1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/parser/expat/lib/xmltok.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,316 @@
1.4 +/* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
1.5 + See the file COPYING for copying permission.
1.6 +*/
1.7 +
1.8 +#ifndef XmlTok_INCLUDED
1.9 +#define XmlTok_INCLUDED 1
1.10 +
1.11 +#ifdef __cplusplus
1.12 +extern "C" {
1.13 +#endif
1.14 +
1.15 +/* The following token may be returned by XmlContentTok */
1.16 +#define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be
1.17 + start of illegal ]]> sequence */
1.18 +/* The following tokens may be returned by both XmlPrologTok and
1.19 + XmlContentTok.
1.20 +*/
1.21 +#define XML_TOK_NONE -4 /* The string to be scanned is empty */
1.22 +#define XML_TOK_TRAILING_CR -3 /* A CR at the end of the scan;
1.23 + might be part of CRLF sequence */
1.24 +#define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */
1.25 +#define XML_TOK_PARTIAL -1 /* only part of a token */
1.26 +#define XML_TOK_INVALID 0
1.27 +
1.28 +/* The following tokens are returned by XmlContentTok; some are also
1.29 + returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
1.30 +*/
1.31 +#define XML_TOK_START_TAG_WITH_ATTS 1
1.32 +#define XML_TOK_START_TAG_NO_ATTS 2
1.33 +#define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
1.34 +#define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
1.35 +#define XML_TOK_END_TAG 5
1.36 +#define XML_TOK_DATA_CHARS 6
1.37 +#define XML_TOK_DATA_NEWLINE 7
1.38 +#define XML_TOK_CDATA_SECT_OPEN 8
1.39 +#define XML_TOK_ENTITY_REF 9
1.40 +#define XML_TOK_CHAR_REF 10 /* numeric character reference */
1.41 +
1.42 +/* The following tokens may be returned by both XmlPrologTok and
1.43 + XmlContentTok.
1.44 +*/
1.45 +#define XML_TOK_PI 11 /* processing instruction */
1.46 +#define XML_TOK_XML_DECL 12 /* XML decl or text decl */
1.47 +#define XML_TOK_COMMENT 13
1.48 +#define XML_TOK_BOM 14 /* Byte order mark */
1.49 +
1.50 +/* The following tokens are returned only by XmlPrologTok */
1.51 +#define XML_TOK_PROLOG_S 15
1.52 +#define XML_TOK_DECL_OPEN 16 /* <!foo */
1.53 +#define XML_TOK_DECL_CLOSE 17 /* > */
1.54 +#define XML_TOK_NAME 18
1.55 +#define XML_TOK_NMTOKEN 19
1.56 +#define XML_TOK_POUND_NAME 20 /* #name */
1.57 +#define XML_TOK_OR 21 /* | */
1.58 +#define XML_TOK_PERCENT 22
1.59 +#define XML_TOK_OPEN_PAREN 23
1.60 +#define XML_TOK_CLOSE_PAREN 24
1.61 +#define XML_TOK_OPEN_BRACKET 25
1.62 +#define XML_TOK_CLOSE_BRACKET 26
1.63 +#define XML_TOK_LITERAL 27
1.64 +#define XML_TOK_PARAM_ENTITY_REF 28
1.65 +#define XML_TOK_INSTANCE_START 29
1.66 +
1.67 +/* The following occur only in element type declarations */
1.68 +#define XML_TOK_NAME_QUESTION 30 /* name? */
1.69 +#define XML_TOK_NAME_ASTERISK 31 /* name* */
1.70 +#define XML_TOK_NAME_PLUS 32 /* name+ */
1.71 +#define XML_TOK_COND_SECT_OPEN 33 /* <![ */
1.72 +#define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */
1.73 +#define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */
1.74 +#define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */
1.75 +#define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */
1.76 +#define XML_TOK_COMMA 38
1.77 +
1.78 +/* The following token is returned only by XmlAttributeValueTok */
1.79 +#define XML_TOK_ATTRIBUTE_VALUE_S 39
1.80 +
1.81 +/* The following token is returned only by XmlCdataSectionTok */
1.82 +#define XML_TOK_CDATA_SECT_CLOSE 40
1.83 +
1.84 +/* With namespace processing this is returned by XmlPrologTok for a
1.85 + name with a colon.
1.86 +*/
1.87 +#define XML_TOK_PREFIXED_NAME 41
1.88 +
1.89 +#ifdef XML_DTD
1.90 +#define XML_TOK_IGNORE_SECT 42
1.91 +#endif /* XML_DTD */
1.92 +
1.93 +#ifdef XML_DTD
1.94 +#define XML_N_STATES 4
1.95 +#else /* not XML_DTD */
1.96 +#define XML_N_STATES 3
1.97 +#endif /* not XML_DTD */
1.98 +
1.99 +#define XML_PROLOG_STATE 0
1.100 +#define XML_CONTENT_STATE 1
1.101 +#define XML_CDATA_SECTION_STATE 2
1.102 +#ifdef XML_DTD
1.103 +#define XML_IGNORE_SECTION_STATE 3
1.104 +#endif /* XML_DTD */
1.105 +
1.106 +#define XML_N_LITERAL_TYPES 2
1.107 +#define XML_ATTRIBUTE_VALUE_LITERAL 0
1.108 +#define XML_ENTITY_VALUE_LITERAL 1
1.109 +
1.110 +/* The size of the buffer passed to XmlUtf8Encode must be at least this. */
1.111 +#define XML_UTF8_ENCODE_MAX 4
1.112 +/* The size of the buffer passed to XmlUtf16Encode must be at least this. */
1.113 +#define XML_UTF16_ENCODE_MAX 2
1.114 +
1.115 +typedef struct position {
1.116 + /* first line and first column are 0 not 1 */
1.117 + XML_Size lineNumber;
1.118 + XML_Size columnNumber;
1.119 +} POSITION;
1.120 +
1.121 +typedef struct {
1.122 + const char *name;
1.123 + const char *valuePtr;
1.124 + const char *valueEnd;
1.125 + char normalized;
1.126 +} ATTRIBUTE;
1.127 +
1.128 +struct encoding;
1.129 +typedef struct encoding ENCODING;
1.130 +
1.131 +typedef int (PTRCALL *SCANNER)(const ENCODING *,
1.132 + const char *,
1.133 + const char *,
1.134 + const char **);
1.135 +
1.136 +struct encoding {
1.137 + SCANNER scanners[XML_N_STATES];
1.138 + SCANNER literalScanners[XML_N_LITERAL_TYPES];
1.139 + int (PTRCALL *sameName)(const ENCODING *,
1.140 + const char *,
1.141 + const char *);
1.142 + int (PTRCALL *nameMatchesAscii)(const ENCODING *,
1.143 + const char *,
1.144 + const char *,
1.145 + const char *);
1.146 + int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
1.147 + const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
1.148 + int (PTRCALL *getAtts)(const ENCODING *enc,
1.149 + const char *ptr,
1.150 + int attsMax,
1.151 + ATTRIBUTE *atts);
1.152 + int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
1.153 + int (PTRCALL *predefinedEntityName)(const ENCODING *,
1.154 + const char *,
1.155 + const char *);
1.156 + void (PTRCALL *updatePosition)(const ENCODING *,
1.157 + const char *ptr,
1.158 + const char *end,
1.159 + POSITION *);
1.160 + int (PTRCALL *isPublicId)(const ENCODING *enc,
1.161 + const char *ptr,
1.162 + const char *end,
1.163 + const char **badPtr);
1.164 + void (PTRCALL *utf8Convert)(const ENCODING *enc,
1.165 + const char **fromP,
1.166 + const char *fromLim,
1.167 + char **toP,
1.168 + const char *toLim);
1.169 + void (PTRCALL *utf16Convert)(const ENCODING *enc,
1.170 + const char **fromP,
1.171 + const char *fromLim,
1.172 + unsigned short **toP,
1.173 + const unsigned short *toLim);
1.174 + int minBytesPerChar;
1.175 + char isUtf8;
1.176 + char isUtf16;
1.177 +};
1.178 +
1.179 +/* Scan the string starting at ptr until the end of the next complete
1.180 + token, but do not scan past eptr. Return an integer giving the
1.181 + type of token.
1.182 +
1.183 + Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
1.184 +
1.185 + Return XML_TOK_PARTIAL when the string does not contain a complete
1.186 + token; nextTokPtr will not be set.
1.187 +
1.188 + Return XML_TOK_INVALID when the string does not start a valid
1.189 + token; nextTokPtr will be set to point to the character which made
1.190 + the token invalid.
1.191 +
1.192 + Otherwise the string starts with a valid token; nextTokPtr will be
1.193 + set to point to the character following the end of that token.
1.194 +
1.195 + Each data character counts as a single token, but adjacent data
1.196 + characters may be returned together. Similarly for characters in
1.197 + the prolog outside literals, comments and processing instructions.
1.198 +*/
1.199 +
1.200 +
1.201 +#define XmlTok(enc, state, ptr, end, nextTokPtr) \
1.202 + (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
1.203 +
1.204 +#define XmlPrologTok(enc, ptr, end, nextTokPtr) \
1.205 + XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
1.206 +
1.207 +#define XmlContentTok(enc, ptr, end, nextTokPtr) \
1.208 + XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
1.209 +
1.210 +#define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
1.211 + XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
1.212 +
1.213 +#ifdef XML_DTD
1.214 +
1.215 +#define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
1.216 + XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
1.217 +
1.218 +#endif /* XML_DTD */
1.219 +
1.220 +/* This is used for performing a 2nd-level tokenization on the content
1.221 + of a literal that has already been returned by XmlTok.
1.222 +*/
1.223 +#define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
1.224 + (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
1.225 +
1.226 +#define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
1.227 + XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
1.228 +
1.229 +#define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
1.230 + XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
1.231 +
1.232 +#define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
1.233 +
1.234 +#define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
1.235 + (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
1.236 +
1.237 +#define XmlNameLength(enc, ptr) \
1.238 + (((enc)->nameLength)(enc, ptr))
1.239 +
1.240 +#define XmlSkipS(enc, ptr) \
1.241 + (((enc)->skipS)(enc, ptr))
1.242 +
1.243 +#define XmlGetAttributes(enc, ptr, attsMax, atts) \
1.244 + (((enc)->getAtts)(enc, ptr, attsMax, atts))
1.245 +
1.246 +#define XmlCharRefNumber(enc, ptr) \
1.247 + (((enc)->charRefNumber)(enc, ptr))
1.248 +
1.249 +#define XmlPredefinedEntityName(enc, ptr, end) \
1.250 + (((enc)->predefinedEntityName)(enc, ptr, end))
1.251 +
1.252 +#define XmlUpdatePosition(enc, ptr, end, pos) \
1.253 + (((enc)->updatePosition)(enc, ptr, end, pos))
1.254 +
1.255 +#define XmlIsPublicId(enc, ptr, end, badPtr) \
1.256 + (((enc)->isPublicId)(enc, ptr, end, badPtr))
1.257 +
1.258 +#define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
1.259 + (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
1.260 +
1.261 +#define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
1.262 + (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
1.263 +
1.264 +typedef struct {
1.265 + ENCODING initEnc;
1.266 + const ENCODING **encPtr;
1.267 +} INIT_ENCODING;
1.268 +
1.269 +int XmlParseXmlDecl(int isGeneralTextEntity,
1.270 + const ENCODING *enc,
1.271 + const char *ptr,
1.272 + const char *end,
1.273 + const char **badPtr,
1.274 + const char **versionPtr,
1.275 + const char **versionEndPtr,
1.276 + const char **encodingNamePtr,
1.277 + const ENCODING **namedEncodingPtr,
1.278 + int *standalonePtr);
1.279 +
1.280 +int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
1.281 +const ENCODING *XmlGetUtf8InternalEncoding(void);
1.282 +const ENCODING *XmlGetUtf16InternalEncoding(void);
1.283 +int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
1.284 +int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
1.285 +int XmlSizeOfUnknownEncoding(void);
1.286 +
1.287 +
1.288 +typedef int (XMLCALL *CONVERTER) (void *userData, const char *p);
1.289 +
1.290 +ENCODING *
1.291 +XmlInitUnknownEncoding(void *mem,
1.292 + int *table,
1.293 + CONVERTER convert,
1.294 + void *userData);
1.295 +
1.296 +int XmlParseXmlDeclNS(int isGeneralTextEntity,
1.297 + const ENCODING *enc,
1.298 + const char *ptr,
1.299 + const char *end,
1.300 + const char **badPtr,
1.301 + const char **versionPtr,
1.302 + const char **versionEndPtr,
1.303 + const char **encodingNamePtr,
1.304 + const ENCODING **namedEncodingPtr,
1.305 + int *standalonePtr);
1.306 +
1.307 +int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
1.308 +const ENCODING *XmlGetUtf8InternalEncodingNS(void);
1.309 +const ENCODING *XmlGetUtf16InternalEncodingNS(void);
1.310 +ENCODING *
1.311 +XmlInitUnknownEncodingNS(void *mem,
1.312 + int *table,
1.313 + CONVERTER convert,
1.314 + void *userData);
1.315 +#ifdef __cplusplus
1.316 +}
1.317 +#endif
1.318 +
1.319 +#endif /* not XmlTok_INCLUDED */
