contrib/src/xml/expat/xmltok/xmltok.h

   1 /*
   2 Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   3 See the file copying.txt for copying permission.
   4 */
   5
   6 #ifndef XmlTok_INCLUDED
   7 #define XmlTok_INCLUDED 1
   8
   9 #ifdef __cplusplus
  10 extern "C" {
  11 #endif
  12
  13 #ifndef XMLTOKAPI
  14 #define XMLTOKAPI /* as nothing */
  15 #endif
  16
  17 /* The following token may be returned by XmlContentTok */
  18 #define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be start of
  19                                     illegal ]]> sequence */
  20 /* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
  21 #define XML_TOK_NONE -4    /* The string to be scanned is empty */
  22 #define XML_TOK_TRAILING_CR -3 /* A CR at the end of the scan;
  23                                   might be part of CRLF sequence */
  24 #define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */
  25 #define XML_TOK_PARTIAL -1 /* only part of a token */
  26 #define XML_TOK_INVALID 0
  27
  28 /* The following tokens are returned by XmlContentTok; some are also
  29   returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok */
  30
  31 #define XML_TOK_START_TAG_WITH_ATTS 1
  32 #define XML_TOK_START_TAG_NO_ATTS 2
  33 #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
  34 #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
  35 #define XML_TOK_END_TAG 5
  36 #define XML_TOK_DATA_CHARS 6
  37 #define XML_TOK_DATA_NEWLINE 7
  38 #define XML_TOK_CDATA_SECT_OPEN 8
  39 #define XML_TOK_ENTITY_REF 9
  40 #define XML_TOK_CHAR_REF 10     /* numeric character reference */
  41
  42 /* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
  43 #define XML_TOK_PI 11      /* processing instruction */
  44 #define XML_TOK_XML_DECL 12 /* XML decl or text decl */
  45 #define XML_TOK_COMMENT 13
  46 #define XML_TOK_BOM 14     /* Byte order mark */
  47
  48 /* The following tokens are returned only by XmlPrologTok */
  49 #define XML_TOK_PROLOG_S 15
  50 #define XML_TOK_DECL_OPEN 16 /* <!foo */
  51 #define XML_TOK_DECL_CLOSE 17 /* > */
  52 #define XML_TOK_NAME 18
  53 #define XML_TOK_NMTOKEN 19
  54 #define XML_TOK_POUND_NAME 20 /* #name */
  55 #define XML_TOK_OR 21 /* | */
  56 #define XML_TOK_PERCENT 22
  57 #define XML_TOK_OPEN_PAREN 23
  58 #define XML_TOK_CLOSE_PAREN 24
  59 #define XML_TOK_OPEN_BRACKET 25
  60 #define XML_TOK_CLOSE_BRACKET 26
  61 #define XML_TOK_LITERAL 27
  62 #define XML_TOK_PARAM_ENTITY_REF 28
  63 #define XML_TOK_INSTANCE_START 29
  64
  65 /* The following occur only in element type declarations */
  66 #define XML_TOK_NAME_QUESTION 30 /* name? */
  67 #define XML_TOK_NAME_ASTERISK 31 /* name* */
  68 #define XML_TOK_NAME_PLUS 32 /* name+ */
  69 #define XML_TOK_COND_SECT_OPEN 33 /* <![ */
  70 #define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */
  71 #define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */
  72 #define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */
  73 #define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */
  74 #define XML_TOK_COMMA 38
  75
  76 /* The following token is returned only by XmlAttributeValueTok */
  77 #define XML_TOK_ATTRIBUTE_VALUE_S 39
  78
  79 /* The following token is returned only by XmlCdataSectionTok */
  80 #define XML_TOK_CDATA_SECT_CLOSE 40
  81
  82 /* With namespace processing this is returned by XmlPrologTok
  83    for a name with a colon. */
  84 #define XML_TOK_PREFIXED_NAME 41
  85
  86 #ifdef XML_DTD
  87 #define XML_TOK_IGNORE_SECT 42
  88 #endif /* XML_DTD */
  89
  90 #ifdef XML_DTD
  91 #define XML_N_STATES 4
  92 #else /* not XML_DTD */
  93 #define XML_N_STATES 3
  94 #endif /* not XML_DTD */
  95
  96 #define XML_PROLOG_STATE 0
  97 #define XML_CONTENT_STATE 1
  98 #define XML_CDATA_SECTION_STATE 2
  99 #ifdef XML_DTD
 100 #define XML_IGNORE_SECTION_STATE 3
 101 #endif /* XML_DTD */
 102
 103 #define XML_N_LITERAL_TYPES 2
 104 #define XML_ATTRIBUTE_VALUE_LITERAL 0
 105 #define XML_ENTITY_VALUE_LITERAL 1
 106
 107 /* The size of the buffer passed to XmlUtf8Encode must be at least this. */
 108 #define XML_UTF8_ENCODE_MAX 4
 109 /* The size of the buffer passed to XmlUtf16Encode must be at least this. */
 110 #define XML_UTF16_ENCODE_MAX 2
 111
 112 typedef struct position {
 113   /* first line and first column are 0 not 1 */
 114   unsigned long lineNumber;
 115   unsigned long columnNumber;
 116 } POSITION;
 117
 118 typedef struct {
 119   const char *name;
 120   const char *valuePtr;
 121   const char *valueEnd;
 122   char normalized;
 123 } ATTRIBUTE;
 124
 125 struct encoding;
 126 typedef struct encoding ENCODING;
 127
 128 struct encoding {
 129   int (*scanners[XML_N_STATES])(const ENCODING *,
 130                                 const char *,
 131                                 const char *,
 132                                 const char **);
 133   int (*literalScanners[XML_N_LITERAL_TYPES])(const ENCODING *,
 134                                               const char *,
 135                                               const char *,
 136                                               const char **);
 137   int (*sameName)(const ENCODING *,
 138                   const char *, const char *);
 139   int (*nameMatchesAscii)(const ENCODING *,
 140                           const char *, const char *, const char *);
 141   int (*nameLength)(const ENCODING *, const char *);
 142   const char *(*skipS)(const ENCODING *, const char *);
 143   int (*getAtts)(const ENCODING *enc, const char *ptr,
 144                  int attsMax, ATTRIBUTE *atts);
 145   int (*charRefNumber)(const ENCODING *enc, const char *ptr);
 146   int (*predefinedEntityName)(const ENCODING *, const char *, const char *);
 147   void (*updatePosition)(const ENCODING *,
 148                          const char *ptr,
 149                          const char *end,
 150                          POSITION *);
 151   int (*isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
 152                     const char **badPtr);
 153   void (*utf8Convert)(const ENCODING *enc,
 154                       const char **fromP,
 155                       const char *fromLim,
 156                       char **toP,
 157                       const char *toLim);
 158   void (*utf16Convert)(const ENCODING *enc,
 159                        const char **fromP,
 160                        const char *fromLim,
 161                        unsigned short **toP,
 162                        const unsigned short *toLim);
 163   int minBytesPerChar;
 164   char isUtf8;
 165   char isUtf16;
 166 };
 167
 168 /*
 169 Scan the string starting at ptr until the end of the next complete token,
 170 but do not scan past eptr.  Return an integer giving the type of token.
 171
 172 Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
 173
 174 Return XML_TOK_PARTIAL when the string does not contain a complete token;
 175 nextTokPtr will not be set.
 176
 177 Return XML_TOK_INVALID when the string does not start a valid token; nextTokPtr
 178 will be set to point to the character which made the token invalid.
 179
 180 Otherwise the string starts with a valid token; nextTokPtr will be set to point
 181 to the character following the end of that token.
 182
 183 Each data character counts as a single token, but adjacent data characters
 184 may be returned together.  Similarly for characters in the prolog outside
 185 literals, comments and processing instructions.
 186 */
 187
 188
 189 #define XmlTok(enc, state, ptr, end, nextTokPtr) \
 190   (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
 191
 192 #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
 193    XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
 194
 195 #define XmlContentTok(enc, ptr, end, nextTokPtr) \
 196    XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
 197
 198 #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
 199    XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
 200
 201 #ifdef XML_DTD
 202
 203 #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
 204    XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
 205
 206 #endif /* XML_DTD */
 207
 208 /* This is used for performing a 2nd-level tokenization on
 209 the content of a literal that has already been returned by XmlTok. */
 210
 211 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
 212   (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
 213
 214 #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
 215    XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
 216
 217 #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
 218    XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
 219
 220 #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
 221
 222 #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
 223   (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
 224
 225 #define XmlNameLength(enc, ptr) \
 226   (((enc)->nameLength)(enc, ptr))
 227
 228 #define XmlSkipS(enc, ptr) \
 229   (((enc)->skipS)(enc, ptr))
 230
 231 #define XmlGetAttributes(enc, ptr, attsMax, atts) \
 232   (((enc)->getAtts)(enc, ptr, attsMax, atts))
 233
 234 #define XmlCharRefNumber(enc, ptr) \
 235   (((enc)->charRefNumber)(enc, ptr))
 236
 237 #define XmlPredefinedEntityName(enc, ptr, end) \
 238   (((enc)->predefinedEntityName)(enc, ptr, end))
 239
 240 #define XmlUpdatePosition(enc, ptr, end, pos) \
 241   (((enc)->updatePosition)(enc, ptr, end, pos))
 242
 243 #define XmlIsPublicId(enc, ptr, end, badPtr) \
 244   (((enc)->isPublicId)(enc, ptr, end, badPtr))
 245
 246 #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
 247   (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
 248
 249 #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
 250   (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
 251
 252 typedef struct {
 253   ENCODING initEnc;
 254   const ENCODING **encPtr;
 255 } INIT_ENCODING;
 256
 257 int XMLTOKAPI XmlParseXmlDecl(int isGeneralTextEntity,
 258                               const ENCODING *enc,
 259                               const char *ptr,
 260                               const char *end,
 261                               const char **badPtr,
 262                               const char **versionPtr,
 263                               const char **encodingNamePtr,
 264                               const ENCODING **namedEncodingPtr,
 265                               int *standalonePtr);
 266
 267 int XMLTOKAPI XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
 268 const ENCODING XMLTOKAPI *XmlGetUtf8InternalEncoding(void);
 269 const ENCODING XMLTOKAPI *XmlGetUtf16InternalEncoding(void);
 270 int XMLTOKAPI XmlUtf8Encode(int charNumber, char *buf);
 271 int XMLTOKAPI XmlUtf16Encode(int charNumber, unsigned short *buf);
 272
 273 int XMLTOKAPI XmlSizeOfUnknownEncoding(void);
 274 ENCODING XMLTOKAPI *
 275 XmlInitUnknownEncoding(void *mem,
 276                        int *table,
 277                        int (*conv)(void *userData, const char *p),
 278                        void *userData);
 279
 280 int XMLTOKAPI XmlParseXmlDeclNS(int isGeneralTextEntity,
 281                                 const ENCODING *enc,
 282                                 const char *ptr,
 283                                 const char *end,
 284                                 const char **badPtr,
 285                                 const char **versionPtr,
 286                                 const char **encodingNamePtr,
 287                                 const ENCODING **namedEncodingPtr,
 288                                 int *standalonePtr);
 289 int XMLTOKAPI XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
 290 const ENCODING XMLTOKAPI *XmlGetUtf8InternalEncodingNS(void);
 291 const ENCODING XMLTOKAPI *XmlGetUtf16InternalEncodingNS(void);
 292 ENCODING XMLTOKAPI *
 293 XmlInitUnknownEncodingNS(void *mem,
 294                          int *table,
 295                          int (*conv)(void *userData, const char *p),
 296                          void *userData);
 297 #ifdef __cplusplus
 298 }
 299 #endif
 300
 301 #endif /* not XmlTok_INCLUDED */