1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
2 See the file COPYING for copying permission.
5 #ifndef XmlTok_INCLUDED
6 #define XmlTok_INCLUDED 1
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
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
25 /* The following tokens are returned by XmlContentTok; some are also
26 returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
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 */
39 /* The following tokens may be returned by both XmlPrologTok and
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 */
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
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
75 /* The following token is returned only by XmlAttributeValueTok */
76 #define XML_TOK_ATTRIBUTE_VALUE_S 39
78 /* The following token is returned only by XmlCdataSectionTok */
79 #define XML_TOK_CDATA_SECT_CLOSE 40
81 /* With namespace processing this is returned by XmlPrologTok for a
84 #define XML_TOK_PREFIXED_NAME 41
87 #define XML_TOK_IGNORE_SECT 42
91 #define XML_N_STATES 4
92 #else /* not XML_DTD */
93 #define XML_N_STATES 3
94 #endif /* not XML_DTD */
96 #define XML_PROLOG_STATE 0
97 #define XML_CONTENT_STATE 1
98 #define XML_CDATA_SECTION_STATE 2
100 #define XML_IGNORE_SECTION_STATE 3
103 #define XML_N_LITERAL_TYPES 2
104 #define XML_ATTRIBUTE_VALUE_LITERAL 0
105 #define XML_ENTITY_VALUE_LITERAL 1
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
112 typedef struct position
{
113 /* first line and first column are 0 not 1 */
114 unsigned long lineNumber
;
115 unsigned long columnNumber
;
120 const char *valuePtr
;
121 const char *valueEnd
;
126 typedef struct encoding ENCODING
;
128 typedef int (FASTCALL
* SCANNER
) (const ENCODING
*, const char *, const char *, const char **);
131 SCANNER scanners
[XML_N_STATES
];
132 SCANNER literalScanners
[XML_N_LITERAL_TYPES
];
133 int (FASTCALL
* sameName
) (const ENCODING
*, const char *, const char *);
134 int (FASTCALL
* nameMatchesAscii
) (const ENCODING
*, const char *, const char *, const char *);
135 int (FASTCALL
* nameLength
) (const ENCODING
*, const char *);
136 const char *(FASTCALL
* skipS
) (const ENCODING
*, const char *);
137 int (FASTCALL
* getAtts
) (const ENCODING
* enc
, const char *ptr
, int attsMax
, ATTRIBUTE
* atts
);
138 int (FASTCALL
* charRefNumber
) (const ENCODING
* enc
, const char *ptr
);
139 int (FASTCALL
* predefinedEntityName
) (const ENCODING
*, const char *, const char *);
140 void (FASTCALL
* updatePosition
) (const ENCODING
*, const char *ptr
, const char *end
, POSITION
*);
141 int (FASTCALL
* isPublicId
) (const ENCODING
* enc
, const char *ptr
, const char *end
, const char **badPtr
);
142 void (FASTCALL
* utf8Convert
) (const ENCODING
* enc
, const char **fromP
, const char *fromLim
, char **toP
, const char *toLim
);
143 void (FASTCALL
* utf16Convert
) (const ENCODING
* enc
, const char **fromP
, const char *fromLim
, unsigned short **toP
, const unsigned short *toLim
);
149 /* Scan the string starting at ptr until the end of the next complete
150 token, but do not scan past eptr. Return an integer giving the
153 Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
155 Return XML_TOK_PARTIAL when the string does not contain a complete
156 token; nextTokPtr will not be set.
158 Return XML_TOK_INVALID when the string does not start a valid
159 token; nextTokPtr will be set to point to the character which made
162 Otherwise the string starts with a valid token; nextTokPtr will be
163 set to point to the character following the end of that token.
165 Each data character counts as a single token, but adjacent data
166 characters may be returned together. Similarly for characters in
167 the prolog outside literals, comments and processing instructions.
170 #define XmlTok(enc, state, ptr, end, nextTokPtr) \
171 (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
173 #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
174 XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
176 #define XmlContentTok(enc, ptr, end, nextTokPtr) \
177 XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
179 #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
180 XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
184 #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
185 XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
189 /* This is used for performing a 2nd-level tokenization on the content
190 of a literal that has already been returned by XmlTok.
192 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
193 (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
195 #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
196 XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
198 #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
199 XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
201 #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
203 #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
204 (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
206 #define XmlNameLength(enc, ptr) \
207 (((enc)->nameLength)(enc, ptr))
209 #define XmlSkipS(enc, ptr) \
210 (((enc)->skipS)(enc, ptr))
212 #define XmlGetAttributes(enc, ptr, attsMax, atts) \
213 (((enc)->getAtts)(enc, ptr, attsMax, atts))
215 #define XmlCharRefNumber(enc, ptr) \
216 (((enc)->charRefNumber)(enc, ptr))
218 #define XmlPredefinedEntityName(enc, ptr, end) \
219 (((enc)->predefinedEntityName)(enc, ptr, end))
221 #define XmlUpdatePosition(enc, ptr, end, pos) \
222 (((enc)->updatePosition)(enc, ptr, end, pos))
224 #define XmlIsPublicId(enc, ptr, end, badPtr) \
225 (((enc)->isPublicId)(enc, ptr, end, badPtr))
227 #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
228 (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
230 #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
231 (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
235 const ENCODING
**encPtr
;
238 int XmlParseXmlDecl(int isGeneralTextEntity
, const ENCODING
* enc
, const char *ptr
, const char *end
, const char **badPtr
, const char **versionPtr
, const char **versionEndPtr
, const char **encodingNamePtr
, const ENCODING
** namedEncodingPtr
, int *standalonePtr
);
240 int XmlInitEncoding(INIT_ENCODING
*, const ENCODING
**, const char *name
);
241 const ENCODING
*XmlGetUtf8InternalEncoding(void);
242 const ENCODING
*XmlGetUtf16InternalEncoding(void);
243 int XmlUtf8Encode(int charNumber
, char *buf
);
244 int XmlUtf16Encode(int charNumber
, unsigned short *buf
);
246 int XmlSizeOfUnknownEncoding(void);
247 ENCODING
*XmlInitUnknownEncoding(void *mem
, int *table
, int (*conv
) (void *userData
, const char *p
), void *userData
);
249 int XmlParseXmlDeclNS(int isGeneralTextEntity
, const ENCODING
* enc
, const char *ptr
, const char *end
, const char **badPtr
, const char **versionPtr
, const char **versionEndPtr
, const char **encodingNamePtr
, const ENCODING
** namedEncodingPtr
, int *standalonePtr
);
250 int XmlInitEncodingNS(INIT_ENCODING
*, const ENCODING
**, const char *name
);
251 const ENCODING
*XmlGetUtf8InternalEncodingNS(void);
252 const ENCODING
*XmlGetUtf16InternalEncodingNS(void);
253 ENCODING
*XmlInitUnknownEncodingNS(void *mem
, int *table
, int (*conv
) (void *userData
, const char *p
), void *userData
);
257 #endif /* not XmlTok_INCLUDED */