */ 00202 typedef void (*TParserSpecial) (struct TParser *); /* special handler for 00203 * special cases... */ 00204 00205 typedef struct 00206 { 00207 TParserCharTest isclass; 00208 char c; 00209 uint16 flags; 00210 TParserState tostate; 00211 int type; 00212 TParserSpecial special; 00213 } TParserStateActionItem; 00214 00215 /* Flag bits in TParserStateActionItem.flags */ 00216 #define A_NEXT 0x0000 00217 #define A_BINGO 0x0001 00218 #define A_POP 0x0002 00219 #define A_PUSH 0x0004 00220 #define A_RERUN 0x0008 00221 #define A_CLEAR 0x0010 00222 #define A_MERGE 0x0020 00223 #define A_CLRALL 0x0040 00224 00225 typedef struct TParserPosition 00226 { 00227 int posbyte; /* position of parser in bytes */ 00228 int poschar; /* position of parser in characters */ 00229 int charlen; /* length of current char */ 00230 int lenbytetoken; /* length of token-so-far in bytes */ 00231 int lenchartoken; /* and in chars */ 00232 TParserState state; 00233 struct TParserPosition *prev; 00234 const TParserStateActionItem *pushedAtAction; 00235 } TParserPosition; 00236 00237 typedef struct TParser 00238 { 00239 /* string and position information */ 00240 char *str; /* multibyte string */ 00241 int lenstr; /* length of mbstring */ 00242 #ifdef USE_WIDE_UPPER_LOWER 00243 wchar_t *wstr; /* wide character string */ 00244 pg_wchar *pgwstr; /* wide character string for C-locale */ 00245 bool usewide; 00246 #endif 00247 00248 /* State of parse */ 00249 int charmaxlen; 00250 TParserPosition *state; 00251 bool ignore; 00252 bool wanthost; 00253 00254 /* silly char */ 00255 char c; 00256 00257 /* out */ 00258 char *token; 00259 int lenbytetoken; 00260 int lenchartoken; 00261 int type; 00262 } TParser; 00263 00264 00265 /* forward decls here */ 00266 static bool TParserGet(TParser *prs); 00267 00268 00269 static TParserPosition * 00270 newTParserPosition(TParserPosition *prev) 00271 { 00272 TParserPosition *res = (TParserPosition *) palloc(sizeof(TParserPosition)); 00273 00274 if (prev) 00275 memcpy(res, prev, sizeof(TParserPosition)); 00276 else 00277 memset(res, 0, sizeof(TParserPosition)); 00278 00279 res->prev = prev; 00280 00281 res->pushedAtAction = NULL; 00282 00283 return res; 00284 } 00285 00286 static TParser * 00287 TParserInit(char *str, int len) 00288 { 00289 TParser *prs = (TParser *) palloc0(sizeof(TParser)); 00290 00291 prs->charmaxlen = pg_database_encoding_max_length(); 00292 prs->str = str; 00293 prs->lenstr = len; 00294 00295 #ifdef USE_WIDE_UPPER_LOWER 00296 00297 /* 00298 * Use wide char code only when max encoding length > 1. 00299 */ 00300 if (prs->charmaxlen > 1) 00301 { 00302 Oid collation = DEFAULT_COLLATION_OID; /* TODO */ 00303 pg_locale_t mylocale = 0; /* TODO */ 00304 00305 prs->usewide = true; 00306 if (lc_ctype_is_c(collation)) 00307 { 00308 /* 00309 * char2wchar doesn't work for C-locale and sizeof(pg_wchar) could 00310 * be different from sizeof(wchar_t) 00311 */ 00312 prs->pgwstr = (pg_wchar *) palloc(sizeof(pg_wchar) * (prs->lenstr + 1)); 00313 pg_mb2wchar_with_len(prs->str, prs->pgwstr, prs->lenstr); 00314 } 00315 else 00316 { 00317 prs->wstr = (wchar_t *) palloc(sizeof(wchar_t) * (prs->lenstr + 1)); 00318 char2wchar(prs->wstr, prs->lenstr + 1, prs->str, prs->lenstr, 00319 mylocale); 00320 } 00321 } 00322 else 00323 prs->usewide = false; 00324 #endif 00325 00326 prs->state = newTParserPosition(NULL); 00327 prs->state->state = TPS_Base; 00328 00329 #ifdef WPARSER_TRACE 00330 00331 /* 00332 * Use of %.*s here is a bit risky since it can misbehave if the data is 00333 * not in what libc thinks is the prevailing encoding. However, since 00334 * this is just a debugging aid, we choose to live with that. 00335 */ 00336 fprintf(stderr, "parsing \"%.*s\"\n", len, str); 00337 #endif 00338 00339 return prs; 00340 } 00341 00342 /* 00343 * As an alternative to a full TParserInit one can create a 00344 * TParserCopy which basically is a regular TParser without a private 00345 * copy of the string - instead it uses the one from another TParser. 00346 * This is useful because at some places TParsers are created 00347 * recursively and the repeated copying around of the strings can 00348 * cause major inefficiency if the source string is long. 00349 * The new parser starts parsing at the original's current position. 00350 * 00351 * Obviously one must not close the original TParser before the copy. 00352 */ 00353 static TParser * 00354 TParserCopyInit(const TParser *orig) 00355 { 00356 TParser *prs = (TParser *) palloc0(sizeof(TParser)); 00357 00358 prs->charmaxlen = orig->charmaxlen; 00359 prs->str = orig->str + orig->state->posbyte; 00360 prs->lenstr = orig->lenstr - orig->state->posbyte; 00361 00362 #ifdef USE_WIDE_UPPER_LOWER 00363 prs->usewide = orig->usewide; 00364 00365 if (orig->pgwstr) 00366 prs->pgwstr = orig->pgwstr + orig->state->poschar; 00367 if (orig->wstr) 00368 prs->wstr = orig->wstr + orig->state->poschar; 00369 #endif 00370 00371 prs->state = newTParserPosition(NULL); 00372 prs->state->state = TPS_Base; 00373 00374 #ifdef WPARSER_TRACE 00375 /* See note above about %.*s */ 00376 fprintf(stderr, "parsing copy of \"%.*s\"\n", prs->lenstr, prs->str); 00377 #endif 00378 00379 return prs; 00380 } 00381 00382 00383 static void 00384 TParserClose(TParser *prs) 00385 { 00386 while (prs->state) 00387 { 00388 TParserPosition *ptr = prs->state->prev; 00389 00390 pfree(prs->state); 00391 prs->state = ptr; 00392 } 00393 00394 #ifdef USE_WIDE_UPPER_LOWER 00395 if (prs->wstr) 00396 pfree(prs->wstr); 00397 if (prs->pgwstr) 00398 pfree(prs->pgwstr); 00399 #endif 00400 00401 #ifdef WPARSER_TRACE 00402 fprintf(stderr, "closing parser\n"); 00403 #endif 00404 pfree(prs); 00405 } 00406 00407 /* 00408 * Close a parser created with TParserCopyInit 00409 */ 00410 static void 00411 TParserCopyClose(TParser *prs) 00412 { 00413 while (prs->state) 00414 { 00415 TParserPosition *ptr = prs->state->prev; 00416 00417 pfree(prs->state); 00418 prs->state = ptr; 00419 } 00420 00421 #ifdef WPARSER_TRACE 00422 fprintf(stderr, "closing parser copy\n"); 00423 #endif 00424 pfree(prs); 00425 } 00426 00427 00428 /* 00429 * Character-type support functions, equivalent to is* macros, but 00430 * working with any possible encodings and locales. Notes: 00431 * - with multibyte encoding and C-locale isw* function may fail 00432 * or give wrong result. 00433 * - multibyte encoding and C-locale often are used for 00434 * Asian languages. 00435 * - if locale is C the we use pgwstr instead of wstr 00436 */ 00437 00438 #ifdef USE_WIDE_UPPER_LOWER 00439 00440 #define p_iswhat(type) \ 00441 static int \ 00442 p_is##type(TParser *prs) { \ 00443 Assert( prs->state ); \ 00444 if ( prs->usewide ) \ 00445 { \ 00446 if ( prs->pgwstr ) \ 00447 return is##type( 0xff & *( prs->pgwstr + prs->state->poschar) );\ 00448 \ 00449 return isw##type( *(wint_t*)( prs->wstr + prs->state->poschar ) ); \ 00450 } \ 00451 \ 00452 return is##type( *(unsigned char*)( prs->str + prs->state->posbyte ) ); \ 00453 } \ 00454 \ 00455 static int \ 00456 p_isnot##type(TParser *prs) { \ 00457 return !p_is##type(prs); \ 00458 } 00459 00460 static int 00461 p_isalnum(TParser *prs) 00462 { 00463 Assert(prs->state); 00464 00465 if (prs->usewide) 00466 { 00467 if (prs->pgwstr) 00468 { 00469 unsigned int c = *(prs->pgwstr + prs->state->poschar); 00470 00471 /* 00472 * any non-ascii symbol with multibyte encoding with C-locale is 00473 * an alpha character 00474 */ 00475 if (c > 0x7f) 00476 return 1; 00477 00478 return isalnum(0xff & c); 00479 } 00480 00481 return iswalnum((wint_t) *(prs->wstr + prs->state->poschar)); 00482 } 00483 00484 return isalnum(*(unsigned char *) (prs->str + prs->state->posbyte)); 00485 } 00486 static int 00487 p_isnotalnum(TParser *prs) 00488 { 00489 return !p_isalnum(prs); 00490 } 00491 00492 static int 00493 p_isalpha(TParser *prs) 00494 { 00495 Assert(prs->state); 00496 00497 if (prs->usewide) 00498 { 00499 if (prs->pgwstr) 00500 { 00501 unsigned int c = *(prs->pgwstr + prs->state->poschar); 00502 00503 /* 00504 * any non-ascii symbol with multibyte encoding with C-locale is 00505 * an alpha character 00506 */ 00507 if (c > 0x7f) 00508 return 1; 00509 00510 return isalpha(0xff & c); 00511 } 00512 00513 return iswalpha((wint_t) *(prs->wstr + prs->state->poschar)); 00514 } 00515 00516 return isalpha(*(unsigned char *) (prs->str + prs->state->posbyte)); 00517 } 00518 00519 static int 00520 p_isnotalpha(TParser *prs) 00521 { 00522 return !p_isalpha(prs); 00523 } 00524 00525 /* p_iseq should be used only for ascii symbols */ 00526 00527 static int 00528 p_iseq(TParser *prs, char c) 00529 { 00530 Assert(prs->state); 00531 return ((prs->state->charlen == 1 && *(prs->str + prs->state->posbyte) == c)) ? 1 : 0; 00532 } 00533 #else /* USE_WIDE_UPPER_LOWER */ 00534 00535 #define p_iswhat(type) \ 00536 static int \ 00537 p_is##type(TParser *prs) { \ 00538 Assert( prs->state ); \ 00539 return is##type( (unsigned char)*( prs->str + prs->state->posbyte ) ); \ 00540 } \ 00541 \ 00542 static int \ 00543 p_isnot##type(TParser *prs) { \ 00544 return !p_is##type(prs); \ 00545 } 00546 00547 00548 static int 00549 p_iseq(TParser *prs, char c) 00550 { 00551 Assert(prs->state); 00552 return (*(prs->str + prs->state->posbyte) == c) ? 1 : 0; 00553 } 00554 00555 p_iswhat(alnum) 00556 p_iswhat(alpha) 00557 #endif /* USE_WIDE_UPPER_LOWER */ 00558 00559 p_iswhat(digit) 00560 p_iswhat(lower) 00561 p_iswhat(print) 00562 p_iswhat(punct) 00563 p_iswhat(space) 00564 p_iswhat(upper) 00565 p_iswhat(xdigit) 00566 00567 static int 00568 p_isEOF(TParser *prs) 00569 { 00570 Assert(prs->state); 00571 return (prs->state->posbyte == prs->lenstr || prs->state->charlen == 0) ? 1 : 0; 00572 } 00573 00574 static int 00575 p_iseqC(TParser *prs) 00576 { 00577 return p_iseq(prs, prs->c); 00578 } 00579 00580 static int 00581 p_isneC(TParser *prs) 00582 { 00583 return !p_iseq(prs, prs->c); 00584 } 00585 00586 static int 00587 p_isascii(TParser *prs) 00588 { 00589 return (prs->state->charlen == 1 && isascii((unsigned char) *(prs->str + prs->state->posbyte))) ? 1 : 0; 00590 } 00591 00592 static int 00593 p_isasclet(TParser *prs) 00594 { 00595 return (p_isascii(prs) && p_isalpha(prs)) ? 1 : 0; 00596 } 00597 00598 static int 00599 p_isurlchar(TParser *prs) 00600 { 00601 char ch; 00602 00603 /* no non-ASCII need apply */ 00604 if (prs->state->charlen != 1) 00605 return 0; 00606 ch = *(prs->str + prs->state->posbyte); 00607 /* no spaces or control characters */ 00608 if (ch <= 0x20 || ch >= 0x7F) 00609 return 0; 00610 /* reject characters disallowed by RFC 3986 */ 00611 switch (ch) 00612 { 00613 case '"': 00614 case '<': 00615 case '>': 00616 case '\\': 00617 case '^': 00618 case '`': 00619 case '{': 00620 case '|': 00621 case '}': 00622 return 0; 00623 } 00624 return 1; 00625 } 00626 00627 00628 /* deliberately suppress unused-function complaints for the above */ 00629 void _make_compiler_happy(void); 00630 void 00631 _make_compiler_happy(void) 00632 { 00633 p_isalnum(NULL); 00634 p_isnotalnum(NULL); 00635 p_isalpha(NULL); 00636 p_isnotalpha(NULL); 00637 p_isdigit(NULL); 00638 p_isnotdigit(NULL); 00639 p_islower(NULL); 00640 p_isnotlower(NULL); 00641 p_isprint(NULL); 00642 p_isnotprint(NULL); 00643 p_ispunct(NULL); 00644 p_isnotpunct(NULL); 00645 p_isspace(NULL); 00646 p_isnotspace(NULL); 00647 p_isupper(NULL); 00648 p_isnotupper(NULL); 00649 p_isxdigit(NULL); 00650 p_isnotxdigit(NULL); 00651 p_isEOF(NULL); 00652 p_iseqC(NULL); 00653 p_isneC(NULL); 00654 } 00655 00656 00657 static void 00658 SpecialTags(TParser *prs) 00659 { 00660 switch (prs->state->lenchartoken) 00661 { 00662 case 8: /* </script */ 00663 if (pg_strncasecmp(prs->token, "</script", 8) == 0) 00664 prs->ignore = false; 00665 break; 00666 case 7: /* <script || </style */ 00667 if (pg_strncasecmp(prs->token, "</style", 7) == 0) 00668 prs->ignore = false; 00669 else if (pg_strncasecmp(prs->token, "<script", 7) == 0) 00670 prs->ignore = true; 00671 break; 00672 case 6: /* <style */ 00673 if (pg_strncasecmp(prs->token, "<style", 6) == 0) 00674 prs->ignore = true; 00675 break; 00676 default: 00677 break; 00678 } 00679 } 00680 00681 static void 00682 SpecialFURL(TParser *prs) 00683 { 00684 prs->wanthost = true; 00685 prs->state->posbyte -= prs->state->lenbytetoken; 00686 prs->state->poschar -= prs->state->lenchartoken; 00687 } 00688 00689 static void 00690 SpecialHyphen(TParser *prs) 00691 { 00692 prs->state->posbyte -= prs->state->lenbytetoken; 00693 prs->state->poschar -= prs->state->lenchartoken; 00694 } 00695 00696 static void 00697 SpecialVerVersion(TParser *prs) 00698 { 00699 prs->state->posbyte -= prs->state->lenbytetoken; 00700 prs->state->poschar -= prs->state->lenchartoken; 00701 prs->state->lenbytetoken = 0; 00702 prs->state->lenchartoken = 0; 00703 } 00704 00705 static int 00706 p_isstophost(TParser *prs) 00707 { 00708 if (prs->wanthost) 00709 { 00710 prs->wanthost = false; 00711 return 1; 00712 } 00713 return 0; 00714 } 00715 00716 static int 00717 p_isignore(TParser *prs) 00718 { 00719 return (prs->ignore) ? 1 : 0; 00720 } 00721 00722 static int 00723 p_ishost(TParser *prs) 00724 { 00725 TParser *tmpprs = TParserCopyInit(prs); 00726 int res = 0; 00727 00728 tmpprs->wanthost = true; 00729 00730 if (TParserGet(tmpprs) && tmpprs->type == HOST) 00731 { 00732 prs->state->posbyte += tmpprs->lenbytetoken; 00733 prs->state->poschar += tmpprs->lenchartoken; 00734 prs->state->lenbytetoken += tmpprs->lenbytetoken; 00735 prs->state->lenchartoken += tmpprs->lenchartoken; 00736 prs->state->charlen = tmpprs->state->charlen; 00737 res = 1; 00738 } 00739 TParserCopyClose(tmpprs); 00740 00741 return res; 00742 } 00743 00744 static int 00745 p_isURLPath(TParser *prs) 00746 { 00747 TParser *tmpprs = TParserCopyInit(prs); 00748 int res = 0; 00749 00750 tmpprs->state = newTParserPosition(tmpprs->state); 00751 tmpprs->state->state = TPS_InURLPathFirst; 00752 00753 if (TParserGet(tmpprs) && tmpprs->type == URLPATH) 00754 { 00755 prs->state->posbyte += tmpprs->lenbytetoken; 00756 prs->state->poschar += tmpprs->lenchartoken; 00757 prs->state->lenbytetoken += tmpprs->lenbytetoken; 00758 prs->state->lenchartoken += tmpprs->lenchartoken; 00759 prs->state->charlen = tmpprs->state->charlen; 00760 res = 1; 00761 } 00762 TParserCopyClose(tmpprs); 00763 00764 return res; 00765 } 00766 00767 /* 00768 * returns true if current character has zero display length or 00769 * it's a special sign in several languages. Such characters 00770 * aren't a word-breaker although they aren't an isalpha. 00771 * In beginning of word they aren't a part of it. 00772 */ 00773 static int 00774 p_isspecial(TParser *prs) 00775 { 00776 /* 00777 * pg_dsplen could return -1 which means error or control character 00778 */ 00779 if (pg_dsplen(prs->str + prs->state->posbyte) == 0) 00780 return 1; 00781 00782 #ifdef USE_WIDE_UPPER_LOWER 00783 00784 /* 00785 * Unicode Characters in the 'Mark, Spacing Combining' Category That 00786 * characters are not alpha although they are not breakers of word too. 00787 * Check that only in utf encoding, because other encodings aren't 00788 * supported by postgres or even exists. 00789 */ 00790 if (GetDatabaseEncoding() == PG_UTF8 && prs->usewide) 00791 { 00792 static pg_wchar strange_letter[] = { 00793 /* 00794 * use binary search, so elements should be ordered 00795 */ 00796 0x0903, /* DEVANAGARI SIGN VISARGA */ 00797 0x093E, /* DEVANAGARI VOWEL SIGN AA */ 00798 0x093F, /* DEVANAGARI VOWEL SIGN I */ 00799 0x0940, /* DEVANAGARI VOWEL SIGN II */ 00800 0x0949, /* DEVANAGARI VOWEL SIGN CANDRA O */ 00801 0x094A, /* DEVANAGARI VOWEL SIGN SHORT O */ 00802 0x094B, /* DEVANAGARI VOWEL SIGN O */ 00803 0x094C, /* DEVANAGARI VOWEL SIGN AU */ 00804 0x0982, /* BENGALI SIGN ANUSVARA */ 00805 0x0983, /* BENGALI SIGN VISARGA */ 00806 0x09BE, /* BENGALI VOWEL SIGN AA */ 00807 0x09BF, /* BENGALI VOWEL SIGN I */ 00808 0x09C0, /* BENGALI VOWEL SIGN II */ 00809 0x09C7, /* BENGALI VOWEL SIGN E */ 00810 0x09C8, /* BENGALI VOWEL SIGN AI */ 00811 0x09CB, /* BENGALI VOWEL SIGN O */ 00812 0x09CC, /* BENGALI VOWEL SIGN AU */ 00813 0x09D7, /* BENGALI AU LENGTH MARK */ 00814 0x0A03, /* GURMUKHI SIGN VISARGA */ 00815 0x0A3E, /* GURMUKHI VOWEL SIGN AA */ 00816 0x0A3F, /* GURMUKHI VOWEL SIGN I */ 00817 0x0A40, /* GURMUKHI VOWEL SIGN II */ 00818 0x0A83, /* GUJARATI SIGN VISARGA */ 00819 0x0ABE, /* GUJARATI VOWEL SIGN AA */ 00820 0x0ABF, /* GUJARATI VOWEL SIGN I */ 00821 0x0AC0, /* GUJARATI VOWEL SIGN II */ 00822 0x0AC9, /* GUJARATI VOWEL SIGN CANDRA O */ 00823 0x0ACB, /* GUJARATI VOWEL SIGN O */ 00824 0x0ACC, /* GUJARATI VOWEL SIGN AU */ 00825 0x0B02, /* ORIYA SIGN ANUSVARA */ 00826 0x0B03, /* ORIYA SIGN VISARGA */ 00827 0x0B3E, /* ORIYA VOWEL SIGN AA */ 00828 0x0B40, /* ORIYA VOWEL SIGN II */ 00829 0x0B47, /* ORIYA VOWEL SIGN E */ 00830 0x0B48, /* ORIYA VOWEL SIGN AI */ 00831 0x0B4B, /* ORIYA VOWEL SIGN O */ 00832 0x0B4C, /* ORIYA VOWEL SIGN AU */ 00833 0x0B57, /* ORIYA AU LENGTH MARK */ 00834 0x0BBE, /* TAMIL VOWEL SIGN AA */ 00835 0x0BBF, /* TAMIL VOWEL SIGN I */ 00836 0x0BC1, /* TAMIL VOWEL SIGN U */ 00837 0x0BC2, /* TAMIL VOWEL SIGN UU */ 00838 0x0BC6, /* TAMIL VOWEL SIGN E */ 00839 0x0BC7, /* TAMIL VOWEL SIGN EE */ 00840 0x0BC8, /* TAMIL VOWEL SIGN AI */ 00841 0x0BCA, /* TAMIL VOWEL SIGN O */ 00842 0x0BCB, /* TAMIL VOWEL SIGN OO */ 00843 0x0BCC, /* TAMIL VOWEL SIGN AU */ 00844 0x0BD7, /* TAMIL AU LENGTH MARK */ 00845 0x0C01, /* TELUGU SIGN CANDRABINDU */ 00846 0x0C02, /* TELUGU SIGN ANUSVARA */ 00847 0x0C03, /* TELUGU SIGN VISARGA */ 00848 0x0C41, /* TELUGU VOWEL SIGN U */ 00849 0x0C42, /* TELUGU VOWEL SIGN UU */ 00850 0x0C43, /* TELUGU VOWEL SIGN VOCALIC R */ 00851 0x0C44, /* TELUGU VOWEL SIGN VOCALIC RR */ 00852 0x0C82, /* KANNADA SIGN ANUSVARA */ 00853 0x0C83, /* KANNADA SIGN VISARGA */ 00854 0x0CBE, /* KANNADA VOWEL SIGN AA */ 00855 0x0CC0, /* KANNADA VOWEL SIGN II */ 00856 0x0CC1, /* KANNADA VOWEL SIGN U */ 00857 0x0CC2, /* KANNADA VOWEL SIGN UU */ 00858 0x0CC3, /* KANNADA VOWEL SIGN VOCALIC R */ 00859 0x0CC4, /* KANNADA VOWEL SIGN VOCALIC RR */ 00860 0x0CC7, /* KANNADA VOWEL SIGN EE */ 00861 0x0CC8, /* KANNADA VOWEL SIGN AI */ 00862 0x0CCA, /* KANNADA VOWEL SIGN O */ 00863 0x0CCB, /* KANNADA VOWEL SIGN OO */ 00864 0x0CD5, /* KANNADA LENGTH MARK */ 00865 0x0CD6, /* KANNADA AI LENGTH MARK */ 00866 0x0D02, /* MALAYALAM SIGN ANUSVARA */ 00867 0x0D03, /* MALAYALAM SIGN VISARGA */ 00868 0x0D3E, /* MALAYALAM VOWEL SIGN AA */ 00869 0x0D3F, /* MALAYALAM VOWEL SIGN I */ 00870 0x0D40, /* MALAYALAM VOWEL SIGN II */ 00871 0x0D46, /* MALAYALAM VOWEL SIGN E */ 00872 0x0D47, /* MALAYALAM VOWEL SIGN EE */ 00873 0x0D48, /* MALAYALAM VOWEL SIGN AI */ 00874 0x0D4A, /* MALAYALAM VOWEL SIGN O */ 00875 0x0D4B, /* MALAYALAM VOWEL SIGN OO */ 00876 0x0D4C, /* MALAYALAM VOWEL SIGN AU */ 00877 0x0D57, /* MALAYALAM AU LENGTH MARK */ 00878 0x0D82, /* SINHALA SIGN ANUSVARAYA */ 00879 0x0D83, /* SINHALA SIGN VISARGAYA */ 00880 0x0DCF, /* SINHALA VOWEL SIGN AELA-PILLA */ 00881 0x0DD0, /* SINHALA VOWEL SIGN KETTI AEDA-PILLA */ 00882 0x0DD1, /* SINHALA VOWEL SIGN DIGA AEDA-PILLA */ 00883 0x0DD8, /* SINHALA VOWEL SIGN GAETTA-PILLA */ 00884 0x0DD9, /* SINHALA VOWEL SIGN KOMBUVA */ 00885 0x0DDA, /* SINHALA VOWEL SIGN DIGA KOMBUVA */ 00886 0x0DDB, /* SINHALA VOWEL SIGN KOMBU DEKA */ 00887 0x0DDC, /* SINHALA VOWEL SIGN KOMBUVA HAA AELA-PILLA */ 00888 0x0DDD, /* SINHALA VOWEL SIGN KOMBUVA HAA DIGA 00889 * AELA-PILLA */ 00890 0x0DDE, /* SINHALA VOWEL SIGN KOMBUVA HAA GAYANUKITTA */ 00891 0x0DDF, /* SINHALA VOWEL SIGN GAYANUKITTA */ 00892 0x0DF2, /* SINHALA VOWEL SIGN DIGA GAETTA-PILLA */ 00893 0x0DF3, /* SINHALA VOWEL SIGN DIGA GAYANUKITTA */ 00894 0x0F3E, /* TIBETAN SIGN YAR TSHES */ 00895 0x0F3F, /* TIBETAN SIGN MAR TSHES */ 00896 0x0F7F, /* TIBETAN SIGN RNAM BCAD */ 00897 0x102B, /* MYANMAR VOWEL SIGN TALL AA */ 00898 0x102C, /* MYANMAR VOWEL SIGN AA */ 00899 0x1031, /* MYANMAR VOWEL SIGN E */ 00900 0x1038, /* MYANMAR SIGN VISARGA */ 00901 0x103B, /* MYANMAR CONSONANT SIGN MEDIAL YA */ 00902 0x103C, /* MYANMAR CONSONANT SIGN MEDIAL RA */ 00903 0x1056, /* MYANMAR VOWEL SIGN VOCALIC R */ 00904 0x1057, /* MYANMAR VOWEL SIGN VOCALIC RR */ 00905 0x1062, /* MYANMAR VOWEL SIGN SGAW KAREN EU */ 00906 0x1063, /* MYANMAR TONE MARK SGAW KAREN HATHI */ 00907 0x1064, /* MYANMAR TONE MARK SGAW KAREN KE PHO */ 00908 0x1067, /* MYANMAR VOWEL SIGN WESTERN PWO KAREN EU */ 00909 0x1068, /* MYANMAR VOWEL SIGN WESTERN PWO KAREN UE */ 00910 0x1069, /* MYANMAR SIGN WESTERN PWO KAREN TONE-1 */ 00911 0x106A, /* MYANMAR SIGN WESTERN PWO KAREN TONE-2 */ 00912 0x106B, /* MYANMAR SIGN WESTERN PWO KAREN TONE-3 */ 00913 0x106C, /* MYANMAR SIGN WESTERN PWO KAREN TONE-4 */ 00914 0x106D, /* MYANMAR SIGN WESTERN PWO KAREN TONE-5 */ 00915 0x1083, /* MYANMAR VOWEL SIGN SHAN AA */ 00916 0x1084, /* MYANMAR VOWEL SIGN SHAN E */ 00917 0x1087, /* MYANMAR SIGN SHAN TONE-2 */ 00918 0x1088, /* MYANMAR SIGN SHAN TONE-3 */ 00919 0x1089, /* MYANMAR SIGN SHAN TONE-5 */ 00920 0x108A, /* MYANMAR SIGN SHAN TONE-6 */ 00921 0x108B, /* MYANMAR SIGN SHAN COUNCIL TONE-2 */ 00922 0x108C, /* MYANMAR SIGN SHAN COUNCIL TONE-3 */ 00923 0x108F, /* MYANMAR SIGN RUMAI PALAUNG TONE-5 */ 00924 0x17B6, /* KHMER VOWEL SIGN AA */ 00925 0x17BE, /* KHMER VOWEL SIGN OE */ 00926 0x17BF, /* KHMER VOWEL SIGN YA */ 00927 0x17C0, /* KHMER VOWEL SIGN IE */ 00928 0x17C1, /* KHMER VOWEL SIGN E */ 00929 0x17C2, /* KHMER VOWEL SIGN AE */ 00930 0x17C3, /* KHMER VOWEL SIGN AI */ 00931 0x17C4, /* KHMER VOWEL SIGN OO */ 00932 0x17C5, /* KHMER VOWEL SIGN AU */ 00933 0x17C7, /* KHMER SIGN REAHMUK */ 00934 0x17C8, /* KHMER SIGN YUUKALEAPINTU */ 00935 0x1923, /* LIMBU VOWEL SIGN EE */ 00936 0x1924, /* LIMBU VOWEL SIGN AI */ 00937 0x1925, /* LIMBU VOWEL SIGN OO */ 00938 0x1926, /* LIMBU VOWEL SIGN AU */ 00939 0x1929, /* LIMBU SUBJOINED LETTER YA */ 00940 0x192A, /* LIMBU SUBJOINED LETTER RA */ 00941 0x192B, /* LIMBU SUBJOINED LETTER WA */ 00942 0x1930, /* LIMBU SMALL LETTER KA */ 00943 0x1931, /* LIMBU SMALL LETTER NGA */ 00944 0x1933, /* LIMBU SMALL LETTER TA */ 00945 0x1934, /* LIMBU SMALL LETTER NA */ 00946 0x1935, /* LIMBU SMALL LETTER PA */ 00947 0x1936, /* LIMBU SMALL LETTER MA */ 00948 0x1937, /* LIMBU SMALL LETTER RA */ 00949 0x1938, /* LIMBU SMALL LETTER LA */ 00950 0x19B0, /* NEW TAI LUE VOWEL SIGN VOWEL SHORTENER */ 00951 0x19B1, /* NEW TAI LUE VOWEL SIGN AA */ 00952 0x19B2, /* NEW TAI LUE VOWEL SIGN II */ 00953 0x19B3, /* NEW TAI LUE VOWEL SIGN U */ 00954 0x19B4, /* NEW TAI LUE VOWEL SIGN UU */ 00955 0x19B5, /* NEW TAI LUE VOWEL SIGN E */ 00956 0x19B6, /* NEW TAI LUE VOWEL SIGN AE */ 00957 0x19B7, /* NEW TAI LUE VOWEL SIGN O */ 00958 0x19B8, /* NEW TAI LUE VOWEL SIGN OA */ 00959 0x19B9, /* NEW TAI LUE VOWEL SIGN UE */ 00960 0x19BA, /* NEW TAI LUE VOWEL SIGN AY */ 00961 0x19BB, /* NEW TAI LUE VOWEL SIGN AAY */ 00962 0x19BC, /* NEW TAI LUE VOWEL SIGN UY */ 00963 0x19BD, /* NEW TAI LUE VOWEL SIGN OY */ 00964 0x19BE, /* NEW TAI LUE VOWEL SIGN OAY */ 00965 0x19BF, /* NEW TAI LUE VOWEL SIGN UEY */ 00966 0x19C0, /* NEW TAI LUE VOWEL SIGN IY */ 00967 0x19C8, /* NEW TAI LUE TONE MARK-1 */ 00968 0x19C9, /* NEW TAI LUE TONE MARK-2 */ 00969 0x1A19, /* BUGINESE VOWEL SIGN E */ 00970 0x1A1A, /* BUGINESE VOWEL SIGN O */ 00971 0x1A1B, /* BUGINESE VOWEL SIGN AE */ 00972 0x1B04, /* BALINESE SIGN BISAH */ 00973 0x1B35, /* BALINESE VOWEL SIGN TEDUNG */ 00974 0x1B3B, /* BALINESE VOWEL SIGN RA REPA TEDUNG */ 00975 0x1B3D, /* BALINESE VOWEL SIGN LA LENGA TEDUNG */ 00976 0x1B3E, /* BALINESE VOWEL SIGN TALING */ 00977 0x1B3F, /* BALINESE VOWEL SIGN TALING REPA */ 00978 0x1B40, /* BALINESE VOWEL SIGN TALING TEDUNG */ 00979 0x1B41, /* BALINESE VOWEL SIGN TALING REPA TEDUNG */ 00980 0x1B43, /* BALINESE VOWEL SIGN PEPET TEDUNG */ 00981 0x1B44, /* BALINESE ADEG ADEG */ 00982 0x1B82, /* SUNDANESE SIGN PANGWISAD */ 00983 0x1BA1, /* SUNDANESE CONSONANT SIGN PAMINGKAL */ 00984 0x1BA6, /* SUNDANESE VOWEL SIGN PANAELAENG */ 00985 0x1BA7, /* SUNDANESE VOWEL SIGN PANOLONG */ 00986 0x1BAA, /* SUNDANESE SIGN PAMAAEH */ 00987 0x1C24, /* LEPCHA SUBJOINED LETTER YA */ 00988 0x1C25, /* LEPCHA SUBJOINED LETTER RA */ 00989 0x1C26, /* LEPCHA VOWEL SIGN AA */ 00990 0x1C27, /* LEPCHA VOWEL SIGN I */ 00991 0x1C28, /* LEPCHA VOWEL SIGN O */ 00992 0x1C29, /* LEPCHA VOWEL SIGN OO */ 00993 0x1C2A, /* LEPCHA VOWEL SIGN U */ 00994 0x1C2B, /* LEPCHA VOWEL SIGN UU */ 00995 0x1C34, /* LEPCHA CONSONANT SIGN NYIN-DO */ 00996 0x1C35, /* LEPCHA CONSONANT SIGN KANG */ 00997 0xA823, /* SYLOTI NAGRI VOWEL SIGN A */ 00998 0xA824, /* SYLOTI NAGRI VOWEL SIGN I */ 00999 0xA827, /* SYLOTI NAGRI VOWEL SIGN OO */ 01000 0xA880, /* SAURASHTRA SIGN ANUSVARA */ 01001 0xA881, /* SAURASHTRA SIGN VISARGA */ 01002 0xA8B4, /* SAURASHTRA CONSONANT SIGN HAARU */ 01003 0xA8B5, /* SAURASHTRA VOWEL SIGN AA */ 01004 0xA8B6, /* SAURASHTRA VOWEL SIGN I */ 01005 0xA8B7, /* SAURASHTRA VOWEL SIGN II */ 01006 0xA8B8, /* SAURASHTRA VOWEL SIGN U */ 01007 0xA8B9, /* SAURASHTRA VOWEL SIGN UU */ 01008 0xA8BA, /* SAURASHTRA VOWEL SIGN VOCALIC R */ 01009 0xA8BB, /* SAURASHTRA VOWEL SIGN VOCALIC RR */ 01010 0xA8BC, /* SAURASHTRA VOWEL SIGN VOCALIC L */ 01011 0xA8BD, /* SAURASHTRA VOWEL SIGN VOCALIC LL */ 01012 0xA8BE, /* SAURASHTRA VOWEL SIGN E */ 01013 0xA8BF, /* SAURASHTRA VOWEL SIGN EE */ 01014 0xA8C0, /* SAURASHTRA VOWEL SIGN AI */ 01015 0xA8C1, /* SAURASHTRA VOWEL SIGN O */ 01016 0xA8C2, /* SAURASHTRA VOWEL SIGN OO */ 01017 0xA8C3, /* SAURASHTRA VOWEL SIGN AU */ 01018 0xA952, /* REJANG CONSONANT SIGN H */ 01019 0xA953, /* REJANG VIRAMA */ 01020 0xAA2F, /* CHAM VOWEL SIGN O */ 01021 0xAA30, /* CHAM VOWEL SIGN AI */ 01022 0xAA33, /* CHAM CONSONANT SIGN YA */ 01023 0xAA34, /* CHAM CONSONANT SIGN RA */ 01024 0xAA4D /* CHAM CONSONANT SIGN FINAL H */ 01025 }; 01026 pg_wchar *StopLow = strange_letter, 01027 *StopHigh = strange_letter + lengthof(strange_letter), 01028 *StopMiddle; 01029 pg_wchar c; 01030 01031 if (prs->pgwstr) 01032 c = *(prs->pgwstr + prs->state->poschar); 01033 else 01034 c = (pg_wchar) *(prs->wstr + prs->state->poschar); 01035 01036 while (StopLow < StopHigh) 01037 { 01038 StopMiddle = StopLow + ((StopHigh - StopLow) >> 1); 01039 if (*StopMiddle == c) 01040 return 1; 01041 else if (*StopMiddle < c) 01042 StopLow = StopMiddle + 1; 01043 else 01044 StopHigh = StopMiddle; 01045 } 01046 } 01047 #endif 01048 01049 return 0; 01050 } 01051 01052 /* 01053 * Table of state/action of parser 01054 */ 01055 01056 static const TParserStateActionItem actionTPS_Base[] = { 01057 {p_isEOF, 0, A_NEXT, TPS_Null, 0, NULL}, 01058 {p_iseqC, '<', A_PUSH, TPS_InTagFirst, 0, NULL}, 01059 {p_isignore, 0, A_NEXT, TPS_InSpace, 0, NULL}, 01060 {p_isasclet, 0, A_NEXT, TPS_InAsciiWord, 0, NULL}, 01061 {p_isalpha, 0, A_NEXT, TPS_InWord, 0, NULL}, 01062 {p_isdigit, 0, A_NEXT, TPS_InUnsignedInt, 0, NULL}, 01063 {p_iseqC, '-', A_PUSH, TPS_InSignedIntFirst, 0, NULL}, 01064 {p_iseqC, '+', A_PUSH, TPS_InSignedIntFirst, 0, NULL}, 01065 {p_iseqC, '&', A_PUSH, TPS_InXMLEntityFirst, 0, NULL}, 01066 {p_iseqC, '~', A_PUSH, TPS_InFileTwiddle, 0, NULL}, 01067 {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL}, 01068 {p_iseqC, '.', A_PUSH, TPS_InPathFirstFirst, 0, NULL}, 01069 {NULL, 0, A_NEXT, TPS_InSpace, 0, NULL} 01070 }; 01071 01072 01073 static const TParserStateActionItem actionTPS_InNumWord[] = { 01074 {p_isEOF, 0, A_BINGO, TPS_Base, NUMWORD, NULL}, 01075 {p_isalnum, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01076 {p_isspecial, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01077 {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL}, 01078 {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL}, 01079 {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL}, 01080 {p_iseqC, '-', A_PUSH, TPS_InHyphenNumWordFirst, 0, NULL}, 01081 {NULL, 0, A_BINGO, TPS_Base, NUMWORD, NULL} 01082 }; 01083 01084 static const TParserStateActionItem actionTPS_InAsciiWord[] = { 01085 {p_isEOF, 0, A_BINGO, TPS_Base, ASCIIWORD, NULL}, 01086 {p_isasclet, 0, A_NEXT, TPS_Null, 0, NULL}, 01087 {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL}, 01088 {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL}, 01089 {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01090 {p_iseqC, '-', A_PUSH, TPS_InHyphenAsciiWordFirst, 0, NULL}, 01091 {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01092 {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL}, 01093 {p_iseqC, ':', A_PUSH, TPS_InProtocolFirst, 0, NULL}, 01094 {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL}, 01095 {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL}, 01096 {p_isdigit, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01097 {p_isalpha, 0, A_NEXT, TPS_InWord, 0, NULL}, 01098 {p_isspecial, 0, A_NEXT, TPS_InWord, 0, NULL}, 01099 {NULL, 0, A_BINGO, TPS_Base, ASCIIWORD, NULL} 01100 }; 01101 01102 static const TParserStateActionItem actionTPS_InWord[] = { 01103 {p_isEOF, 0, A_BINGO, TPS_Base, WORD_T, NULL}, 01104 {p_isalpha, 0, A_NEXT, TPS_Null, 0, NULL}, 01105 {p_isspecial, 0, A_NEXT, TPS_Null, 0, NULL}, 01106 {p_isdigit, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01107 {p_iseqC, '-', A_PUSH, TPS_InHyphenWordFirst, 0, NULL}, 01108 {NULL, 0, A_BINGO, TPS_Base, WORD_T, NULL} 01109 }; 01110 01111 static const TParserStateActionItem actionTPS_InUnsignedInt[] = { 01112 {p_isEOF, 0, A_BINGO, TPS_Base, UNSIGNEDINT, NULL}, 01113 {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL}, 01114 {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL}, 01115 {p_iseqC, '.', A_PUSH, TPS_InUDecimalFirst, 0, NULL}, 01116 {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01117 {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01118 {p_isasclet, 0, A_PUSH, TPS_InHost, 0, NULL}, 01119 {p_isalpha, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01120 {p_isspecial, 0, A_NEXT, TPS_InNumWord, 0, NULL}, 01121 {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL}, 01122 {NULL, 0, A_BINGO, TPS_Base, UNSIGNEDINT, NULL} 01123 }; 01124 01125 static const TParserStateActionItem actionTPS_InSignedIntFirst[] = { 01126 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01127 {p_isdigit, 0, A_NEXT | A_CLEAR, TPS_InSignedInt, 0, NULL}, 01128 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01129 }; 01130 01131 static const TParserStateActionItem actionTPS_InSignedInt[] = { 01132 {p_isEOF, 0, A_BINGO, TPS_Base, SIGNEDINT, NULL}, 01133 {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL}, 01134 {p_iseqC, '.', A_PUSH, TPS_InDecimalFirst, 0, NULL}, 01135 {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01136 {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01137 {NULL, 0, A_BINGO, TPS_Base, SIGNEDINT, NULL} 01138 }; 01139 01140 static const TParserStateActionItem actionTPS_InSpace[] = { 01141 {p_isEOF, 0, A_BINGO, TPS_Base, SPACE, NULL}, 01142 {p_iseqC, '<', A_BINGO, TPS_Base, SPACE, NULL}, 01143 {p_isignore, 0, A_NEXT, TPS_Null, 0, NULL}, 01144 {p_iseqC, '-', A_BINGO, TPS_Base, SPACE, NULL}, 01145 {p_iseqC, '+', A_BINGO, TPS_Base, SPACE, NULL}, 01146 {p_iseqC, '&', A_BINGO, TPS_Base, SPACE, NULL}, 01147 {p_iseqC, '/', A_BINGO, TPS_Base, SPACE, NULL}, 01148 {p_isnotalnum, 0, A_NEXT, TPS_InSpace, 0, NULL}, 01149 {NULL, 0, A_BINGO, TPS_Base, SPACE, NULL} 01150 }; 01151 01152 static const TParserStateActionItem actionTPS_InUDecimalFirst[] = { 01153 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01154 {p_isdigit, 0, A_CLEAR, TPS_InUDecimal, 0, NULL}, 01155 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01156 }; 01157 01158 static const TParserStateActionItem actionTPS_InUDecimal[] = { 01159 {p_isEOF, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL}, 01160 {p_isdigit, 0, A_NEXT, TPS_InUDecimal, 0, NULL}, 01161 {p_iseqC, '.', A_PUSH, TPS_InVersionFirst, 0, NULL}, 01162 {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01163 {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01164 {NULL, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL} 01165 }; 01166 01167 static const TParserStateActionItem actionTPS_InDecimalFirst[] = { 01168 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01169 {p_isdigit, 0, A_CLEAR, TPS_InDecimal, 0, NULL}, 01170 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01171 }; 01172 01173 static const TParserStateActionItem actionTPS_InDecimal[] = { 01174 {p_isEOF, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL}, 01175 {p_isdigit, 0, A_NEXT, TPS_InDecimal, 0, NULL}, 01176 {p_iseqC, '.', A_PUSH, TPS_InVerVersion, 0, NULL}, 01177 {p_iseqC, 'e', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01178 {p_iseqC, 'E', A_PUSH, TPS_InMantissaFirst, 0, NULL}, 01179 {NULL, 0, A_BINGO, TPS_Base, DECIMAL_T, NULL} 01180 }; 01181 01182 static const TParserStateActionItem actionTPS_InVerVersion[] = { 01183 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01184 {p_isdigit, 0, A_RERUN, TPS_InSVerVersion, 0, SpecialVerVersion}, 01185 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01186 }; 01187 01188 static const TParserStateActionItem actionTPS_InSVerVersion[] = { 01189 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01190 {p_isdigit, 0, A_BINGO | A_CLRALL, TPS_InUnsignedInt, SPACE, NULL}, 01191 {NULL, 0, A_NEXT, TPS_Null, 0, NULL} 01192 }; 01193 01194 01195 static const TParserStateActionItem actionTPS_InVersionFirst[] = { 01196 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01197 {p_isdigit, 0, A_CLEAR, TPS_InVersion, 0, NULL}, 01198 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01199 }; 01200 01201 static const TParserStateActionItem actionTPS_InVersion[] = { 01202 {p_isEOF, 0, A_BINGO, TPS_Base, VERSIONNUMBER, NULL}, 01203 {p_isdigit, 0, A_NEXT, TPS_InVersion, 0, NULL}, 01204 {p_iseqC, '.', A_PUSH, TPS_InVersionFirst, 0, NULL}, 01205 {NULL, 0, A_BINGO, TPS_Base, VERSIONNUMBER, NULL} 01206 }; 01207 01208 static const TParserStateActionItem actionTPS_InMantissaFirst[] = { 01209 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01210 {p_isdigit, 0, A_CLEAR, TPS_InMantissa, 0, NULL}, 01211 {p_iseqC, '+', A_NEXT, TPS_InMantissaSign, 0, NULL}, 01212 {p_iseqC, '-', A_NEXT, TPS_InMantissaSign, 0, NULL}, 01213 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01214 }; 01215 01216 static const TParserStateActionItem actionTPS_InMantissaSign[] = { 01217 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01218 {p_isdigit, 0, A_CLEAR, TPS_InMantissa, 0, NULL}, 01219 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01220 }; 01221 01222 static const TParserStateActionItem actionTPS_InMantissa[] = { 01223 {p_isEOF, 0, A_BINGO, TPS_Base, SCIENTIFIC, NULL}, 01224 {p_isdigit, 0, A_NEXT, TPS_InMantissa, 0, NULL}, 01225 {NULL, 0, A_BINGO, TPS_Base, SCIENTIFIC, NULL} 01226 }; 01227 01228 static const TParserStateActionItem actionTPS_InXMLEntityFirst[] = { 01229 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01230 {p_iseqC, '#', A_NEXT, TPS_InXMLEntityNumFirst, 0, NULL}, 01231 {p_isasclet, 0, A_NEXT, TPS_InXMLEntity, 0, NULL}, 01232 {p_iseqC, ':', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01233 {p_iseqC, '_', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01234 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01235 }; 01236 01237 static const TParserStateActionItem actionTPS_InXMLEntity[] = { 01238 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01239 {p_isalnum, 0, A_NEXT, TPS_InXMLEntity, 0, NULL}, 01240 {p_iseqC, ':', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01241 {p_iseqC, '_', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01242 {p_iseqC, '.', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01243 {p_iseqC, '-', A_NEXT, TPS_InXMLEntity, 0, NULL}, 01244 {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL}, 01245 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01246 }; 01247 01248 static const TParserStateActionItem actionTPS_InXMLEntityNumFirst[] = { 01249 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01250 {p_iseqC, 'x', A_NEXT, TPS_InXMLEntityHexNumFirst, 0, NULL}, 01251 {p_iseqC, 'X', A_NEXT, TPS_InXMLEntityHexNumFirst, 0, NULL}, 01252 {p_isdigit, 0, A_NEXT, TPS_InXMLEntityNum, 0, NULL}, 01253 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01254 }; 01255 01256 static const TParserStateActionItem actionTPS_InXMLEntityHexNumFirst[] = { 01257 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01258 {p_isxdigit, 0, A_NEXT, TPS_InXMLEntityHexNum, 0, NULL}, 01259 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01260 }; 01261 01262 static const TParserStateActionItem actionTPS_InXMLEntityNum[] = { 01263 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01264 {p_isdigit, 0, A_NEXT, TPS_InXMLEntityNum, 0, NULL}, 01265 {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL}, 01266 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01267 }; 01268 01269 static const TParserStateActionItem actionTPS_InXMLEntityHexNum[] = { 01270 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01271 {p_isxdigit, 0, A_NEXT, TPS_InXMLEntityHexNum, 0, NULL}, 01272 {p_iseqC, ';', A_NEXT, TPS_InXMLEntityEnd, 0, NULL}, 01273 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01274 }; 01275 01276 static const TParserStateActionItem actionTPS_InXMLEntityEnd[] = { 01277 {NULL, 0, A_BINGO | A_CLEAR, TPS_Base, XMLENTITY, NULL} 01278 }; 01279 01280 static const TParserStateActionItem actionTPS_InTagFirst[] = { 01281 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01282 {p_iseqC, '/', A_PUSH, TPS_InTagCloseFirst, 0, NULL}, 01283 {p_iseqC, '!', A_PUSH, TPS_InCommentFirst, 0, NULL}, 01284 {p_iseqC, '?', A_PUSH, TPS_InXMLBegin, 0, NULL}, 01285 {p_isasclet, 0, A_PUSH, TPS_InTagName, 0, NULL}, 01286 {p_iseqC, ':', A_PUSH, TPS_InTagName, 0, NULL}, 01287 {p_iseqC, '_', A_PUSH, TPS_InTagName, 0, NULL}, 01288 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01289 }; 01290 01291 static const TParserStateActionItem actionTPS_InXMLBegin[] = { 01292 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01293 /* <?xml ... */ 01294 /* XXX do we wants states for the m and l ? Right now this accepts <?xZ */ 01295 {p_iseqC, 'x', A_NEXT, TPS_InTag, 0, NULL}, 01296 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01297 }; 01298 01299 static const TParserStateActionItem actionTPS_InTagCloseFirst[] = { 01300 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01301 {p_isasclet, 0, A_NEXT, TPS_InTagName, 0, NULL}, 01302 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01303 }; 01304 01305 static const TParserStateActionItem actionTPS_InTagName[] = { 01306 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01307 /* <br/> case */ 01308 {p_iseqC, '/', A_NEXT, TPS_InTagBeginEnd, 0, NULL}, 01309 {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, SpecialTags}, 01310 {p_isspace, 0, A_NEXT, TPS_InTag, 0, SpecialTags}, 01311 {p_isalnum, 0, A_NEXT, TPS_Null, 0, NULL}, 01312 {p_iseqC, ':', A_NEXT, TPS_Null, 0, NULL}, 01313 {p_iseqC, '_', A_NEXT, TPS_Null, 0, NULL}, 01314 {p_iseqC, '.', A_NEXT, TPS_Null, 0, NULL}, 01315 {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL}, 01316 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01317 }; 01318 01319 static const TParserStateActionItem actionTPS_InTagBeginEnd[] = { 01320 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01321 {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, NULL}, 01322 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01323 }; 01324 01325 static const TParserStateActionItem actionTPS_InTag[] = { 01326 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01327 {p_iseqC, '>', A_NEXT, TPS_InTagEnd, 0, SpecialTags}, 01328 {p_iseqC, '\'', A_NEXT, TPS_InTagEscapeK, 0, NULL}, 01329 {p_iseqC, '"', A_NEXT, TPS_InTagEscapeKK, 0, NULL}, 01330 {p_isasclet, 0, A_NEXT, TPS_Null, 0, NULL}, 01331 {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL}, 01332 {p_iseqC, '=', A_NEXT, TPS_Null, 0, NULL}, 01333 {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL}, 01334 {p_iseqC, '_', A_NEXT, TPS_Null, 0, NULL}, 01335 {p_iseqC, '#', A_NEXT, TPS_Null, 0, NULL}, 01336 {p_iseqC, '/', A_NEXT, TPS_Null, 0, NULL}, 01337 {p_iseqC, ':', A_NEXT, TPS_Null, 0, NULL}, 01338 {p_iseqC, '.', A_NEXT, TPS_Null, 0, NULL}, 01339 {p_iseqC, '&', A_NEXT, TPS_Null, 0, NULL}, 01340 {p_iseqC, '?', A_NEXT, TPS_Null, 0, NULL}, 01341 {p_iseqC, '%', A_NEXT, TPS_Null, 0, NULL}, 01342 {p_iseqC, '~', A_NEXT, TPS_Null, 0, NULL}, 01343 {p_isspace, 0, A_NEXT, TPS_Null, 0, SpecialTags}, 01344 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01345 }; 01346 01347 static const TParserStateActionItem actionTPS_InTagEscapeK[] = { 01348 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01349 {p_iseqC, '\\', A_PUSH, TPS_InTagBackSleshed, 0, NULL}, 01350 {p_iseqC, '\'', A_NEXT, TPS_InTag, 0, NULL}, 01351 {NULL, 0, A_NEXT, TPS_InTagEscapeK, 0, NULL} 01352 }; 01353 01354 static const TParserStateActionItem actionTPS_InTagEscapeKK[] = { 01355 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01356 {p_iseqC, '\\', A_PUSH, TPS_InTagBackSleshed, 0, NULL}, 01357 {p_iseqC, '"', A_NEXT, TPS_InTag, 0, NULL}, 01358 {NULL, 0, A_NEXT, TPS_InTagEscapeKK, 0, NULL} 01359 }; 01360 01361 static const TParserStateActionItem actionTPS_InTagBackSleshed[] = { 01362 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01363 {NULL, 0, A_MERGE, TPS_Null, 0, NULL} 01364 }; 01365 01366 static const TParserStateActionItem actionTPS_InTagEnd[] = { 01367 {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, TAG_T, NULL} 01368 }; 01369 01370 static const TParserStateActionItem actionTPS_InCommentFirst[] = { 01371 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01372 {p_iseqC, '-', A_NEXT, TPS_InCommentLast, 0, NULL}, 01373 /* <!DOCTYPE ...> */ 01374 {p_iseqC, 'D', A_NEXT, TPS_InTag, 0, NULL}, 01375 {p_iseqC, 'd', A_NEXT, TPS_InTag, 0, NULL}, 01376 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01377 }; 01378 01379 static const TParserStateActionItem actionTPS_InCommentLast[] = { 01380 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01381 {p_iseqC, '-', A_NEXT, TPS_InComment, 0, NULL}, 01382 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01383 }; 01384 01385 static const TParserStateActionItem actionTPS_InComment[] = { 01386 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01387 {p_iseqC, '-', A_NEXT, TPS_InCloseCommentFirst, 0, NULL}, 01388 {NULL, 0, A_NEXT, TPS_Null, 0, NULL} 01389 }; 01390 01391 static const TParserStateActionItem actionTPS_InCloseCommentFirst[] = { 01392 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01393 {p_iseqC, '-', A_NEXT, TPS_InCloseCommentLast, 0, NULL}, 01394 {NULL, 0, A_NEXT, TPS_InComment, 0, NULL} 01395 }; 01396 01397 static const TParserStateActionItem actionTPS_InCloseCommentLast[] = { 01398 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01399 {p_iseqC, '-', A_NEXT, TPS_Null, 0, NULL}, 01400 {p_iseqC, '>', A_NEXT, TPS_InCommentEnd, 0, NULL}, 01401 {NULL, 0, A_NEXT, TPS_InComment, 0, NULL} 01402 }; 01403 01404 static const TParserStateActionItem actionTPS_InCommentEnd[] = { 01405 {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, TAG_T, NULL} 01406 }; 01407 01408 static const TParserStateActionItem actionTPS_InHostFirstDomain[] = { 01409 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01410 {p_isasclet, 0, A_NEXT, TPS_InHostDomainSecond, 0, NULL}, 01411 {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL}, 01412 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01413 }; 01414 01415 static const TParserStateActionItem actionTPS_InHostDomainSecond[] = { 01416 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01417 {p_isasclet, 0, A_NEXT, TPS_InHostDomain, 0, NULL}, 01418 {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL}, 01419 {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01420 {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01421 {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL}, 01422 {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL}, 01423 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01424 }; 01425 01426 static const TParserStateActionItem actionTPS_InHostDomain[] = { 01427 {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL}, 01428 {p_isasclet, 0, A_NEXT, TPS_InHostDomain, 0, NULL}, 01429 {p_isdigit, 0, A_PUSH, TPS_InHost, 0, NULL}, 01430 {p_iseqC, ':', A_PUSH, TPS_InPortFirst, 0, NULL}, 01431 {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01432 {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01433 {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL}, 01434 {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL}, 01435 {p_isdigit, 0, A_POP, TPS_Null, 0, NULL}, 01436 {p_isstophost, 0, A_BINGO | A_CLRALL, TPS_InURLPathStart, HOST, NULL}, 01437 {p_iseqC, '/', A_PUSH, TPS_InFURL, 0, NULL}, 01438 {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL} 01439 }; 01440 01441 static const TParserStateActionItem actionTPS_InPortFirst[] = { 01442 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01443 {p_isdigit, 0, A_NEXT, TPS_InPort, 0, NULL}, 01444 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01445 }; 01446 01447 static const TParserStateActionItem actionTPS_InPort[] = { 01448 {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL}, 01449 {p_isdigit, 0, A_NEXT, TPS_InPort, 0, NULL}, 01450 {p_isstophost, 0, A_BINGO | A_CLRALL, TPS_InURLPathStart, HOST, NULL}, 01451 {p_iseqC, '/', A_PUSH, TPS_InFURL, 0, NULL}, 01452 {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, HOST, NULL} 01453 }; 01454 01455 static const TParserStateActionItem actionTPS_InHostFirstAN[] = { 01456 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01457 {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL}, 01458 {p_isasclet, 0, A_NEXT, TPS_InHost, 0, NULL}, 01459 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01460 }; 01461 01462 static const TParserStateActionItem actionTPS_InHost[] = { 01463 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01464 {p_isdigit, 0, A_NEXT, TPS_InHost, 0, NULL}, 01465 {p_isasclet, 0, A_NEXT, TPS_InHost, 0, NULL}, 01466 {p_iseqC, '@', A_PUSH, TPS_InEmail, 0, NULL}, 01467 {p_iseqC, '.', A_PUSH, TPS_InHostFirstDomain, 0, NULL}, 01468 {p_iseqC, '-', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01469 {p_iseqC, '_', A_PUSH, TPS_InHostFirstAN, 0, NULL}, 01470 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01471 }; 01472 01473 static const TParserStateActionItem actionTPS_InEmail[] = { 01474 {p_isstophost, 0, A_POP, TPS_Null, 0, NULL}, 01475 {p_ishost, 0, A_BINGO | A_CLRALL, TPS_Base, EMAIL, NULL}, 01476 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01477 }; 01478 01479 static const TParserStateActionItem actionTPS_InFileFirst[] = { 01480 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01481 {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL}, 01482 {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL}, 01483 {p_iseqC, '.', A_NEXT, TPS_InPathFirst, 0, NULL}, 01484 {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL}, 01485 {p_iseqC, '~', A_PUSH, TPS_InFileTwiddle, 0, NULL}, 01486 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01487 }; 01488 01489 static const TParserStateActionItem actionTPS_InFileTwiddle[] = { 01490 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01491 {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL}, 01492 {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL}, 01493 {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL}, 01494 {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL}, 01495 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01496 }; 01497 01498 static const TParserStateActionItem actionTPS_InPathFirst[] = { 01499 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01500 {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL}, 01501 {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL}, 01502 {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL}, 01503 {p_iseqC, '.', A_NEXT, TPS_InPathSecond, 0, NULL}, 01504 {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL}, 01505 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01506 }; 01507 01508 static const TParserStateActionItem actionTPS_InPathFirstFirst[] = { 01509 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01510 {p_iseqC, '.', A_NEXT, TPS_InPathSecond, 0, NULL}, 01511 {p_iseqC, '/', A_NEXT, TPS_InFileFirst, 0, NULL}, 01512 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01513 }; 01514 01515 static const TParserStateActionItem actionTPS_InPathSecond[] = { 01516 {p_isEOF, 0, A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL}, 01517 {p_iseqC, '/', A_NEXT | A_PUSH, TPS_InFileFirst, 0, NULL}, 01518 {p_iseqC, '/', A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL}, 01519 {p_isspace, 0, A_BINGO | A_CLEAR, TPS_Base, FILEPATH, NULL}, 01520 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01521 }; 01522 01523 static const TParserStateActionItem actionTPS_InFile[] = { 01524 {p_isEOF, 0, A_BINGO, TPS_Base, FILEPATH, NULL}, 01525 {p_isasclet, 0, A_NEXT, TPS_InFile, 0, NULL}, 01526 {p_isdigit, 0, A_NEXT, TPS_InFile, 0, NULL}, 01527 {p_iseqC, '.', A_PUSH, TPS_InFileNext, 0, NULL}, 01528 {p_iseqC, '_', A_NEXT, TPS_InFile, 0, NULL}, 01529 {p_iseqC, '-', A_NEXT, TPS_InFile, 0, NULL}, 01530 {p_iseqC, '/', A_PUSH, TPS_InFileFirst, 0, NULL}, 01531 {NULL, 0, A_BINGO, TPS_Base, FILEPATH, NULL} 01532 }; 01533 01534 static const TParserStateActionItem actionTPS_InFileNext[] = { 01535 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01536 {p_isasclet, 0, A_CLEAR, TPS_InFile, 0, NULL}, 01537 {p_isdigit, 0, A_CLEAR, TPS_InFile, 0, NULL}, 01538 {p_iseqC, '_', A_CLEAR, TPS_InFile, 0, NULL}, 01539 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01540 }; 01541 01542 static const TParserStateActionItem actionTPS_InURLPathFirst[] = { 01543 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01544 {p_isurlchar, 0, A_NEXT, TPS_InURLPath, 0, NULL}, 01545 {NULL, 0, A_POP, TPS_Null, 0, NULL}, 01546 }; 01547 01548 static const TParserStateActionItem actionTPS_InURLPathStart[] = { 01549 {NULL, 0, A_NEXT, TPS_InURLPath, 0, NULL} 01550 }; 01551 01552 static const TParserStateActionItem actionTPS_InURLPath[] = { 01553 {p_isEOF, 0, A_BINGO, TPS_Base, URLPATH, NULL}, 01554 {p_isurlchar, 0, A_NEXT, TPS_InURLPath, 0, NULL}, 01555 {NULL, 0, A_BINGO, TPS_Base, URLPATH, NULL} 01556 }; 01557 01558 static const TParserStateActionItem actionTPS_InFURL[] = { 01559 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01560 {p_isURLPath, 0, A_BINGO | A_CLRALL, TPS_Base, URL_T, SpecialFURL}, 01561 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01562 }; 01563 01564 static const TParserStateActionItem actionTPS_InProtocolFirst[] = { 01565 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01566 {p_iseqC, '/', A_NEXT, TPS_InProtocolSecond, 0, NULL}, 01567 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01568 }; 01569 01570 static const TParserStateActionItem actionTPS_InProtocolSecond[] = { 01571 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01572 {p_iseqC, '/', A_NEXT, TPS_InProtocolEnd, 0, NULL}, 01573 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01574 }; 01575 01576 static const TParserStateActionItem actionTPS_InProtocolEnd[] = { 01577 {NULL, 0, A_BINGO | A_CLRALL, TPS_Base, PROTOCOL, NULL} 01578 }; 01579 01580 static const TParserStateActionItem actionTPS_InHyphenAsciiWordFirst[] = { 01581 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01582 {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWord, 0, NULL}, 01583 {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01584 {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL}, 01585 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01586 }; 01587 01588 static const TParserStateActionItem actionTPS_InHyphenAsciiWord[] = { 01589 {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, ASCIIHWORD, SpecialHyphen}, 01590 {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWord, 0, NULL}, 01591 {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01592 {p_isspecial, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01593 {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01594 {p_iseqC, '-', A_PUSH, TPS_InHyphenAsciiWordFirst, 0, NULL}, 01595 {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, ASCIIHWORD, SpecialHyphen} 01596 }; 01597 01598 static const TParserStateActionItem actionTPS_InHyphenWordFirst[] = { 01599 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01600 {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01601 {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL}, 01602 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01603 }; 01604 01605 static const TParserStateActionItem actionTPS_InHyphenWord[] = { 01606 {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, HWORD, SpecialHyphen}, 01607 {p_isalpha, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01608 {p_isspecial, 0, A_NEXT, TPS_InHyphenWord, 0, NULL}, 01609 {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01610 {p_iseqC, '-', A_PUSH, TPS_InHyphenWordFirst, 0, NULL}, 01611 {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, HWORD, SpecialHyphen} 01612 }; 01613 01614 static const TParserStateActionItem actionTPS_InHyphenNumWordFirst[] = { 01615 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01616 {p_isalpha, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01617 {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL}, 01618 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01619 }; 01620 01621 static const TParserStateActionItem actionTPS_InHyphenNumWord[] = { 01622 {p_isEOF, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, NUMHWORD, SpecialHyphen}, 01623 {p_isalnum, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01624 {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01625 {p_iseqC, '-', A_PUSH, TPS_InHyphenNumWordFirst, 0, NULL}, 01626 {NULL, 0, A_BINGO | A_CLRALL, TPS_InParseHyphen, NUMHWORD, SpecialHyphen} 01627 }; 01628 01629 static const TParserStateActionItem actionTPS_InHyphenDigitLookahead[] = { 01630 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01631 {p_isdigit, 0, A_NEXT, TPS_InHyphenDigitLookahead, 0, NULL}, 01632 {p_isalpha, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01633 {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWord, 0, NULL}, 01634 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01635 }; 01636 01637 static const TParserStateActionItem actionTPS_InParseHyphen[] = { 01638 {p_isEOF, 0, A_RERUN, TPS_Base, 0, NULL}, 01639 {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWordPart, 0, NULL}, 01640 {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL}, 01641 {p_isdigit, 0, A_PUSH, TPS_InHyphenUnsignedInt, 0, NULL}, 01642 {p_iseqC, '-', A_PUSH, TPS_InParseHyphenHyphen, 0, NULL}, 01643 {NULL, 0, A_RERUN, TPS_Base, 0, NULL} 01644 }; 01645 01646 static const TParserStateActionItem actionTPS_InParseHyphenHyphen[] = { 01647 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01648 {p_isalnum, 0, A_BINGO | A_CLEAR, TPS_InParseHyphen, SPACE, NULL}, 01649 {p_isspecial, 0, A_BINGO | A_CLEAR, TPS_InParseHyphen, SPACE, NULL}, 01650 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01651 }; 01652 01653 static const TParserStateActionItem actionTPS_InHyphenWordPart[] = { 01654 {p_isEOF, 0, A_BINGO, TPS_Base, PARTHWORD, NULL}, 01655 {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL}, 01656 {p_isspecial, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL}, 01657 {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL}, 01658 {NULL, 0, A_BINGO, TPS_InParseHyphen, PARTHWORD, NULL} 01659 }; 01660 01661 static const TParserStateActionItem actionTPS_InHyphenAsciiWordPart[] = { 01662 {p_isEOF, 0, A_BINGO, TPS_Base, ASCIIPARTHWORD, NULL}, 01663 {p_isasclet, 0, A_NEXT, TPS_InHyphenAsciiWordPart, 0, NULL}, 01664 {p_isalpha, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL}, 01665 {p_isspecial, 0, A_NEXT, TPS_InHyphenWordPart, 0, NULL}, 01666 {p_isdigit, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL}, 01667 {NULL, 0, A_BINGO, TPS_InParseHyphen, ASCIIPARTHWORD, NULL} 01668 }; 01669 01670 static const TParserStateActionItem actionTPS_InHyphenNumWordPart[] = { 01671 {p_isEOF, 0, A_BINGO, TPS_Base, NUMPARTHWORD, NULL}, 01672 {p_isalnum, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL}, 01673 {p_isspecial, 0, A_NEXT, TPS_InHyphenNumWordPart, 0, NULL}, 01674 {NULL, 0, A_BINGO, TPS_InParseHyphen, NUMPARTHWORD, NULL} 01675 }; 01676 01677 static const TParserStateActionItem actionTPS_InHyphenUnsignedInt[] = { 01678 {p_isEOF, 0, A_POP, TPS_Null, 0, NULL}, 01679 {p_isdigit, 0, A_NEXT, TPS_Null, 0, NULL}, 01680 {p_isalpha, 0, A_CLEAR, TPS_InHyphenNumWordPart, 0, NULL}, 01681 {p_isspecial, 0, A_CLEAR, TPS_InHyphenNumWordPart, 0, NULL}, 01682 {NULL, 0, A_POP, TPS_Null, 0, NULL} 01683 }; 01684 01685 01686 /* 01687 * main table of per-state parser actions 01688 */ 01689 typedef struct 01690 { 01691 const TParserStateActionItem *action; /* the actual state info */ 01692 TParserState state; /* only for Assert crosscheck */ 01693 #ifdef WPARSER_TRACE 01694 const char *state_name; /* only for debug printout */ 01695 #endif 01696 } TParserStateAction; 01697 01698 #ifdef WPARSER_TRACE 01699 #define TPARSERSTATEACTION(state) \ 01700 { CppConcat(action,state), state, CppAsString(state) } 01701 #else 01702 #define TPARSERSTATEACTION(state) \ 01703 { CppConcat(action,state), state } 01704 #endif 01705 01706 /* 01707 * order must be the same as in typedef enum {} TParserState!! 01708 */ 01709 01710 static const TParserStateAction Actions[] = { 01711 TPARSERSTATEACTION(TPS_Base), 01712 TPARSERSTATEACTION(TPS_InNumWord), 01713 TPARSERSTATEACTION(TPS_InAsciiWord), 01714 TPARSERSTATEACTION(TPS_InWord), 01715 TPARSERSTATEACTION(TPS_InUnsignedInt), 01716 TPARSERSTATEACTION(TPS_InSignedIntFirst), 01717 TPARSERSTATEACTION(TPS_InSignedInt), 01718 TPARSERSTATEACTION(TPS_InSpace), 01719 TPARSERSTATEACTION(TPS_InUDecimalFirst), 01720 TPARSERSTATEACTION(TPS_InUDecimal), 01721 TPARSERSTATEACTION(TPS_InDecimalFirst), 01722 TPARSERSTATEACTION(TPS_InDecimal), 01723 TPARSERSTATEACTION(TPS_InVerVersion), 01724 TPARSERSTATEACTION(TPS_InSVerVersion), 01725 TPARSERSTATEACTION(TPS_InVersionFirst), 01726 TPARSERSTATEACTION(TPS_InVersion), 01727 TPARSERSTATEACTION(TPS_InMantissaFirst), 01728 TPARSERSTATEACTION(TPS_InMantissaSign), 01729 TPARSERSTATEACTION(TPS_InMantissa), 01730 TPARSERSTATEACTION(TPS_InXMLEntityFirst), 01731 TPARSERSTATEACTION(TPS_InXMLEntity), 01732 TPARSERSTATEACTION(TPS_InXMLEntityNumFirst), 01733 TPARSERSTATEACTION(TPS_InXMLEntityNum), 01734 TPARSERSTATEACTION(TPS_InXMLEntityHexNumFirst), 01735 TPARSERSTATEACTION(TPS_InXMLEntityHexNum), 01736 TPARSERSTATEACTION(TPS_InXMLEntityEnd), 01737 TPARSERSTATEACTION(TPS_InTagFirst), 01738 TPARSERSTATEACTION(TPS_InXMLBegin), 01739 TPARSERSTATEACTION(TPS_InTagCloseFirst), 01740 TPARSERSTATEACTION(TPS_InTagName), 01741 TPARSERSTATEACTION(TPS_InTagBeginEnd), 01742 TPARSERSTATEACTION(TPS_InTag), 01743 TPARSERSTATEACTION(TPS_InTagEscapeK), 01744 TPARSERSTATEACTION(TPS_InTagEscapeKK), 01745 TPARSERSTATEACTION(TPS_InTagBackSleshed), 01746 TPARSERSTATEACTION(TPS_InTagEnd), 01747 TPARSERSTATEACTION(TPS_InCommentFirst), 01748 TPARSERSTATEACTION(TPS_InCommentLast), 01749 TPARSERSTATEACTION(TPS_InComment), 01750 TPARSERSTATEACTION(TPS_InCloseCommentFirst), 01751 TPARSERSTATEACTION(TPS_InCloseCommentLast), 01752 TPARSERSTATEACTION(TPS_InCommentEnd), 01753 TPARSERSTATEACTION(TPS_InHostFirstDomain), 01754 TPARSERSTATEACTION(TPS_InHostDomainSecond), 01755 TPARSERSTATEACTION(TPS_InHostDomain), 01756 TPARSERSTATEACTION(TPS_InPortFirst), 01757 TPARSERSTATEACTION(TPS_InPort), 01758 TPARSERSTATEACTION(TPS_InHostFirstAN), 01759 TPARSERSTATEACTION(TPS_InHost), 01760 TPARSERSTATEACTION(TPS_InEmail), 01761 TPARSERSTATEACTION(TPS_InFileFirst), 01762 TPARSERSTATEACTION(TPS_InFileTwiddle), 01763 TPARSERSTATEACTION(TPS_InPathFirst), 01764 TPARSERSTATEACTION(TPS_InPathFirstFirst), 01765 TPARSERSTATEACTION(TPS_InPathSecond), 01766 TPARSERSTATEACTION(TPS_InFile), 01767 TPARSERSTATEACTION(TPS_InFileNext), 01768 TPARSERSTATEACTION(TPS_InURLPathFirst), 01769 TPARSERSTATEACTION(TPS_InURLPathStart), 01770 TPARSERSTATEACTION(TPS_InURLPath), 01771 TPARSERSTATEACTION(TPS_InFURL), 01772 TPARSERSTATEACTION(TPS_InProtocolFirst), 01773 TPARSERSTATEACTION(TPS_InProtocolSecond), 01774 TPARSERSTATEACTION(TPS_InProtocolEnd), 01775 TPARSERSTATEACTION(TPS_InHyphenAsciiWordFirst), 01776 TPARSERSTATEACTION(TPS_InHyphenAsciiWord), 01777 TPARSERSTATEACTION(TPS_InHyphenWordFirst), 01778 TPARSERSTATEACTION(TPS_InHyphenWord), 01779 TPARSERSTATEACTION(TPS_InHyphenNumWordFirst), 01780 TPARSERSTATEACTION(TPS_InHyphenNumWord), 01781 TPARSERSTATEACTION(TPS_InHyphenDigitLookahead), 01782 TPARSERSTATEACTION(TPS_InParseHyphen), 01783 TPARSERSTATEACTION(TPS_InParseHyphenHyphen), 01784 TPARSERSTATEACTION(TPS_InHyphenWordPart), 01785 TPARSERSTATEACTION(TPS_InHyphenAsciiWordPart), 01786 TPARSERSTATEACTION(TPS_InHyphenNumWordPart), 01787 TPARSERSTATEACTION(TPS_InHyphenUnsignedInt) 01788 }; 01789 01790 01791 static bool 01792 TParserGet(TParser *prs) 01793 { 01794 const TParserStateActionItem *item = NULL; 01795 01796 Assert(prs->state); 01797 01798 if (prs->state->posbyte >= prs->lenstr) 01799 return false; 01800 01801 prs->token = prs->str + prs->state->posbyte; 01802 prs->state->pushedAtAction = NULL; 01803 01804 /* look at string */ 01805 while (prs->state->posbyte <= prs->lenstr) 01806 { 01807 if (prs->state->posbyte == prs->lenstr) 01808 prs->state->charlen = 0; 01809 else 01810 prs->state->charlen = (prs->charmaxlen == 1) ? prs->charmaxlen : 01811 pg_mblen(prs->str + prs->state->posbyte); 01812 01813 Assert(prs->state->posbyte + prs->state->charlen <= prs->lenstr); 01814 Assert(prs->state->state >= TPS_Base && prs->state->state < TPS_Null); 01815 Assert(Actions[prs->state->state].state == prs->state->state); 01816 01817 if (prs->state->pushedAtAction) 01818 { 01819 /* After a POP, pick up at the next test */ 01820 item = prs->state->pushedAtAction + 1; 01821 prs->state->pushedAtAction = NULL; 01822 } 01823 else 01824 { 01825 item = Actions[prs->state->state].action; 01826 Assert(item != NULL); 01827 } 01828 01829 /* find action by character class */ 01830 while (item->isclass) 01831 { 01832 prs->c = item->c; 01833 if (item->isclass(prs) != 0) 01834 break; 01835 item++; 01836 } 01837 01838 #ifdef WPARSER_TRACE 01839 { 01840 TParserPosition *ptr; 01841 01842 fprintf(stderr, "state "); 01843 /* indent according to stack depth */ 01844 for (ptr = prs->state->prev; ptr; ptr = ptr->prev) 01845 fprintf(stderr, " "); 01846 fprintf(stderr, "%s ", Actions[prs->state->state].state_name); 01847 if (prs->state->posbyte < prs->lenstr) 01848 fprintf(stderr, "at %c", *(prs->str + prs->state->posbyte)); 01849 else 01850 fprintf(stderr, "at EOF"); 01851 fprintf(stderr, " matched rule %d flags%s%s%s%s%s%s%s%s%s%s%s\n", 01852 (int) (item - Actions[prs->state->state].action), 01853 (item->flags & A_BINGO) ? " BINGO" : "", 01854 (item->flags & A_POP) ? " POP" : "", 01855 (item->flags & A_PUSH) ? " PUSH" : "", 01856 (item->flags & A_RERUN) ? " RERUN" : "", 01857 (item->flags & A_CLEAR) ? " CLEAR" : "", 01858 (item->flags & A_MERGE) ? " MERGE" : "", 01859 (item->flags & A_CLRALL) ? " CLRALL" : "", 01860 (item->tostate != TPS_Null) ? " tostate " : "", 01861 (item->tostate != TPS_Null) ? Actions[item->tostate].state_name : "", 01862 (item->type > 0) ? " type " : "", 01863 tok_alias[item->type]); 01864 } 01865 #endif 01866 01867 /* call special handler if exists */ 01868 if (item->special) 01869 item->special(prs); 01870 01871 /* BINGO, token is found */ 01872 if (item->flags & A_BINGO) 01873 { 01874 Assert(item->type > 0); 01875 prs->lenbytetoken = prs->state->lenbytetoken; 01876 prs->lenchartoken = prs->state->lenchartoken; 01877 prs->state->lenbytetoken = prs->state->lenchartoken = 0; 01878 prs->type = item->type; 01879 } 01880 01881 /* do various actions by flags */ 01882 if (item->flags & A_POP) 01883 { /* pop stored state in stack */ 01884 TParserPosition *ptr = prs->state->prev; 01885 01886 pfree(prs->state); 01887 prs->state = ptr; 01888 Assert(prs->state); 01889 } 01890 else if (item->flags & A_PUSH) 01891 { /* push (store) state in stack */ 01892 prs->state->pushedAtAction = item; /* remember where we push */ 01893 prs->state = newTParserPosition(prs->state); 01894 } 01895 else if (item->flags & A_CLEAR) 01896 { /* clear previous pushed state */ 01897 TParserPosition *ptr; 01898 01899 Assert(prs->state->prev); 01900 ptr = prs->state->prev->prev; 01901 pfree(prs->state->prev); 01902 prs->state->prev = ptr; 01903 } 01904 else if (item->flags & A_CLRALL) 01905 { /* clear all previous pushed state */ 01906 TParserPosition *ptr; 01907 01908 while (prs->state->prev) 01909 { 01910 ptr = prs->state->prev->prev; 01911 pfree(prs->state->prev); 01912 prs->state->prev = ptr; 01913 } 01914 } 01915 else if (item->flags & A_MERGE) 01916 { /* merge posinfo with current and pushed state */ 01917 TParserPosition *ptr = prs->state; 01918 01919 Assert(prs->state->prev); 01920 prs->state = prs->state->prev; 01921 01922 prs->state->posbyte = ptr->posbyte; 01923 prs->state->poschar = ptr->poschar; 01924 prs->state->charlen = ptr->charlen; 01925 prs->state->lenbytetoken = ptr->lenbytetoken; 01926 prs->state->lenchartoken = ptr->lenchartoken; 01927 pfree(ptr); 01928 } 01929 01930 /* set new state if pointed */ 01931 if (item->tostate != TPS_Null) 01932 prs->state->state = item->tostate; 01933 01934 /* check for go away */ 01935 if ((item->flags & A_BINGO) || 01936 (prs->state->posbyte >= prs->lenstr && 01937 (item->flags & A_RERUN) == 0)) 01938 break; 01939 01940 /* go to beginning of loop if we should rerun or we just restore state */ 01941 if (item->flags & (A_RERUN | A_POP)) 01942 continue; 01943 01944 /* move forward */ 01945 if (prs->state->charlen) 01946 { 01947 prs->state->posbyte += prs->state->charlen; 01948 prs->state->lenbytetoken += prs->state->charlen; 01949 prs->state->poschar++; 01950 prs->state->lenchartoken++; 01951 } 01952 } 01953 01954 return (item && (item->flags & A_BINGO)) ? true : false; 01955 } 01956 01957 Datum 01958 prsd_lextype(PG_FUNCTION_ARGS) 01959 { 01960 LexDescr *descr = (LexDescr *) palloc(sizeof(LexDescr) * (LASTNUM + 1)); 01961 int i; 01962 01963 for (i = 1; i <= LASTNUM; i++) 01964 { 01965 descr[i - 1].lexid = i; 01966 descr[i - 1].alias = pstrdup(tok_alias[i]); 01967 descr[i - 1].descr = pstrdup(lex_descr[i]); 01968 } 01969 01970 descr[LASTNUM].lexid = 0; 01971 01972 PG_RETURN_POINTER(descr); 01973 } 01974 01975 Datum 01976 prsd_start(PG_FUNCTION_ARGS) 01977 { 01978 PG_RETURN_POINTER(TParserInit((char *) PG_GETARG_POINTER(0), PG_GETARG_INT32(1))); 01979 } 01980 01981 Datum 01982 prsd_nexttoken(PG_FUNCTION_ARGS) 01983 { 01984 TParser *p = (TParser *) PG_GETARG_POINTER(0); 01985 char **t = (char **) PG_GETARG_POINTER(1); 01986 int *tlen = (int *) PG_GETARG_POINTER(2); 01987 01988 if (!TParserGet(p)) 01989 PG_RETURN_INT32(0); 01990 01991 *t = p->token; 01992 *tlen = p->lenbytetoken; 01993 01994 PG_RETURN_INT32(p->type); 01995 } 01996 01997 Datum 01998 prsd_end(PG_FUNCTION_ARGS) 01999 { 02000 TParser *p = (TParser *) PG_GETARG_POINTER(0); 02001 02002 TParserClose(p); 02003 PG_RETURN_VOID(); 02004 } 02005 02006 #define LEAVETOKEN(x) ( (x)==SPACE ) 02007 #define COMPLEXTOKEN(x) ( (x)==URL_T || (x)==NUMHWORD || (x)==ASCIIHWORD || (x)==HWORD ) 02008 #define ENDPUNCTOKEN(x) ( (x)==SPACE ) 02009 02010 #define TS_IDIGNORE(x) ( (x)==TAG_T || (x)==PROTOCOL || (x)==SPACE || (x)==XMLENTITY ) 02011 #define HLIDREPLACE(x) ( (x)==TAG_T ) 02012 #define HLIDSKIP(x) ( (x)==URL_T || (x)==NUMHWORD || (x)==ASCIIHWORD || (x)==HWORD ) 02013 #define XMLHLIDSKIP(x) ( (x)==URL_T || (x)==NUMHWORD || (x)==ASCIIHWORD || (x)==HWORD ) 02014 #define NONWORDTOKEN(x) ( (x)==SPACE || HLIDREPLACE(x) || HLIDSKIP(x) ) 02015 #define NOENDTOKEN(x) ( NONWORDTOKEN(x) || (x)==SCIENTIFIC || (x)==VERSIONNUMBER || (x)==DECIMAL_T || (x)==SIGNEDINT || (x)==UNSIGNEDINT || TS_IDIGNORE(x) ) 02016 02017 typedef struct 02018 { 02019 HeadlineWordEntry *words; 02020 int len; 02021 } hlCheck; 02022 02023 static bool 02024 checkcondition_HL(void *checkval, QueryOperand *val) 02025 { 02026 int i; 02027 02028 for (i = 0; i < ((hlCheck *) checkval)->len; i++) 02029 { 02030 if (((hlCheck *) checkval)->words[i].item == val) 02031 return true; 02032 } 02033 return false; 02034 } 02035 02036 02037 static bool 02038 hlCover(HeadlineParsedText *prs, TSQuery query, int *p, int *q) 02039 { 02040 int i, 02041 j; 02042 QueryItem *item = GETQUERY(query); 02043 int pos = *p; 02044 02045 *q = -1; 02046 *p = 0x7fffffff; 02047 02048 for (j = 0; j < query->size; j++) 02049 { 02050 if (item->type != QI_VAL) 02051 { 02052 item++; 02053 continue; 02054 } 02055 for (i = pos; i < prs->curwords; i++) 02056 { 02057 if (prs->words[i].item == &item->qoperand) 02058 { 02059 if (i > *q) 02060 *q = i; 02061 break; 02062 } 02063 } 02064 item++; 02065 } 02066 02067 if (*q < 0) 02068 return false; 02069 02070 item = GETQUERY(query); 02071 for (j = 0; j < query->size; j++) 02072 { 02073 if (item->type != QI_VAL) 02074 { 02075 item++; 02076 continue; 02077 } 02078 for (i = *q; i >= pos; i--) 02079 { 02080 if (prs->words[i].item == &item->qoperand) 02081 { 02082 if (i < *p) 02083 *p = i; 02084 break; 02085 } 02086 } 02087 item++; 02088 } 02089 02090 if (*p <= *q) 02091 { 02092 hlCheck ch; 02093 02094 ch.words = &(prs->words[*p]); 02095 ch.len = *q - *p + 1; 02096 if (TS_execute(GETQUERY(query), &ch, false, checkcondition_HL)) 02097 return true; 02098 else 02099 { 02100 (*p)++; 02101 return hlCover(prs, query, p, q); 02102 } 02103 } 02104 02105 return false; 02106 } 02107 02108 static void 02109 mark_fragment(HeadlineParsedText *prs, int highlight, int startpos, int endpos) 02110 { 02111 int i; 02112 02113 for (i = startpos; i <= endpos; i++) 02114 { 02115 if (prs->words[i].item) 02116 prs->words[i].selected = 1; 02117 if (highlight == 0) 02118 { 02119 if (HLIDREPLACE(prs->words[i].type)) 02120 prs->words[i].replace = 1; 02121 else if (HLIDSKIP(prs->words[i].type)) 02122 prs->words[i].skip = 1; 02123 } 02124 else 02125 { 02126 if (XMLHLIDSKIP(prs->words[i].type)) 02127 prs->words[i].skip = 1; 02128 } 02129 02130 prs->words[i].in = (prs->words[i].repeated) ? 0 : 1; 02131 } 02132 } 02133 02134 typedef struct 02135 { 02136 int32 startpos; 02137 int32 endpos; 02138 int32 poslen; 02139 int32 curlen; 02140 int16 in; 02141 int16 excluded; 02142 } CoverPos; 02143 02144 static void 02145 get_next_fragment(HeadlineParsedText *prs, int *startpos, int *endpos, 02146 int *curlen, int *poslen, int max_words) 02147 { 02148 int i; 02149 02150 /* 02151 * Objective: Generate a fragment of words between startpos and endpos 02152 * such that it has at most max_words and both ends has query words. If 02153 * the startpos and endpos are the endpoints of the cover and the cover 02154 * has fewer words than max_words, then this function should just return 02155 * the cover 02156 */ 02157 /* first move startpos to an item */ 02158 for (i = *startpos; i <= *endpos; i++) 02159 { 02160 *startpos = i; 02161 if (prs->words[i].item && !prs->words[i].repeated) 02162 break; 02163 } 02164 /* cut endpos to have only max_words */ 02165 *curlen = 0; 02166 *poslen = 0; 02167 for (i = *startpos; i <= *endpos && *curlen < max_words; i++) 02168 { 02169 if (!NONWORDTOKEN(prs->words[i].type)) 02170 *curlen += 1; 02171 if (prs->words[i].item && !prs->words[i].repeated) 02172 *poslen += 1; 02173 } 02174 /* if the cover was cut then move back endpos to a query item */ 02175 if (*endpos > i) 02176 { 02177 *endpos = i; 02178 for (i = *endpos; i >= *startpos; i--) 02179 { 02180 *endpos = i; 02181 if (prs->words[i].item && !prs->words[i].repeated) 02182 break; 02183 if (!NONWORDTOKEN(prs->words[i].type)) 02184 *curlen -= 1; 02185 } 02186 } 02187 } 02188 02189 static void 02190 mark_hl_fragments(HeadlineParsedText *prs, TSQuery query, int highlight, 02191 int shortword, int min_words, 02192 int max_words, int max_fragments) 02193 { 02194 int32 poslen, 02195 curlen, 02196 i, 02197 f, 02198 num_f = 0; 02199 int32 stretch, 02200 maxstretch, 02201 posmarker; 02202 02203 int32 startpos = 0, 02204 endpos = 0, 02205 p = 0, 02206 q = 0; 02207 02208 int32 numcovers = 0, 02209 maxcovers = 32; 02210 02211 int32 minI, 02212 minwords, 02213 maxitems; 02214 CoverPos *covers; 02215 02216 covers = palloc(maxcovers * sizeof(CoverPos)); 02217 02218 /* get all covers */ 02219 while (hlCover(prs, query, &p, &q)) 02220 { 02221 startpos = p; 02222 endpos = q; 02223 02224 /* 02225 * Break the cover into smaller fragments such that each fragment has 02226 * at most max_words. Also ensure that each end of the fragment is a 02227 * query word. This will allow us to stretch the fragment in either 02228 * direction 02229 */ 02230 02231 while (startpos <= endpos) 02232 { 02233 get_next_fragment(prs, &startpos, &endpos, &curlen, &poslen, max_words); 02234 if (numcovers >= maxcovers) 02235 { 02236 maxcovers *= 2; 02237 covers = repalloc(covers, sizeof(CoverPos) * maxcovers); 02238 } 02239 covers[numcovers].startpos = startpos; 02240 covers[numcovers].endpos = endpos; 02241 covers[numcovers].curlen = curlen; 02242 covers[numcovers].poslen = poslen; 02243 covers[numcovers].in = 0; 02244 covers[numcovers].excluded = 0; 02245 numcovers++; 02246 startpos = endpos + 1; 02247 endpos = q; 02248 } 02249 /* move p to generate the next cover */ 02250 p++; 02251 } 02252 02253 /* choose best covers */ 02254 for (f = 0; f < max_fragments; f++) 02255 { 02256 maxitems = 0; 02257 minwords = 0x7fffffff; 02258 minI = -1; 02259 02260 /* 02261 * Choose the cover that contains max items. In case of tie choose the 02262 * one with smaller number of words. 02263 */ 02264 for (i = 0; i < numcovers; i++) 02265 { 02266 if (!covers[i].in && !covers[i].excluded && 02267 (maxitems < covers[i].poslen || (maxitems == covers[i].poslen 02268 && minwords > covers[i].curlen))) 02269 { 02270 maxitems = covers[i].poslen; 02271 minwords = covers[i].curlen; 02272 minI = i; 02273 } 02274 } 02275 /* if a cover was found mark it */ 02276 if (minI >= 0) 02277 { 02278 covers[minI].in = 1; 02279 /* adjust the size of cover */ 02280 startpos = covers[minI].startpos; 02281 endpos = covers[minI].endpos; 02282 curlen = covers[minI].curlen; 02283 /* stretch the cover if cover size is lower than max_words */ 02284 if (curlen < max_words) 02285 { 02286 /* divide the stretch on both sides of cover */ 02287 maxstretch = (max_words - curlen) / 2; 02288 02289 /* 02290 * first stretch the startpos stop stretching if 1. we hit the 02291 * beginning of document 2. exceed maxstretch 3. we hit an 02292 * already marked fragment 02293 */ 02294 stretch = 0; 02295 posmarker = startpos; 02296 for (i = startpos - 1; i >= 0 && stretch < maxstretch && !prs->words[i].in; i--) 02297 { 02298 if (!NONWORDTOKEN(prs->words[i].type)) 02299 { 02300 curlen++; 02301 stretch++; 02302 } 02303 posmarker = i; 02304 } 02305 /* cut back startpos till we find a non short token */ 02306 for (i = posmarker; i < startpos && (NOENDTOKEN(prs->words[i].type) || prs->words[i].len <= shortword); i++) 02307 { 02308 if (!NONWORDTOKEN(prs->words[i].type)) 02309 curlen--; 02310 } 02311 startpos = i; 02312 /* now stretch the endpos as much as possible */ 02313 posmarker = endpos; 02314 for (i = endpos + 1; i < prs->curwords && curlen < max_words && !prs->words[i].in; i++) 02315 { 02316 if (!NONWORDTOKEN(prs->words[i].type)) 02317 curlen++; 02318 posmarker = i; 02319 } 02320 /* cut back endpos till we find a non-short token */ 02321 for (i = posmarker; i > endpos && (NOENDTOKEN(prs->words[i].type) || prs->words[i].len <= shortword); i--) 02322 { 02323 if (!NONWORDTOKEN(prs->words[i].type)) 02324 curlen--; 02325 } 02326 endpos = i; 02327 } 02328 covers[minI].startpos = startpos; 02329 covers[minI].endpos = endpos; 02330 covers[minI].curlen = curlen; 02331 /* Mark the chosen fragments (covers) */ 02332 mark_fragment(prs, highlight, startpos, endpos); 02333 num_f++; 02334 /* exclude overlapping covers */ 02335 for (i = 0; i < numcovers; i++) 02336 { 02337 if (i != minI && ((covers[i].startpos >= covers[minI].startpos && covers[i].startpos <= covers[minI].endpos) || (covers[i].endpos >= covers[minI].startpos && covers[i].endpos <= covers[minI].endpos))) 02338 covers[i].excluded = 1; 02339 } 02340 } 02341 else 02342 break; 02343 } 02344 02345 /* show at least min_words we have not marked anything */ 02346 if (num_f <= 0) 02347 { 02348 startpos = endpos = curlen = 0; 02349 for (i = 0; i < prs->curwords && curlen < min_words; i++) 02350 { 02351 if (!NONWORDTOKEN(prs->words[i].type)) 02352 curlen++; 02353 endpos = i; 02354 } 02355 mark_fragment(prs, highlight, startpos, endpos); 02356 } 02357 pfree(covers); 02358 } 02359 02360 static void 02361 mark_hl_words(HeadlineParsedText *prs, TSQuery query, int highlight, 02362 int shortword, int min_words, int max_words) 02363 { 02364 int p = 0, 02365 q = 0; 02366 int bestb = -1, 02367 beste = -1; 02368 int bestlen = -1; 02369 int pose = 0, 02370 posb, 02371 poslen, 02372 curlen; 02373 02374 int i; 02375 02376 if (highlight == 0) 02377 { 02378 while (hlCover(prs, query, &p, &q)) 02379 { 02380 /* find cover len in words */ 02381 curlen = 0; 02382 poslen = 0; 02383 for (i = p; i <= q && curlen < max_words; i++) 02384 { 02385 if (!NONWORDTOKEN(prs->words[i].type)) 02386 curlen++; 02387 if (prs->words[i].item && !prs->words[i].repeated) 02388 poslen++; 02389 pose = i; 02390 } 02391 02392 if (poslen < bestlen && !(NOENDTOKEN(prs->words[beste].type) || prs->words[beste].len <= shortword)) 02393 { 02394 /* best already finded, so try one more cover */ 02395 p++; 02396 continue; 02397 } 02398 02399 posb = p; 02400 if (curlen < max_words) 02401 { /* find good end */ 02402 for (i = i - 1; i < prs->curwords && curlen < max_words; i++) 02403 { 02404 if (i != q) 02405 { 02406 if (!NONWORDTOKEN(prs->words[i].type)) 02407 curlen++; 02408 if (prs->words[i].item && !prs->words[i].repeated) 02409 poslen++; 02410 } 02411 pose = i; 02412 if (NOENDTOKEN(prs->words[i].type) || prs->words[i].len <= shortword) 02413 continue; 02414 if (curlen >= min_words) 02415 break; 02416 } 02417 if (curlen < min_words && i >= prs->curwords) 02418 { /* got end of text and our cover is shoter 02419 * than min_words */ 02420 for (i = p - 1; i >= 0; i--) 02421 { 02422 if (!NONWORDTOKEN(prs->words[i].type)) 02423 curlen++; 02424 if (prs->words[i].item && !prs->words[i].repeated) 02425 poslen++; 02426 if (curlen >= max_words) 02427 break; 02428 if (NOENDTOKEN(prs->words[i].type) || prs->words[i].len <= shortword) 02429 continue; 02430 if (curlen >= min_words) 02431 break; 02432 } 02433 posb = (i >= 0) ? i : 0; 02434 } 02435 } 02436 else 02437 { /* shorter cover :((( */ 02438 for (; curlen > min_words; i--) 02439 { 02440 if (!NONWORDTOKEN(prs->words[i].type)) 02441 curlen--; 02442 if (prs->words[i].item && !prs->words[i].repeated) 02443 poslen--; 02444 pose = i; 02445 if (NOENDTOKEN(prs->words[i].type) || prs->words[i].len <= shortword) 02446 continue; 02447 break; 02448 } 02449 } 02450 02451 if (bestlen < 0 || (poslen > bestlen && !(NOENDTOKEN(prs->words[pose].type) || prs->words[pose].len <= shortword)) || 02452 (bestlen >= 0 && !(NOENDTOKEN(prs->words[pose].type) || prs->words[pose].len <= shortword) && 02453 (NOENDTOKEN(prs->words[beste].type) || prs->words[beste].len <= shortword))) 02454 { 02455 bestb = posb; 02456 beste = pose; 02457 bestlen = poslen; 02458 } 02459 02460 p++; 02461 } 02462 02463 if (bestlen < 0) 02464 { 02465 curlen = 0; 02466 for (i = 0; i < prs->curwords && curlen < min_words; i++) 02467 { 02468 if (!NONWORDTOKEN(prs->words[i].type)) 02469 curlen++; 02470 pose = i; 02471 } 02472 bestb = 0; 02473 beste = pose; 02474 } 02475 } 02476 else 02477 { 02478 bestb = 0; 02479 beste = prs->curwords - 1; 02480 } 02481 02482 for (i = bestb; i <= beste; i++) 02483 { 02484 if (prs->words[i].item) 02485 prs->words[i].selected = 1; 02486 if (highlight == 0) 02487 { 02488 if (HLIDREPLACE(prs->words[i].type)) 02489 prs->words[i].replace = 1; 02490 else if (HLIDSKIP(prs->words[i].type)) 02491 prs->words[i].skip = 1; 02492 } 02493 else 02494 { 02495 if (XMLHLIDSKIP(prs->words[i].type)) 02496 prs->words[i].skip = 1; 02497 } 02498 02499 prs->words[i].in = (prs->words[i].repeated) ? 0 : 1; 02500 } 02501 02502 } 02503 02504 Datum 02505 prsd_headline(PG_FUNCTION_ARGS) 02506 { 02507 HeadlineParsedText *prs = (HeadlineParsedText *) PG_GETARG_POINTER(0); 02508 List *prsoptions = (List *) PG_GETARG_POINTER(1); 02509 TSQuery query = PG_GETARG_TSQUERY(2); 02510 02511 /* from opt + start and end tag */ 02512 int min_words = 15; 02513 int max_words = 35; 02514 int shortword = 3; 02515 int max_fragments = 0; 02516 int highlight = 0; 02517 ListCell *l; 02518 02519 /* config */ 02520 prs->startsel = NULL; 02521 prs->stopsel = NULL; 02522 foreach(l, prsoptions) 02523 { 02524 DefElem *defel = (DefElem *) lfirst(l); 02525 char *val = defGetString(defel); 02526 02527 if (pg_strcasecmp(defel->defname, "MaxWords") == 0) 02528 max_words = pg_atoi(val, sizeof(int32), 0); 02529 else if (pg_strcasecmp(defel->defname, "MinWords") == 0) 02530 min_words = pg_atoi(val, sizeof(int32), 0); 02531 else if (pg_strcasecmp(defel->defname, "ShortWord") == 0) 02532 shortword = pg_atoi(val, sizeof(int32), 0); 02533 else if (pg_strcasecmp(defel->defname, "MaxFragments") == 0) 02534 max_fragments = pg_atoi(val, sizeof(int32), 0); 02535 else if (pg_strcasecmp(defel->defname, "StartSel") == 0) 02536 prs->startsel = pstrdup(val); 02537 else if (pg_strcasecmp(defel->defname, "StopSel") == 0) 02538 prs->stopsel = pstrdup(val); 02539 else if (pg_strcasecmp(defel->defname, "FragmentDelimiter") == 0) 02540 prs->fragdelim = pstrdup(val); 02541 else if (pg_strcasecmp(defel->defname, "HighlightAll") == 0) 02542 highlight = (pg_strcasecmp(val, "1") == 0 || 02543 pg_strcasecmp(val, "on") == 0 || 02544 pg_strcasecmp(val, "true") == 0 || 02545 pg_strcasecmp(val, "t") == 0 || 02546 pg_strcasecmp(val, "y") == 0 || 02547 pg_strcasecmp(val, "yes") == 0); 02548 else 02549 ereport(ERROR, 02550 (errcode(ERRCODE_INVALID_PARAMETER_VALUE), 02551 errmsg("unrecognized headline parameter: \"%s\"", 02552 defel->defname))); 02553 } 02554 02555 if (highlight == 0) 02556 { 02557 if (min_words >= max_words) 02558 ereport(ERROR, 02559 (errcode(ERRCODE_INVALID_PARAMETER_VALUE), 02560 errmsg("MinWords should be less than MaxWords"))); 02561 if (min_words <= 0) 02562 ereport(ERROR, 02563 (errcode(ERRCODE_INVALID_PARAMETER_VALUE), 02564 errmsg("MinWords should be positive"))); 02565 if (shortword < 0) 02566 ereport(ERROR, 02567 (errcode(ERRCODE_INVALID_PARAMETER_VALUE), 02568 errmsg("ShortWord should be >= 0"))); 02569 if (max_fragments < 0) 02570 ereport(ERROR, 02571 (errcode(ERRCODE_INVALID_PARAMETER_VALUE), 02572 errmsg("MaxFragments should be >= 0"))); 02573 } 02574 02575 if (max_fragments == 0) 02576 /* call the default headline generator */ 02577 mark_hl_words(prs, query, highlight, shortword, min_words, max_words); 02578 else 02579 mark_hl_fragments(prs, query, highlight, shortword, min_words, max_words, max_fragments); 02580 02581 if (!prs->startsel) 02582 prs->startsel = pstrdup("<b>"); 02583 if (!prs->stopsel) 02584 prs->stopsel = pstrdup("</b>"); 02585 if (!prs->fragdelim) 02586 prs->fragdelim = pstrdup(" ... "); 02587 prs->startsellen = strlen(prs->startsel); 02588 prs->stopsellen = strlen(prs->stopsel); 02589 prs->fragdelimlen = strlen(prs->fragdelim); 02590 02591 PG_RETURN_POINTER(prs); 02592 }
1.7.1