1 | /* |
---|
2 | * Licensed to the Apache Software Foundation (ASF) under one or more |
---|
3 | * contributor license agreements. See the NOTICE file distributed with |
---|
4 | * this work for additional information regarding copyright ownership. |
---|
5 | * The ASF licenses this file to You under the Apache License, Version 2.0 |
---|
6 | * (the "License"); you may not use this file except in compliance with |
---|
7 | * the License. You may obtain a copy of the License at |
---|
8 | * |
---|
9 | * http://www.apache.org/licenses/LICENSE-2.0 |
---|
10 | * |
---|
11 | * Unless required by applicable law or agreed to in writing, software |
---|
12 | * distributed under the License is distributed on an "AS IS" BASIS, |
---|
13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
---|
14 | * See the License for the specific language governing permissions and |
---|
15 | * limitations under the License. |
---|
16 | */ |
---|
17 | |
---|
18 | /* |
---|
19 | * $Id: XMLChar.hpp 568078 2007-08-21 11:43:25Z amassari $ |
---|
20 | */ |
---|
21 | |
---|
22 | #if !defined(XMLCHAR_HPP) |
---|
23 | #define XMLCHAR_HPP |
---|
24 | |
---|
25 | #include <xercesc/util/XMLUniDefs.hpp> |
---|
26 | |
---|
27 | XERCES_CPP_NAMESPACE_BEGIN |
---|
28 | |
---|
29 | // --------------------------------------------------------------------------- |
---|
30 | // This file defines Char and utility that conforms to XML 1.0 and XML 1.1 |
---|
31 | // --------------------------------------------------------------------------- |
---|
32 | // Masks for the fgCharCharsTable1_0 array |
---|
33 | const XMLByte gNCNameCharMask = 0x1; |
---|
34 | const XMLByte gFirstNameCharMask = 0x2; |
---|
35 | const XMLByte gNameCharMask = 0x4; |
---|
36 | const XMLByte gPlainContentCharMask = 0x8; |
---|
37 | const XMLByte gSpecialStartTagCharMask = 0x10; |
---|
38 | const XMLByte gControlCharMask = 0x20; |
---|
39 | const XMLByte gXMLCharMask = 0x40; |
---|
40 | const XMLByte gWhitespaceCharMask = 0x80; |
---|
41 | |
---|
42 | // --------------------------------------------------------------------------- |
---|
43 | // This class is for XML 1.0 |
---|
44 | // --------------------------------------------------------------------------- |
---|
45 | class XMLUTIL_EXPORT XMLChar1_0 |
---|
46 | { |
---|
47 | public: |
---|
48 | // ----------------------------------------------------------------------- |
---|
49 | // Public, static methods, check the string |
---|
50 | // ----------------------------------------------------------------------- |
---|
51 | static bool isAllSpaces |
---|
52 | ( |
---|
53 | const XMLCh* const toCheck |
---|
54 | , const unsigned int count |
---|
55 | ); |
---|
56 | |
---|
57 | static bool containsWhiteSpace |
---|
58 | ( |
---|
59 | const XMLCh* const toCheck |
---|
60 | , const unsigned int count |
---|
61 | ); |
---|
62 | |
---|
63 | static bool isValidNmtoken |
---|
64 | ( |
---|
65 | const XMLCh* const toCheck |
---|
66 | , const unsigned int count |
---|
67 | ); |
---|
68 | |
---|
69 | static bool isValidName |
---|
70 | ( |
---|
71 | const XMLCh* const toCheck |
---|
72 | , const unsigned int count |
---|
73 | ); |
---|
74 | |
---|
75 | static bool isValidName |
---|
76 | ( |
---|
77 | const XMLCh* const toCheck |
---|
78 | ); |
---|
79 | |
---|
80 | static bool isValidNCName |
---|
81 | ( |
---|
82 | const XMLCh* const toCheck |
---|
83 | , const unsigned int count |
---|
84 | ); |
---|
85 | |
---|
86 | static bool isValidQName |
---|
87 | ( |
---|
88 | const XMLCh* const toCheck |
---|
89 | , const unsigned int count |
---|
90 | ); |
---|
91 | |
---|
92 | // ----------------------------------------------------------------------- |
---|
93 | // Public, static methods, check the XMLCh |
---|
94 | // surrogate pair is assumed if second parameter is not null |
---|
95 | // ----------------------------------------------------------------------- |
---|
96 | static bool isXMLLetter(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
97 | static bool isFirstNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
98 | static bool isNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
99 | static bool isPlainContentChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
100 | static bool isSpecialStartTagChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
101 | static bool isXMLChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
102 | static bool isWhitespace(const XMLCh toCheck); |
---|
103 | static bool isWhitespace(const XMLCh toCheck, const XMLCh toCheck2); |
---|
104 | static bool isControlChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
105 | |
---|
106 | static bool isPublicIdChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
107 | static bool isFirstNCNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
108 | static bool isNCNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
109 | |
---|
110 | // ----------------------------------------------------------------------- |
---|
111 | // Special Non-conformant Public, static methods |
---|
112 | // ----------------------------------------------------------------------- |
---|
113 | /** |
---|
114 | * Return true if NEL (0x85) and LSEP (0x2028) to be treated as white space char. |
---|
115 | */ |
---|
116 | static bool isNELRecognized(); |
---|
117 | |
---|
118 | /** |
---|
119 | * Method to enable NEL (0x85) and LSEP (0x2028) to be treated as white space char. |
---|
120 | */ |
---|
121 | static void enableNELWS(); |
---|
122 | |
---|
123 | private: |
---|
124 | // ----------------------------------------------------------------------- |
---|
125 | // Unimplemented constructors and operators |
---|
126 | // ----------------------------------------------------------------------- |
---|
127 | XMLChar1_0(); |
---|
128 | |
---|
129 | // ----------------------------------------------------------------------- |
---|
130 | // Static data members |
---|
131 | // |
---|
132 | // fgCharCharsTable1_0 |
---|
133 | // The character characteristics table. Bits in each byte, represent |
---|
134 | // the characteristics of each character. It is generated via some |
---|
135 | // code and then hard coded into the cpp file for speed. |
---|
136 | // |
---|
137 | // fNEL |
---|
138 | // Flag to respresents whether NEL and LSEP newline recognition is enabled |
---|
139 | // or disabled |
---|
140 | // ----------------------------------------------------------------------- |
---|
141 | static XMLByte fgCharCharsTable1_0[0x10000]; |
---|
142 | static bool enableNEL; |
---|
143 | |
---|
144 | friend class XMLReader; |
---|
145 | }; |
---|
146 | |
---|
147 | |
---|
148 | // --------------------------------------------------------------------------- |
---|
149 | // XMLReader: Public, static methods |
---|
150 | // --------------------------------------------------------------------------- |
---|
151 | inline bool XMLChar1_0::isXMLLetter(const XMLCh toCheck, const XMLCh toCheck2) |
---|
152 | { |
---|
153 | // An XML letter is a FirstNameChar minus ':' and '_'. |
---|
154 | if (!toCheck2) { |
---|
155 | return (((fgCharCharsTable1_0[toCheck] & gFirstNameCharMask) != 0) |
---|
156 | && (toCheck != chColon) && (toCheck != chUnderscore)); |
---|
157 | } |
---|
158 | return false; |
---|
159 | } |
---|
160 | |
---|
161 | inline bool XMLChar1_0::isFirstNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
162 | { |
---|
163 | if (!toCheck2) |
---|
164 | return ((fgCharCharsTable1_0[toCheck] & gFirstNameCharMask) != 0); |
---|
165 | return false; |
---|
166 | } |
---|
167 | |
---|
168 | inline bool XMLChar1_0::isFirstNCNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
169 | { |
---|
170 | if (!toCheck2) { |
---|
171 | return (((fgCharCharsTable1_0[toCheck] & gFirstNameCharMask) != 0) && (toCheck != chColon)); |
---|
172 | } |
---|
173 | |
---|
174 | return false; |
---|
175 | } |
---|
176 | |
---|
177 | inline bool XMLChar1_0::isNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
178 | { |
---|
179 | if (!toCheck2) |
---|
180 | return ((fgCharCharsTable1_0[toCheck] & gNameCharMask) != 0); |
---|
181 | return false; |
---|
182 | } |
---|
183 | |
---|
184 | inline bool XMLChar1_0::isNCNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
185 | { |
---|
186 | if (!toCheck2) |
---|
187 | return ((fgCharCharsTable1_0[toCheck] & gNCNameCharMask) != 0); |
---|
188 | return false; |
---|
189 | } |
---|
190 | |
---|
191 | inline bool XMLChar1_0::isPlainContentChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
192 | { |
---|
193 | if (!toCheck2) |
---|
194 | return ((fgCharCharsTable1_0[toCheck] & gPlainContentCharMask) != 0); |
---|
195 | else { |
---|
196 | if ((toCheck >= 0xD800) && (toCheck <= 0xDBFF)) |
---|
197 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
198 | return true; |
---|
199 | } |
---|
200 | return false; |
---|
201 | } |
---|
202 | |
---|
203 | |
---|
204 | inline bool XMLChar1_0::isSpecialStartTagChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
205 | { |
---|
206 | if (!toCheck2) |
---|
207 | return ((fgCharCharsTable1_0[toCheck] & gSpecialStartTagCharMask) != 0); |
---|
208 | return false; |
---|
209 | } |
---|
210 | |
---|
211 | inline bool XMLChar1_0::isXMLChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
212 | { |
---|
213 | if (!toCheck2) |
---|
214 | return ((fgCharCharsTable1_0[toCheck] & gXMLCharMask) != 0); |
---|
215 | else { |
---|
216 | if ((toCheck >= 0xD800) && (toCheck <= 0xDBFF)) |
---|
217 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
218 | return true; |
---|
219 | } |
---|
220 | return false; |
---|
221 | } |
---|
222 | |
---|
223 | inline bool XMLChar1_0::isWhitespace(const XMLCh toCheck) |
---|
224 | { |
---|
225 | return ((fgCharCharsTable1_0[toCheck] & gWhitespaceCharMask) != 0); |
---|
226 | } |
---|
227 | |
---|
228 | inline bool XMLChar1_0::isWhitespace(const XMLCh toCheck, const XMLCh toCheck2) |
---|
229 | { |
---|
230 | if (!toCheck2) |
---|
231 | return ((fgCharCharsTable1_0[toCheck] & gWhitespaceCharMask) != 0); |
---|
232 | return false; |
---|
233 | } |
---|
234 | |
---|
235 | inline bool XMLChar1_0::isControlChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
236 | { |
---|
237 | if (!toCheck2) |
---|
238 | return ((fgCharCharsTable1_0[toCheck] & gControlCharMask) != 0); |
---|
239 | return false; |
---|
240 | } |
---|
241 | |
---|
242 | inline bool XMLChar1_0::isNELRecognized() { |
---|
243 | |
---|
244 | return enableNEL; |
---|
245 | } |
---|
246 | |
---|
247 | |
---|
248 | // --------------------------------------------------------------------------- |
---|
249 | // This class is for XML 1.1 |
---|
250 | // --------------------------------------------------------------------------- |
---|
251 | class XMLUTIL_EXPORT XMLChar1_1 |
---|
252 | { |
---|
253 | public: |
---|
254 | // ----------------------------------------------------------------------- |
---|
255 | // Public, static methods, check the string |
---|
256 | // ----------------------------------------------------------------------- |
---|
257 | static bool isAllSpaces |
---|
258 | ( |
---|
259 | const XMLCh* const toCheck |
---|
260 | , const unsigned int count |
---|
261 | ); |
---|
262 | |
---|
263 | static bool containsWhiteSpace |
---|
264 | ( |
---|
265 | const XMLCh* const toCheck |
---|
266 | , const unsigned int count |
---|
267 | ); |
---|
268 | |
---|
269 | static bool isValidNmtoken |
---|
270 | ( |
---|
271 | const XMLCh* const toCheck |
---|
272 | , const unsigned int count |
---|
273 | ); |
---|
274 | |
---|
275 | static bool isValidName |
---|
276 | ( |
---|
277 | const XMLCh* const toCheck |
---|
278 | , const unsigned int count |
---|
279 | ); |
---|
280 | |
---|
281 | static bool isValidName |
---|
282 | ( |
---|
283 | const XMLCh* const toCheck |
---|
284 | ); |
---|
285 | |
---|
286 | static bool isValidNCName |
---|
287 | ( |
---|
288 | const XMLCh* const toCheck |
---|
289 | , const unsigned int count |
---|
290 | ); |
---|
291 | |
---|
292 | static bool isValidQName |
---|
293 | ( |
---|
294 | const XMLCh* const toCheck |
---|
295 | , const unsigned int count |
---|
296 | ); |
---|
297 | |
---|
298 | // ----------------------------------------------------------------------- |
---|
299 | // Public, static methods, check the XMLCh |
---|
300 | // ----------------------------------------------------------------------- |
---|
301 | static bool isXMLLetter(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
302 | static bool isFirstNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
303 | static bool isNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
304 | static bool isPlainContentChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
305 | static bool isSpecialStartTagChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
306 | static bool isXMLChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
307 | static bool isWhitespace(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
308 | static bool isControlChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
309 | |
---|
310 | static bool isPublicIdChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
311 | static bool isFirstNCNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
312 | static bool isNCNameChar(const XMLCh toCheck, const XMLCh toCheck2 = 0); |
---|
313 | |
---|
314 | private: |
---|
315 | // ----------------------------------------------------------------------- |
---|
316 | // Unimplemented constructors and operators |
---|
317 | // ----------------------------------------------------------------------- |
---|
318 | XMLChar1_1(); |
---|
319 | |
---|
320 | // ----------------------------------------------------------------------- |
---|
321 | // Static data members |
---|
322 | // |
---|
323 | // fgCharCharsTable1_1 |
---|
324 | // The character characteristics table. Bits in each byte, represent |
---|
325 | // the characteristics of each character. It is generated via some |
---|
326 | // code and then hard coded into the cpp file for speed. |
---|
327 | // |
---|
328 | // ----------------------------------------------------------------------- |
---|
329 | static XMLByte fgCharCharsTable1_1[0x10000]; |
---|
330 | |
---|
331 | friend class XMLReader; |
---|
332 | }; |
---|
333 | |
---|
334 | |
---|
335 | // --------------------------------------------------------------------------- |
---|
336 | // XMLReader: Public, static methods |
---|
337 | // --------------------------------------------------------------------------- |
---|
338 | inline bool XMLChar1_1::isXMLLetter(const XMLCh toCheck, const XMLCh toCheck2) |
---|
339 | { |
---|
340 | /** XML 1.1 does not define a letter, so we use the 1.0 definition */ |
---|
341 | return XMLChar1_0::isXMLLetter(toCheck, toCheck2); |
---|
342 | } |
---|
343 | |
---|
344 | inline bool XMLChar1_1::isFirstNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
345 | { |
---|
346 | if (!toCheck2) |
---|
347 | return ((fgCharCharsTable1_1[toCheck] & gFirstNameCharMask) != 0); |
---|
348 | else { |
---|
349 | if ((toCheck >= 0xD800) && (toCheck <= 0xDB7F)) |
---|
350 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
351 | return true; |
---|
352 | } |
---|
353 | return false; |
---|
354 | } |
---|
355 | |
---|
356 | inline bool XMLChar1_1::isFirstNCNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
357 | { |
---|
358 | if (!toCheck2) { |
---|
359 | return (((fgCharCharsTable1_1[toCheck] & gFirstNameCharMask) != 0) && (toCheck != chColon)); |
---|
360 | } |
---|
361 | else { |
---|
362 | if ((toCheck >= 0xD800) && (toCheck <= 0xDB7F)) |
---|
363 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
364 | return true; |
---|
365 | } |
---|
366 | return false; |
---|
367 | } |
---|
368 | |
---|
369 | inline bool XMLChar1_1::isNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
370 | { |
---|
371 | if (!toCheck2) |
---|
372 | return ((fgCharCharsTable1_1[toCheck] & gNameCharMask) != 0); |
---|
373 | else { |
---|
374 | if ((toCheck >= 0xD800) && (toCheck <= 0xDB7F)) |
---|
375 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
376 | return true; |
---|
377 | } |
---|
378 | return false; |
---|
379 | } |
---|
380 | |
---|
381 | inline bool XMLChar1_1::isNCNameChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
382 | { |
---|
383 | if (!toCheck2) |
---|
384 | return ((fgCharCharsTable1_1[toCheck] & gNCNameCharMask) != 0); |
---|
385 | else { |
---|
386 | if ((toCheck >= 0xD800) && (toCheck <= 0xDB7F)) |
---|
387 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
388 | return true; |
---|
389 | } |
---|
390 | return false; |
---|
391 | } |
---|
392 | |
---|
393 | inline bool XMLChar1_1::isPlainContentChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
394 | { |
---|
395 | if (!toCheck2) |
---|
396 | return ((fgCharCharsTable1_1[toCheck] & gPlainContentCharMask) != 0); |
---|
397 | else { |
---|
398 | if ((toCheck >= 0xD800) && (toCheck <= 0xDBFF)) |
---|
399 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
400 | return true; |
---|
401 | } |
---|
402 | return false; |
---|
403 | } |
---|
404 | |
---|
405 | |
---|
406 | inline bool XMLChar1_1::isSpecialStartTagChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
407 | { |
---|
408 | if (!toCheck2) |
---|
409 | return ((fgCharCharsTable1_1[toCheck] & gSpecialStartTagCharMask) != 0); |
---|
410 | return false; |
---|
411 | } |
---|
412 | |
---|
413 | inline bool XMLChar1_1::isXMLChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
414 | { |
---|
415 | if (!toCheck2) |
---|
416 | return ((fgCharCharsTable1_1[toCheck] & gXMLCharMask) != 0); |
---|
417 | else { |
---|
418 | if ((toCheck >= 0xD800) && (toCheck <= 0xDBFF)) |
---|
419 | if ((toCheck2 >= 0xDC00) && (toCheck2 <= 0xDFFF)) |
---|
420 | return true; |
---|
421 | } |
---|
422 | return false; |
---|
423 | } |
---|
424 | |
---|
425 | inline bool XMLChar1_1::isWhitespace(const XMLCh toCheck, const XMLCh toCheck2) |
---|
426 | { |
---|
427 | if (!toCheck2) |
---|
428 | return ((fgCharCharsTable1_1[toCheck] & gWhitespaceCharMask) != 0); |
---|
429 | return false; |
---|
430 | } |
---|
431 | |
---|
432 | inline bool XMLChar1_1::isControlChar(const XMLCh toCheck, const XMLCh toCheck2) |
---|
433 | { |
---|
434 | if (!toCheck2) |
---|
435 | return ((fgCharCharsTable1_1[toCheck] & gControlCharMask) != 0); |
---|
436 | return false; |
---|
437 | } |
---|
438 | |
---|
439 | |
---|
440 | XERCES_CPP_NAMESPACE_END |
---|
441 | |
---|
442 | #endif |
---|