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: XMLFormatter.hpp 568078 2007-08-21 11:43:25Z amassari $ |
---|
20 | */ |
---|
21 | |
---|
22 | #if !defined(XMLFORMATTER_HPP) |
---|
23 | #define XMLFORMATTER_HPP |
---|
24 | |
---|
25 | #include <xercesc/util/PlatformUtils.hpp> |
---|
26 | |
---|
27 | XERCES_CPP_NAMESPACE_BEGIN |
---|
28 | |
---|
29 | class XMLFormatTarget; |
---|
30 | class XMLTranscoder; |
---|
31 | |
---|
32 | /** |
---|
33 | * This class provides the basic formatting capabilities that are required |
---|
34 | * to turn the Unicode based XML data from the parsers into a form that can |
---|
35 | * be used on non-Unicode based systems, that is, into local or generic text |
---|
36 | * encodings. |
---|
37 | * |
---|
38 | * A number of flags are provided to control whether various optional |
---|
39 | * formatting operations are performed. |
---|
40 | */ |
---|
41 | class XMLPARSER_EXPORT XMLFormatter : public XMemory |
---|
42 | { |
---|
43 | public: |
---|
44 | // ----------------------------------------------------------------------- |
---|
45 | // Class types |
---|
46 | // ----------------------------------------------------------------------- |
---|
47 | /** @name Public Contants */ |
---|
48 | //@{ |
---|
49 | /** |
---|
50 | * EscapeFlags - Different styles of escape flags to control various formatting. |
---|
51 | * |
---|
52 | * <p><code>NoEscapes:</code> |
---|
53 | * No character needs to be escaped. Just write them out as is.</p> |
---|
54 | * <p><code>StdEscapes:</code> |
---|
55 | * The following characters need to be escaped:</p> |
---|
56 | * <table border='1'> |
---|
57 | * <tr> |
---|
58 | * <td>character</td> |
---|
59 | * <td>should be escaped and written as</td> |
---|
60 | * </tr> |
---|
61 | * <tr> |
---|
62 | * <td valign='top' rowspan='1' colspan='1'>&</td> |
---|
63 | * <td valign='top' rowspan='1' colspan='1'>&amp;</td> |
---|
64 | * </tr> |
---|
65 | * <tr> |
---|
66 | * <td valign='top' rowspan='1' colspan='1'>></td> |
---|
67 | * <td valign='top' rowspan='1' colspan='1'>&gt;</td> |
---|
68 | * </tr> |
---|
69 | * <tr> |
---|
70 | * <td valign='top' rowspan='1' colspan='1'>"</td> |
---|
71 | * <td valign='top' rowspan='1' colspan='1'>&quot;</td> |
---|
72 | * </tr> |
---|
73 | * <tr> |
---|
74 | * <td valign='top' rowspan='1' colspan='1'><</td> |
---|
75 | * <td valign='top' rowspan='1' colspan='1'>&lt;</td> |
---|
76 | * </tr> |
---|
77 | * <tr> |
---|
78 | * <td valign='top' rowspan='1' colspan='1'>'</td> |
---|
79 | * <td valign='top' rowspan='1' colspan='1'>&apos;</td> |
---|
80 | * </tr> |
---|
81 | * </table> |
---|
82 | * <p><code>AttrEscapes:</code> |
---|
83 | * The following characters need to be escaped:</p> |
---|
84 | * <table border='1'> |
---|
85 | * <tr> |
---|
86 | * <td>character</td> |
---|
87 | * <td>should be escaped and written as</td> |
---|
88 | * </tr> |
---|
89 | * <tr> |
---|
90 | * <td valign='top' rowspan='1' colspan='1'>&</td> |
---|
91 | * <td valign='top' rowspan='1' colspan='1'>&amp;</td> |
---|
92 | * </tr> |
---|
93 | * <tr> |
---|
94 | * <td valign='top' rowspan='1' colspan='1'>></td> |
---|
95 | * <td valign='top' rowspan='1' colspan='1'>&gt;</td> |
---|
96 | * </tr> |
---|
97 | * <tr> |
---|
98 | * <td valign='top' rowspan='1' colspan='1'>"</td> |
---|
99 | * <td valign='top' rowspan='1' colspan='1'>&quot;</td> |
---|
100 | * </tr> |
---|
101 | * </table> |
---|
102 | * <p><code>CharEscapes:</code> |
---|
103 | * The following characters need to be escaped:</p> |
---|
104 | * <table border='1'> |
---|
105 | * <tr> |
---|
106 | * <td>character</td> |
---|
107 | * <td>should be escaped and written as</td> |
---|
108 | * </tr> |
---|
109 | * <tr> |
---|
110 | * <td valign='top' rowspan='1' colspan='1'>&</td> |
---|
111 | * <td valign='top' rowspan='1' colspan='1'>&amp;</td> |
---|
112 | * </tr> |
---|
113 | * <tr> |
---|
114 | * <td valign='top' rowspan='1' colspan='1'>></td> |
---|
115 | * <td valign='top' rowspan='1' colspan='1'>&gt;</td> |
---|
116 | * </tr> |
---|
117 | * </table> |
---|
118 | * <p><code>EscapeFlags_Count:</code> |
---|
119 | * Special value, do not use directly.</p> |
---|
120 | * <p><code>DefaultEscape:</code> |
---|
121 | * Special value, do not use directly.</p> |
---|
122 | * |
---|
123 | */ |
---|
124 | enum EscapeFlags |
---|
125 | { |
---|
126 | NoEscapes |
---|
127 | , StdEscapes |
---|
128 | , AttrEscapes |
---|
129 | , CharEscapes |
---|
130 | |
---|
131 | // Special values, don't use directly |
---|
132 | , EscapeFlags_Count |
---|
133 | , DefaultEscape = 999 |
---|
134 | }; |
---|
135 | |
---|
136 | /** |
---|
137 | * UnRepFlags |
---|
138 | * |
---|
139 | * The unrepresentable flags that indicate how to react when a |
---|
140 | * character cannot be represented in the target encoding. |
---|
141 | * |
---|
142 | * <p><code>UnRep_Fail:</code> |
---|
143 | * Fail the operation.</p> |
---|
144 | * <p><code>UnRep_CharRef:</code> |
---|
145 | * Display the unrepresented character as reference.</p> |
---|
146 | * <p><code>UnRep_Replace:</code> |
---|
147 | * Replace the unrepresented character with the replacement character.</p> |
---|
148 | * <p><code>DefaultUnRep:</code> |
---|
149 | * Special value, do not use directly.</p> |
---|
150 | * |
---|
151 | */ |
---|
152 | enum UnRepFlags |
---|
153 | { |
---|
154 | UnRep_Fail |
---|
155 | , UnRep_CharRef |
---|
156 | , UnRep_Replace |
---|
157 | |
---|
158 | , DefaultUnRep = 999 |
---|
159 | }; |
---|
160 | //@} |
---|
161 | |
---|
162 | |
---|
163 | // ----------------------------------------------------------------------- |
---|
164 | // Constructors and Destructor |
---|
165 | // ----------------------------------------------------------------------- |
---|
166 | /** @name Constructor and Destructor */ |
---|
167 | //@{ |
---|
168 | /** |
---|
169 | * @param outEncoding the encoding for the formatted content |
---|
170 | * @param docVersion |
---|
171 | * @param target the formatTarget where the formatted content is written to |
---|
172 | * @param escapeFlags the escape style for certain character |
---|
173 | * @param unrepFlags the reaction to unrepresentable character |
---|
174 | * @param manager Pointer to the memory manager to be used to |
---|
175 | * allocate objects. |
---|
176 | */ |
---|
177 | XMLFormatter |
---|
178 | ( |
---|
179 | const XMLCh* const outEncoding |
---|
180 | , const XMLCh* const docVersion |
---|
181 | , XMLFormatTarget* const target |
---|
182 | , const EscapeFlags escapeFlags = NoEscapes |
---|
183 | , const UnRepFlags unrepFlags = UnRep_Fail |
---|
184 | , MemoryManager* const manager = XMLPlatformUtils::fgMemoryManager |
---|
185 | ); |
---|
186 | |
---|
187 | XMLFormatter |
---|
188 | ( |
---|
189 | const char* const outEncoding |
---|
190 | , const char* const docVersion |
---|
191 | , XMLFormatTarget* const target |
---|
192 | , const EscapeFlags escapeFlags = NoEscapes |
---|
193 | , const UnRepFlags unrepFlags = UnRep_Fail |
---|
194 | , MemoryManager* const manager = XMLPlatformUtils::fgMemoryManager |
---|
195 | ); |
---|
196 | |
---|
197 | XMLFormatter |
---|
198 | ( |
---|
199 | const XMLCh* const outEncoding |
---|
200 | , XMLFormatTarget* const target |
---|
201 | , const EscapeFlags escapeFlags = NoEscapes |
---|
202 | , const UnRepFlags unrepFlags = UnRep_Fail |
---|
203 | , MemoryManager* const manager = XMLPlatformUtils::fgMemoryManager |
---|
204 | ); |
---|
205 | |
---|
206 | XMLFormatter |
---|
207 | ( |
---|
208 | const char* const outEncoding |
---|
209 | , XMLFormatTarget* const target |
---|
210 | , const EscapeFlags escapeFlags = NoEscapes |
---|
211 | , const UnRepFlags unrepFlags = UnRep_Fail |
---|
212 | , MemoryManager* const manager = XMLPlatformUtils::fgMemoryManager |
---|
213 | ); |
---|
214 | |
---|
215 | ~XMLFormatter(); |
---|
216 | //@} |
---|
217 | |
---|
218 | |
---|
219 | // ----------------------------------------------------------------------- |
---|
220 | // Formatting methods |
---|
221 | // ----------------------------------------------------------------------- |
---|
222 | /** @name Formatting methods */ |
---|
223 | //@{ |
---|
224 | /** |
---|
225 | * @param toFormat the string to be formatted |
---|
226 | * @param count length of the string |
---|
227 | * @param escapeFlags the escape style for formatting toFormat |
---|
228 | * @param unrepFlags the reaction for any unrepresentable character in toFormat |
---|
229 | * |
---|
230 | */ |
---|
231 | void formatBuf |
---|
232 | ( |
---|
233 | const XMLCh* const toFormat |
---|
234 | , const unsigned int count |
---|
235 | , const EscapeFlags escapeFlags = DefaultEscape |
---|
236 | , const UnRepFlags unrepFlags = DefaultUnRep |
---|
237 | ); |
---|
238 | |
---|
239 | /** |
---|
240 | * @see formatBuf |
---|
241 | */ |
---|
242 | XMLFormatter& operator<< |
---|
243 | ( |
---|
244 | const XMLCh* const toFormat |
---|
245 | ); |
---|
246 | |
---|
247 | XMLFormatter& operator<< |
---|
248 | ( |
---|
249 | const XMLCh toFormat |
---|
250 | ); |
---|
251 | |
---|
252 | void writeBOM(const XMLByte* const toFormat |
---|
253 | , const unsigned int count); |
---|
254 | |
---|
255 | //@} |
---|
256 | |
---|
257 | // ----------------------------------------------------------------------- |
---|
258 | // Getter methods |
---|
259 | // ----------------------------------------------------------------------- |
---|
260 | /** @name Getter methods */ |
---|
261 | //@{ |
---|
262 | /** |
---|
263 | * @return return the encoding set for the formatted content |
---|
264 | */ |
---|
265 | |
---|
266 | const XMLCh* getEncodingName() const; |
---|
267 | |
---|
268 | /** |
---|
269 | * @return return the transcoder used internally for transcoding the formatter conent |
---|
270 | */ |
---|
271 | inline const XMLTranscoder* getTranscoder() const; |
---|
272 | |
---|
273 | //@} |
---|
274 | |
---|
275 | // ----------------------------------------------------------------------- |
---|
276 | // Setter methods |
---|
277 | // ----------------------------------------------------------------------- |
---|
278 | /** @name Setter methods */ |
---|
279 | //@{ |
---|
280 | /** |
---|
281 | * @param newFlags set the escape style for the follow-on formatted content |
---|
282 | */ |
---|
283 | void setEscapeFlags |
---|
284 | ( |
---|
285 | const EscapeFlags newFlags |
---|
286 | ); |
---|
287 | |
---|
288 | /** |
---|
289 | * @param newFlags set the reaction for unrepresentable character |
---|
290 | */ |
---|
291 | void setUnRepFlags |
---|
292 | ( |
---|
293 | const UnRepFlags newFlags |
---|
294 | ); |
---|
295 | |
---|
296 | /** |
---|
297 | * @param newFlags set the escape style for the follow-on formatted content |
---|
298 | * @see setEscapeFlags |
---|
299 | */ |
---|
300 | XMLFormatter& operator<< |
---|
301 | ( |
---|
302 | const EscapeFlags newFlags |
---|
303 | ); |
---|
304 | |
---|
305 | /** |
---|
306 | * @param newFlags set the reaction for unrepresentable character |
---|
307 | * @see setUnRepFlags |
---|
308 | */ |
---|
309 | XMLFormatter& operator<< |
---|
310 | ( |
---|
311 | const UnRepFlags newFlags |
---|
312 | ); |
---|
313 | //@} |
---|
314 | |
---|
315 | |
---|
316 | // ----------------------------------------------------------------------- |
---|
317 | // Getter methods |
---|
318 | // ----------------------------------------------------------------------- |
---|
319 | /** @name Setter methods */ |
---|
320 | //@{ |
---|
321 | /** |
---|
322 | * @return return the escape style for the formatted content |
---|
323 | */ |
---|
324 | EscapeFlags getEscapeFlags() const; |
---|
325 | |
---|
326 | /** |
---|
327 | * @return return the reaction for unrepresentable character |
---|
328 | */ |
---|
329 | UnRepFlags getUnRepFlags() const; |
---|
330 | //@} |
---|
331 | |
---|
332 | private : |
---|
333 | // ----------------------------------------------------------------------- |
---|
334 | // Unimplemented constructors and operators |
---|
335 | // ----------------------------------------------------------------------- |
---|
336 | XMLFormatter(); |
---|
337 | XMLFormatter(const XMLFormatter&); |
---|
338 | XMLFormatter& operator=(const XMLFormatter&); |
---|
339 | |
---|
340 | |
---|
341 | // ----------------------------------------------------------------------- |
---|
342 | // Private class constants |
---|
343 | // ----------------------------------------------------------------------- |
---|
344 | enum Constants |
---|
345 | { |
---|
346 | kTmpBufSize = 16 * 1024 |
---|
347 | }; |
---|
348 | |
---|
349 | |
---|
350 | // ----------------------------------------------------------------------- |
---|
351 | // Private helper methods |
---|
352 | // ----------------------------------------------------------------------- |
---|
353 | const XMLByte* getCharRef(unsigned int & count, |
---|
354 | XMLByte* &ref, |
---|
355 | const XMLCh * stdRef); |
---|
356 | |
---|
357 | void writeCharRef(const XMLCh &toWrite); |
---|
358 | void writeCharRef(unsigned long toWrite); |
---|
359 | |
---|
360 | bool inEscapeList(const XMLFormatter::EscapeFlags escStyle |
---|
361 | , const XMLCh toCheck); |
---|
362 | |
---|
363 | |
---|
364 | unsigned int handleUnEscapedChars(const XMLCh * srcPtr, |
---|
365 | const unsigned int count, |
---|
366 | const UnRepFlags unrepFlags); |
---|
367 | |
---|
368 | void specialFormat |
---|
369 | ( |
---|
370 | const XMLCh* const toFormat |
---|
371 | , const unsigned int count |
---|
372 | , const EscapeFlags escapeFlags |
---|
373 | ); |
---|
374 | |
---|
375 | |
---|
376 | // ----------------------------------------------------------------------- |
---|
377 | // Private, non-virtual methods |
---|
378 | // |
---|
379 | // fEscapeFlags |
---|
380 | // The escape flags we were told to use in formatting. These are |
---|
381 | // defaults set in the ctor, which can be overridden on a particular |
---|
382 | // call. |
---|
383 | // |
---|
384 | // fOutEncoding |
---|
385 | // This the name of the output encoding. Saved mainly for meaningful |
---|
386 | // error messages. |
---|
387 | // |
---|
388 | // fTarget |
---|
389 | // This is the target object for the formatting operation. |
---|
390 | // |
---|
391 | // fUnRepFlags |
---|
392 | // The unrepresentable flags that indicate how to react when a |
---|
393 | // character cannot be represented in the target encoding. |
---|
394 | // |
---|
395 | // fXCoder |
---|
396 | // This the transcoder that we will use. It is created using the |
---|
397 | // encoding name we were told to use. |
---|
398 | // |
---|
399 | // fTmpBuf |
---|
400 | // An output buffer that we use to transcode chars into before we |
---|
401 | // send them off to be output. |
---|
402 | // |
---|
403 | // fAposRef |
---|
404 | // fAmpRef |
---|
405 | // fGTRef |
---|
406 | // fLTRef |
---|
407 | // fQuoteRef |
---|
408 | // These are character refs for the standard char refs, in the |
---|
409 | // output encoding. They are faulted in as required, by transcoding |
---|
410 | // them from fixed Unicode versions. |
---|
411 | // |
---|
412 | // fIsXML11 |
---|
413 | // for performance reason, we do not store the actual version string |
---|
414 | // and do the string comparison again and again. |
---|
415 | // |
---|
416 | // ----------------------------------------------------------------------- |
---|
417 | EscapeFlags fEscapeFlags; |
---|
418 | XMLCh* fOutEncoding; |
---|
419 | XMLFormatTarget* fTarget; |
---|
420 | UnRepFlags fUnRepFlags; |
---|
421 | XMLTranscoder* fXCoder; |
---|
422 | XMLByte fTmpBuf[kTmpBufSize + 4]; |
---|
423 | XMLByte* fAposRef; |
---|
424 | unsigned int fAposLen; |
---|
425 | XMLByte* fAmpRef; |
---|
426 | unsigned int fAmpLen; |
---|
427 | XMLByte* fGTRef; |
---|
428 | unsigned int fGTLen; |
---|
429 | XMLByte* fLTRef; |
---|
430 | unsigned int fLTLen; |
---|
431 | XMLByte* fQuoteRef; |
---|
432 | unsigned int fQuoteLen; |
---|
433 | bool fIsXML11; |
---|
434 | MemoryManager* fMemoryManager; |
---|
435 | }; |
---|
436 | |
---|
437 | |
---|
438 | class XMLPARSER_EXPORT XMLFormatTarget : public XMemory |
---|
439 | { |
---|
440 | public: |
---|
441 | // ----------------------------------------------------------------------- |
---|
442 | // Constructors and Destructor |
---|
443 | // ----------------------------------------------------------------------- |
---|
444 | virtual ~XMLFormatTarget() {} |
---|
445 | |
---|
446 | |
---|
447 | // ----------------------------------------------------------------------- |
---|
448 | // Virtual interface |
---|
449 | // ----------------------------------------------------------------------- |
---|
450 | virtual void writeChars |
---|
451 | ( |
---|
452 | const XMLByte* const toWrite |
---|
453 | , const unsigned int count |
---|
454 | , XMLFormatter* const formatter |
---|
455 | ) = 0; |
---|
456 | |
---|
457 | virtual void flush() {}; |
---|
458 | |
---|
459 | |
---|
460 | protected : |
---|
461 | // ----------------------------------------------------------------------- |
---|
462 | // Hidden constructors and operators |
---|
463 | // ----------------------------------------------------------------------- |
---|
464 | XMLFormatTarget() {}; |
---|
465 | |
---|
466 | private: |
---|
467 | // ----------------------------------------------------------------------- |
---|
468 | // Unimplemented constructors and operators |
---|
469 | // ----------------------------------------------------------------------- |
---|
470 | XMLFormatTarget(const XMLFormatTarget&); |
---|
471 | XMLFormatTarget& operator=(const XMLFormatTarget&); |
---|
472 | }; |
---|
473 | |
---|
474 | |
---|
475 | // --------------------------------------------------------------------------- |
---|
476 | // XMLFormatter: Getter methods |
---|
477 | // --------------------------------------------------------------------------- |
---|
478 | inline const XMLCh* XMLFormatter::getEncodingName() const |
---|
479 | { |
---|
480 | return fOutEncoding; |
---|
481 | } |
---|
482 | |
---|
483 | inline const XMLTranscoder* XMLFormatter::getTranscoder() const |
---|
484 | { |
---|
485 | return fXCoder; |
---|
486 | } |
---|
487 | |
---|
488 | // --------------------------------------------------------------------------- |
---|
489 | // XMLFormatter: Setter methods |
---|
490 | // --------------------------------------------------------------------------- |
---|
491 | inline void XMLFormatter::setEscapeFlags(const EscapeFlags newFlags) |
---|
492 | { |
---|
493 | fEscapeFlags = newFlags; |
---|
494 | } |
---|
495 | |
---|
496 | inline void XMLFormatter::setUnRepFlags(const UnRepFlags newFlags) |
---|
497 | { |
---|
498 | fUnRepFlags = newFlags; |
---|
499 | } |
---|
500 | |
---|
501 | |
---|
502 | inline XMLFormatter& XMLFormatter::operator<<(const EscapeFlags newFlags) |
---|
503 | { |
---|
504 | fEscapeFlags = newFlags; |
---|
505 | return *this; |
---|
506 | } |
---|
507 | |
---|
508 | inline XMLFormatter& XMLFormatter::operator<<(const UnRepFlags newFlags) |
---|
509 | { |
---|
510 | fUnRepFlags = newFlags; |
---|
511 | return *this; |
---|
512 | } |
---|
513 | |
---|
514 | // --------------------------------------------------------------------------- |
---|
515 | // XMLFormatter: Getter methods |
---|
516 | // --------------------------------------------------------------------------- |
---|
517 | inline XMLFormatter::EscapeFlags XMLFormatter::getEscapeFlags() const |
---|
518 | { |
---|
519 | return fEscapeFlags; |
---|
520 | } |
---|
521 | |
---|
522 | inline XMLFormatter::UnRepFlags XMLFormatter::getUnRepFlags() const |
---|
523 | { |
---|
524 | return fUnRepFlags; |
---|
525 | } |
---|
526 | |
---|
527 | XERCES_CPP_NAMESPACE_END |
---|
528 | |
---|
529 | #endif |
---|