Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 1 | /* |
| 2 | * testHTML.c : a small tester program for HTML input. |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * Daniel.Veillard@w3.org |
| 7 | */ |
| 8 | |
| 9 | #ifdef WIN32 |
Daniel Veillard | 3c558c3 | 1999-12-22 11:30:41 +0000 | [diff] [blame] | 10 | #include "win32config.h" |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 11 | #else |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 12 | #include "config.h" |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 13 | #endif |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 14 | |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 15 | #include "xmlversion.h" |
| 16 | #ifdef LIBXML_HTML_ENABLED |
| 17 | |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 18 | #include <stdio.h> |
| 19 | #include <string.h> |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 20 | #include <stdarg.h> |
| 21 | |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 22 | |
| 23 | #ifdef HAVE_SYS_TYPES_H |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 24 | #include <sys/types.h> |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 25 | #endif |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 26 | #ifdef HAVE_SYS_STAT_H |
| 27 | #include <sys/stat.h> |
| 28 | #endif |
| 29 | #ifdef HAVE_FCNTL_H |
| 30 | #include <fcntl.h> |
| 31 | #endif |
| 32 | #ifdef HAVE_UNISTD_H |
| 33 | #include <unistd.h> |
| 34 | #endif |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 35 | #ifdef HAVE_STDLIB_H |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 36 | #include <stdlib.h> |
Daniel Veillard | 7f7d111 | 1999-09-22 09:46:25 +0000 | [diff] [blame] | 37 | #endif |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 38 | |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 39 | #include <libxml/xmlmemory.h> |
| 40 | #include <libxml/HTMLparser.h> |
| 41 | #include <libxml/HTMLtree.h> |
| 42 | #include <libxml/debugXML.h> |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 43 | |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 44 | #ifdef LIBXML_DEBUG_ENABLED |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 45 | static int debug = 0; |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 46 | #endif |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 47 | static int copy = 0; |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 48 | static int sax = 0; |
| 49 | static int repeat = 0; |
| 50 | static int noout = 0; |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 51 | static int push = 0; |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 52 | static char *encoding = NULL; |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 53 | |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 54 | xmlSAXHandler emptySAXHandlerStruct = { |
| 55 | NULL, /* internalSubset */ |
| 56 | NULL, /* isStandalone */ |
| 57 | NULL, /* hasInternalSubset */ |
| 58 | NULL, /* hasExternalSubset */ |
| 59 | NULL, /* resolveEntity */ |
| 60 | NULL, /* getEntity */ |
| 61 | NULL, /* entityDecl */ |
| 62 | NULL, /* notationDecl */ |
| 63 | NULL, /* attributeDecl */ |
| 64 | NULL, /* elementDecl */ |
| 65 | NULL, /* unparsedEntityDecl */ |
| 66 | NULL, /* setDocumentLocator */ |
| 67 | NULL, /* startDocument */ |
| 68 | NULL, /* endDocument */ |
| 69 | NULL, /* startElement */ |
| 70 | NULL, /* endElement */ |
| 71 | NULL, /* reference */ |
| 72 | NULL, /* characters */ |
| 73 | NULL, /* ignorableWhitespace */ |
| 74 | NULL, /* processingInstruction */ |
| 75 | NULL, /* comment */ |
| 76 | NULL, /* xmlParserWarning */ |
| 77 | NULL, /* xmlParserError */ |
| 78 | NULL, /* xmlParserError */ |
| 79 | NULL, /* getParameterEntity */ |
| 80 | }; |
| 81 | |
| 82 | xmlSAXHandlerPtr emptySAXHandler = &emptySAXHandlerStruct; |
| 83 | extern xmlSAXHandlerPtr debugSAXHandler; |
| 84 | |
| 85 | /************************************************************************ |
| 86 | * * |
| 87 | * Debug Handlers * |
| 88 | * * |
| 89 | ************************************************************************/ |
| 90 | |
| 91 | /** |
| 92 | * isStandaloneDebug: |
| 93 | * @ctxt: An XML parser context |
| 94 | * |
| 95 | * Is this document tagged standalone ? |
| 96 | * |
| 97 | * Returns 1 if true |
| 98 | */ |
| 99 | int |
| 100 | isStandaloneDebug(void *ctx) |
| 101 | { |
| 102 | fprintf(stdout, "SAX.isStandalone()\n"); |
| 103 | return(0); |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * hasInternalSubsetDebug: |
| 108 | * @ctxt: An XML parser context |
| 109 | * |
| 110 | * Does this document has an internal subset |
| 111 | * |
| 112 | * Returns 1 if true |
| 113 | */ |
| 114 | int |
| 115 | hasInternalSubsetDebug(void *ctx) |
| 116 | { |
| 117 | fprintf(stdout, "SAX.hasInternalSubset()\n"); |
| 118 | return(0); |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * hasExternalSubsetDebug: |
| 123 | * @ctxt: An XML parser context |
| 124 | * |
| 125 | * Does this document has an external subset |
| 126 | * |
| 127 | * Returns 1 if true |
| 128 | */ |
| 129 | int |
| 130 | hasExternalSubsetDebug(void *ctx) |
| 131 | { |
| 132 | fprintf(stdout, "SAX.hasExternalSubset()\n"); |
| 133 | return(0); |
| 134 | } |
| 135 | |
| 136 | /** |
| 137 | * hasInternalSubsetDebug: |
| 138 | * @ctxt: An XML parser context |
| 139 | * |
| 140 | * Does this document has an internal subset |
| 141 | */ |
| 142 | void |
| 143 | internalSubsetDebug(void *ctx, const xmlChar *name, |
| 144 | const xmlChar *ExternalID, const xmlChar *SystemID) |
| 145 | { |
Daniel Veillard | 808a3f1 | 2000-08-17 13:50:51 +0000 | [diff] [blame] | 146 | fprintf(stdout, "SAX.internalSubset(%s,", name); |
| 147 | if (ExternalID == NULL) |
| 148 | fprintf(stdout, " ,"); |
| 149 | else |
| 150 | fprintf(stdout, " %s,", ExternalID); |
| 151 | if (SystemID == NULL) |
| 152 | fprintf(stdout, " )\n"); |
| 153 | else |
| 154 | fprintf(stdout, " %s)\n", SystemID); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 155 | } |
| 156 | |
| 157 | /** |
| 158 | * resolveEntityDebug: |
| 159 | * @ctxt: An XML parser context |
| 160 | * @publicId: The public ID of the entity |
| 161 | * @systemId: The system ID of the entity |
| 162 | * |
| 163 | * Special entity resolver, better left to the parser, it has |
| 164 | * more context than the application layer. |
| 165 | * The default behaviour is to NOT resolve the entities, in that case |
| 166 | * the ENTITY_REF nodes are built in the structure (and the parameter |
| 167 | * values). |
| 168 | * |
| 169 | * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour. |
| 170 | */ |
| 171 | xmlParserInputPtr |
| 172 | resolveEntityDebug(void *ctx, const xmlChar *publicId, const xmlChar *systemId) |
| 173 | { |
| 174 | /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ |
| 175 | |
| 176 | |
| 177 | fprintf(stdout, "SAX.resolveEntity("); |
| 178 | if (publicId != NULL) |
| 179 | fprintf(stdout, "%s", (char *)publicId); |
| 180 | else |
| 181 | fprintf(stdout, " "); |
| 182 | if (systemId != NULL) |
| 183 | fprintf(stdout, ", %s)\n", (char *)systemId); |
| 184 | else |
| 185 | fprintf(stdout, ", )\n"); |
| 186 | /********* |
| 187 | if (systemId != NULL) { |
| 188 | return(xmlNewInputFromFile(ctxt, (char *) systemId)); |
| 189 | } |
| 190 | *********/ |
| 191 | return(NULL); |
| 192 | } |
| 193 | |
| 194 | /** |
| 195 | * getEntityDebug: |
| 196 | * @ctxt: An XML parser context |
| 197 | * @name: The entity name |
| 198 | * |
| 199 | * Get an entity by name |
| 200 | * |
| 201 | * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour. |
| 202 | */ |
| 203 | xmlEntityPtr |
| 204 | getEntityDebug(void *ctx, const xmlChar *name) |
| 205 | { |
| 206 | fprintf(stdout, "SAX.getEntity(%s)\n", name); |
| 207 | return(NULL); |
| 208 | } |
| 209 | |
| 210 | /** |
| 211 | * getParameterEntityDebug: |
| 212 | * @ctxt: An XML parser context |
| 213 | * @name: The entity name |
| 214 | * |
| 215 | * Get a parameter entity by name |
| 216 | * |
| 217 | * Returns the xmlParserInputPtr |
| 218 | */ |
| 219 | xmlEntityPtr |
| 220 | getParameterEntityDebug(void *ctx, const xmlChar *name) |
| 221 | { |
| 222 | fprintf(stdout, "SAX.getParameterEntity(%s)\n", name); |
| 223 | return(NULL); |
| 224 | } |
| 225 | |
| 226 | |
| 227 | /** |
| 228 | * entityDeclDebug: |
| 229 | * @ctxt: An XML parser context |
| 230 | * @name: the entity name |
| 231 | * @type: the entity type |
| 232 | * @publicId: The public ID of the entity |
| 233 | * @systemId: The system ID of the entity |
| 234 | * @content: the entity value (without processing). |
| 235 | * |
| 236 | * An entity definition has been parsed |
| 237 | */ |
| 238 | void |
| 239 | entityDeclDebug(void *ctx, const xmlChar *name, int type, |
| 240 | const xmlChar *publicId, const xmlChar *systemId, xmlChar *content) |
| 241 | { |
| 242 | fprintf(stdout, "SAX.entityDecl(%s, %d, %s, %s, %s)\n", |
| 243 | name, type, publicId, systemId, content); |
| 244 | } |
| 245 | |
| 246 | /** |
| 247 | * attributeDeclDebug: |
| 248 | * @ctxt: An XML parser context |
| 249 | * @name: the attribute name |
| 250 | * @type: the attribute type |
| 251 | * |
| 252 | * An attribute definition has been parsed |
| 253 | */ |
| 254 | void |
| 255 | attributeDeclDebug(void *ctx, const xmlChar *elem, const xmlChar *name, |
| 256 | int type, int def, const xmlChar *defaultValue, |
| 257 | xmlEnumerationPtr tree) |
| 258 | { |
| 259 | fprintf(stdout, "SAX.attributeDecl(%s, %s, %d, %d, %s, ...)\n", |
| 260 | elem, name, type, def, defaultValue); |
| 261 | } |
| 262 | |
| 263 | /** |
| 264 | * elementDeclDebug: |
| 265 | * @ctxt: An XML parser context |
| 266 | * @name: the element name |
| 267 | * @type: the element type |
| 268 | * @content: the element value (without processing). |
| 269 | * |
| 270 | * An element definition has been parsed |
| 271 | */ |
| 272 | void |
| 273 | elementDeclDebug(void *ctx, const xmlChar *name, int type, |
| 274 | xmlElementContentPtr content) |
| 275 | { |
| 276 | fprintf(stdout, "SAX.elementDecl(%s, %d, ...)\n", |
| 277 | name, type); |
| 278 | } |
| 279 | |
| 280 | /** |
| 281 | * notationDeclDebug: |
| 282 | * @ctxt: An XML parser context |
| 283 | * @name: The name of the notation |
| 284 | * @publicId: The public ID of the entity |
| 285 | * @systemId: The system ID of the entity |
| 286 | * |
| 287 | * What to do when a notation declaration has been parsed. |
| 288 | */ |
| 289 | void |
| 290 | notationDeclDebug(void *ctx, const xmlChar *name, |
| 291 | const xmlChar *publicId, const xmlChar *systemId) |
| 292 | { |
| 293 | fprintf(stdout, "SAX.notationDecl(%s, %s, %s)\n", |
| 294 | (char *) name, (char *) publicId, (char *) systemId); |
| 295 | } |
| 296 | |
| 297 | /** |
| 298 | * unparsedEntityDeclDebug: |
| 299 | * @ctxt: An XML parser context |
| 300 | * @name: The name of the entity |
| 301 | * @publicId: The public ID of the entity |
| 302 | * @systemId: The system ID of the entity |
| 303 | * @notationName: the name of the notation |
| 304 | * |
| 305 | * What to do when an unparsed entity declaration is parsed |
| 306 | */ |
| 307 | void |
| 308 | unparsedEntityDeclDebug(void *ctx, const xmlChar *name, |
| 309 | const xmlChar *publicId, const xmlChar *systemId, |
| 310 | const xmlChar *notationName) |
| 311 | { |
| 312 | fprintf(stdout, "SAX.unparsedEntityDecl(%s, %s, %s, %s)\n", |
| 313 | (char *) name, (char *) publicId, (char *) systemId, |
| 314 | (char *) notationName); |
| 315 | } |
| 316 | |
| 317 | /** |
| 318 | * setDocumentLocatorDebug: |
| 319 | * @ctxt: An XML parser context |
| 320 | * @loc: A SAX Locator |
| 321 | * |
| 322 | * Receive the document locator at startup, actually xmlDefaultSAXLocator |
| 323 | * Everything is available on the context, so this is useless in our case. |
| 324 | */ |
| 325 | void |
| 326 | setDocumentLocatorDebug(void *ctx, xmlSAXLocatorPtr loc) |
| 327 | { |
| 328 | fprintf(stdout, "SAX.setDocumentLocator()\n"); |
| 329 | } |
| 330 | |
| 331 | /** |
| 332 | * startDocumentDebug: |
| 333 | * @ctxt: An XML parser context |
| 334 | * |
| 335 | * called when the document start being processed. |
| 336 | */ |
| 337 | void |
| 338 | startDocumentDebug(void *ctx) |
| 339 | { |
| 340 | fprintf(stdout, "SAX.startDocument()\n"); |
| 341 | } |
| 342 | |
| 343 | /** |
| 344 | * endDocumentDebug: |
| 345 | * @ctxt: An XML parser context |
| 346 | * |
| 347 | * called when the document end has been detected. |
| 348 | */ |
| 349 | void |
| 350 | endDocumentDebug(void *ctx) |
| 351 | { |
| 352 | fprintf(stdout, "SAX.endDocument()\n"); |
| 353 | } |
| 354 | |
| 355 | /** |
| 356 | * startElementDebug: |
| 357 | * @ctxt: An XML parser context |
| 358 | * @name: The element name |
| 359 | * |
| 360 | * called when an opening tag has been processed. |
| 361 | */ |
| 362 | void |
| 363 | startElementDebug(void *ctx, const xmlChar *name, const xmlChar **atts) |
| 364 | { |
| 365 | int i; |
| 366 | |
| 367 | fprintf(stdout, "SAX.startElement(%s", (char *) name); |
| 368 | if (atts != NULL) { |
| 369 | for (i = 0;(atts[i] != NULL);i++) { |
Daniel Veillard | 808a3f1 | 2000-08-17 13:50:51 +0000 | [diff] [blame] | 370 | fprintf(stdout, ", %s", atts[i++]); |
Daniel Veillard | e010c17 | 2000-08-28 10:04:51 +0000 | [diff] [blame] | 371 | if (atts[i] != NULL) { |
| 372 | unsigned char output[40]; |
| 373 | const unsigned char *att = atts[i]; |
| 374 | int outlen, attlen; |
| 375 | fprintf(stdout, "='"); |
| 376 | while ((attlen = strlen((char*)att)) > 0) { |
| 377 | outlen = sizeof output - 1; |
| 378 | htmlEncodeEntities(output, &outlen, att, &attlen, '\''); |
| 379 | fprintf(stdout, "%.*s", outlen, output); |
| 380 | att += attlen; |
| 381 | } |
| 382 | fprintf(stdout, "'"); |
| 383 | } |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 384 | } |
| 385 | } |
| 386 | fprintf(stdout, ")\n"); |
| 387 | } |
| 388 | |
| 389 | /** |
| 390 | * endElementDebug: |
| 391 | * @ctxt: An XML parser context |
| 392 | * @name: The element name |
| 393 | * |
| 394 | * called when the end of an element has been detected. |
| 395 | */ |
| 396 | void |
| 397 | endElementDebug(void *ctx, const xmlChar *name) |
| 398 | { |
| 399 | fprintf(stdout, "SAX.endElement(%s)\n", (char *) name); |
| 400 | } |
| 401 | |
| 402 | /** |
| 403 | * charactersDebug: |
| 404 | * @ctxt: An XML parser context |
| 405 | * @ch: a xmlChar string |
| 406 | * @len: the number of xmlChar |
| 407 | * |
| 408 | * receiving some chars from the parser. |
| 409 | * Question: how much at a time ??? |
| 410 | */ |
| 411 | void |
| 412 | charactersDebug(void *ctx, const xmlChar *ch, int len) |
| 413 | { |
Daniel Veillard | e010c17 | 2000-08-28 10:04:51 +0000 | [diff] [blame] | 414 | unsigned char output[40]; |
Daniel Veillard | 4948eb4 | 2000-08-29 09:41:15 +0000 | [diff] [blame] | 415 | int inlen = len, outlen = 30; |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 416 | |
Daniel Veillard | 4948eb4 | 2000-08-29 09:41:15 +0000 | [diff] [blame] | 417 | htmlEncodeEntities(output, &outlen, ch, &inlen, 0); |
Daniel Veillard | e010c17 | 2000-08-28 10:04:51 +0000 | [diff] [blame] | 418 | output[outlen] = 0; |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 419 | |
| 420 | fprintf(stdout, "SAX.characters(%s, %d)\n", output, len); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 421 | } |
| 422 | |
| 423 | /** |
| 424 | * referenceDebug: |
| 425 | * @ctxt: An XML parser context |
| 426 | * @name: The entity name |
| 427 | * |
| 428 | * called when an entity reference is detected. |
| 429 | */ |
| 430 | void |
| 431 | referenceDebug(void *ctx, const xmlChar *name) |
| 432 | { |
| 433 | fprintf(stdout, "SAX.reference(%s)\n", name); |
| 434 | } |
| 435 | |
| 436 | /** |
| 437 | * ignorableWhitespaceDebug: |
| 438 | * @ctxt: An XML parser context |
| 439 | * @ch: a xmlChar string |
| 440 | * @start: the first char in the string |
| 441 | * @len: the number of xmlChar |
| 442 | * |
| 443 | * receiving some ignorable whitespaces from the parser. |
| 444 | * Question: how much at a time ??? |
| 445 | */ |
| 446 | void |
| 447 | ignorableWhitespaceDebug(void *ctx, const xmlChar *ch, int len) |
| 448 | { |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 449 | char output[40]; |
| 450 | int i; |
| 451 | |
| 452 | for (i = 0;(i<len) && (i < 30);i++) |
| 453 | output[i] = ch[i]; |
| 454 | output[i] = 0; |
| 455 | |
| 456 | fprintf(stdout, "SAX.ignorableWhitespace(%s, %d)\n", output, len); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 457 | } |
| 458 | |
| 459 | /** |
| 460 | * processingInstructionDebug: |
| 461 | * @ctxt: An XML parser context |
| 462 | * @target: the target name |
| 463 | * @data: the PI data's |
| 464 | * @len: the number of xmlChar |
| 465 | * |
| 466 | * A processing instruction has been parsed. |
| 467 | */ |
| 468 | void |
| 469 | processingInstructionDebug(void *ctx, const xmlChar *target, |
| 470 | const xmlChar *data) |
| 471 | { |
| 472 | fprintf(stdout, "SAX.processingInstruction(%s, %s)\n", |
| 473 | (char *) target, (char *) data); |
| 474 | } |
| 475 | |
| 476 | /** |
| 477 | * commentDebug: |
| 478 | * @ctxt: An XML parser context |
| 479 | * @value: the comment content |
| 480 | * |
| 481 | * A comment has been parsed. |
| 482 | */ |
| 483 | void |
| 484 | commentDebug(void *ctx, const xmlChar *value) |
| 485 | { |
| 486 | fprintf(stdout, "SAX.comment(%s)\n", value); |
| 487 | } |
| 488 | |
| 489 | /** |
| 490 | * warningDebug: |
| 491 | * @ctxt: An XML parser context |
| 492 | * @msg: the message to display/transmit |
| 493 | * @...: extra parameters for the message display |
| 494 | * |
| 495 | * Display and format a warning messages, gives file, line, position and |
| 496 | * extra parameters. |
| 497 | */ |
| 498 | void |
| 499 | warningDebug(void *ctx, const char *msg, ...) |
| 500 | { |
| 501 | va_list args; |
| 502 | |
| 503 | va_start(args, msg); |
| 504 | fprintf(stdout, "SAX.warning: "); |
| 505 | vfprintf(stdout, msg, args); |
| 506 | va_end(args); |
| 507 | } |
| 508 | |
| 509 | /** |
| 510 | * errorDebug: |
| 511 | * @ctxt: An XML parser context |
| 512 | * @msg: the message to display/transmit |
| 513 | * @...: extra parameters for the message display |
| 514 | * |
| 515 | * Display and format a error messages, gives file, line, position and |
| 516 | * extra parameters. |
| 517 | */ |
| 518 | void |
| 519 | errorDebug(void *ctx, const char *msg, ...) |
| 520 | { |
| 521 | va_list args; |
| 522 | |
| 523 | va_start(args, msg); |
| 524 | fprintf(stdout, "SAX.error: "); |
| 525 | vfprintf(stdout, msg, args); |
| 526 | va_end(args); |
| 527 | } |
| 528 | |
| 529 | /** |
| 530 | * fatalErrorDebug: |
| 531 | * @ctxt: An XML parser context |
| 532 | * @msg: the message to display/transmit |
| 533 | * @...: extra parameters for the message display |
| 534 | * |
| 535 | * Display and format a fatalError messages, gives file, line, position and |
| 536 | * extra parameters. |
| 537 | */ |
| 538 | void |
| 539 | fatalErrorDebug(void *ctx, const char *msg, ...) |
| 540 | { |
| 541 | va_list args; |
| 542 | |
| 543 | va_start(args, msg); |
| 544 | fprintf(stdout, "SAX.fatalError: "); |
| 545 | vfprintf(stdout, msg, args); |
| 546 | va_end(args); |
| 547 | } |
| 548 | |
| 549 | xmlSAXHandler debugSAXHandlerStruct = { |
| 550 | internalSubsetDebug, |
| 551 | isStandaloneDebug, |
| 552 | hasInternalSubsetDebug, |
| 553 | hasExternalSubsetDebug, |
| 554 | resolveEntityDebug, |
| 555 | getEntityDebug, |
| 556 | entityDeclDebug, |
| 557 | notationDeclDebug, |
| 558 | attributeDeclDebug, |
| 559 | elementDeclDebug, |
| 560 | unparsedEntityDeclDebug, |
| 561 | setDocumentLocatorDebug, |
| 562 | startDocumentDebug, |
| 563 | endDocumentDebug, |
| 564 | startElementDebug, |
| 565 | endElementDebug, |
| 566 | referenceDebug, |
| 567 | charactersDebug, |
| 568 | ignorableWhitespaceDebug, |
| 569 | processingInstructionDebug, |
| 570 | commentDebug, |
| 571 | warningDebug, |
| 572 | errorDebug, |
| 573 | fatalErrorDebug, |
| 574 | getParameterEntityDebug, |
| 575 | }; |
| 576 | |
| 577 | xmlSAXHandlerPtr debugSAXHandler = &debugSAXHandlerStruct; |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 578 | /************************************************************************ |
| 579 | * * |
| 580 | * Debug * |
| 581 | * * |
| 582 | ************************************************************************/ |
| 583 | |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 584 | void parseSAXFile(char *filename) { |
Daniel Veillard | f0cc7cc | 2000-08-26 21:40:43 +0000 | [diff] [blame] | 585 | htmlDocPtr doc = NULL; |
| 586 | |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 587 | /* |
| 588 | * Empty callbacks for checking |
| 589 | */ |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 590 | if (push) { |
| 591 | FILE *f; |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 592 | |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 593 | f = fopen(filename, "r"); |
| 594 | if (f != NULL) { |
| 595 | int res, size = 3; |
| 596 | char chars[4096]; |
| 597 | htmlParserCtxtPtr ctxt; |
| 598 | |
| 599 | /* if (repeat) */ |
| 600 | size = 4096; |
| 601 | res = fread(chars, 1, 4, f); |
| 602 | if (res > 0) { |
| 603 | ctxt = htmlCreatePushParserCtxt(emptySAXHandler, NULL, |
| 604 | chars, res, filename, 0); |
| 605 | while ((res = fread(chars, 1, size, f)) > 0) { |
| 606 | htmlParseChunk(ctxt, chars, res, 0); |
| 607 | } |
| 608 | htmlParseChunk(ctxt, chars, 0, 1); |
| 609 | doc = ctxt->myDoc; |
| 610 | htmlFreeParserCtxt(ctxt); |
| 611 | } |
| 612 | if (doc != NULL) { |
| 613 | fprintf(stdout, "htmlSAXParseFile returned non-NULL\n"); |
| 614 | xmlFreeDoc(doc); |
| 615 | } |
| 616 | fclose(f); |
| 617 | } |
| 618 | if (!noout) { |
| 619 | f = fopen(filename, "r"); |
| 620 | if (f != NULL) { |
| 621 | int res, size = 3; |
| 622 | char chars[4096]; |
| 623 | htmlParserCtxtPtr ctxt; |
| 624 | |
| 625 | /* if (repeat) */ |
| 626 | size = 4096; |
| 627 | res = fread(chars, 1, 4, f); |
| 628 | if (res > 0) { |
| 629 | ctxt = htmlCreatePushParserCtxt(debugSAXHandler, NULL, |
| 630 | chars, res, filename, 0); |
| 631 | while ((res = fread(chars, 1, size, f)) > 0) { |
| 632 | htmlParseChunk(ctxt, chars, res, 0); |
| 633 | } |
| 634 | htmlParseChunk(ctxt, chars, 0, 1); |
| 635 | doc = ctxt->myDoc; |
| 636 | htmlFreeParserCtxt(ctxt); |
| 637 | } |
| 638 | if (doc != NULL) { |
| 639 | fprintf(stdout, "htmlSAXParseFile returned non-NULL\n"); |
| 640 | xmlFreeDoc(doc); |
| 641 | } |
| 642 | fclose(f); |
| 643 | } |
| 644 | } |
| 645 | } else { |
| 646 | doc = htmlSAXParseFile(filename, NULL, emptySAXHandler, NULL); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 647 | if (doc != NULL) { |
| 648 | fprintf(stdout, "htmlSAXParseFile returned non-NULL\n"); |
| 649 | xmlFreeDoc(doc); |
| 650 | } |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 651 | |
| 652 | if (!noout) { |
| 653 | /* |
| 654 | * Debug callback |
| 655 | */ |
| 656 | doc = htmlSAXParseFile(filename, NULL, debugSAXHandler, NULL); |
| 657 | if (doc != NULL) { |
| 658 | fprintf(stdout, "htmlSAXParseFile returned non-NULL\n"); |
| 659 | xmlFreeDoc(doc); |
| 660 | } |
| 661 | } |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 662 | } |
| 663 | } |
| 664 | |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 665 | void parseAndPrintFile(char *filename) { |
Daniel Veillard | 2eac503 | 2000-01-09 21:08:56 +0000 | [diff] [blame] | 666 | htmlDocPtr doc = NULL, tmp; |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 667 | |
| 668 | /* |
| 669 | * build an HTML tree from a string; |
| 670 | */ |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 671 | if (push) { |
| 672 | FILE *f; |
| 673 | |
| 674 | f = fopen(filename, "r"); |
| 675 | if (f != NULL) { |
| 676 | int res, size = 3; |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 677 | char chars[4096]; |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 678 | htmlParserCtxtPtr ctxt; |
| 679 | |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 680 | /* if (repeat) */ |
| 681 | size = 4096; |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 682 | res = fread(chars, 1, 4, f); |
| 683 | if (res > 0) { |
| 684 | ctxt = htmlCreatePushParserCtxt(NULL, NULL, |
| 685 | chars, res, filename, 0); |
| 686 | while ((res = fread(chars, 1, size, f)) > 0) { |
| 687 | htmlParseChunk(ctxt, chars, res, 0); |
| 688 | } |
| 689 | htmlParseChunk(ctxt, chars, 0, 1); |
| 690 | doc = ctxt->myDoc; |
| 691 | htmlFreeParserCtxt(ctxt); |
| 692 | } |
Daniel Veillard | 87b9539 | 2000-08-12 21:12:04 +0000 | [diff] [blame] | 693 | fclose(f); |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 694 | } |
| 695 | } else { |
| 696 | doc = htmlParseFile(filename, NULL); |
| 697 | } |
| 698 | if (doc == NULL) { |
| 699 | fprintf(stderr, "Could not parse %s\n", filename); |
| 700 | } |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 701 | |
| 702 | /* |
| 703 | * test intermediate copy if needed. |
| 704 | */ |
| 705 | if (copy) { |
| 706 | tmp = doc; |
| 707 | doc = xmlCopyDoc(doc, 1); |
| 708 | xmlFreeDoc(tmp); |
| 709 | } |
| 710 | |
| 711 | /* |
| 712 | * print it. |
| 713 | */ |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 714 | if (!noout) { |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 715 | #ifdef LIBXML_DEBUG_ENABLED |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 716 | if (!debug) { |
| 717 | if (encoding) |
| 718 | htmlSaveFileEnc("-", doc, encoding); |
| 719 | else |
| 720 | htmlDocDump(stdout, doc); |
| 721 | } else |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 722 | xmlDebugDumpDocument(stdout, doc); |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 723 | #else |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 724 | if (encoding) |
| 725 | htmlSaveFileEnc("-", doc, encoding); |
| 726 | else |
| 727 | htmlDocDump(stdout, doc); |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 728 | #endif |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 729 | } |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 730 | |
| 731 | /* |
| 732 | * free it. |
| 733 | */ |
| 734 | xmlFreeDoc(doc); |
| 735 | } |
| 736 | |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 737 | int main(int argc, char **argv) { |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 738 | int i, count; |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 739 | int files = 0; |
| 740 | |
| 741 | for (i = 1; i < argc ; i++) { |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 742 | #ifdef LIBXML_DEBUG_ENABLED |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 743 | if ((!strcmp(argv[i], "-debug")) || (!strcmp(argv[i], "--debug"))) |
| 744 | debug++; |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 745 | else |
| 746 | #endif |
| 747 | if ((!strcmp(argv[i], "-copy")) || (!strcmp(argv[i], "--copy"))) |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 748 | copy++; |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 749 | else if ((!strcmp(argv[i], "-push")) || (!strcmp(argv[i], "--push"))) |
| 750 | push++; |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 751 | else if ((!strcmp(argv[i], "-sax")) || (!strcmp(argv[i], "--sax"))) |
| 752 | sax++; |
| 753 | else if ((!strcmp(argv[i], "-noout")) || (!strcmp(argv[i], "--noout"))) |
| 754 | noout++; |
| 755 | else if ((!strcmp(argv[i], "-repeat")) || |
| 756 | (!strcmp(argv[i], "--repeat"))) |
| 757 | repeat++; |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 758 | else if ((!strcmp(argv[i], "-encode")) || |
| 759 | (!strcmp(argv[i], "--encode"))) { |
| 760 | i++; |
| 761 | encoding = argv[i]; |
| 762 | } |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 763 | } |
| 764 | for (i = 1; i < argc ; i++) { |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 765 | if ((!strcmp(argv[i], "-encode")) || |
| 766 | (!strcmp(argv[i], "--encode"))) { |
| 767 | i++; |
| 768 | continue; |
| 769 | } |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 770 | if (argv[i][0] != '-') { |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 771 | if (repeat) { |
| 772 | for (count = 0;count < 100 * repeat;count++) { |
| 773 | if (sax) |
| 774 | parseSAXFile(argv[i]); |
| 775 | else |
| 776 | parseAndPrintFile(argv[i]); |
| 777 | } |
| 778 | } else { |
| 779 | if (sax) |
| 780 | parseSAXFile(argv[i]); |
| 781 | else |
| 782 | parseAndPrintFile(argv[i]); |
| 783 | } |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 784 | files ++; |
| 785 | } |
| 786 | } |
| 787 | if (files == 0) { |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 788 | printf("Usage : %s [--debug] [--copy] [--copy] HTMLfiles ...\n", |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 789 | argv[0]); |
| 790 | printf("\tParse the HTML files and output the result of the parsing\n"); |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 791 | #ifdef LIBXML_DEBUG_ENABLED |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 792 | printf("\t--debug : dump a debug tree of the in-memory document\n"); |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 793 | #endif |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 794 | printf("\t--copy : used to test the internal copy implementation\n"); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 795 | printf("\t--sax : debug the sequence of SAX callbacks\n"); |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 796 | printf("\t--repeat : parse the file 100 times, for timing\n"); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 797 | printf("\t--noout : do not print the result\n"); |
Daniel Veillard | 5e5c623 | 1999-12-29 12:49:06 +0000 | [diff] [blame] | 798 | printf("\t--push : use the push mode parser\n"); |
Daniel Veillard | 32bc74e | 2000-07-14 14:49:25 +0000 | [diff] [blame] | 799 | printf("\t--encode encoding : output in the given encoding\n"); |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 800 | } |
Daniel Veillard | f5c2c87 | 1999-12-01 09:51:45 +0000 | [diff] [blame] | 801 | xmlCleanupParser(); |
Daniel Veillard | 7c1206f | 1999-10-14 09:10:25 +0000 | [diff] [blame] | 802 | xmlMemoryDump(); |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 803 | |
| 804 | return(0); |
| 805 | } |
Daniel Veillard | 361d845 | 2000-04-03 19:48:13 +0000 | [diff] [blame] | 806 | #else /* !LIBXML_HTML_ENABLED */ |
| 807 | #include <stdio.h> |
| 808 | int main(int argc, char **argv) { |
| 809 | printf("%s : HTML support not compiled in\n", argv[0]); |
| 810 | return(0); |
| 811 | } |
| 812 | #endif |