1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd 2 See the file COPYING for copying permission. 3 */ 4 5 #ifndef XmlTok_INCLUDED 6 #define XmlTok_INCLUDED 1 7 8 #ifdef __cplusplus 9 extern "C" { 10 #endif 11 12 /* The following token may be returned by XmlContentTok */ 13 #define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be 14 start of illegal ]]> sequence */ 15 /* The following tokens may be returned by both XmlPrologTok and 16 XmlContentTok. 17 */ 18 #define XML_TOK_NONE -4 /* The string to be scanned is empty */ 19 #define XML_TOK_TRAILING_CR -3 /* A CR at the end of the scan; 20 might be part of CRLF sequence */ 21 #define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */ 22 #define XML_TOK_PARTIAL -1 /* only part of a token */ 23 #define XML_TOK_INVALID 0 24 25 /* The following tokens are returned by XmlContentTok; some are also 26 returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok. 27 */ 28 #define XML_TOK_START_TAG_WITH_ATTS 1 29 #define XML_TOK_START_TAG_NO_ATTS 2 30 #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */ 31 #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4 32 #define XML_TOK_END_TAG 5 33 #define XML_TOK_DATA_CHARS 6 34 #define XML_TOK_DATA_NEWLINE 7 35 #define XML_TOK_CDATA_SECT_OPEN 8 36 #define XML_TOK_ENTITY_REF 9 37 #define XML_TOK_CHAR_REF 10 /* numeric character reference */ 38 39 /* The following tokens may be returned by both XmlPrologTok and 40 XmlContentTok. 41 */ 42 #define XML_TOK_PI 11 /* processing instruction */ 43 #define XML_TOK_XML_DECL 12 /* XML decl or text decl */ 44 #define XML_TOK_COMMENT 13 45 #define XML_TOK_BOM 14 /* Byte order mark */ 46 47 /* The following tokens are returned only by XmlPrologTok */ 48 #define XML_TOK_PROLOG_S 15 49 #define XML_TOK_DECL_OPEN 16 /* <!foo */ 50 #define XML_TOK_DECL_CLOSE 17 /* > */ 51 #define XML_TOK_NAME 18 52 #define XML_TOK_NMTOKEN 19 53 #define XML_TOK_POUND_NAME 20 /* #name */ 54 #define XML_TOK_OR 21 /* | */ 55 #define XML_TOK_PERCENT 22 56 #define XML_TOK_OPEN_PAREN 23 57 #define XML_TOK_CLOSE_PAREN 24 58 #define XML_TOK_OPEN_BRACKET 25 59 #define XML_TOK_CLOSE_BRACKET 26 60 #define XML_TOK_LITERAL 27 61 #define XML_TOK_PARAM_ENTITY_REF 28 62 #define XML_TOK_INSTANCE_START 29 63 64 /* The following occur only in element type declarations */ 65 #define XML_TOK_NAME_QUESTION 30 /* name? */ 66 #define XML_TOK_NAME_ASTERISK 31 /* name* */ 67 #define XML_TOK_NAME_PLUS 32 /* name+ */ 68 #define XML_TOK_COND_SECT_OPEN 33 /* <![ */ 69 #define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */ 70 #define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */ 71 #define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */ 72 #define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */ 73 #define XML_TOK_COMMA 38 74 75 /* The following token is returned only by XmlAttributeValueTok */ 76 #define XML_TOK_ATTRIBUTE_VALUE_S 39 77 78 /* The following token is returned only by XmlCdataSectionTok */ 79 #define XML_TOK_CDATA_SECT_CLOSE 40 80 81 /* With namespace processing this is returned by XmlPrologTok for a 82 name with a colon. 83 */ 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 XML_Size lineNumber; 115 XML_Size 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 typedef int (PTRCALL *SCANNER)(const ENCODING *, 129 const char *, 130 const char *, 131 const char **); 132 133 enum XML_Convert_Result { 134 XML_CONVERT_COMPLETED = 0, 135 XML_CONVERT_INPUT_INCOMPLETE = 1, 136 XML_CONVERT_OUTPUT_EXHAUSTED = 2 /* and therefore potentially input remaining as well */ 137 }; 138 139 struct encoding { 140 SCANNER scanners[XML_N_STATES]; 141 SCANNER literalScanners[XML_N_LITERAL_TYPES]; 142 int (PTRCALL *sameName)(const ENCODING *, 143 const char *, 144 const char *); 145 int (PTRCALL *nameMatchesAscii)(const ENCODING *, 146 const char *, 147 const char *, 148 const char *); 149 int (PTRFASTCALL *nameLength)(const ENCODING *, const char *); 150 const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *); 151 int (PTRCALL *getAtts)(const ENCODING *enc, 152 const char *ptr, 153 int attsMax, 154 ATTRIBUTE *atts); 155 int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr); 156 int (PTRCALL *predefinedEntityName)(const ENCODING *, 157 const char *, 158 const char *); 159 void (PTRCALL *updatePosition)(const ENCODING *, 160 const char *ptr, 161 const char *end, 162 POSITION *); 163 int (PTRCALL *isPublicId)(const ENCODING *enc, 164 const char *ptr, 165 const char *end, 166 const char **badPtr); 167 enum XML_Convert_Result (PTRCALL *utf8Convert)(const ENCODING *enc, 168 const char **fromP, 169 const char *fromLim, 170 char **toP, 171 const char *toLim); 172 enum XML_Convert_Result (PTRCALL *utf16Convert)(const ENCODING *enc, 173 const char **fromP, 174 const char *fromLim, 175 unsigned short **toP, 176 const unsigned short *toLim); 177 int minBytesPerChar; 178 char isUtf8; 179 char isUtf16; 180 }; 181 182 /* Scan the string starting at ptr until the end of the next complete 183 token, but do not scan past eptr. Return an integer giving the 184 type of token. 185 186 Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set. 187 188 Return XML_TOK_PARTIAL when the string does not contain a complete 189 token; nextTokPtr will not be set. 190 191 Return XML_TOK_INVALID when the string does not start a valid 192 token; nextTokPtr will be set to point to the character which made 193 the token invalid. 194 195 Otherwise the string starts with a valid token; nextTokPtr will be 196 set to point to the character following the end of that token. 197 198 Each data character counts as a single token, but adjacent data 199 characters may be returned together. Similarly for characters in 200 the prolog outside literals, comments and processing instructions. 201 */ 202 203 204 #define XmlTok(enc, state, ptr, end, nextTokPtr) \ 205 (((enc)->scanners[state])(enc, ptr, end, nextTokPtr)) 206 207 #define XmlPrologTok(enc, ptr, end, nextTokPtr) \ 208 XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr) 209 210 #define XmlContentTok(enc, ptr, end, nextTokPtr) \ 211 XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr) 212 213 #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \ 214 XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr) 215 216 #ifdef XML_DTD 217 218 #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \ 219 XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr) 220 221 #endif /* XML_DTD */ 222 223 /* This is used for performing a 2nd-level tokenization on the content 224 of a literal that has already been returned by XmlTok. 225 */ 226 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \ 227 (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr)) 228 229 #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \ 230 XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr) 231 232 #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \ 233 XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr) 234 235 #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2)) 236 237 #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \ 238 (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2)) 239 240 #define XmlNameLength(enc, ptr) \ 241 (((enc)->nameLength)(enc, ptr)) 242 243 #define XmlSkipS(enc, ptr) \ 244 (((enc)->skipS)(enc, ptr)) 245 246 #define XmlGetAttributes(enc, ptr, attsMax, atts) \ 247 (((enc)->getAtts)(enc, ptr, attsMax, atts)) 248 249 #define XmlCharRefNumber(enc, ptr) \ 250 (((enc)->charRefNumber)(enc, ptr)) 251 252 #define XmlPredefinedEntityName(enc, ptr, end) \ 253 (((enc)->predefinedEntityName)(enc, ptr, end)) 254 255 #define XmlUpdatePosition(enc, ptr, end, pos) \ 256 (((enc)->updatePosition)(enc, ptr, end, pos)) 257 258 #define XmlIsPublicId(enc, ptr, end, badPtr) \ 259 (((enc)->isPublicId)(enc, ptr, end, badPtr)) 260 261 #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \ 262 (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim)) 263 264 #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \ 265 (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim)) 266 267 typedef struct { 268 ENCODING initEnc; 269 const ENCODING **encPtr; 270 } INIT_ENCODING; 271 272 int XmlParseXmlDecl(int isGeneralTextEntity, 273 const ENCODING *enc, 274 const char *ptr, 275 const char *end, 276 const char **badPtr, 277 const char **versionPtr, 278 const char **versionEndPtr, 279 const char **encodingNamePtr, 280 const ENCODING **namedEncodingPtr, 281 int *standalonePtr); 282 283 int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name); 284 const ENCODING *XmlGetUtf8InternalEncoding(void); 285 const ENCODING *XmlGetUtf16InternalEncoding(void); 286 int FASTCALL XmlUtf8Encode(int charNumber, char *buf); 287 int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf); 288 int XmlSizeOfUnknownEncoding(void); 289 290 291 typedef int (XMLCALL *CONVERTER) (void *userData, const char *p); 292 293 ENCODING * 294 XmlInitUnknownEncoding(void *mem, 295 int *table, 296 CONVERTER convert, 297 void *userData); 298 299 int XmlParseXmlDeclNS(int isGeneralTextEntity, 300 const ENCODING *enc, 301 const char *ptr, 302 const char *end, 303 const char **badPtr, 304 const char **versionPtr, 305 const char **versionEndPtr, 306 const char **encodingNamePtr, 307 const ENCODING **namedEncodingPtr, 308 int *standalonePtr); 309 310 int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name); 311 const ENCODING *XmlGetUtf8InternalEncodingNS(void); 312 const ENCODING *XmlGetUtf16InternalEncodingNS(void); 313 ENCODING * 314 XmlInitUnknownEncodingNS(void *mem, 315 int *table, 316 CONVERTER convert, 317 void *userData); 318 #ifdef __cplusplus 319 } 320 #endif 321 322 #endif /* not XmlTok_INCLUDED */ 323