xmltok.h revision 1320f92c476a1ad9d19dba2a48c72b75566198e9
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
9extern "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
112typedef struct position {
113  /* first line and first column are 0 not 1 */
114  XML_Size lineNumber;
115  XML_Size columnNumber;
116} POSITION;
117
118typedef struct {
119  const char *name;
120  const char *valuePtr;
121  const char *valueEnd;
122  char normalized;
123} ATTRIBUTE;
124
125struct encoding;
126typedef struct encoding ENCODING;
127
128typedef int (PTRCALL *SCANNER)(const ENCODING *,
129                               const char *,
130                               const char *,
131                               const char **);
132
133struct encoding {
134  SCANNER scanners[XML_N_STATES];
135  SCANNER literalScanners[XML_N_LITERAL_TYPES];
136  int (PTRCALL *sameName)(const ENCODING *,
137                          const char *,
138                          const char *);
139  int (PTRCALL *nameMatchesAscii)(const ENCODING *,
140                                  const char *,
141                                  const char *,
142                                  const char *);
143  int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
144  const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
145  int (PTRCALL *getAtts)(const ENCODING *enc,
146                         const char *ptr,
147                         int attsMax,
148                         ATTRIBUTE *atts);
149  int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
150  int (PTRCALL *predefinedEntityName)(const ENCODING *,
151                                      const char *,
152                                      const char *);
153  void (PTRCALL *updatePosition)(const ENCODING *,
154                                 const char *ptr,
155                                 const char *end,
156                                 POSITION *);
157  int (PTRCALL *isPublicId)(const ENCODING *enc,
158                            const char *ptr,
159                            const char *end,
160                            const char **badPtr);
161  void (PTRCALL *utf8Convert)(const ENCODING *enc,
162                              const char **fromP,
163                              const char *fromLim,
164                              char **toP,
165                              const char *toLim);
166  void (PTRCALL *utf16Convert)(const ENCODING *enc,
167                               const char **fromP,
168                               const char *fromLim,
169                               unsigned short **toP,
170                               const unsigned short *toLim);
171  int minBytesPerChar;
172  char isUtf8;
173  char isUtf16;
174};
175
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
178   type of token.
179
180   Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
181
182   Return XML_TOK_PARTIAL when the string does not contain a complete
183   token; nextTokPtr will not be set.
184
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
187   the token invalid.
188
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.
191
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.
195*/
196
197
198#define XmlTok(enc, state, ptr, end, nextTokPtr) \
199  (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
200
201#define XmlPrologTok(enc, ptr, end, nextTokPtr) \
202   XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
203
204#define XmlContentTok(enc, ptr, end, nextTokPtr) \
205   XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
206
207#define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
208   XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
209
210#ifdef XML_DTD
211
212#define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
213   XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
214
215#endif /* XML_DTD */
216
217/* This is used for performing a 2nd-level tokenization on the content
218   of a literal that has already been returned by XmlTok.
219*/
220#define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
221  (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
222
223#define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
224   XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
225
226#define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
227   XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
228
229#define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
230
231#define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
232  (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
233
234#define XmlNameLength(enc, ptr) \
235  (((enc)->nameLength)(enc, ptr))
236
237#define XmlSkipS(enc, ptr) \
238  (((enc)->skipS)(enc, ptr))
239
240#define XmlGetAttributes(enc, ptr, attsMax, atts) \
241  (((enc)->getAtts)(enc, ptr, attsMax, atts))
242
243#define XmlCharRefNumber(enc, ptr) \
244  (((enc)->charRefNumber)(enc, ptr))
245
246#define XmlPredefinedEntityName(enc, ptr, end) \
247  (((enc)->predefinedEntityName)(enc, ptr, end))
248
249#define XmlUpdatePosition(enc, ptr, end, pos) \
250  (((enc)->updatePosition)(enc, ptr, end, pos))
251
252#define XmlIsPublicId(enc, ptr, end, badPtr) \
253  (((enc)->isPublicId)(enc, ptr, end, badPtr))
254
255#define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
256  (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
257
258#define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
259  (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
260
261typedef struct {
262  ENCODING initEnc;
263  const ENCODING **encPtr;
264} INIT_ENCODING;
265
266int XmlParseXmlDecl(int isGeneralTextEntity,
267                    const ENCODING *enc,
268                    const char *ptr,
269                    const char *end,
270                    const char **badPtr,
271                    const char **versionPtr,
272                    const char **versionEndPtr,
273                    const char **encodingNamePtr,
274                    const ENCODING **namedEncodingPtr,
275                    int *standalonePtr);
276
277int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
278const ENCODING *XmlGetUtf8InternalEncoding(void);
279const ENCODING *XmlGetUtf16InternalEncoding(void);
280int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
281int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
282int XmlSizeOfUnknownEncoding(void);
283
284
285typedef int (XMLCALL *CONVERTER) (void *userData, const char *p);
286
287ENCODING *
288XmlInitUnknownEncoding(void *mem,
289                       int *table,
290                       CONVERTER convert,
291                       void *userData);
292
293int XmlParseXmlDeclNS(int isGeneralTextEntity,
294                      const ENCODING *enc,
295                      const char *ptr,
296                      const char *end,
297                      const char **badPtr,
298                      const char **versionPtr,
299                      const char **versionEndPtr,
300                      const char **encodingNamePtr,
301                      const ENCODING **namedEncodingPtr,
302                      int *standalonePtr);
303
304int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
305const ENCODING *XmlGetUtf8InternalEncodingNS(void);
306const ENCODING *XmlGetUtf16InternalEncodingNS(void);
307ENCODING *
308XmlInitUnknownEncodingNS(void *mem,
309                         int *table,
310                         CONVERTER convert,
311                         void *userData);
312#ifdef __cplusplus
313}
314#endif
315
316#endif /* not XmlTok_INCLUDED */
317