2 * testSAX.c : a small tester program for parsing using the SAX API.
4 * See Copyright for the status of this software.
14 #ifdef HAVE_SYS_TYPES_H
15 #include <sys/types.h>
17 #ifdef HAVE_SYS_STAT_H
34 #include <libxml/globals.h>
35 #include <libxml/xmlerror.h>
36 #include <libxml/parser.h>
37 #include <libxml/parserInternals.h> /* only for xmlNewInputFromFile() */
38 #include <libxml/tree.h>
39 #include <libxml/debugXML.h>
40 #include <libxml/xmlmemory.h>
44 static int recovery = 0;
49 static int callbacks = 0;
51 xmlSAXHandler emptySAXHandlerStruct = {
52 NULL, /* internalSubset */
53 NULL, /* isStandalone */
54 NULL, /* hasInternalSubset */
55 NULL, /* hasExternalSubset */
56 NULL, /* resolveEntity */
58 NULL, /* entityDecl */
59 NULL, /* notationDecl */
60 NULL, /* attributeDecl */
61 NULL, /* elementDecl */
62 NULL, /* unparsedEntityDecl */
63 NULL, /* setDocumentLocator */
64 NULL, /* startDocument */
65 NULL, /* endDocument */
66 NULL, /* startElement */
67 NULL, /* endElement */
69 NULL, /* characters */
70 NULL, /* ignorableWhitespace */
71 NULL, /* processingInstruction */
73 NULL, /* xmlParserWarning */
74 NULL, /* xmlParserError */
75 NULL, /* xmlParserError */
76 NULL, /* getParameterEntity */
77 NULL, /* cdataBlock; */
78 NULL, /* externalSubset; */
82 xmlSAXHandlerPtr emptySAXHandler = &emptySAXHandlerStruct;
83 extern xmlSAXHandlerPtr debugSAXHandler;
85 /************************************************************************
89 ************************************************************************/
93 * @ctxt: An XML parser context
95 * Is this document tagged standalone ?
100 isStandaloneDebug(void *ctx ATTRIBUTE_UNUSED)
105 fprintf(stdout, "SAX.isStandalone()\n");
110 * hasInternalSubsetDebug:
111 * @ctxt: An XML parser context
113 * Does this document has an internal subset
118 hasInternalSubsetDebug(void *ctx ATTRIBUTE_UNUSED)
123 fprintf(stdout, "SAX.hasInternalSubset()\n");
128 * hasExternalSubsetDebug:
129 * @ctxt: An XML parser context
131 * Does this document has an external subset
136 hasExternalSubsetDebug(void *ctx ATTRIBUTE_UNUSED)
141 fprintf(stdout, "SAX.hasExternalSubset()\n");
146 * internalSubsetDebug:
147 * @ctxt: An XML parser context
149 * Does this document has an internal subset
152 internalSubsetDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name,
153 const xmlChar *ExternalID, const xmlChar *SystemID)
158 fprintf(stdout, "SAX.internalSubset(%s,", name);
159 if (ExternalID == NULL)
160 fprintf(stdout, " ,");
162 fprintf(stdout, " %s,", ExternalID);
163 if (SystemID == NULL)
164 fprintf(stdout, " )\n");
166 fprintf(stdout, " %s)\n", SystemID);
170 * externalSubsetDebug:
171 * @ctxt: An XML parser context
173 * Does this document has an external subset
176 externalSubsetDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name,
177 const xmlChar *ExternalID, const xmlChar *SystemID)
182 fprintf(stdout, "SAX.externalSubset(%s,", name);
183 if (ExternalID == NULL)
184 fprintf(stdout, " ,");
186 fprintf(stdout, " %s,", ExternalID);
187 if (SystemID == NULL)
188 fprintf(stdout, " )\n");
190 fprintf(stdout, " %s)\n", SystemID);
194 * resolveEntityDebug:
195 * @ctxt: An XML parser context
196 * @publicId: The public ID of the entity
197 * @systemId: The system ID of the entity
199 * Special entity resolver, better left to the parser, it has
200 * more context than the application layer.
201 * The default behaviour is to NOT resolve the entities, in that case
202 * the ENTITY_REF nodes are built in the structure (and the parameter
205 * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour.
207 static xmlParserInputPtr
208 resolveEntityDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *publicId, const xmlChar *systemId)
213 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
216 fprintf(stdout, "SAX.resolveEntity(");
217 if (publicId != NULL)
218 fprintf(stdout, "%s", (char *)publicId);
220 fprintf(stdout, " ");
221 if (systemId != NULL)
222 fprintf(stdout, ", %s)\n", (char *)systemId);
224 fprintf(stdout, ", )\n");
226 if (systemId != NULL) {
227 return(xmlNewInputFromFile(ctxt, (char *) systemId));
235 * @ctxt: An XML parser context
236 * @name: The entity name
238 * Get an entity by name
240 * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour.
243 getEntityDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name)
248 fprintf(stdout, "SAX.getEntity(%s)\n", name);
253 * getParameterEntityDebug:
254 * @ctxt: An XML parser context
255 * @name: The entity name
257 * Get a parameter entity by name
259 * Returns the xmlParserInputPtr
262 getParameterEntityDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name)
267 fprintf(stdout, "SAX.getParameterEntity(%s)\n", name);
274 * @ctxt: An XML parser context
275 * @name: the entity name
276 * @type: the entity type
277 * @publicId: The public ID of the entity
278 * @systemId: The system ID of the entity
279 * @content: the entity value (without processing).
281 * An entity definition has been parsed
284 entityDeclDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name, int type,
285 const xmlChar *publicId, const xmlChar *systemId, xmlChar *content)
290 fprintf(stdout, "SAX.entityDecl(%s, %d, %s, %s, %s)\n",
291 name, type, publicId, systemId, content);
295 * attributeDeclDebug:
296 * @ctxt: An XML parser context
297 * @name: the attribute name
298 * @type: the attribute type
300 * An attribute definition has been parsed
303 attributeDeclDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *elem, const xmlChar *name,
304 int type, int def, const xmlChar *defaultValue,
305 xmlEnumerationPtr tree ATTRIBUTE_UNUSED)
310 if (defaultValue == NULL)
311 fprintf(stdout, "SAX.attributeDecl(%s, %s, %d, %d, NULL, ...)\n",
312 elem, name, type, def);
314 fprintf(stdout, "SAX.attributeDecl(%s, %s, %d, %d, %s, ...)\n",
315 elem, name, type, def, defaultValue);
320 * @ctxt: An XML parser context
321 * @name: the element name
322 * @type: the element type
323 * @content: the element value (without processing).
325 * An element definition has been parsed
328 elementDeclDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name, int type,
329 xmlElementContentPtr content ATTRIBUTE_UNUSED)
334 fprintf(stdout, "SAX.elementDecl(%s, %d, ...)\n",
340 * @ctxt: An XML parser context
341 * @name: The name of the notation
342 * @publicId: The public ID of the entity
343 * @systemId: The system ID of the entity
345 * What to do when a notation declaration has been parsed.
348 notationDeclDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name,
349 const xmlChar *publicId, const xmlChar *systemId)
354 fprintf(stdout, "SAX.notationDecl(%s, %s, %s)\n",
355 (char *) name, (char *) publicId, (char *) systemId);
359 * unparsedEntityDeclDebug:
360 * @ctxt: An XML parser context
361 * @name: The name of the entity
362 * @publicId: The public ID of the entity
363 * @systemId: The system ID of the entity
364 * @notationName: the name of the notation
366 * What to do when an unparsed entity declaration is parsed
369 unparsedEntityDeclDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name,
370 const xmlChar *publicId, const xmlChar *systemId,
371 const xmlChar *notationName)
376 fprintf(stdout, "SAX.unparsedEntityDecl(%s, %s, %s, %s)\n",
377 (char *) name, (char *) publicId, (char *) systemId,
378 (char *) notationName);
382 * setDocumentLocatorDebug:
383 * @ctxt: An XML parser context
384 * @loc: A SAX Locator
386 * Receive the document locator at startup, actually xmlDefaultSAXLocator
387 * Everything is available on the context, so this is useless in our case.
390 setDocumentLocatorDebug(void *ctx ATTRIBUTE_UNUSED, xmlSAXLocatorPtr loc ATTRIBUTE_UNUSED)
395 fprintf(stdout, "SAX.setDocumentLocator()\n");
399 * startDocumentDebug:
400 * @ctxt: An XML parser context
402 * called when the document start being processed.
405 startDocumentDebug(void *ctx ATTRIBUTE_UNUSED)
410 fprintf(stdout, "SAX.startDocument()\n");
415 * @ctxt: An XML parser context
417 * called when the document end has been detected.
420 endDocumentDebug(void *ctx ATTRIBUTE_UNUSED)
425 fprintf(stdout, "SAX.endDocument()\n");
430 * @ctxt: An XML parser context
431 * @name: The element name
433 * called when an opening tag has been processed.
436 startElementDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name, const xmlChar **atts)
443 fprintf(stdout, "SAX.startElement(%s", (char *) name);
445 for (i = 0;(atts[i] != NULL);i++) {
446 fprintf(stdout, ", %s='", atts[i++]);
448 fprintf(stdout, "%s'", atts[i]);
451 fprintf(stdout, ")\n");
456 * @ctxt: An XML parser context
457 * @name: The element name
459 * called when the end of an element has been detected.
462 endElementDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name)
467 fprintf(stdout, "SAX.endElement(%s)\n", (char *) name);
472 * @ctxt: An XML parser context
473 * @ch: a xmlChar string
474 * @len: the number of xmlChar
476 * receiving some chars from the parser.
477 * Question: how much at a time ???
480 charactersDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch, int len)
488 for (i = 0;(i<len) && (i < 30);i++)
492 fprintf(stdout, "SAX.characters(%s, %d)\n", output, len);
497 * @ctxt: An XML parser context
498 * @name: The entity name
500 * called when an entity reference is detected.
503 referenceDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *name)
508 fprintf(stdout, "SAX.reference(%s)\n", name);
512 * ignorableWhitespaceDebug:
513 * @ctxt: An XML parser context
514 * @ch: a xmlChar string
515 * @start: the first char in the string
516 * @len: the number of xmlChar
518 * receiving some ignorable whitespaces from the parser.
519 * Question: how much at a time ???
522 ignorableWhitespaceDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch, int len)
530 for (i = 0;(i<len) && (i < 30);i++)
533 fprintf(stdout, "SAX.ignorableWhitespace(%s, %d)\n", output, len);
537 * processingInstructionDebug:
538 * @ctxt: An XML parser context
539 * @target: the target name
540 * @data: the PI data's
541 * @len: the number of xmlChar
543 * A processing instruction has been parsed.
546 processingInstructionDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *target,
552 fprintf(stdout, "SAX.processingInstruction(%s, %s)\n",
553 (char *) target, (char *) data);
558 * @ctx: the user data (XML parser context)
559 * @value: The pcdata content
560 * @len: the block length
562 * called when a pcdata block has been parsed
565 cdataBlockDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *value, int len)
570 fprintf(stdout, "SAX.pcdata(%.20s, %d)\n",
571 (char *) value, len);
576 * @ctxt: An XML parser context
577 * @value: the comment content
579 * A comment has been parsed.
582 commentDebug(void *ctx ATTRIBUTE_UNUSED, const xmlChar *value)
587 fprintf(stdout, "SAX.comment(%s)\n", value);
592 * @ctxt: An XML parser context
593 * @msg: the message to display/transmit
594 * @...: extra parameters for the message display
596 * Display and format a warning messages, gives file, line, position and
600 warningDebug(void *ctx ATTRIBUTE_UNUSED, const char *msg, ...)
608 fprintf(stdout, "SAX.warning: ");
609 vfprintf(stdout, msg, args);
615 * @ctxt: An XML parser context
616 * @msg: the message to display/transmit
617 * @...: extra parameters for the message display
619 * Display and format a error messages, gives file, line, position and
623 errorDebug(void *ctx ATTRIBUTE_UNUSED, const char *msg, ...)
631 fprintf(stdout, "SAX.error: ");
632 vfprintf(stdout, msg, args);
638 * @ctxt: An XML parser context
639 * @msg: the message to display/transmit
640 * @...: extra parameters for the message display
642 * Display and format a fatalError messages, gives file, line, position and
646 fatalErrorDebug(void *ctx ATTRIBUTE_UNUSED, const char *msg, ...)
654 fprintf(stdout, "SAX.fatalError: ");
655 vfprintf(stdout, msg, args);
659 xmlSAXHandler debugSAXHandlerStruct = {
662 hasInternalSubsetDebug,
663 hasExternalSubsetDebug,
670 unparsedEntityDeclDebug,
671 setDocumentLocatorDebug,
678 ignorableWhitespaceDebug,
679 processingInstructionDebug,
684 getParameterEntityDebug,
690 xmlSAXHandlerPtr debugSAXHandler = &debugSAXHandlerStruct;
692 /************************************************************************
696 ************************************************************************/
699 parseAndPrintFile(char *filename) {
706 * Empty callbacks for checking
708 f = fopen(filename, "r");
712 xmlParserCtxtPtr ctxt;
714 ret = fread(chars, 1, 4, f);
716 ctxt = xmlCreatePushParserCtxt(emptySAXHandler, NULL,
717 chars, ret, filename);
718 while ((ret = fread(chars, 1, 3, f)) > 0) {
719 xmlParseChunk(ctxt, chars, ret, 0);
721 xmlParseChunk(ctxt, chars, 0, 1);
722 xmlFreeParserCtxt(ctxt);
726 xmlGenericError(xmlGenericErrorContext,
727 "Cannot read file %s\n", filename);
732 f = fopen(filename, "r");
736 xmlParserCtxtPtr ctxt;
738 ret = fread(chars, 1, 4, f);
740 ctxt = xmlCreatePushParserCtxt(debugSAXHandler, NULL,
741 chars, ret, filename);
742 while ((ret = fread(chars, 1, 3, f)) > 0) {
743 xmlParseChunk(ctxt, chars, ret, 0);
745 ret = xmlParseChunk(ctxt, chars, 0, 1);
746 xmlFreeParserCtxt(ctxt);
749 "xmlSAXUserParseFile returned error %d\n", ret);
757 * Empty callbacks for checking
760 res = xmlSAXUserParseFile(emptySAXHandler, NULL, filename);
762 fprintf(stdout, "xmlSAXUserParseFile returned error %d\n", res);
770 res = xmlSAXUserParseFile(debugSAXHandler, NULL, filename);
772 fprintf(stdout, "xmlSAXUserParseFile returned error %d\n", res);
775 fprintf(stdout, "%d callbacks generated\n", callbacks);
778 * test 100x the SAX parse
782 for (i = 0; i<100;i++)
783 res = xmlSAXUserParseFile(emptySAXHandler, NULL, filename);
785 fprintf(stdout, "xmlSAXUserParseFile returned error %d\n", res);
792 int main(int argc, char **argv) {
796 for (i = 1; i < argc ; i++) {
797 if ((!strcmp(argv[i], "-debug")) || (!strcmp(argv[i], "--debug")))
799 else if ((!strcmp(argv[i], "-copy")) || (!strcmp(argv[i], "--copy")))
801 else if ((!strcmp(argv[i], "-recover")) ||
802 (!strcmp(argv[i], "--recover")))
804 else if ((!strcmp(argv[i], "-push")) ||
805 (!strcmp(argv[i], "--push")))
807 else if ((!strcmp(argv[i], "-speed")) ||
808 (!strcmp(argv[i], "--speed")))
810 else if ((!strcmp(argv[i], "-noent")) ||
811 (!strcmp(argv[i], "--noent")))
813 else if ((!strcmp(argv[i], "-quiet")) ||
814 (!strcmp(argv[i], "--quiet")))
817 if (noent != 0) xmlSubstituteEntitiesDefault(1);
818 for (i = 1; i < argc ; i++) {
819 if (argv[i][0] != '-') {
820 parseAndPrintFile(argv[i]);