| 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 | #include "expat\expat_setup.h"
|
|---|
| 10 | // V0.9.9 (2001-02-10) [umoeller]
|
|---|
| 11 | // to save the app from having to include this as well
|
|---|
| 12 |
|
|---|
| 13 | #ifdef __cplusplus
|
|---|
| 14 | extern "C" {
|
|---|
| 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 | typedef int EXPATENTRY SCANNERFUNC(const ENCODING *,
|
|---|
| 129 | const char *,
|
|---|
| 130 | const char *,
|
|---|
| 131 | const char **);
|
|---|
| 132 | typedef SCANNERFUNC *PSCANNERFUNC;
|
|---|
| 133 |
|
|---|
| 134 | struct encoding {
|
|---|
| 135 | PSCANNERFUNC scanners[XML_N_STATES];
|
|---|
| 136 | PSCANNERFUNC literalScanners[XML_N_LITERAL_TYPES];
|
|---|
| 137 | /* int (* EXPATENTRY literalScanners[XML_N_LITERAL_TYPES])(const ENCODING *,
|
|---|
| 138 | const char *,
|
|---|
| 139 | const char *,
|
|---|
| 140 | const char **); */
|
|---|
| 141 | int (* EXPATENTRY sameName)(const ENCODING *,
|
|---|
| 142 | const char *, const char *);
|
|---|
| 143 | int (* EXPATENTRY nameMatchesAscii)(const ENCODING *,
|
|---|
| 144 | const char *, const char *, const char *);
|
|---|
| 145 | int (* EXPATENTRY nameLength)(const ENCODING *, const char *);
|
|---|
| 146 | const char *(* EXPATENTRY skipS)(const ENCODING *, const char *);
|
|---|
| 147 | int (* EXPATENTRY getAtts)(const ENCODING *enc, const char *ptr,
|
|---|
| 148 | int attsMax, ATTRIBUTE *atts);
|
|---|
| 149 | int (* EXPATENTRY charRefNumber)(const ENCODING *enc, const char *ptr);
|
|---|
| 150 | int (* EXPATENTRY predefinedEntityName)(const ENCODING *, const char *, const char *);
|
|---|
| 151 | void (* EXPATENTRY updatePosition)(const ENCODING *,
|
|---|
| 152 | const char *ptr,
|
|---|
| 153 | const char *end,
|
|---|
| 154 | POSITION *);
|
|---|
| 155 | int (* EXPATENTRY isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
|
|---|
| 156 | const char **badPtr);
|
|---|
| 157 | void (* EXPATENTRY utf8Convert)(const ENCODING *enc,
|
|---|
| 158 | const char **fromP,
|
|---|
| 159 | const char *fromLim,
|
|---|
| 160 | char **toP,
|
|---|
| 161 | const char *toLim);
|
|---|
| 162 | void (* EXPATENTRY utf16Convert)(const ENCODING *enc,
|
|---|
| 163 | const char **fromP,
|
|---|
| 164 | const char *fromLim,
|
|---|
| 165 | unsigned short **toP,
|
|---|
| 166 | const unsigned short *toLim);
|
|---|
| 167 | int minBytesPerChar;
|
|---|
| 168 | char isUtf8;
|
|---|
| 169 | char isUtf16;
|
|---|
| 170 | };
|
|---|
| 171 |
|
|---|
| 172 | /*
|
|---|
| 173 | Scan the string starting at ptr until the end of the next complete token,
|
|---|
| 174 | but do not scan past eptr. Return an integer giving the type of token.
|
|---|
| 175 |
|
|---|
| 176 | Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
|
|---|
| 177 |
|
|---|
| 178 | Return XML_TOK_PARTIAL when the string does not contain a complete token;
|
|---|
| 179 | nextTokPtr will not be set.
|
|---|
| 180 |
|
|---|
| 181 | Return XML_TOK_INVALID when the string does not start a valid token; nextTokPtr
|
|---|
| 182 | will be set to point to the character which made the token invalid.
|
|---|
| 183 |
|
|---|
| 184 | Otherwise the string starts with a valid token; nextTokPtr will be set to point
|
|---|
| 185 | to the character following the end of that token.
|
|---|
| 186 |
|
|---|
| 187 | Each data character counts as a single token, but adjacent data characters
|
|---|
| 188 | may be returned together. Similarly for characters in the prolog outside
|
|---|
| 189 | literals, comments and processing instructions.
|
|---|
| 190 | */
|
|---|
| 191 |
|
|---|
| 192 |
|
|---|
| 193 | #define XmlTok(enc, state, ptr, end, nextTokPtr) \
|
|---|
| 194 | (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
|
|---|
| 195 |
|
|---|
| 196 | #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 197 | XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
|
|---|
| 198 |
|
|---|
| 199 | #define XmlContentTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 200 | XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
|
|---|
| 201 |
|
|---|
| 202 | #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 203 | XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
|
|---|
| 204 |
|
|---|
| 205 | #ifdef XML_DTD
|
|---|
| 206 |
|
|---|
| 207 | #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 208 | XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
|
|---|
| 209 |
|
|---|
| 210 | #endif /* XML_DTD */
|
|---|
| 211 |
|
|---|
| 212 | /* This is used for performing a 2nd-level tokenization on
|
|---|
| 213 | the content of a literal that has already been returned by XmlTok. */
|
|---|
| 214 |
|
|---|
| 215 | #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
|
|---|
| 216 | (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
|
|---|
| 217 |
|
|---|
| 218 | #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 219 | XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
|
|---|
| 220 |
|
|---|
| 221 | #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
|
|---|
| 222 | XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
|
|---|
| 223 |
|
|---|
| 224 | #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
|
|---|
| 225 |
|
|---|
| 226 | #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
|
|---|
| 227 | (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
|
|---|
| 228 |
|
|---|
| 229 | #define XmlNameLength(enc, ptr) \
|
|---|
| 230 | (((enc)->nameLength)(enc, ptr))
|
|---|
| 231 |
|
|---|
| 232 | #define XmlSkipS(enc, ptr) \
|
|---|
| 233 | (((enc)->skipS)(enc, ptr))
|
|---|
| 234 |
|
|---|
| 235 | #define XmlGetAttributes(enc, ptr, attsMax, atts) \
|
|---|
| 236 | (((enc)->getAtts)(enc, ptr, attsMax, atts))
|
|---|
| 237 |
|
|---|
| 238 | #define XmlCharRefNumber(enc, ptr) \
|
|---|
| 239 | (((enc)->charRefNumber)(enc, ptr))
|
|---|
| 240 |
|
|---|
| 241 | #define XmlPredefinedEntityName(enc, ptr, end) \
|
|---|
| 242 | (((enc)->predefinedEntityName)(enc, ptr, end))
|
|---|
| 243 |
|
|---|
| 244 | #define XmlUpdatePosition(enc, ptr, end, pos) \
|
|---|
| 245 | (((enc)->updatePosition)(enc, ptr, end, pos))
|
|---|
| 246 |
|
|---|
| 247 | #define XmlIsPublicId(enc, ptr, end, badPtr) \
|
|---|
| 248 | (((enc)->isPublicId)(enc, ptr, end, badPtr))
|
|---|
| 249 |
|
|---|
| 250 | #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
|
|---|
| 251 | (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
|
|---|
| 252 |
|
|---|
| 253 | #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
|
|---|
| 254 | (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
|
|---|
| 255 |
|
|---|
| 256 | typedef struct {
|
|---|
| 257 | ENCODING initEnc;
|
|---|
| 258 | const ENCODING **encPtr;
|
|---|
| 259 | } INIT_ENCODING;
|
|---|
| 260 |
|
|---|
| 261 | int XmlParseXmlDecl(int isGeneralTextEntity,
|
|---|
| 262 | const ENCODING *enc,
|
|---|
| 263 | const char *ptr,
|
|---|
| 264 | const char *end,
|
|---|
| 265 | const char **badPtr,
|
|---|
| 266 | const char **versionPtr,
|
|---|
| 267 | const char **versionEndPtr,
|
|---|
| 268 | const char **encodingNamePtr,
|
|---|
| 269 | const ENCODING **namedEncodingPtr,
|
|---|
| 270 | int *standalonePtr);
|
|---|
| 271 |
|
|---|
| 272 | int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
|
|---|
| 273 | const ENCODING *XmlGetUtf8InternalEncoding(void);
|
|---|
| 274 | const ENCODING *XmlGetUtf16InternalEncoding(void);
|
|---|
| 275 | int XmlUtf8Encode(int charNumber, char *buf);
|
|---|
| 276 | int XmlUtf16Encode(int charNumber, unsigned short *buf);
|
|---|
| 277 |
|
|---|
| 278 | int XmlSizeOfUnknownEncoding(void);
|
|---|
| 279 | ENCODING *
|
|---|
| 280 | XmlInitUnknownEncoding(void *mem,
|
|---|
| 281 | int *table,
|
|---|
| 282 | int (*conv)(void *userData, const char *p),
|
|---|
| 283 | void *userData);
|
|---|
| 284 |
|
|---|
| 285 | int XmlParseXmlDeclNS(int isGeneralTextEntity,
|
|---|
| 286 | const ENCODING *enc,
|
|---|
| 287 | const char *ptr,
|
|---|
| 288 | const char *end,
|
|---|
| 289 | const char **badPtr,
|
|---|
| 290 | const char **versionPtr,
|
|---|
| 291 | const char **versionEndPtr,
|
|---|
| 292 | const char **encodingNamePtr,
|
|---|
| 293 | const ENCODING **namedEncodingPtr,
|
|---|
| 294 | int *standalonePtr);
|
|---|
| 295 | int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
|
|---|
| 296 | const ENCODING *XmlGetUtf8InternalEncodingNS(void);
|
|---|
| 297 | const ENCODING *XmlGetUtf16InternalEncodingNS(void);
|
|---|
| 298 | ENCODING *
|
|---|
| 299 | XmlInitUnknownEncodingNS(void *mem,
|
|---|
| 300 | int *table,
|
|---|
| 301 | int (*conv)(void *userData, const char *p),
|
|---|
| 302 | void *userData);
|
|---|
| 303 | #ifdef __cplusplus
|
|---|
| 304 | }
|
|---|
| 305 | #endif
|
|---|
| 306 |
|
|---|
| 307 | #endif /* not XmlTok_INCLUDED */
|
|---|