1 /*
2 __ __ _
3 ___\ \/ /_ __ __ _| |_
4 / _ \\ /| '_ \ / _` | __|
5 | __// \| |_) | (_| | |_
6 \___/_/\_\ .__/ \__,_|\__|
7 |_| XML parser
8
9 Copyright (c) 1997-2000 Thai Open Source Software Center Ltd
10 Copyright (c) 2000 Clark Cooper <coopercc@users.sourceforge.net>
11 Copyright (c) 2002 Fred L. Drake, Jr. <fdrake@users.sourceforge.net>
12 Copyright (c) 2002-2005 Karl Waclawek <karl@waclawek.net>
13 Copyright (c) 2016-2017 Sebastian Pipping <sebastian@pipping.org>
14 Copyright (c) 2017 Rhodri James <rhodri@wildebeest.org.uk>
15 Licensed under the MIT license:
16
17 Permission is hereby granted, free of charge, to any person obtaining
18 a copy of this software and associated documentation files (the
19 "Software"), to deal in the Software without restriction, including
20 without limitation the rights to use, copy, modify, merge, publish,
21 distribute, sublicense, and/or sell copies of the Software, and to permit
22 persons to whom the Software is furnished to do so, subject to the
23 following conditions:
24
25 The above copyright notice and this permission notice shall be included
26 in all copies or substantial portions of the Software.
27
28 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
29 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
30 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN
31 NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
32 DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
33 OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
34 USE OR OTHER DEALINGS IN THE SOFTWARE.
35 */
36
37 #ifndef XmlTok_INCLUDED
38 #define XmlTok_INCLUDED 1
39
40 #ifdef __cplusplus
41 extern "C" {
42 #endif
43
44 /* The following token may be returned by XmlContentTok */
45 #define XML_TOK_TRAILING_RSQB \
46 -5 /* ] or ]] at the end of the scan; might be \
47 start of illegal ]]> sequence */
48 /* The following tokens may be returned by both XmlPrologTok and
49 XmlContentTok.
50 */
51 #define XML_TOK_NONE -4 /* The string to be scanned is empty */
52 #define XML_TOK_TRAILING_CR \
53 -3 /* A CR at the end of the scan; \
54 might be part of CRLF sequence */
55 #define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */
56 #define XML_TOK_PARTIAL -1 /* only part of a token */
57 #define XML_TOK_INVALID 0
58
59 /* The following tokens are returned by XmlContentTok; some are also
60 returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
61 */
62 #define XML_TOK_START_TAG_WITH_ATTS 1
63 #define XML_TOK_START_TAG_NO_ATTS 2
64 #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
65 #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
66 #define XML_TOK_END_TAG 5
67 #define XML_TOK_DATA_CHARS 6
68 #define XML_TOK_DATA_NEWLINE 7
69 #define XML_TOK_CDATA_SECT_OPEN 8
70 #define XML_TOK_ENTITY_REF 9
71 #define XML_TOK_CHAR_REF 10 /* numeric character reference */
72
73 /* The following tokens may be returned by both XmlPrologTok and
74 XmlContentTok.
75 */
76 #define XML_TOK_PI 11 /* processing instruction */
77 #define XML_TOK_XML_DECL 12 /* XML decl or text decl */
78 #define XML_TOK_COMMENT 13
79 #define XML_TOK_BOM 14 /* Byte order mark */
80
81 /* The following tokens are returned only by XmlPrologTok */
82 #define XML_TOK_PROLOG_S 15
83 #define XML_TOK_DECL_OPEN 16 /* <!foo */
84 #define XML_TOK_DECL_CLOSE 17 /* > */
85 #define XML_TOK_NAME 18
86 #define XML_TOK_NMTOKEN 19
87 #define XML_TOK_POUND_NAME 20 /* #name */
88 #define XML_TOK_OR 21 /* | */
89 #define XML_TOK_PERCENT 22
90 #define XML_TOK_OPEN_PAREN 23
91 #define XML_TOK_CLOSE_PAREN 24
92 #define XML_TOK_OPEN_BRACKET 25
93 #define XML_TOK_CLOSE_BRACKET 26
94 #define XML_TOK_LITERAL 27
95 #define XML_TOK_PARAM_ENTITY_REF 28
96 #define XML_TOK_INSTANCE_START 29
97
98 /* The following occur only in element type declarations */
99 #define XML_TOK_NAME_QUESTION 30 /* name? */
100 #define XML_TOK_NAME_ASTERISK 31 /* name* */
101 #define XML_TOK_NAME_PLUS 32 /* name+ */
102 #define XML_TOK_COND_SECT_OPEN 33 /* <![ */
103 #define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */
104 #define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */
105 #define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */
106 #define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */
107 #define XML_TOK_COMMA 38
108
109 /* The following token is returned only by XmlAttributeValueTok */
110 #define XML_TOK_ATTRIBUTE_VALUE_S 39
111
112 /* The following token is returned only by XmlCdataSectionTok */
113 #define XML_TOK_CDATA_SECT_CLOSE 40
114
115 /* With namespace processing this is returned by XmlPrologTok for a
116 name with a colon.
117 */
118 #define XML_TOK_PREFIXED_NAME 41
119
120 #ifdef XML_DTD
121 # define XML_TOK_IGNORE_SECT 42
122 #endif /* XML_DTD */
123
124 #ifdef XML_DTD
125 # define XML_N_STATES 4
126 #else /* not XML_DTD */
127 # define XML_N_STATES 3
128 #endif /* not XML_DTD */
129
130 #define XML_PROLOG_STATE 0
131 #define XML_CONTENT_STATE 1
132 #define XML_CDATA_SECTION_STATE 2
133 #ifdef XML_DTD
134 # define XML_IGNORE_SECTION_STATE 3
135 #endif /* XML_DTD */
136
137 #define XML_N_LITERAL_TYPES 2
138 #define XML_ATTRIBUTE_VALUE_LITERAL 0
139 #define XML_ENTITY_VALUE_LITERAL 1
140
141 /* The size of the buffer passed to XmlUtf8Encode must be at least this. */
142 #define XML_UTF8_ENCODE_MAX 4
143 /* The size of the buffer passed to XmlUtf16Encode must be at least this. */
144 #define XML_UTF16_ENCODE_MAX 2
145
146 typedef struct position {
147 /* first line and first column are 0 not 1 */
148 XML_Size lineNumber;
149 XML_Size columnNumber;
150 } POSITION;
151
152 typedef struct {
153 const char *name;
154 const char *valuePtr;
155 const char *valueEnd;
156 char normalized;
157 } ATTRIBUTE;
158
159 struct encoding;
160 typedef struct encoding ENCODING;
161
162 typedef int(PTRCALL *SCANNER)(const ENCODING *, const char *, const char *,
163 const char **);
164
165 enum XML_Convert_Result {
166 XML_CONVERT_COMPLETED = 0,
167 XML_CONVERT_INPUT_INCOMPLETE = 1,
168 XML_CONVERT_OUTPUT_EXHAUSTED
169 = 2 /* and therefore potentially input remaining as well */
170 };
171
172 struct encoding {
173 SCANNER scanners[XML_N_STATES];
174 SCANNER literalScanners[XML_N_LITERAL_TYPES];
175 int(PTRCALL *nameMatchesAscii)(const ENCODING *, const char *, const char *,
176 const char *);
177 int(PTRFASTCALL *nameLength)(const ENCODING *, const char *);
178 const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
179 int(PTRCALL *getAtts)(const ENCODING *enc, const char *ptr, int attsMax,
180 ATTRIBUTE *atts);
181 int(PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
182 int(PTRCALL *predefinedEntityName)(const ENCODING *, const char *,
183 const char *);
184 void(PTRCALL *updatePosition)(const ENCODING *, const char *ptr,
185 const char *end, POSITION *);
186 int(PTRCALL *isPublicId)(const ENCODING *enc, const char *ptr,
187 const char *end, const char **badPtr);
188 enum XML_Convert_Result(PTRCALL *utf8Convert)(const ENCODING *enc,
189 const char **fromP,
190 const char *fromLim, char **toP,
191 const char *toLim);
192 enum XML_Convert_Result(PTRCALL *utf16Convert)(const ENCODING *enc,
193 const char **fromP,
194 const char *fromLim,
195 unsigned short **toP,
196 const unsigned short *toLim);
197 int minBytesPerChar;
198 char isUtf8;
199 char isUtf16;
200 };
201
202 /* Scan the string starting at ptr until the end of the next complete
203 token, but do not scan past eptr. Return an integer giving the
204 type of token.
205
206 Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
207
208 Return XML_TOK_PARTIAL when the string does not contain a complete
209 token; nextTokPtr will not be set.
210
211 Return XML_TOK_INVALID when the string does not start a valid
212 token; nextTokPtr will be set to point to the character which made
213 the token invalid.
214
215 Otherwise the string starts with a valid token; nextTokPtr will be
216 set to point to the character following the end of that token.
217
218 Each data character counts as a single token, but adjacent data
219 characters may be returned together. Similarly for characters in
220 the prolog outside literals, comments and processing instructions.
221 */
222
223 #define XmlTok(enc, state, ptr, end, nextTokPtr) \
224 (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
225
226 #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
227 XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
228
229 #define XmlContentTok(enc, ptr, end, nextTokPtr) \
230 XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
231
232 #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
233 XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
234
235 #ifdef XML_DTD
236
237 # define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
238 XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
239
240 #endif /* XML_DTD */
241
242 /* This is used for performing a 2nd-level tokenization on the content
243 of a literal that has already been returned by XmlTok.
244 */
245 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
246 (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
247
248 #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
249 XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
250
251 #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
252 XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
253
254 #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
255 (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
256
257 #define XmlNameLength(enc, ptr) (((enc)->nameLength)(enc, ptr))
258
259 #define XmlSkipS(enc, ptr) (((enc)->skipS)(enc, ptr))
260
261 #define XmlGetAttributes(enc, ptr, attsMax, atts) \
262 (((enc)->getAtts)(enc, ptr, attsMax, atts))
263
264 #define XmlCharRefNumber(enc, ptr) (((enc)->charRefNumber)(enc, ptr))
265
266 #define XmlPredefinedEntityName(enc, ptr, end) \
267 (((enc)->predefinedEntityName)(enc, ptr, end))
268
269 #define XmlUpdatePosition(enc, ptr, end, pos) \
270 (((enc)->updatePosition)(enc, ptr, end, pos))
271
272 #define XmlIsPublicId(enc, ptr, end, badPtr) \
273 (((enc)->isPublicId)(enc, ptr, end, badPtr))
274
275 #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
276 (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
277
278 #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
279 (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
280
281 typedef struct {
282 ENCODING initEnc;
283 const ENCODING **encPtr;
284 } INIT_ENCODING;
285
286 int XmlParseXmlDecl(int isGeneralTextEntity, const ENCODING *enc,
287 const char *ptr, const char *end, const char **badPtr,
288 const char **versionPtr, const char **versionEndPtr,
289 const char **encodingNamePtr,
290 const ENCODING **namedEncodingPtr, int *standalonePtr);
291
292 int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
293 const ENCODING *XmlGetUtf8InternalEncoding(void);
294 const ENCODING *XmlGetUtf16InternalEncoding(void);
295 int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
296 int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
297 int XmlSizeOfUnknownEncoding(void);
298
299 typedef int(XMLCALL *CONVERTER)(void *userData, const char *p);
300
301 ENCODING *XmlInitUnknownEncoding(void *mem, int *table, CONVERTER convert,
302 void *userData);
303
304 int XmlParseXmlDeclNS(int isGeneralTextEntity, const ENCODING *enc,
305 const char *ptr, const char *end, const char **badPtr,
306 const char **versionPtr, const char **versionEndPtr,
307 const char **encodingNamePtr,
308 const ENCODING **namedEncodingPtr, int *standalonePtr);
309
310 int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
311 const ENCODING *XmlGetUtf8InternalEncodingNS(void);
312 const ENCODING *XmlGetUtf16InternalEncodingNS(void);
313 ENCODING *XmlInitUnknownEncodingNS(void *mem, int *table, CONVERTER convert,
314 void *userData);
315 #ifdef __cplusplus
316 }
317 #endif
318
319 #endif /* not XmlTok_INCLUDED */