1 /*
2                             __  __            _
3                          ___\ \/ /_ __   __ _| |_
4                         / _ \\  /| '_ \ / _` | __|
5                        |  __//  \| |_) | (_| | |_
6                         \___/_/\_\ .__/ \__,_|\__|
7                                  |_| XML parser
8 
9    Copyright (c) 1997-2000 Thai Open Source Software Center Ltd
10    Copyright (c) 2000-2017 Expat development team
11    Licensed under the MIT license:
12 
13    Permission is  hereby granted,  free of charge,  to any  person obtaining
14    a  copy  of  this  software   and  associated  documentation  files  (the
15    "Software"),  to  deal in  the  Software  without restriction,  including
16    without  limitation the  rights  to use,  copy,  modify, merge,  publish,
17    distribute, sublicense, and/or sell copies of the Software, and to permit
18    persons  to whom  the Software  is  furnished to  do so,  subject to  the
19    following conditions:
20 
21    The above copyright  notice and this permission notice  shall be included
22    in all copies or substantial portions of the Software.
23 
24    THE  SOFTWARE  IS  PROVIDED  "AS  IS",  WITHOUT  WARRANTY  OF  ANY  KIND,
25    EXPRESS  OR IMPLIED,  INCLUDING  BUT  NOT LIMITED  TO  THE WARRANTIES  OF
26    MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN
27    NO EVENT SHALL THE AUTHORS OR  COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
28    DAMAGES OR  OTHER LIABILITY, WHETHER  IN AN  ACTION OF CONTRACT,  TORT OR
29    OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
30    USE OR OTHER DEALINGS IN THE SOFTWARE.
31 */
32 
33 #ifndef XmlTok_INCLUDED
34 #define XmlTok_INCLUDED 1
35 
36 #ifdef __cplusplus
37 extern "C" {
38 #endif
39 
40 /* The following token may be returned by XmlContentTok */
41 #define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be
42                                     start of illegal ]]> sequence */
43 /* The following tokens may be returned by both XmlPrologTok and
44    XmlContentTok.
45 */
46 #define XML_TOK_NONE -4          /* The string to be scanned is empty */
47 #define XML_TOK_TRAILING_CR -3   /* A CR at the end of the scan;
48                                     might be part of CRLF sequence */
49 #define XML_TOK_PARTIAL_CHAR -2  /* only part of a multibyte sequence */
50 #define XML_TOK_PARTIAL -1       /* only part of a token */
51 #define XML_TOK_INVALID 0
52 
53 /* The following tokens are returned by XmlContentTok; some are also
54    returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
55 */
56 #define XML_TOK_START_TAG_WITH_ATTS 1
57 #define XML_TOK_START_TAG_NO_ATTS 2
58 #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
59 #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
60 #define XML_TOK_END_TAG 5
61 #define XML_TOK_DATA_CHARS 6
62 #define XML_TOK_DATA_NEWLINE 7
63 #define XML_TOK_CDATA_SECT_OPEN 8
64 #define XML_TOK_ENTITY_REF 9
65 #define XML_TOK_CHAR_REF 10               /* numeric character reference */
66 
67 /* The following tokens may be returned by both XmlPrologTok and
68    XmlContentTok.
69 */
70 #define XML_TOK_PI 11                     /* processing instruction */
71 #define XML_TOK_XML_DECL 12               /* XML decl or text decl */
72 #define XML_TOK_COMMENT 13
73 #define XML_TOK_BOM 14                    /* Byte order mark */
74 
75 /* The following tokens are returned only by XmlPrologTok */
76 #define XML_TOK_PROLOG_S 15
77 #define XML_TOK_DECL_OPEN 16              /* <!foo */
78 #define XML_TOK_DECL_CLOSE 17             /* > */
79 #define XML_TOK_NAME 18
80 #define XML_TOK_NMTOKEN 19
81 #define XML_TOK_POUND_NAME 20             /* #name */
82 #define XML_TOK_OR 21                     /* | */
83 #define XML_TOK_PERCENT 22
84 #define XML_TOK_OPEN_PAREN 23
85 #define XML_TOK_CLOSE_PAREN 24
86 #define XML_TOK_OPEN_BRACKET 25
87 #define XML_TOK_CLOSE_BRACKET 26
88 #define XML_TOK_LITERAL 27
89 #define XML_TOK_PARAM_ENTITY_REF 28
90 #define XML_TOK_INSTANCE_START 29
91 
92 /* The following occur only in element type declarations */
93 #define XML_TOK_NAME_QUESTION 30          /* name? */
94 #define XML_TOK_NAME_ASTERISK 31          /* name* */
95 #define XML_TOK_NAME_PLUS 32              /* name+ */
96 #define XML_TOK_COND_SECT_OPEN 33         /* <![ */
97 #define XML_TOK_COND_SECT_CLOSE 34        /* ]]> */
98 #define XML_TOK_CLOSE_PAREN_QUESTION 35   /* )? */
99 #define XML_TOK_CLOSE_PAREN_ASTERISK 36   /* )* */
100 #define XML_TOK_CLOSE_PAREN_PLUS 37       /* )+ */
101 #define XML_TOK_COMMA 38
102 
103 /* The following token is returned only by XmlAttributeValueTok */
104 #define XML_TOK_ATTRIBUTE_VALUE_S 39
105 
106 /* The following token is returned only by XmlCdataSectionTok */
107 #define XML_TOK_CDATA_SECT_CLOSE 40
108 
109 /* With namespace processing this is returned by XmlPrologTok for a
110    name with a colon.
111 */
112 #define XML_TOK_PREFIXED_NAME 41
113 
114 #ifdef XML_DTD
115 #define XML_TOK_IGNORE_SECT 42
116 #endif /* XML_DTD */
117 
118 #ifdef XML_DTD
119 #define XML_N_STATES 4
120 #else /* not XML_DTD */
121 #define XML_N_STATES 3
122 #endif /* not XML_DTD */
123 
124 #define XML_PROLOG_STATE 0
125 #define XML_CONTENT_STATE 1
126 #define XML_CDATA_SECTION_STATE 2
127 #ifdef XML_DTD
128 #define XML_IGNORE_SECTION_STATE 3
129 #endif /* XML_DTD */
130 
131 #define XML_N_LITERAL_TYPES 2
132 #define XML_ATTRIBUTE_VALUE_LITERAL 0
133 #define XML_ENTITY_VALUE_LITERAL 1
134 
135 /* The size of the buffer passed to XmlUtf8Encode must be at least this. */
136 #define XML_UTF8_ENCODE_MAX 4
137 /* The size of the buffer passed to XmlUtf16Encode must be at least this. */
138 #define XML_UTF16_ENCODE_MAX 2
139 
140 typedef struct position {
141   /* first line and first column are 0 not 1 */
142   XML_Size lineNumber;
143   XML_Size columnNumber;
144 } POSITION;
145 
146 typedef struct {
147   const char *name;
148   const char *valuePtr;
149   const char *valueEnd;
150   char normalized;
151 } ATTRIBUTE;
152 
153 struct encoding;
154 typedef struct encoding ENCODING;
155 
156 typedef int (PTRCALL *SCANNER)(const ENCODING *,
157                                const char *,
158                                const char *,
159                                const char **);
160 
161 enum XML_Convert_Result {
162   XML_CONVERT_COMPLETED = 0,
163   XML_CONVERT_INPUT_INCOMPLETE = 1,
164   XML_CONVERT_OUTPUT_EXHAUSTED = 2  /* and therefore potentially input remaining as well */
165 };
166 
167 struct encoding {
168   SCANNER scanners[XML_N_STATES];
169   SCANNER literalScanners[XML_N_LITERAL_TYPES];
170   int (PTRCALL *nameMatchesAscii)(const ENCODING *,
171                                   const char *,
172                                   const char *,
173                                   const char *);
174   int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
175   const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
176   int (PTRCALL *getAtts)(const ENCODING *enc,
177                          const char *ptr,
178                          int attsMax,
179                          ATTRIBUTE *atts);
180   int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
181   int (PTRCALL *predefinedEntityName)(const ENCODING *,
182                                       const char *,
183                                       const char *);
184   void (PTRCALL *updatePosition)(const ENCODING *,
185                                  const char *ptr,
186                                  const char *end,
187                                  POSITION *);
188   int (PTRCALL *isPublicId)(const ENCODING *enc,
189                             const char *ptr,
190                             const char *end,
191                             const char **badPtr);
192   enum XML_Convert_Result (PTRCALL *utf8Convert)(const ENCODING *enc,
193                               const char **fromP,
194                               const char *fromLim,
195                               char **toP,
196                               const char *toLim);
197   enum XML_Convert_Result (PTRCALL *utf16Convert)(const ENCODING *enc,
198                                const char **fromP,
199                                const char *fromLim,
200                                unsigned short **toP,
201                                const unsigned short *toLim);
202   int minBytesPerChar;
203   char isUtf8;
204   char isUtf16;
205 };
206 
207 /* Scan the string starting at ptr until the end of the next complete
208    token, but do not scan past eptr.  Return an integer giving the
209    type of token.
210 
211    Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
212 
213    Return XML_TOK_PARTIAL when the string does not contain a complete
214    token; nextTokPtr will not be set.
215 
216    Return XML_TOK_INVALID when the string does not start a valid
217    token; nextTokPtr will be set to point to the character which made
218    the token invalid.
219 
220    Otherwise the string starts with a valid token; nextTokPtr will be
221    set to point to the character following the end of that token.
222 
223    Each data character counts as a single token, but adjacent data
224    characters may be returned together.  Similarly for characters in
225    the prolog outside literals, comments and processing instructions.
226 */
227 
228 
229 #define XmlTok(enc, state, ptr, end, nextTokPtr) \
230   (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
231 
232 #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
233    XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
234 
235 #define XmlContentTok(enc, ptr, end, nextTokPtr) \
236    XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
237 
238 #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
239    XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
240 
241 #ifdef XML_DTD
242 
243 #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
244    XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
245 
246 #endif /* XML_DTD */
247 
248 /* This is used for performing a 2nd-level tokenization on the content
249    of a literal that has already been returned by XmlTok.
250 */
251 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
252   (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
253 
254 #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
255    XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
256 
257 #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
258    XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
259 
260 #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
261   (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
262 
263 #define XmlNameLength(enc, ptr) \
264   (((enc)->nameLength)(enc, ptr))
265 
266 #define XmlSkipS(enc, ptr) \
267   (((enc)->skipS)(enc, ptr))
268 
269 #define XmlGetAttributes(enc, ptr, attsMax, atts) \
270   (((enc)->getAtts)(enc, ptr, attsMax, atts))
271 
272 #define XmlCharRefNumber(enc, ptr) \
273   (((enc)->charRefNumber)(enc, ptr))
274 
275 #define XmlPredefinedEntityName(enc, ptr, end) \
276   (((enc)->predefinedEntityName)(enc, ptr, end))
277 
278 #define XmlUpdatePosition(enc, ptr, end, pos) \
279   (((enc)->updatePosition)(enc, ptr, end, pos))
280 
281 #define XmlIsPublicId(enc, ptr, end, badPtr) \
282   (((enc)->isPublicId)(enc, ptr, end, badPtr))
283 
284 #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
285   (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
286 
287 #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
288   (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
289 
290 typedef struct {
291   ENCODING initEnc;
292   const ENCODING **encPtr;
293 } INIT_ENCODING;
294 
295 int XmlParseXmlDecl(int isGeneralTextEntity,
296                     const ENCODING *enc,
297                     const char *ptr,
298                     const char *end,
299                     const char **badPtr,
300                     const char **versionPtr,
301                     const char **versionEndPtr,
302                     const char **encodingNamePtr,
303                     const ENCODING **namedEncodingPtr,
304                     int *standalonePtr);
305 
306 int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
307 const ENCODING *XmlGetUtf8InternalEncoding(void);
308 const ENCODING *XmlGetUtf16InternalEncoding(void);
309 int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
310 int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
311 int XmlSizeOfUnknownEncoding(void);
312 
313 
314 typedef int (XMLCALL *CONVERTER) (void *userData, const char *p);
315 
316 ENCODING *
317 XmlInitUnknownEncoding(void *mem,
318                        int *table,
319                        CONVERTER convert,
320                        void *userData);
321 
322 int XmlParseXmlDeclNS(int isGeneralTextEntity,
323                       const ENCODING *enc,
324                       const char *ptr,
325                       const char *end,
326                       const char **badPtr,
327                       const char **versionPtr,
328                       const char **versionEndPtr,
329                       const char **encodingNamePtr,
330                       const ENCODING **namedEncodingPtr,
331                       int *standalonePtr);
332 
333 int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
334 const ENCODING *XmlGetUtf8InternalEncodingNS(void);
335 const ENCODING *XmlGetUtf16InternalEncodingNS(void);
336 ENCODING *
337 XmlInitUnknownEncodingNS(void *mem,
338                          int *table,
339                          CONVERTER convert,
340                          void *userData);
341 #ifdef __cplusplus
342 }
343 #endif
344 
345 #endif /* not XmlTok_INCLUDED */
346