SimGear: xmltok_impl.c Source File

00001 /*
00002 The contents of this file are subject to the Mozilla Public License
00003 Version 1.1 (the "License"); you may not use this file except in
00004 compliance with the License. You may obtain a copy of the License at
00005 http://www.mozilla.org/MPL/
00006 
00007 Software distributed under the License is distributed on an "AS IS"
00008 basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the
00009 License for the specific language governing rights and limitations
00010 under the License.
00011 
00012 The Original Code is expat.
00013 
00014 The Initial Developer of the Original Code is James Clark.
00015 Portions created by James Clark are Copyright (C) 1998, 1999
00016 James Clark. All Rights Reserved.
00017 
00018 Contributor(s):
00019 
00020 Alternatively, the contents of this file may be used under the terms
00021 of the GNU General Public License (the "GPL"), in which case the
00022 provisions of the GPL are applicable instead of those above.  If you
00023 wish to allow use of your version of this file only under the terms of
00024 the GPL and not to allow others to use your version of this file under
00025 the MPL, indicate your decision by deleting the provisions above and
00026 replace them with the notice and other provisions required by the
00027 GPL. If you do not delete the provisions above, a recipient may use
00028 your version of this file under either the MPL or the GPL.
00029 */
00030 
00031 #ifndef IS_INVALID_CHAR
00032 #define IS_INVALID_CHAR(enc, ptr, n) (0)
00033 #endif
00034 
00035 #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
00036     case BT_LEAD ## n: \
00037       if (end - ptr < n) \
00038         return XML_TOK_PARTIAL_CHAR; \
00039       if (IS_INVALID_CHAR(enc, ptr, n)) { \
00040         *(nextTokPtr) = (ptr); \
00041         return XML_TOK_INVALID; \
00042       } \
00043       ptr += n; \
00044       break;
00045 
00046 #define INVALID_CASES(ptr, nextTokPtr) \
00047   INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
00048   INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
00049   INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
00050   case BT_NONXML: \
00051   case BT_MALFORM: \
00052   case BT_TRAIL: \
00053     *(nextTokPtr) = (ptr); \
00054     return XML_TOK_INVALID;
00055 
00056 #define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
00057    case BT_LEAD ## n: \
00058      if (end - ptr < n) \
00059        return XML_TOK_PARTIAL_CHAR; \
00060      if (!IS_NAME_CHAR(enc, ptr, n)) { \
00061        *nextTokPtr = ptr; \
00062        return XML_TOK_INVALID; \
00063      } \
00064      ptr += n; \
00065      break;
00066 
00067 #define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
00068   case BT_NONASCII: \
00069     if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \
00070       *nextTokPtr = ptr; \
00071       return XML_TOK_INVALID; \
00072     } \
00073   case BT_NMSTRT: \
00074   case BT_HEX: \
00075   case BT_DIGIT: \
00076   case BT_NAME: \
00077   case BT_MINUS: \
00078     ptr += MINBPC(enc); \
00079     break; \
00080   CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
00081   CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
00082   CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)
00083 
00084 #define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
00085    case BT_LEAD ## n: \
00086      if (end - ptr < n) \
00087        return XML_TOK_PARTIAL_CHAR; \
00088      if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \
00089        *nextTokPtr = ptr; \
00090        return XML_TOK_INVALID; \
00091      } \
00092      ptr += n; \
00093      break;
00094 
00095 #define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
00096   case BT_NONASCII: \
00097     if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
00098       *nextTokPtr = ptr; \
00099       return XML_TOK_INVALID; \
00100     } \
00101   case BT_NMSTRT: \
00102   case BT_HEX: \
00103     ptr += MINBPC(enc); \
00104     break; \
00105   CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
00106   CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
00107   CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
00108 
00109 #ifndef PREFIX
00110 #define PREFIX(ident) ident
00111 #endif
00112 
00113 /* ptr points to character following "<!-" */
00114 
00115 static
00116 int PREFIX(scanComment)(const ENCODING *enc, const char *ptr, const char *end,
00117                         const char **nextTokPtr)
00118 {
00119   if (ptr != end) {
00120     if (!CHAR_MATCHES(enc, ptr, '-')) {
00121       *nextTokPtr = ptr;
00122       return XML_TOK_INVALID;
00123     }
00124     ptr += MINBPC(enc);
00125     while (ptr != end) {
00126       switch (BYTE_TYPE(enc, ptr)) {
00127       INVALID_CASES(ptr, nextTokPtr)
00128       case BT_MINUS:
00129         if ((ptr += MINBPC(enc)) == end)
00130           return XML_TOK_PARTIAL;
00131         if (CHAR_MATCHES(enc, ptr, '-')) {
00132           if ((ptr += MINBPC(enc)) == end)
00133             return XML_TOK_PARTIAL;
00134           if (!CHAR_MATCHES(enc, ptr, '>')) {
00135             *nextTokPtr = ptr;
00136             return XML_TOK_INVALID;
00137           }
00138           *nextTokPtr = ptr + MINBPC(enc);
00139           return XML_TOK_COMMENT;
00140         }
00141         break;
00142       default:
00143         ptr += MINBPC(enc);
00144         break;
00145       }
00146     }
00147   }
00148   return XML_TOK_PARTIAL;
00149 }
00150 
00151 /* ptr points to character following "<!" */
00152 
00153 static
00154 int PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end,
00155                      const char **nextTokPtr)
00156 {
00157   if (ptr == end)
00158     return XML_TOK_PARTIAL;
00159   switch (BYTE_TYPE(enc, ptr)) {
00160   case BT_MINUS:
00161     return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00162   case BT_LSQB:
00163     *nextTokPtr = ptr + MINBPC(enc);
00164     return XML_TOK_COND_SECT_OPEN;
00165   case BT_NMSTRT:
00166   case BT_HEX:
00167     ptr += MINBPC(enc);
00168     break;
00169   default:
00170     *nextTokPtr = ptr;
00171     return XML_TOK_INVALID;
00172   }
00173   while (ptr != end) {
00174     switch (BYTE_TYPE(enc, ptr)) {
00175     case BT_PERCNT:
00176       if (ptr + MINBPC(enc) == end)
00177         return XML_TOK_PARTIAL;
00178       /* don't allow <!ENTITY% foo "whatever"> */
00179       switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
00180       case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
00181         *nextTokPtr = ptr;
00182         return XML_TOK_INVALID;
00183       }
00184       /* fall through */
00185     case BT_S: case BT_CR: case BT_LF:
00186       *nextTokPtr = ptr;
00187       return XML_TOK_DECL_OPEN;
00188     case BT_NMSTRT:
00189     case BT_HEX:
00190       ptr += MINBPC(enc);
00191       break;
00192     default:
00193       *nextTokPtr = ptr;
00194       return XML_TOK_INVALID;
00195     }
00196   }
00197   return XML_TOK_PARTIAL;
00198 }
00199 
00200 static
00201 int PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end, int *tokPtr)
00202 {
00203   int upper = 0;
00204   *tokPtr = XML_TOK_PI;
00205   if (end - ptr != MINBPC(enc)*3)
00206     return 1;
00207   switch (BYTE_TO_ASCII(enc, ptr)) {
00208   case 'x':
00209     break;
00210   case 'X':
00211     upper = 1;
00212     break;
00213   default:
00214     return 1;
00215   }
00216   ptr += MINBPC(enc);
00217   switch (BYTE_TO_ASCII(enc, ptr)) {
00218   case 'm':
00219     break;
00220   case 'M':
00221     upper = 1;
00222     break;
00223   default:
00224     return 1;
00225   }
00226   ptr += MINBPC(enc);
00227   switch (BYTE_TO_ASCII(enc, ptr)) {
00228   case 'l':
00229     break;
00230   case 'L':
00231     upper = 1;
00232     break;
00233   default:
00234     return 1;
00235   }
00236   if (upper)
00237     return 0;
00238   *tokPtr = XML_TOK_XML_DECL;
00239   return 1;
00240 }
00241 
00242 /* ptr points to character following "<?" */
00243 
00244 static
00245 int PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end,
00246                    const char **nextTokPtr)
00247 {
00248   int tok;
00249   const char *target = ptr;
00250   if (ptr == end)
00251     return XML_TOK_PARTIAL;
00252   switch (BYTE_TYPE(enc, ptr)) {
00253   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00254   default:
00255     *nextTokPtr = ptr;
00256     return XML_TOK_INVALID;
00257   }
00258   while (ptr != end) {
00259     switch (BYTE_TYPE(enc, ptr)) {
00260     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00261     case BT_S: case BT_CR: case BT_LF:
00262       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
00263         *nextTokPtr = ptr;
00264         return XML_TOK_INVALID;
00265       }
00266       ptr += MINBPC(enc);
00267       while (ptr != end) {
00268         switch (BYTE_TYPE(enc, ptr)) {
00269         INVALID_CASES(ptr, nextTokPtr)
00270         case BT_QUEST:
00271           ptr += MINBPC(enc);
00272           if (ptr == end)
00273             return XML_TOK_PARTIAL;
00274           if (CHAR_MATCHES(enc, ptr, '>')) {
00275             *nextTokPtr = ptr + MINBPC(enc);
00276             return tok;
00277           }
00278           break;
00279         default:
00280           ptr += MINBPC(enc);
00281           break;
00282         }
00283       }
00284       return XML_TOK_PARTIAL;
00285     case BT_QUEST:
00286       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
00287         *nextTokPtr = ptr;
00288         return XML_TOK_INVALID;
00289       }
00290       ptr += MINBPC(enc);
00291       if (ptr == end)
00292         return XML_TOK_PARTIAL;
00293       if (CHAR_MATCHES(enc, ptr, '>')) {
00294         *nextTokPtr = ptr + MINBPC(enc);
00295         return tok;
00296       }
00297       /* fall through */
00298     default:
00299       *nextTokPtr = ptr;
00300       return XML_TOK_INVALID;
00301     }
00302   }
00303   return XML_TOK_PARTIAL;
00304 }
00305 
00306 
00307 static
00308 int PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end,
00309                              const char **nextTokPtr)
00310 {
00311   int i;
00312   /* CDATA[ */
00313   if (end - ptr < 6 * MINBPC(enc))
00314     return XML_TOK_PARTIAL;
00315   for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
00316     if (!CHAR_MATCHES(enc, ptr, "CDATA["[i])) {
00317       *nextTokPtr = ptr;
00318       return XML_TOK_INVALID;
00319     }
00320   }
00321   *nextTokPtr = ptr;
00322   return XML_TOK_CDATA_SECT_OPEN;
00323 }
00324 
00325 static
00326 int PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
00327                             const char **nextTokPtr)
00328 {
00329   if (ptr == end)
00330     return XML_TOK_NONE;
00331   if (MINBPC(enc) > 1) {
00332     size_t n = end - ptr;
00333     if (n & (MINBPC(enc) - 1)) {
00334       n &= ~(MINBPC(enc) - 1);
00335       if (n == 0)
00336         return XML_TOK_PARTIAL;
00337       end = ptr + n;
00338     }
00339   }
00340   switch (BYTE_TYPE(enc, ptr)) {
00341   case BT_RSQB:
00342     ptr += MINBPC(enc);
00343     if (ptr == end)
00344       return XML_TOK_PARTIAL;
00345     if (!CHAR_MATCHES(enc, ptr, ']'))
00346       break;
00347     ptr += MINBPC(enc);
00348     if (ptr == end)
00349       return XML_TOK_PARTIAL;
00350     if (!CHAR_MATCHES(enc, ptr, '>')) {
00351       ptr -= MINBPC(enc);
00352       break;
00353     }
00354     *nextTokPtr = ptr + MINBPC(enc);
00355     return XML_TOK_CDATA_SECT_CLOSE;
00356   case BT_CR:
00357     ptr += MINBPC(enc);
00358     if (ptr == end)
00359       return XML_TOK_PARTIAL;
00360     if (BYTE_TYPE(enc, ptr) == BT_LF)
00361       ptr += MINBPC(enc);
00362     *nextTokPtr = ptr;
00363     return XML_TOK_DATA_NEWLINE;
00364   case BT_LF:
00365     *nextTokPtr = ptr + MINBPC(enc);
00366     return XML_TOK_DATA_NEWLINE;
00367   INVALID_CASES(ptr, nextTokPtr)
00368   default:
00369     ptr += MINBPC(enc);
00370     break;
00371   }
00372   while (ptr != end) {
00373     switch (BYTE_TYPE(enc, ptr)) {
00374 #define LEAD_CASE(n) \
00375     case BT_LEAD ## n: \
00376       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
00377         *nextTokPtr = ptr; \
00378         return XML_TOK_DATA_CHARS; \
00379       } \
00380       ptr += n; \
00381       break;
00382     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
00383 #undef LEAD_CASE
00384     case BT_NONXML:
00385     case BT_MALFORM:
00386     case BT_TRAIL:
00387     case BT_CR:
00388     case BT_LF:
00389     case BT_RSQB:
00390       *nextTokPtr = ptr;
00391       return XML_TOK_DATA_CHARS;
00392     default:
00393       ptr += MINBPC(enc);
00394       break;
00395     }
00396   }
00397   *nextTokPtr = ptr;
00398   return XML_TOK_DATA_CHARS;
00399 }
00400 
00401 /* ptr points to character following "</" */
00402 
00403 static
00404 int PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end,
00405                        const char **nextTokPtr)
00406 {
00407   if (ptr == end)
00408     return XML_TOK_PARTIAL;
00409   switch (BYTE_TYPE(enc, ptr)) {
00410   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00411   default:
00412     *nextTokPtr = ptr;
00413     return XML_TOK_INVALID;
00414   }
00415   while (ptr != end) {
00416     switch (BYTE_TYPE(enc, ptr)) {
00417     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00418     case BT_S: case BT_CR: case BT_LF:
00419       for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
00420         switch (BYTE_TYPE(enc, ptr)) {
00421         case BT_S: case BT_CR: case BT_LF:
00422           break;
00423         case BT_GT:
00424           *nextTokPtr = ptr + MINBPC(enc);
00425           return XML_TOK_END_TAG;
00426         default:
00427           *nextTokPtr = ptr;
00428           return XML_TOK_INVALID;
00429         }
00430       }
00431       return XML_TOK_PARTIAL;
00432 #ifdef XML_NS
00433     case BT_COLON:
00434       /* no need to check qname syntax here, since end-tag must match exactly */
00435       ptr += MINBPC(enc);
00436       break;
00437 #endif
00438     case BT_GT:
00439       *nextTokPtr = ptr + MINBPC(enc);
00440       return XML_TOK_END_TAG;
00441     default:
00442       *nextTokPtr = ptr;
00443       return XML_TOK_INVALID;
00444     }
00445   }
00446   return XML_TOK_PARTIAL;
00447 }
00448 
00449 /* ptr points to character following "&#X" */
00450 
00451 static
00452 int PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end,
00453                            const char **nextTokPtr)
00454 {
00455   if (ptr != end) {
00456     switch (BYTE_TYPE(enc, ptr)) {
00457     case BT_DIGIT:
00458     case BT_HEX:
00459       break;
00460     default:
00461       *nextTokPtr = ptr;
00462       return XML_TOK_INVALID;
00463     }
00464     for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
00465       switch (BYTE_TYPE(enc, ptr)) {
00466       case BT_DIGIT:
00467       case BT_HEX:
00468         break;
00469       case BT_SEMI:
00470         *nextTokPtr = ptr + MINBPC(enc);
00471         return XML_TOK_CHAR_REF;
00472       default:
00473         *nextTokPtr = ptr;
00474         return XML_TOK_INVALID;
00475       }
00476     }
00477   }
00478   return XML_TOK_PARTIAL;
00479 }
00480 
00481 /* ptr points to character following "&#" */
00482 
00483 static
00484 int PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end,
00485                         const char **nextTokPtr)
00486 {
00487   if (ptr != end) {
00488     if (CHAR_MATCHES(enc, ptr, 'x'))
00489       return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00490     switch (BYTE_TYPE(enc, ptr)) {
00491     case BT_DIGIT:
00492       break;
00493     default:
00494       *nextTokPtr = ptr;
00495       return XML_TOK_INVALID;
00496     }
00497     for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
00498       switch (BYTE_TYPE(enc, ptr)) {
00499       case BT_DIGIT:
00500         break;
00501       case BT_SEMI:
00502         *nextTokPtr = ptr + MINBPC(enc);
00503         return XML_TOK_CHAR_REF;
00504       default:
00505         *nextTokPtr = ptr;
00506         return XML_TOK_INVALID;
00507       }
00508     }
00509   }
00510   return XML_TOK_PARTIAL;
00511 }
00512 
00513 /* ptr points to character following "&" */
00514 
00515 static
00516 int PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
00517                     const char **nextTokPtr)
00518 {
00519   if (ptr == end)
00520     return XML_TOK_PARTIAL;
00521   switch (BYTE_TYPE(enc, ptr)) {
00522   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00523   case BT_NUM:
00524     return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00525   default:
00526     *nextTokPtr = ptr;
00527     return XML_TOK_INVALID;
00528   }
00529   while (ptr != end) {
00530     switch (BYTE_TYPE(enc, ptr)) {
00531     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00532     case BT_SEMI:
00533       *nextTokPtr = ptr + MINBPC(enc);
00534       return XML_TOK_ENTITY_REF;
00535     default:
00536       *nextTokPtr = ptr;
00537       return XML_TOK_INVALID;
00538     }
00539   }
00540   return XML_TOK_PARTIAL;
00541 }
00542 
00543 /* ptr points to character following first character of attribute name */
00544 
00545 static
00546 int PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
00547                      const char **nextTokPtr)
00548 {
00549 #ifdef XML_NS
00550   int hadColon = 0;
00551 #endif
00552   while (ptr != end) {
00553     switch (BYTE_TYPE(enc, ptr)) {
00554     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00555 #ifdef XML_NS
00556     case BT_COLON:
00557       if (hadColon) {
00558         *nextTokPtr = ptr;
00559         return XML_TOK_INVALID;
00560       }
00561       hadColon = 1;
00562       ptr += MINBPC(enc);
00563       if (ptr == end)
00564         return XML_TOK_PARTIAL;
00565       switch (BYTE_TYPE(enc, ptr)) {
00566       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00567       default:
00568         *nextTokPtr = ptr;
00569         return XML_TOK_INVALID;
00570       }
00571       break;
00572 #endif
00573     case BT_S: case BT_CR: case BT_LF:
00574       for (;;) {
00575         int t;
00576 
00577         ptr += MINBPC(enc);
00578         if (ptr == end)
00579           return XML_TOK_PARTIAL;
00580         t = BYTE_TYPE(enc, ptr);
00581         if (t == BT_EQUALS)
00582           break;
00583         switch (t) {
00584         case BT_S:
00585         case BT_LF:
00586         case BT_CR:
00587           break;
00588         default:
00589           *nextTokPtr = ptr;
00590           return XML_TOK_INVALID;
00591         }
00592       }
00593     /* fall through */
00594     case BT_EQUALS:
00595       {
00596         int open;
00597 #ifdef XML_NS
00598         hadColon = 0;
00599 #endif
00600         for (;;) {
00601           
00602           ptr += MINBPC(enc);
00603           if (ptr == end)
00604             return XML_TOK_PARTIAL;
00605           open = BYTE_TYPE(enc, ptr);
00606           if (open == BT_QUOT || open == BT_APOS)
00607             break;
00608           switch (open) {
00609           case BT_S:
00610           case BT_LF:
00611           case BT_CR:
00612             break;
00613           default:
00614             *nextTokPtr = ptr;
00615             return XML_TOK_INVALID;
00616           }
00617         }
00618         ptr += MINBPC(enc);
00619         /* in attribute value */
00620         for (;;) {
00621           int t;
00622           if (ptr == end)
00623             return XML_TOK_PARTIAL;
00624           t = BYTE_TYPE(enc, ptr);
00625           if (t == open)
00626             break;
00627           switch (t) {
00628           INVALID_CASES(ptr, nextTokPtr)
00629           case BT_AMP:
00630             {
00631               int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
00632               if (tok <= 0) {
00633                 if (tok == XML_TOK_INVALID)
00634                   *nextTokPtr = ptr;
00635                 return tok;
00636               }
00637               break;
00638             }
00639           case BT_LT:
00640             *nextTokPtr = ptr;
00641             return XML_TOK_INVALID;
00642           default:
00643             ptr += MINBPC(enc);
00644             break;
00645           }
00646         }
00647         ptr += MINBPC(enc);
00648         if (ptr == end)
00649           return XML_TOK_PARTIAL;
00650         switch (BYTE_TYPE(enc, ptr)) {
00651         case BT_S:
00652         case BT_CR:
00653         case BT_LF:
00654           break;
00655         case BT_SOL:
00656           goto sol;
00657         case BT_GT:
00658           goto gt;
00659         default:
00660           *nextTokPtr = ptr;
00661           return XML_TOK_INVALID;
00662         }
00663         /* ptr points to closing quote */
00664         for (;;) {
00665           ptr += MINBPC(enc);
00666           if (ptr == end)
00667             return XML_TOK_PARTIAL;
00668           switch (BYTE_TYPE(enc, ptr)) {
00669           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00670           case BT_S: case BT_CR: case BT_LF:
00671             continue;
00672           case BT_GT:
00673           gt:
00674             *nextTokPtr = ptr + MINBPC(enc);
00675             return XML_TOK_START_TAG_WITH_ATTS;
00676           case BT_SOL:
00677           sol:
00678             ptr += MINBPC(enc);
00679             if (ptr == end)
00680               return XML_TOK_PARTIAL;
00681             if (!CHAR_MATCHES(enc, ptr, '>')) {
00682               *nextTokPtr = ptr;
00683               return XML_TOK_INVALID;
00684             }
00685             *nextTokPtr = ptr + MINBPC(enc);
00686             return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
00687           default:
00688             *nextTokPtr = ptr;
00689             return XML_TOK_INVALID;
00690           }
00691           break;
00692         }
00693         break;
00694       }
00695     default:
00696       *nextTokPtr = ptr;
00697       return XML_TOK_INVALID;
00698     }
00699   }
00700   return XML_TOK_PARTIAL;
00701 }
00702 
00703 /* ptr points to character following "<" */
00704 
00705 static
00706 int PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
00707                    const char **nextTokPtr)
00708 {
00709 #ifdef XML_NS
00710   int hadColon;
00711 #endif
00712   if (ptr == end)
00713     return XML_TOK_PARTIAL;
00714   switch (BYTE_TYPE(enc, ptr)) {
00715   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00716   case BT_EXCL:
00717     if ((ptr += MINBPC(enc)) == end)
00718       return XML_TOK_PARTIAL;
00719     switch (BYTE_TYPE(enc, ptr)) {
00720     case BT_MINUS:
00721       return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00722     case BT_LSQB:
00723       return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00724     }
00725     *nextTokPtr = ptr;
00726     return XML_TOK_INVALID;
00727   case BT_QUEST:
00728     return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00729   case BT_SOL:
00730     return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00731   default:
00732     *nextTokPtr = ptr;
00733     return XML_TOK_INVALID;
00734   }
00735 #ifdef XML_NS
00736   hadColon = 0;
00737 #endif
00738   /* we have a start-tag */
00739   while (ptr != end) {
00740     switch (BYTE_TYPE(enc, ptr)) {
00741     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00742 #ifdef XML_NS
00743     case BT_COLON:
00744       if (hadColon) {
00745         *nextTokPtr = ptr;
00746         return XML_TOK_INVALID;
00747       }
00748       hadColon = 1;
00749       ptr += MINBPC(enc);
00750       if (ptr == end)
00751         return XML_TOK_PARTIAL;
00752       switch (BYTE_TYPE(enc, ptr)) {
00753       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00754       default:
00755         *nextTokPtr = ptr;
00756         return XML_TOK_INVALID;
00757       }
00758       break;
00759 #endif
00760     case BT_S: case BT_CR: case BT_LF:
00761       {
00762         ptr += MINBPC(enc);
00763         while (ptr != end) {
00764           switch (BYTE_TYPE(enc, ptr)) {
00765           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00766           case BT_GT:
00767             goto gt;
00768           case BT_SOL:
00769             goto sol;
00770           case BT_S: case BT_CR: case BT_LF:
00771             ptr += MINBPC(enc);
00772             continue;
00773           default:
00774             *nextTokPtr = ptr;
00775             return XML_TOK_INVALID;
00776           }
00777           return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
00778         }
00779         return XML_TOK_PARTIAL;
00780       }
00781     case BT_GT:
00782     gt:
00783       *nextTokPtr = ptr + MINBPC(enc);
00784       return XML_TOK_START_TAG_NO_ATTS;
00785     case BT_SOL:
00786     sol:
00787       ptr += MINBPC(enc);
00788       if (ptr == end)
00789         return XML_TOK_PARTIAL;
00790       if (!CHAR_MATCHES(enc, ptr, '>')) {
00791         *nextTokPtr = ptr;
00792         return XML_TOK_INVALID;
00793       }
00794       *nextTokPtr = ptr + MINBPC(enc);
00795       return XML_TOK_EMPTY_ELEMENT_NO_ATTS;
00796     default:
00797       *nextTokPtr = ptr;
00798       return XML_TOK_INVALID;
00799     }
00800   }
00801   return XML_TOK_PARTIAL;
00802 }
00803 
00804 static
00805 int PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end,
00806                        const char **nextTokPtr)
00807 {
00808   if (ptr == end)
00809     return XML_TOK_NONE;
00810   if (MINBPC(enc) > 1) {
00811     size_t n = end - ptr;
00812     if (n & (MINBPC(enc) - 1)) {
00813       n &= ~(MINBPC(enc) - 1);
00814       if (n == 0)
00815         return XML_TOK_PARTIAL;
00816       end = ptr + n;
00817     }
00818   }
00819   switch (BYTE_TYPE(enc, ptr)) {
00820   case BT_LT:
00821     return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00822   case BT_AMP:
00823     return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
00824   case BT_CR:
00825     ptr += MINBPC(enc);
00826     if (ptr == end)
00827       return XML_TOK_TRAILING_CR;
00828     if (BYTE_TYPE(enc, ptr) == BT_LF)
00829       ptr += MINBPC(enc);
00830     *nextTokPtr = ptr;
00831     return XML_TOK_DATA_NEWLINE;
00832   case BT_LF:
00833     *nextTokPtr = ptr + MINBPC(enc);
00834     return XML_TOK_DATA_NEWLINE;
00835   case BT_RSQB:
00836     ptr += MINBPC(enc);
00837     if (ptr == end)
00838       return XML_TOK_TRAILING_RSQB;
00839     if (!CHAR_MATCHES(enc, ptr, ']'))
00840       break;
00841     ptr += MINBPC(enc);
00842     if (ptr == end)
00843       return XML_TOK_TRAILING_RSQB;
00844     if (!CHAR_MATCHES(enc, ptr, '>')) {
00845       ptr -= MINBPC(enc);
00846       break;
00847     }
00848     *nextTokPtr = ptr;
00849     return XML_TOK_INVALID;
00850   INVALID_CASES(ptr, nextTokPtr)
00851   default:
00852     ptr += MINBPC(enc);
00853     break;
00854   }
00855   while (ptr != end) {
00856     switch (BYTE_TYPE(enc, ptr)) {
00857 #define LEAD_CASE(n) \
00858     case BT_LEAD ## n: \
00859       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
00860         *nextTokPtr = ptr; \
00861         return XML_TOK_DATA_CHARS; \
00862       } \
00863       ptr += n; \
00864       break;
00865     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
00866 #undef LEAD_CASE
00867     case BT_RSQB:
00868       if (ptr + MINBPC(enc) != end) {
00869          if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ']')) {
00870            ptr += MINBPC(enc);
00871            break;
00872          }
00873          if (ptr + 2*MINBPC(enc) != end) {
00874            if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), '>')) {
00875              ptr += MINBPC(enc);
00876              break;
00877            }
00878            *nextTokPtr = ptr + 2*MINBPC(enc);
00879            return XML_TOK_INVALID;
00880          }
00881       }
00882       /* fall through */
00883     case BT_AMP:
00884     case BT_LT:
00885     case BT_NONXML:
00886     case BT_MALFORM:
00887     case BT_TRAIL:
00888     case BT_CR:
00889     case BT_LF:
00890       *nextTokPtr = ptr;
00891       return XML_TOK_DATA_CHARS;
00892     default:
00893       ptr += MINBPC(enc);
00894       break;
00895     }
00896   }
00897   *nextTokPtr = ptr;
00898   return XML_TOK_DATA_CHARS;
00899 }
00900 
00901 /* ptr points to character following "%" */
00902 
00903 static
00904 int PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end,
00905                         const char **nextTokPtr)
00906 {
00907   if (ptr == end)
00908     return XML_TOK_PARTIAL;
00909   switch (BYTE_TYPE(enc, ptr)) {
00910   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00911   case BT_S: case BT_LF: case BT_CR: case BT_PERCNT:
00912     *nextTokPtr = ptr;
00913     return XML_TOK_PERCENT;
00914   default:
00915     *nextTokPtr = ptr;
00916     return XML_TOK_INVALID;
00917   }
00918   while (ptr != end) {
00919     switch (BYTE_TYPE(enc, ptr)) {
00920     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00921     case BT_SEMI:
00922       *nextTokPtr = ptr + MINBPC(enc);
00923       return XML_TOK_PARAM_ENTITY_REF;
00924     default:
00925       *nextTokPtr = ptr;
00926       return XML_TOK_INVALID;
00927     }
00928   }
00929   return XML_TOK_PARTIAL;
00930 }
00931 
00932 static
00933 int PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end,
00934                           const char **nextTokPtr)
00935 {
00936   if (ptr == end)
00937     return XML_TOK_PARTIAL;
00938   switch (BYTE_TYPE(enc, ptr)) {
00939   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
00940   default:
00941     *nextTokPtr = ptr;
00942     return XML_TOK_INVALID;
00943   }
00944   while (ptr != end) {
00945     switch (BYTE_TYPE(enc, ptr)) {
00946     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
00947     case BT_CR: case BT_LF: case BT_S:
00948     case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
00949       *nextTokPtr = ptr;
00950       return XML_TOK_POUND_NAME;
00951     default:
00952       *nextTokPtr = ptr;
00953       return XML_TOK_INVALID;
00954     }
00955   }
00956   return XML_TOK_PARTIAL;
00957 }
00958 
00959 static
00960 int PREFIX(scanLit)(int open, const ENCODING *enc,
00961                     const char *ptr, const char *end,
00962                     const char **nextTokPtr)
00963 {
00964   while (ptr != end) {
00965     int t = BYTE_TYPE(enc, ptr);
00966     switch (t) {
00967     INVALID_CASES(ptr, nextTokPtr)
00968     case BT_QUOT:
00969     case BT_APOS:
00970       ptr += MINBPC(enc);
00971       if (t != open)
00972         break;
00973       if (ptr == end)
00974         return XML_TOK_PARTIAL;
00975       *nextTokPtr = ptr;
00976       switch (BYTE_TYPE(enc, ptr)) {
00977       case BT_S: case BT_CR: case BT_LF:
00978       case BT_GT: case BT_PERCNT: case BT_LSQB:
00979         return XML_TOK_LITERAL;
00980       default:
00981         return XML_TOK_INVALID;
00982       }
00983     default:
00984       ptr += MINBPC(enc);
00985       break;
00986     }
00987   }
00988   return XML_TOK_PARTIAL;
00989 }
00990 
00991 static
00992 int PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
00993                       const char **nextTokPtr)
00994 {
00995   int tok;
00996   if (ptr == end)
00997     return XML_TOK_NONE;
00998   if (MINBPC(enc) > 1) {
00999     size_t n = end - ptr;
01000     if (n & (MINBPC(enc) - 1)) {
01001       n &= ~(MINBPC(enc) - 1);
01002       if (n == 0)
01003         return XML_TOK_PARTIAL;
01004       end = ptr + n;
01005     }
01006   }
01007   switch (BYTE_TYPE(enc, ptr)) {
01008   case BT_QUOT:
01009     return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
01010   case BT_APOS:
01011     return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
01012   case BT_LT:
01013     {
01014       ptr += MINBPC(enc);
01015       if (ptr == end)
01016         return XML_TOK_PARTIAL;
01017       switch (BYTE_TYPE(enc, ptr)) {
01018       case BT_EXCL:
01019         return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01020       case BT_QUEST:
01021         return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01022       case BT_NMSTRT:
01023       case BT_HEX:
01024       case BT_NONASCII:
01025       case BT_LEAD2:
01026       case BT_LEAD3:
01027       case BT_LEAD4:
01028         *nextTokPtr = ptr - MINBPC(enc);
01029         return XML_TOK_INSTANCE_START;
01030       }
01031       *nextTokPtr = ptr;
01032       return XML_TOK_INVALID;
01033     }
01034   case BT_CR:
01035     if (ptr + MINBPC(enc) == end)
01036       return XML_TOK_TRAILING_CR;
01037     /* fall through */
01038   case BT_S: case BT_LF:
01039     for (;;) {
01040       ptr += MINBPC(enc);
01041       if (ptr == end)
01042         break;
01043       switch (BYTE_TYPE(enc, ptr)) {
01044       case BT_S: case BT_LF:
01045         break;
01046       case BT_CR:
01047         /* don't split CR/LF pair */
01048         if (ptr + MINBPC(enc) != end)
01049           break;
01050         /* fall through */
01051       default:
01052         *nextTokPtr = ptr;
01053         return XML_TOK_PROLOG_S;
01054       }
01055     }
01056     *nextTokPtr = ptr;
01057     return XML_TOK_PROLOG_S;
01058   case BT_PERCNT:
01059     return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01060   case BT_COMMA:
01061     *nextTokPtr = ptr + MINBPC(enc);
01062     return XML_TOK_COMMA;
01063   case BT_LSQB:
01064     *nextTokPtr = ptr + MINBPC(enc);
01065     return XML_TOK_OPEN_BRACKET;
01066   case BT_RSQB:
01067     ptr += MINBPC(enc);
01068     if (ptr == end)
01069       return XML_TOK_PARTIAL;
01070     if (CHAR_MATCHES(enc, ptr, ']')) {
01071       if (ptr + MINBPC(enc) == end)
01072         return XML_TOK_PARTIAL;
01073       if (CHAR_MATCHES(enc, ptr + MINBPC(enc), '>')) {
01074         *nextTokPtr = ptr + 2*MINBPC(enc);
01075         return XML_TOK_COND_SECT_CLOSE;
01076       }
01077     }
01078     *nextTokPtr = ptr;
01079     return XML_TOK_CLOSE_BRACKET;
01080   case BT_LPAR:
01081     *nextTokPtr = ptr + MINBPC(enc);
01082     return XML_TOK_OPEN_PAREN;
01083   case BT_RPAR:
01084     ptr += MINBPC(enc);
01085     if (ptr == end)
01086       return XML_TOK_PARTIAL;
01087     switch (BYTE_TYPE(enc, ptr)) {
01088     case BT_AST:
01089       *nextTokPtr = ptr + MINBPC(enc);
01090       return XML_TOK_CLOSE_PAREN_ASTERISK;
01091     case BT_QUEST:
01092       *nextTokPtr = ptr + MINBPC(enc);
01093       return XML_TOK_CLOSE_PAREN_QUESTION;
01094     case BT_PLUS:
01095       *nextTokPtr = ptr + MINBPC(enc);
01096       return XML_TOK_CLOSE_PAREN_PLUS;
01097     case BT_CR: case BT_LF: case BT_S:
01098     case BT_GT: case BT_COMMA: case BT_VERBAR:
01099     case BT_RPAR:
01100       *nextTokPtr = ptr;
01101       return XML_TOK_CLOSE_PAREN;
01102     }
01103     *nextTokPtr = ptr;
01104     return XML_TOK_INVALID;
01105   case BT_VERBAR:
01106     *nextTokPtr = ptr + MINBPC(enc);
01107     return XML_TOK_OR;
01108   case BT_GT:
01109     *nextTokPtr = ptr + MINBPC(enc);
01110     return XML_TOK_DECL_CLOSE;
01111   case BT_NUM:
01112     return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01113 #define LEAD_CASE(n) \
01114   case BT_LEAD ## n: \
01115     if (end - ptr < n) \
01116       return XML_TOK_PARTIAL_CHAR; \
01117     if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
01118       ptr += n; \
01119       tok = XML_TOK_NAME; \
01120       break; \
01121     } \
01122     if (IS_NAME_CHAR(enc, ptr, n)) { \
01123       ptr += n; \
01124       tok = XML_TOK_NMTOKEN; \
01125       break; \
01126     } \
01127     *nextTokPtr = ptr; \
01128     return XML_TOK_INVALID;
01129     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01130 #undef LEAD_CASE
01131   case BT_NMSTRT:
01132   case BT_HEX:
01133     tok = XML_TOK_NAME;
01134     ptr += MINBPC(enc);
01135     break;
01136   case BT_DIGIT:
01137   case BT_NAME:
01138   case BT_MINUS:
01139 #ifdef XML_NS
01140   case BT_COLON:
01141 #endif
01142     tok = XML_TOK_NMTOKEN;
01143     ptr += MINBPC(enc);
01144     break;
01145   case BT_NONASCII:
01146     if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) {
01147       ptr += MINBPC(enc);
01148       tok = XML_TOK_NAME;
01149       break;
01150     }
01151     if (IS_NAME_CHAR_MINBPC(enc, ptr)) {
01152       ptr += MINBPC(enc);
01153       tok = XML_TOK_NMTOKEN;
01154       break;
01155     }
01156     /* fall through */
01157   default:
01158     *nextTokPtr = ptr;
01159     return XML_TOK_INVALID;
01160   }
01161   while (ptr != end) {
01162     switch (BYTE_TYPE(enc, ptr)) {
01163     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
01164     case BT_GT: case BT_RPAR: case BT_COMMA:
01165     case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
01166     case BT_S: case BT_CR: case BT_LF:
01167       *nextTokPtr = ptr;
01168       return tok;
01169 #ifdef XML_NS
01170     case BT_COLON:
01171       ptr += MINBPC(enc);
01172       switch (tok) {
01173       case XML_TOK_NAME:
01174         if (ptr == end)
01175           return XML_TOK_PARTIAL;
01176         tok = XML_TOK_PREFIXED_NAME;
01177         switch (BYTE_TYPE(enc, ptr)) {
01178         CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
01179         default:
01180           tok = XML_TOK_NMTOKEN;
01181           break;
01182         }
01183         break;
01184       case XML_TOK_PREFIXED_NAME:
01185         tok = XML_TOK_NMTOKEN;
01186         break;
01187       }
01188       break;
01189 #endif
01190     case BT_PLUS:
01191       if (tok == XML_TOK_NMTOKEN)  {
01192         *nextTokPtr = ptr;
01193         return XML_TOK_INVALID;
01194       }
01195       *nextTokPtr = ptr + MINBPC(enc);
01196       return XML_TOK_NAME_PLUS;
01197     case BT_AST:
01198       if (tok == XML_TOK_NMTOKEN)  {
01199         *nextTokPtr = ptr;
01200         return XML_TOK_INVALID;
01201       }
01202       *nextTokPtr = ptr + MINBPC(enc);
01203       return XML_TOK_NAME_ASTERISK;
01204     case BT_QUEST:
01205       if (tok == XML_TOK_NMTOKEN)  {
01206         *nextTokPtr = ptr;
01207         return XML_TOK_INVALID;
01208       }
01209       *nextTokPtr = ptr + MINBPC(enc);
01210       return XML_TOK_NAME_QUESTION;
01211     default:
01212       *nextTokPtr = ptr;
01213       return XML_TOK_INVALID;
01214     }
01215   }
01216   return XML_TOK_PARTIAL;
01217 }
01218 
01219 static
01220 int PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, const char *end,
01221                               const char **nextTokPtr)
01222 {
01223   const char *start;
01224   if (ptr == end)
01225     return XML_TOK_NONE;
01226   start = ptr;
01227   while (ptr != end) {
01228     switch (BYTE_TYPE(enc, ptr)) {
01229 #define LEAD_CASE(n) \
01230     case BT_LEAD ## n: ptr += n; break;
01231     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01232 #undef LEAD_CASE
01233     case BT_AMP:
01234       if (ptr == start)
01235         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01236       *nextTokPtr = ptr;
01237       return XML_TOK_DATA_CHARS;
01238     case BT_LT:
01239       /* this is for inside entity references */
01240       *nextTokPtr = ptr;
01241       return XML_TOK_INVALID;
01242     case BT_LF:
01243       if (ptr == start) {
01244         *nextTokPtr = ptr + MINBPC(enc);
01245         return XML_TOK_DATA_NEWLINE;
01246       }
01247       *nextTokPtr = ptr;
01248       return XML_TOK_DATA_CHARS;
01249     case BT_CR:
01250       if (ptr == start) {
01251         ptr += MINBPC(enc);
01252         if (ptr == end)
01253           return XML_TOK_TRAILING_CR;
01254         if (BYTE_TYPE(enc, ptr) == BT_LF)
01255           ptr += MINBPC(enc);
01256         *nextTokPtr = ptr;
01257         return XML_TOK_DATA_NEWLINE;
01258       }
01259       *nextTokPtr = ptr;
01260       return XML_TOK_DATA_CHARS;
01261     case BT_S:
01262       if (ptr == start) {
01263         *nextTokPtr = ptr + MINBPC(enc);
01264         return XML_TOK_ATTRIBUTE_VALUE_S;
01265       }
01266       *nextTokPtr = ptr;
01267       return XML_TOK_DATA_CHARS;
01268     default:
01269       ptr += MINBPC(enc);
01270       break;
01271     }
01272   }
01273   *nextTokPtr = ptr;
01274   return XML_TOK_DATA_CHARS;
01275 }
01276 
01277 static
01278 int PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, const char *end,
01279                            const char **nextTokPtr)
01280 {
01281   const char *start;
01282   if (ptr == end)
01283     return XML_TOK_NONE;
01284   start = ptr;
01285   while (ptr != end) {
01286     switch (BYTE_TYPE(enc, ptr)) {
01287 #define LEAD_CASE(n) \
01288     case BT_LEAD ## n: ptr += n; break;
01289     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01290 #undef LEAD_CASE
01291     case BT_AMP:
01292       if (ptr == start)
01293         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01294       *nextTokPtr = ptr;
01295       return XML_TOK_DATA_CHARS;
01296     case BT_PERCNT:
01297       if (ptr == start)
01298         return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
01299       *nextTokPtr = ptr;
01300       return XML_TOK_DATA_CHARS;
01301     case BT_LF:
01302       if (ptr == start) {
01303         *nextTokPtr = ptr + MINBPC(enc);
01304         return XML_TOK_DATA_NEWLINE;
01305       }
01306       *nextTokPtr = ptr;
01307       return XML_TOK_DATA_CHARS;
01308     case BT_CR:
01309       if (ptr == start) {
01310         ptr += MINBPC(enc);
01311         if (ptr == end)
01312           return XML_TOK_TRAILING_CR;
01313         if (BYTE_TYPE(enc, ptr) == BT_LF)
01314           ptr += MINBPC(enc);
01315         *nextTokPtr = ptr;
01316         return XML_TOK_DATA_NEWLINE;
01317       }
01318       *nextTokPtr = ptr;
01319       return XML_TOK_DATA_CHARS;
01320     default:
01321       ptr += MINBPC(enc);
01322       break;
01323     }
01324   }
01325   *nextTokPtr = ptr;
01326   return XML_TOK_DATA_CHARS;
01327 }
01328 
01329 static
01330 int PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
01331                        const char **badPtr)
01332 {
01333   ptr += MINBPC(enc);
01334   end -= MINBPC(enc);
01335   for (; ptr != end; ptr += MINBPC(enc)) {
01336     switch (BYTE_TYPE(enc, ptr)) {
01337     case BT_DIGIT:
01338     case BT_HEX:
01339     case BT_MINUS:
01340     case BT_APOS:
01341     case BT_LPAR:
01342     case BT_RPAR:
01343     case BT_PLUS:
01344     case BT_COMMA:
01345     case BT_SOL:
01346     case BT_EQUALS:
01347     case BT_QUEST:
01348     case BT_CR:
01349     case BT_LF:
01350     case BT_SEMI:
01351     case BT_EXCL:
01352     case BT_AST:
01353     case BT_PERCNT:
01354     case BT_NUM:
01355 #ifdef XML_NS
01356     case BT_COLON:
01357 #endif
01358       break;
01359     case BT_S:
01360       if (CHAR_MATCHES(enc, ptr, '\t')) {
01361         *badPtr = ptr;
01362         return 0;
01363       }
01364       break;
01365     case BT_NAME:
01366     case BT_NMSTRT:
01367       if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f))
01368         break;
01369     default:
01370       switch (BYTE_TO_ASCII(enc, ptr)) {
01371       case 0x24: /* $ */
01372       case 0x40: /* @ */
01373         break;
01374       default:
01375         *badPtr = ptr;
01376         return 0;
01377       }
01378       break;
01379     }
01380   }
01381   return 1;
01382 }
01383 
01384 /* This must only be called for a well-formed start-tag or empty element tag.
01385 Returns the number of attributes.  Pointers to the first attsMax attributes 
01386 are stored in atts. */
01387 
01388 static
01389 int PREFIX(getAtts)(const ENCODING *enc, const char *ptr,
01390                     int attsMax, ATTRIBUTE *atts)
01391 {
01392   enum { other, inName, inValue } state = inName;
01393   int nAtts = 0;
01394   int open = 0;
01395 
01396   for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) {
01397     switch (BYTE_TYPE(enc, ptr)) {
01398 #define START_NAME \
01399       if (state == other) { \
01400         if (nAtts < attsMax) { \
01401           atts[nAtts].name = ptr; \
01402           atts[nAtts].normalized = 1; \
01403         } \
01404         state = inName; \
01405       }
01406 #define LEAD_CASE(n) \
01407     case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break;
01408     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01409 #undef LEAD_CASE
01410     case BT_NONASCII:
01411     case BT_NMSTRT:
01412     case BT_HEX:
01413       START_NAME
01414       break;
01415 #undef START_NAME
01416     case BT_QUOT:
01417       if (state != inValue) {
01418         if (nAtts < attsMax)
01419           atts[nAtts].valuePtr = ptr + MINBPC(enc);
01420         state = inValue;
01421         open = BT_QUOT;
01422       }
01423       else if (open == BT_QUOT) {
01424         state = other;
01425         if (nAtts < attsMax)
01426           atts[nAtts].valueEnd = ptr;
01427         nAtts++;
01428       }
01429       break;
01430     case BT_APOS:
01431       if (state != inValue) {
01432         if (nAtts < attsMax)
01433           atts[nAtts].valuePtr = ptr + MINBPC(enc);
01434         state = inValue;
01435         open = BT_APOS;
01436       }
01437       else if (open == BT_APOS) {
01438         state = other;
01439         if (nAtts < attsMax)
01440           atts[nAtts].valueEnd = ptr;
01441         nAtts++;
01442       }
01443       break;
01444     case BT_AMP:
01445       if (nAtts < attsMax)
01446         atts[nAtts].normalized = 0;
01447       break;
01448     case BT_S:
01449       if (state == inName)
01450         state = other;
01451       else if (state == inValue
01452                && nAtts < attsMax
01453                && atts[nAtts].normalized
01454                && (ptr == atts[nAtts].valuePtr
01455                    || BYTE_TO_ASCII(enc, ptr) != ' '
01456                    || BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ' '
01457                    || BYTE_TYPE(enc, ptr + MINBPC(enc)) == open))
01458         atts[nAtts].normalized = 0;
01459       break;
01460     case BT_CR: case BT_LF:
01461       /* This case ensures that the first attribute name is counted
01462          Apart from that we could just change state on the quote. */
01463       if (state == inName)
01464         state = other;
01465       else if (state == inValue && nAtts < attsMax)
01466         atts[nAtts].normalized = 0;
01467       break;
01468     case BT_GT:
01469     case BT_SOL:
01470       if (state != inValue)
01471         return nAtts;
01472       break;
01473     default:
01474       break;
01475     }
01476   }
01477   /* not reached */
01478 }
01479 
01480 static
01481 int PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr)
01482 {
01483   int result = 0;
01484   /* skip &# */
01485   ptr += 2*MINBPC(enc);
01486   if (CHAR_MATCHES(enc, ptr, 'x')) {
01487     for (ptr += MINBPC(enc); !CHAR_MATCHES(enc, ptr, ';'); ptr += MINBPC(enc)) {
01488       int c = BYTE_TO_ASCII(enc, ptr);
01489       switch (c) {
01490       case '0': case '1': case '2': case '3': case '4':
01491       case '5': case '6': case '7': case '8': case '9':
01492         result <<= 4;
01493         result |= (c - '0');
01494         break;
01495       case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
01496         result <<= 4;
01497         result += 10 + (c - 'A');
01498         break;
01499       case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
01500         result <<= 4;
01501         result += 10 + (c - 'a');
01502         break;
01503       }
01504       if (result >= 0x110000)
01505         return -1;
01506     }
01507   }
01508   else {
01509     for (; !CHAR_MATCHES(enc, ptr, ';'); ptr += MINBPC(enc)) {
01510       int c = BYTE_TO_ASCII(enc, ptr);
01511       result *= 10;
01512       result += (c - '0');
01513       if (result >= 0x110000)
01514         return -1;
01515     }
01516   }
01517   return checkCharRefNumber(result);
01518 }
01519 
01520 static
01521 int PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr, const char *end)
01522 {
01523   switch ((end - ptr)/MINBPC(enc)) {
01524   case 2:
01525     if (CHAR_MATCHES(enc, ptr + MINBPC(enc), 't')) {
01526       switch (BYTE_TO_ASCII(enc, ptr)) {
01527       case 'l':
01528         return '<';
01529       case 'g':
01530         return '>';
01531       }
01532     }
01533     break;
01534   case 3:
01535     if (CHAR_MATCHES(enc, ptr, 'a')) {
01536       ptr += MINBPC(enc);
01537       if (CHAR_MATCHES(enc, ptr, 'm')) {
01538         ptr += MINBPC(enc);
01539         if (CHAR_MATCHES(enc, ptr, 'p'))
01540           return '&';
01541       }
01542     }
01543     break;
01544   case 4:
01545     switch (BYTE_TO_ASCII(enc, ptr)) {
01546     case 'q':
01547       ptr += MINBPC(enc);
01548       if (CHAR_MATCHES(enc, ptr, 'u')) {
01549         ptr += MINBPC(enc);
01550         if (CHAR_MATCHES(enc, ptr, 'o')) {
01551           ptr += MINBPC(enc);
01552           if (CHAR_MATCHES(enc, ptr, 't'))
01553             return '"';
01554         }
01555       }
01556       break;
01557     case 'a':
01558       ptr += MINBPC(enc);
01559       if (CHAR_MATCHES(enc, ptr, 'p')) {
01560         ptr += MINBPC(enc);
01561         if (CHAR_MATCHES(enc, ptr, 'o')) {
01562           ptr += MINBPC(enc);
01563           if (CHAR_MATCHES(enc, ptr, 's'))
01564             return '\'';
01565         }
01566       }
01567       break;
01568     }
01569   }
01570   return 0;
01571 }
01572 
01573 static
01574 int PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2)
01575 {
01576   for (;;) {
01577     switch (BYTE_TYPE(enc, ptr1)) {
01578 #define LEAD_CASE(n) \
01579     case BT_LEAD ## n: \
01580       if (*ptr1++ != *ptr2++) \
01581         return 0;
01582     LEAD_CASE(4) LEAD_CASE(3) LEAD_CASE(2)
01583 #undef LEAD_CASE
01584       /* fall through */
01585       if (*ptr1++ != *ptr2++)
01586         return 0;
01587       break;
01588     case BT_NONASCII:
01589     case BT_NMSTRT:
01590 #ifdef XML_NS
01591     case BT_COLON:
01592 #endif
01593     case BT_HEX:
01594     case BT_DIGIT:
01595     case BT_NAME:
01596     case BT_MINUS:
01597       if (*ptr2++ != *ptr1++)
01598         return 0;
01599       if (MINBPC(enc) > 1) {
01600         if (*ptr2++ != *ptr1++)
01601           return 0;
01602         if (MINBPC(enc) > 2) {
01603           if (*ptr2++ != *ptr1++)
01604             return 0;
01605           if (MINBPC(enc) > 3) {
01606             if (*ptr2++ != *ptr1++)
01607               return 0;
01608           }
01609         }
01610       }
01611       break;
01612     default:
01613       if (MINBPC(enc) == 1 && *ptr1 == *ptr2)
01614         return 1;
01615       switch (BYTE_TYPE(enc, ptr2)) {
01616       case BT_LEAD2:
01617       case BT_LEAD3:
01618       case BT_LEAD4:
01619       case BT_NONASCII:
01620       case BT_NMSTRT:
01621 #ifdef XML_NS
01622       case BT_COLON:
01623 #endif
01624       case BT_HEX:
01625       case BT_DIGIT:
01626       case BT_NAME:
01627       case BT_MINUS:
01628         return 0;
01629       default:
01630         return 1;
01631       }
01632     }
01633   }
01634   /* not reached */
01635 }
01636 
01637 static
01638 int PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1, const char *ptr2)
01639 {
01640   for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
01641     if (!CHAR_MATCHES(enc, ptr1, *ptr2))
01642       return 0;
01643   }
01644   switch (BYTE_TYPE(enc, ptr1)) {
01645   case BT_LEAD2:
01646   case BT_LEAD3:
01647   case BT_LEAD4:
01648   case BT_NONASCII:
01649   case BT_NMSTRT:
01650 #ifdef XML_NS
01651   case BT_COLON:
01652 #endif
01653   case BT_HEX:
01654   case BT_DIGIT:
01655   case BT_NAME:
01656   case BT_MINUS:
01657     return 0;
01658   default:
01659     return 1;
01660   }
01661 }
01662 
01663 static
01664 int PREFIX(nameLength)(const ENCODING *enc, const char *ptr)
01665 {
01666   const char *start = ptr;
01667   for (;;) {
01668     switch (BYTE_TYPE(enc, ptr)) {
01669 #define LEAD_CASE(n) \
01670     case BT_LEAD ## n: ptr += n; break;
01671     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01672 #undef LEAD_CASE
01673     case BT_NONASCII:
01674     case BT_NMSTRT:
01675 #ifdef XML_NS
01676     case BT_COLON:
01677 #endif
01678     case BT_HEX:
01679     case BT_DIGIT:
01680     case BT_NAME:
01681     case BT_MINUS:
01682       ptr += MINBPC(enc);
01683       break;
01684     default:
01685       return ptr - start;
01686     }
01687   }
01688 }
01689 
01690 static
01691 const char *PREFIX(skipS)(const ENCODING *enc, const char *ptr)
01692 {
01693   for (;;) {
01694     switch (BYTE_TYPE(enc, ptr)) {
01695     case BT_LF:
01696     case BT_CR:
01697     case BT_S:
01698       ptr += MINBPC(enc);
01699       break;
01700     default:
01701       return ptr;
01702     }
01703   }
01704 }
01705 
01706 static
01707 void PREFIX(updatePosition)(const ENCODING *enc,
01708                             const char *ptr,
01709                             const char *end,
01710                             POSITION *pos)
01711 {
01712   while (ptr != end) {
01713     switch (BYTE_TYPE(enc, ptr)) {
01714 #define LEAD_CASE(n) \
01715     case BT_LEAD ## n: \
01716       ptr += n; \
01717       break;
01718     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
01719 #undef LEAD_CASE
01720     case BT_LF:
01721       pos->columnNumber = (unsigned)-1;
01722       pos->lineNumber++;
01723       ptr += MINBPC(enc);
01724       break;
01725     case BT_CR:
01726       pos->lineNumber++;
01727       ptr += MINBPC(enc);
01728       if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF)
01729         ptr += MINBPC(enc);
01730       pos->columnNumber = (unsigned)-1;
01731       break;
01732     default:
01733       ptr += MINBPC(enc);
01734       break;
01735     }
01736     pos->columnNumber++;
01737   }
01738 }
01739 
01740 #undef DO_LEAD_CASE
01741 #undef MULTIBYTE_CASES
01742 #undef INVALID_CASES
01743 #undef CHECK_NAME_CASE
01744 #undef CHECK_NAME_CASES
01745 #undef CHECK_NMSTRT_CASE
01746 #undef CHECK_NMSTRT_CASES