13
12
/* 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 */
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 */
22
23
#define XML_TOK_INVALID 0
24
25
/* The following tokens are returned by XmlContentTok; some are also
25
returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok */
26
returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
27
28
#define XML_TOK_START_TAG_WITH_ATTS 1
28
29
#define XML_TOK_START_TAG_NO_ATTS 2
29
30
#define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
33
34
#define XML_TOK_DATA_NEWLINE 7
34
35
#define XML_TOK_CDATA_SECT_OPEN 8
35
36
#define XML_TOK_ENTITY_REF 9
36
#define XML_TOK_CHAR_REF 10 /* numeric character reference */
37
#define XML_TOK_CHAR_REF 10 /* numeric character reference */
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 */
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 */
41
44
#define XML_TOK_COMMENT 13
42
#define XML_TOK_BOM 14 /* Byte order mark */
45
#define XML_TOK_BOM 14 /* Byte order mark */
44
47
/* The following tokens are returned only by XmlPrologTok */
45
48
#define XML_TOK_PROLOG_S 15
46
#define XML_TOK_DECL_OPEN 16 /* <!foo */
47
#define XML_TOK_DECL_CLOSE 17 /* > */
49
#define XML_TOK_DECL_OPEN 16 /* <!foo */
50
#define XML_TOK_DECL_CLOSE 17 /* > */
48
51
#define XML_TOK_NAME 18
49
52
#define XML_TOK_NMTOKEN 19
50
#define XML_TOK_POUND_NAME 20 /* #name */
51
#define XML_TOK_OR 21 /* | */
53
#define XML_TOK_POUND_NAME 20 /* #name */
54
#define XML_TOK_OR 21 /* | */
52
55
#define XML_TOK_PERCENT 22
53
56
#define XML_TOK_OPEN_PAREN 23
54
57
#define XML_TOK_CLOSE_PAREN 24
59
62
#define XML_TOK_INSTANCE_START 29
61
64
/* 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 /* )+ */
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 /* )+ */
70
73
#define XML_TOK_COMMA 38
72
75
/* The following token is returned only by XmlAttributeValueTok */
122
126
typedef struct encoding ENCODING;
128
typedef int (PTRCALL *SCANNER)(const ENCODING *,
124
133
struct encoding {
125
int (*scanners[XML_N_STATES])(const ENCODING *,
129
int (*literalScanners[XML_N_LITERAL_TYPES])(const ENCODING *,
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 *,
147
int (*isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
148
const char **badPtr);
149
void (*utf8Convert)(const ENCODING *enc,
154
void (*utf16Convert)(const ENCODING *enc,
157
unsigned short **toP,
158
const unsigned short *toLim);
134
SCANNER scanners[XML_N_STATES];
135
SCANNER literalScanners[XML_N_LITERAL_TYPES];
136
int (PTRCALL *sameName)(const ENCODING *,
139
int (PTRCALL *nameMatchesAscii)(const ENCODING *,
143
int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
144
const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
145
int (PTRCALL *getAtts)(const ENCODING *enc,
149
int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
150
int (PTRCALL *predefinedEntityName)(const ENCODING *,
153
void (PTRCALL *updatePosition)(const ENCODING *,
157
int (PTRCALL *isPublicId)(const ENCODING *enc,
160
const char **badPtr);
161
void (PTRCALL *utf8Convert)(const ENCODING *enc,
166
void (PTRCALL *utf16Convert)(const ENCODING *enc,
169
unsigned short **toP,
170
const unsigned short *toLim);
159
171
int minBytesPerChar;
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.
168
Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
170
Return XML_TOK_PARTIAL when the string does not contain a complete token;
171
nextTokPtr will not be set.
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.
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.
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.
176
/* Scan the string starting at ptr until the end of the next complete
177
token, but do not scan past eptr. Return an integer giving the
180
Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
182
Return XML_TOK_PARTIAL when the string does not contain a complete
183
token; nextTokPtr will not be set.
185
Return XML_TOK_INVALID when the string does not start a valid
186
token; nextTokPtr will be set to point to the character which made
189
Otherwise the string starts with a valid token; nextTokPtr will be
190
set to point to the character following the end of that token.
192
Each data character counts as a single token, but adjacent data
193
characters may be returned together. Similarly for characters in
194
the prolog outside literals, comments and processing instructions.
250
263
const ENCODING **encPtr;
253
int XmlParseXmlDecl(int isGeneralTextEntity,
258
const char **versionPtr,
259
const char **versionEndPtr,
260
const char **encodingNamePtr,
261
const ENCODING **namedEncodingPtr,
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);
270
int XmlSizeOfUnknownEncoding(void);
266
int XmlParseXmlDecl(int isGeneralTextEntity,
271
const char **versionPtr,
272
const char **versionEndPtr,
273
const char **encodingNamePtr,
274
const ENCODING **namedEncodingPtr,
277
int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
278
const ENCODING *XmlGetUtf8InternalEncoding(void);
279
const ENCODING *XmlGetUtf16InternalEncoding(void);
280
int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
281
int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
282
int XmlSizeOfUnknownEncoding(void);
284
typedef int (*CONVERTER)(void *userData, const char *p);
272
287
XmlInitUnknownEncoding(void *mem,
274
int (*conv)(void *userData, const char *p),
277
int XmlParseXmlDeclNS(int isGeneralTextEntity,
282
const char **versionPtr,
283
const char **versionEndPtr,
284
const char **encodingNamePtr,
285
const ENCODING **namedEncodingPtr,
287
int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
288
const ENCODING *XmlGetUtf8InternalEncodingNS(void);
289
const ENCODING *XmlGetUtf16InternalEncodingNS(void);
292
int XmlParseXmlDeclNS(int isGeneralTextEntity,
297
const char **versionPtr,
298
const char **versionEndPtr,
299
const char **encodingNamePtr,
300
const ENCODING **namedEncodingPtr,
303
int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
304
const ENCODING *XmlGetUtf8InternalEncodingNS(void);
305
const ENCODING *XmlGetUtf16InternalEncodingNS(void);
291
307
XmlInitUnknownEncodingNS(void *mem,
293
int (*conv)(void *userData, const char *p),
295
311
#ifdef __cplusplus