| Geant4 Cross Reference |
1 // Copyright (C) 2010, Guy Barrand. All rights reserved.
2 // See the file tools.license for terms.
3
4 #ifndef toolx_xml_loader
5 #define toolx_xml_loader
6
7 #include <tools/xml/tree>
8 #include <tools/file>
9 #include <tools/mnmx>
10
11 #include <cstdio>
12 #include <cctype> //iscntrl
13
14 #include <expat.h>
15
16 #ifdef TOOLS_MEM
17 #include <tools/mem>
18 namespace toolx {
19 extern "C" {
20 inline void* xml_malloc(size_t a_size){
21 tools::mem::increment(tools::s_malloc().c_str());
22 return ::malloc(a_size);
23 }
24 inline void* xml_realloc(void* a_ptr,size_t a_size){
25 if(a_ptr==NULL) tools::mem::increment(tools::s_malloc().c_str());
26 return ::realloc(a_ptr,a_size);
27 }
28 inline void xml_free(void* a_ptr){
29 if(a_ptr!=NULL) tools::mem::decrement(tools::s_malloc().c_str());
30 ::free(a_ptr);
31 }
32 }}
33 #endif
34
35 namespace toolx {
36 namespace xml {
37
38 class loader {
39 #ifdef TOOLS_MEM
40 TOOLS_SCLASS(toolx::xml::loader)
41 #endif
42 public:
43 loader(tools::xml::factory& a_factory,
44 std::ostream& a_out,bool a_verbose = false)
45 :m_factory(a_factory)
46 ,m_out(a_out)
47 ,m_verbose(a_verbose)
48 ,m_take_cntrl(false)
49
50 ,m_errors(0)
51 ,m_top(0) // Used to cleanup in case XML parsing failed.
52 ,m_current(0)
53 ,m_compressed_reader(0)
54 ,m_depth(0)
55 ,m_abort(false)
56 {
57 #ifdef TOOLS_MEM
58 tools::mem::increment(s_class().c_str());
59 #endif
60 }
61
62 virtual ~loader(){
63 delete m_compressed_reader;
64 clear();
65 #ifdef TOOLS_MEM
66 tools::mem::decrement(s_class().c_str());
67 #endif
68 }
69
70 protected:
71 loader(const loader& a_from)
72 :m_factory(a_from.m_factory)
73 ,m_out(a_from.m_out)
74 ,m_verbose(a_from.m_verbose)
75 ,m_take_cntrl(a_from.m_take_cntrl)
76
77 ,m_errors(0)
78 ,m_top(0) // Used to cleanup in case XML parsing failed.
79 ,m_current(0)
80 ,m_compressed_reader(0)
81 ,m_depth(0)
82 ,m_abort(false)
83 {
84 #ifdef TOOLS_MEM
85 tools::mem::increment(s_class().c_str());
86 #endif
87 }
88 loader& operator=(const loader& a_from){
89 if(&a_from==this) return *this;
90
91 m_verbose = a_from.m_verbose;
92 m_take_cntrl = a_from.m_take_cntrl;
93
94 m_errors = 0;
95 m_top = 0;
96 m_current = 0;
97 m_compressed_reader = 0;
98 m_depth = 0;
99 m_abort = false;
100
101 return *this;
102 }
103
104 public:
105 virtual bool visit_end_element(tools::xml::tree&,bool& a_keep) {
106 a_keep = true;
107 return true;
108 }
109
110 public:
111 void set_take_cntrl_chars(bool a_value) {m_take_cntrl = a_value;}
112
113 std::ostream& out() const {return m_out;}
114 void set_compressed_reader(tools::file::reader* aReader){
115 delete m_compressed_reader;
116 m_compressed_reader = aReader; //take ownership.
117 }
118
119 unsigned int errors() const {return m_errors;}
120
121 void set_tags(const std::vector<std::string>& a_tags){m_tags=a_tags;}
122 void add_tag(const std::string& a_tag){m_tags.push_back(a_tag);}
123
124 bool load_file(const std::string& a_file,bool a_compressed) {
125 clear();
126 if(!parse_file(a_file,
127 (XML_StartElementHandler)start_element,
128 (XML_EndElementHandler)end_element,
129 this,a_compressed)) {
130 clear();
131 return false;
132 }
133 if(m_current) m_current->set_file(a_file);
134 return true;
135 }
136
137 bool load_string(const std::string& a_string){
138 clear();
139 if(!parse_buffer(a_string.size(),a_string.c_str(),
140 (XML_StartElementHandler)start_element,
141 (XML_EndElementHandler)end_element,
142 this)) {
143 clear();
144 return false;
145 }
146 return true;
147 }
148
149 bool load_buffer(size_t aSize,const char* aBuffer){
150 clear();
151 if(!parse_buffer(aSize,aBuffer,
152 (XML_StartElementHandler)start_element,
153 (XML_EndElementHandler)end_element,
154 this)) {
155 clear();
156 return false;
157 }
158 return true;
159 }
160
161 const tools::xml::tree* top_item() const {return m_current;}
162
163 tools::xml::tree* top_item(){return m_current;}
164
165 void empty(){m_top = 0;m_current = 0;}
166
167 bool is_tag(const std::string& a_string) const {
168 size_t number = m_tags.size();
169 for(size_t index=0;index<number;index++) {
170 if(a_string==m_tags[index]) return true;
171 }
172 return false;
173 }
174
175 protected:
176 void clear(){
177 // In case of problem, deleting m_current is not sufficient.
178 delete m_top;
179 m_top = 0;
180 m_current = 0;
181 }
182
183 bool parse_buffer(size_t aSize,const char* aBuffer,
184 XML_StartElementHandler a_start,XML_EndElementHandler a_end,
185 void* a_tag){
186 m_errors = 0;
187 if(!aSize) return true; //nothing to do.
188 m_depth = 0;
189 m_abort = false;
190
191 #ifdef TOOLS_MEM
192 XML_Memory_Handling_Suite mem;
193 mem.malloc_fcn = xml_malloc;
194 mem.realloc_fcn = xml_realloc;
195 mem.free_fcn = xml_free;
196 XML_Parser _parser = XML_ParserCreate_MM(NULL,&mem,NULL);
197 #else
198 XML_Parser _parser = XML_ParserCreate(NULL);
199 #endif
200
201 XML_SetUserData(_parser,a_tag);
202 XML_SetElementHandler(_parser,a_start,a_end);
203 XML_SetCharacterDataHandler(_parser,(XML_CharacterDataHandler)character_data_handler);
204 //XML_SetProcessingInstructionHandler(_parser,processingInstructionHandler);
205 char* buf = (char*)aBuffer;
206 size_t l = aSize;
207 int done = 0;
208 do {
209 size_t len = tools::mn<size_t>(l,BUFSIZ); //BUFSIZ in cstdio
210 done = len < BUFSIZ ? 1 : 0;
211 if(XML_Parse(_parser, buf, (int)len, done)==XML_STATUS_ERROR) {
212 m_out << "parse_buffer :"
213 << " " << XML_ErrorString(XML_GetErrorCode(_parser))
214 << " at line " << (int)XML_GetCurrentLineNumber(_parser)
215 << " at byte index " << (int)XML_GetCurrentByteIndex(_parser)
216 << std::endl;
217 {XML_Index pos = XML_GetCurrentByteIndex(_parser);
218 XML_Index pmn = tools::mx<XML_Index>(pos-10,0);
219 XML_Index pmx = tools::mn<XML_Index>(pos+10,XML_Index(aSize)-1);
220 std::string c = " ";
221 {for(XML_Index p=pmn;p<=pmx;p++) {c[0] = *(aBuffer+p);m_out << c;}
222 m_out << std::endl;}
223 {for(XML_Index p=pmn;p<pos;p++) m_out << " ";
224 m_out << "^" << std::endl;}}
225 XML_ParserFree(_parser);
226 return false;
227 }
228 if(m_abort) {
229 XML_ParserFree(_parser);
230 return false;
231 }
232 buf += len;
233 l -= len;
234 } while (!done);
235 XML_ParserFree(_parser);
236 return true;
237 }
238
239 bool parse_file(const std::string& a_file,
240 XML_StartElementHandler a_start,XML_EndElementHandler a_end,
241 void* a_tag,bool a_compressed){
242 if(m_verbose) {
243 m_out << "parse_file :"
244 << " parse file " << tools::sout(a_file) << "..." << std::endl;
245 }
246 m_errors = 0;
247
248 bool use_zlib = false;
249 if(a_compressed) {
250 if(m_verbose) {
251 m_out << "parse_file :"
252 << " uncompress requested for file "
253 << tools::sout(a_file) << "."
254 << std::endl;
255 }
256 use_zlib = true;
257 } else {
258 // may be compressed anyway :
259 bool compressed;
260 if(!tools::file::is_gzip(a_file,compressed)) {
261 m_out << "parse_file :"
262 << " tools::file::is_gzip() failed for " << a_file << "."
263 << std::endl;
264 return false;
265 }
266 if(compressed) use_zlib = true;
267 }
268
269 tools::file::reader* freader = 0;
270 bool delete_freader = false;
271 if(use_zlib) {
272 if(!m_compressed_reader) {
273 m_out << "parse_file :"
274 << " no compressed reader given."
275 << std::endl;
276 return false;
277 }
278 freader = m_compressed_reader;
279 } else {
280 freader = new tools::FILE_reader();
281 delete_freader = true;
282 }
283 if(!freader->open(a_file)) {
284 m_out << "parse_file :"
285 << " can't open file " << a_file << std::endl;
286 if(delete_freader) delete freader;
287 return false;
288 }
289
290 m_depth = 0;
291 m_abort = false;
292
293 #ifdef TOOLS_MEM
294 XML_Memory_Handling_Suite mem;
295 mem.malloc_fcn = xml_malloc;
296 mem.realloc_fcn = xml_realloc;
297 mem.free_fcn = xml_free;
298 XML_Parser _parser = XML_ParserCreate_MM(NULL,&mem,NULL);
299 #else
300 XML_Parser _parser = XML_ParserCreate(NULL);
301 #endif
302
303 XML_SetUserData(_parser,a_tag);
304 XML_SetElementHandler(_parser,a_start,a_end);
305 XML_SetCharacterDataHandler(_parser,(XML_CharacterDataHandler)character_data_handler);
306 //XML_SetProcessingInstructionHandler(_parser,
307 // processingInstructionHandler);
308
309
310 //char buf[1024 * BUFSIZ];
311 char buf[BUFSIZ];
312 int done = 0;
313 do {
314 size_t len;
315 if(!freader->read(buf,sizeof(buf),len)) {
316 XML_ParserFree(_parser);
317 freader->close();
318 if(delete_freader) delete freader;
319 return false;
320 }
321 done = len < sizeof(buf) ? 1 : 0;
322 if(XML_Parse(_parser, buf, (int)len, done)==XML_STATUS_ERROR) {
323 m_out << "parse_file :"
324 << " in file " << tools::sout(a_file)
325 << " " << XML_ErrorString(XML_GetErrorCode(_parser))
326 << " at line " << (int)XML_GetCurrentLineNumber(_parser)
327 << std::endl;
328 XML_ParserFree(_parser);
329 freader->close();
330 if(delete_freader) delete freader;
331 return false;
332 }
333 if(m_abort) {
334 XML_ParserFree(_parser);
335 freader->close();
336 if(delete_freader) delete freader;
337 return false;
338 }
339 } while (!done);
340 XML_ParserFree(_parser);
341 freader->close();
342 if(m_verbose) {
343 m_out << "parse_file :"
344 << " parse file " << tools::sout(a_file) << " done." << std::endl;
345 }
346 if(delete_freader) delete freader;
347 return true;
348 }
349
350 protected:
351 static void character_data_handler(void* aUserData,const XML_Char* a_string,int aLength){
352 loader* This = (loader*)aUserData;
353 std::string _s;
354 _s.resize(aLength);
355 size_t count = 0;
356 char* p = (char*)a_string;
357 for (int i = 0; i < aLength; i++, p++) {
358 if(This->m_take_cntrl || (!iscntrl(*p))) {
359 _s[count] = *p;
360 count++;
361 }
362 }
363 if(count) {
364 _s.resize(count);
365 This->m_value += _s;
366 }
367 }
368
369 static void start_element(void* aUserData,const XML_Char* a_name,const XML_Char** a_atbs){
370 loader* This = (loader*)aUserData;
371 if(This->m_abort) return; //Do nothing.
372
373 This->m_depth++;
374 This->m_value = "";
375
376 std::string name = a_name; //Can't be empty
377 if(This->is_tag(name)) {
378
379 if(!This->m_current) {
380 if(This->m_depth==1) {
381 // Ok. Head.
382 } else {
383 This->m_out << "start_element :"
384 << " no tag with a depth of " << This->m_depth
385 << std::endl;
386 This->m_abort = true;
387 return;
388 }
389 } else {
390 int delta = This->m_current->depth() - This->m_depth;
391 if(delta>=1) {
392 This->m_out << "start_element :"
393 << " for element " << tools::sout(name)
394 << " tag with a delta depth of " << delta
395 << std::endl;
396 This->m_abort = true;
397 return;
398 }
399 }
400
401 std::vector<tools::xml::tree::atb> atbs;
402 {const XML_Char** a_atts = a_atbs;
403 while((*a_atts)&&(*(a_atts+1))) {
404 atbs.push_back(tools::xml::tree::atb(*a_atts,*(a_atts+1)));
405 a_atts+=2;
406 }}
407
408 tools::xml::tree* parent = This->m_current;
409 tools::xml::tree* _tree = This->m_factory.create(name,atbs,parent);
410 if(!_tree) {
411 This->m_out << "start_element :"
412 << " can't create a tree for tag " << tools::sout(name)
413 << std::endl;
414 This->m_abort = true;
415 return;
416 }
417
418 //out << "start_element :" << std::endl;
419 //_tree->print_xml(*(This->m_printer),"debug : ");
420
421 if(parent) parent->add_child(_tree);
422
423 /*
424 if(This->m_current && !This->m_current->parent()) {
425 This->m_out << "start_element :"
426 << " warning : current tree without parent."
427 << " Potential mem leak."
428 << std::endl;
429 }
430 */
431
432 This->m_current = _tree;
433 _tree->set_depth(This->m_depth); // Internal only.
434
435 if(!This->m_top) This->m_top = _tree;
436
437 } else {
438
439 if(!This->m_current) {
440
441 // Can't be in a non-tag without a tag !
442 This->m_out << "start_element :"
443 << " for element " << tools::sout(name)
444 << " non-tag without some parent tag."
445 << std::endl;
446 This->m_abort = true;
447 return;
448
449 } else {
450
451 int delta = This->m_depth - This->m_current->depth();
452 if(delta>1) {
453
454 This->m_out << "start_element :"
455 << " for element " << tools::sout(name)
456 << " grand child of a tag."
457 << std::endl;
458 This->m_abort = true;
459 return;
460
461 } else if(delta==1) { //ok
462
463 This->m_atbs.clear();
464 {const XML_Char** a_atts = a_atbs;
465 while((*a_atts)&&(*(a_atts+1))) {
466 This->m_atbs.push_back(tools::xml::tree::atb(*a_atts,*(a_atts+1)));
467 a_atts+=2;
468 }}
469
470 } else {
471
472 This->m_out << "start_element :"
473 << " for element " << tools::sout(name)
474 << " non-tag with a delta depth of " << delta
475 << std::endl;
476 This->m_abort = true;
477 return;
478
479 }
480 }
481
482 }
483 }
484
485
486 static void end_element(void* aUserData,const XML_Char* a_name){
487 loader* This = (loader*)aUserData;
488 if(This->m_abort) return; //Do nothing.
489
490 if(This->m_current) {
491
492 tools::xml::tree* tr = This->m_current;
493 int delta = This->m_depth - tr->depth();
494 if(delta==0) { //Back to a tag :
495
496 tools::xml::tree* parent = tr->parent();
497
498 bool keep = false;
499 bool cont = This->visit_end_element(*tr,keep);
500 if(keep) {
501 if(parent) {
502 /*
503 if(!This->m_current->parent()) {
504 This->m_out << "end_element :"
505 << " warning : current tree without parent (1)."
506 << " Potential mem leak."
507 << std::endl;
508 }
509 */
510 This->m_current = parent;
511 }
512 } else {
513 //FIXME : the top could be recreated !
514 if(This->m_top==tr) This->m_top = 0;
515
516 if(parent) {
517 parent->remove_child(tr); //delete the tr
518 } else {
519 delete tr;
520 }
521
522 /*
523 if(!This->m_current->parent()) {
524 This->m_out << "end_element :"
525 << " warning : current tree without parent (2)."
526 << " Potential mem leak."
527 << std::endl;
528 }
529 */
530
531 This->m_current = parent; //parent could be 0 : ok.
532 }
533
534 if(!cont) This->m_abort = true;
535
536 } else if(delta==1) { //Back to a child of tag :
537
538 //FIXME : correct m_value ? (Can we pick the one of a sub item ?)
539 tr->add_element(std::string(a_name),This->m_atbs,This->m_value);
540 //This->m_value = "";
541
542 } else {
543
544 This->m_out << "end_element :"
545 << " problem for element " << tools::sout(std::string(a_name))
546 << " : delta depth of " << delta
547 << std::endl;
548 This->m_abort = true;
549
550 }
551
552 }
553
554
555 This->m_depth--;
556 }
557
558 protected:
559 tools::xml::factory& m_factory;
560 std::ostream& m_out;
561 protected:
562 bool m_verbose;
563 bool m_take_cntrl;
564 unsigned int m_errors;
565 std::vector<std::string> m_tags;
566 tools::xml::tree* m_top;
567 tools::xml::tree* m_current;
568 //std::vector<tools::xml::tree::atb> m_atbs;
569 std::vector< std::pair<std::string,std::string> > m_atbs;
570 std::string m_value;
571 tools::file::reader* m_compressed_reader;
572 unsigned int m_depth;
573 bool m_abort;
574 };
575
576 }}
577
578 #endif