Utilities/cmexpat/xmltok.h

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