Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1 | /* |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 2 | * HTMLtree.c : implementation of access function for an HTML tree. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
Daniel Veillard | c5d6434 | 2001-06-24 12:13:24 +0000 | [diff] [blame] | 6 | * daniel@veillard.com |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7 | */ |
| 8 | |
| 9 | |
Bjorn Reese | 70a9da5 | 2001-04-21 16:57:29 +0000 | [diff] [blame] | 10 | #include "libxml.h" |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 11 | #ifdef LIBXML_HTML_ENABLED |
| 12 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 13 | #ifdef HAVE_CTYPE_H |
| 14 | #include <ctype.h> |
| 15 | #endif |
| 16 | #ifdef HAVE_STDLIB_H |
| 17 | #include <stdlib.h> |
| 18 | #endif |
| 19 | |
| 20 | #include <libxml/xmlmemory.h> |
| 21 | #include <libxml/HTMLparser.h> |
| 22 | #include <libxml/HTMLtree.h> |
| 23 | #include <libxml/entities.h> |
| 24 | #include <libxml/valid.h> |
| 25 | #include <libxml/xmlerror.h> |
| 26 | #include <libxml/parserInternals.h> |
Daniel Veillard | 3c01b1d | 2001-10-17 15:58:35 +0000 | [diff] [blame] | 27 | #include <libxml/globals.h> |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 28 | |
| 29 | /************************************************************************ |
| 30 | * * |
| 31 | * Getting/Setting encoding meta tags * |
| 32 | * * |
| 33 | ************************************************************************/ |
| 34 | |
| 35 | /** |
| 36 | * htmlGetMetaEncoding: |
| 37 | * @doc: the document |
| 38 | * |
| 39 | * Encoding definition lookup in the Meta tags |
| 40 | * |
| 41 | * Returns the current encoding as flagged in the HTML source |
| 42 | */ |
| 43 | const xmlChar * |
| 44 | htmlGetMetaEncoding(htmlDocPtr doc) { |
| 45 | htmlNodePtr cur; |
| 46 | const xmlChar *content; |
| 47 | const xmlChar *encoding; |
| 48 | |
| 49 | if (doc == NULL) |
| 50 | return(NULL); |
| 51 | cur = doc->children; |
| 52 | |
| 53 | /* |
| 54 | * Search the html |
| 55 | */ |
| 56 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 57 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 58 | if (xmlStrEqual(cur->name, BAD_CAST"html")) |
| 59 | break; |
| 60 | if (xmlStrEqual(cur->name, BAD_CAST"head")) |
| 61 | goto found_head; |
| 62 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) |
| 63 | goto found_meta; |
| 64 | } |
| 65 | cur = cur->next; |
| 66 | } |
| 67 | if (cur == NULL) |
| 68 | return(NULL); |
| 69 | cur = cur->children; |
| 70 | |
| 71 | /* |
| 72 | * Search the head |
| 73 | */ |
| 74 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 75 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 76 | if (xmlStrEqual(cur->name, BAD_CAST"head")) |
| 77 | break; |
| 78 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) |
| 79 | goto found_meta; |
| 80 | } |
| 81 | cur = cur->next; |
| 82 | } |
| 83 | if (cur == NULL) |
| 84 | return(NULL); |
| 85 | found_head: |
| 86 | cur = cur->children; |
| 87 | |
| 88 | /* |
| 89 | * Search the meta elements |
| 90 | */ |
| 91 | found_meta: |
| 92 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 93 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 94 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) { |
| 95 | xmlAttrPtr attr = cur->properties; |
| 96 | int http; |
| 97 | const xmlChar *value; |
| 98 | |
| 99 | content = NULL; |
| 100 | http = 0; |
| 101 | while (attr != NULL) { |
| 102 | if ((attr->children != NULL) && |
| 103 | (attr->children->type == XML_TEXT_NODE) && |
| 104 | (attr->children->next == NULL)) { |
| 105 | #ifndef XML_USE_BUFFER_CONTENT |
| 106 | value = attr->children->content; |
| 107 | #else |
| 108 | value = xmlBufferContent(attr->children->content); |
| 109 | #endif |
| 110 | if ((!xmlStrcasecmp(attr->name, BAD_CAST"http-equiv")) |
| 111 | && (!xmlStrcasecmp(value, BAD_CAST"Content-Type"))) |
| 112 | http = 1; |
| 113 | else if ((value != NULL) |
| 114 | && (!xmlStrcasecmp(attr->name, BAD_CAST"content"))) |
| 115 | content = value; |
| 116 | if ((http != 0) && (content != NULL)) |
| 117 | goto found_content; |
| 118 | } |
| 119 | attr = attr->next; |
| 120 | } |
| 121 | } |
| 122 | } |
| 123 | cur = cur->next; |
| 124 | } |
| 125 | return(NULL); |
| 126 | |
| 127 | found_content: |
| 128 | encoding = xmlStrstr(content, BAD_CAST"charset="); |
| 129 | if (encoding == NULL) |
| 130 | encoding = xmlStrstr(content, BAD_CAST"Charset="); |
| 131 | if (encoding == NULL) |
| 132 | encoding = xmlStrstr(content, BAD_CAST"CHARSET="); |
| 133 | if (encoding != NULL) { |
| 134 | encoding += 8; |
| 135 | } else { |
| 136 | encoding = xmlStrstr(content, BAD_CAST"charset ="); |
| 137 | if (encoding == NULL) |
| 138 | encoding = xmlStrstr(content, BAD_CAST"Charset ="); |
| 139 | if (encoding == NULL) |
| 140 | encoding = xmlStrstr(content, BAD_CAST"CHARSET ="); |
| 141 | if (encoding != NULL) |
| 142 | encoding += 9; |
| 143 | } |
| 144 | if (encoding != NULL) { |
| 145 | while ((*encoding == ' ') || (*encoding == '\t')) encoding++; |
| 146 | } |
| 147 | return(encoding); |
| 148 | } |
| 149 | |
| 150 | /** |
| 151 | * htmlSetMetaEncoding: |
| 152 | * @doc: the document |
| 153 | * @encoding: the encoding string |
| 154 | * |
| 155 | * Sets the current encoding in the Meta tags |
| 156 | * NOTE: this will not change the document content encoding, just |
| 157 | * the META flag associated. |
| 158 | * |
| 159 | * Returns 0 in case of success and -1 in case of error |
| 160 | */ |
| 161 | int |
| 162 | htmlSetMetaEncoding(htmlDocPtr doc, const xmlChar *encoding) { |
| 163 | htmlNodePtr cur, meta; |
| 164 | const xmlChar *content; |
| 165 | char newcontent[100]; |
| 166 | |
| 167 | |
| 168 | if (doc == NULL) |
| 169 | return(-1); |
| 170 | |
| 171 | if (encoding != NULL) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 172 | snprintf(newcontent, sizeof(newcontent), "text/html; charset=%s", |
| 173 | encoding); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 174 | newcontent[sizeof(newcontent) - 1] = 0; |
| 175 | } |
| 176 | |
| 177 | cur = doc->children; |
| 178 | |
| 179 | /* |
| 180 | * Search the html |
| 181 | */ |
| 182 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 183 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 184 | if (xmlStrcasecmp(cur->name, BAD_CAST"html") == 0) |
| 185 | break; |
| 186 | if (xmlStrcasecmp(cur->name, BAD_CAST"head") == 0) |
| 187 | goto found_head; |
| 188 | if (xmlStrcasecmp(cur->name, BAD_CAST"meta") == 0) |
| 189 | goto found_meta; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 190 | } |
| 191 | cur = cur->next; |
| 192 | } |
| 193 | if (cur == NULL) |
| 194 | return(-1); |
| 195 | cur = cur->children; |
| 196 | |
| 197 | /* |
| 198 | * Search the head |
| 199 | */ |
| 200 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 201 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 202 | if (xmlStrcasecmp(cur->name, BAD_CAST"head") == 0) |
| 203 | break; |
| 204 | if (xmlStrcasecmp(cur->name, BAD_CAST"meta") == 0) |
| 205 | goto found_meta; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 206 | } |
| 207 | cur = cur->next; |
| 208 | } |
| 209 | if (cur == NULL) |
| 210 | return(-1); |
| 211 | found_head: |
| 212 | if (cur->children == NULL) { |
| 213 | if (encoding == NULL) |
| 214 | return(0); |
| 215 | meta = xmlNewDocNode(doc, NULL, BAD_CAST"meta", NULL); |
| 216 | xmlAddChild(cur, meta); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 217 | xmlNewProp(meta, BAD_CAST"content", BAD_CAST newcontent); |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 218 | xmlNewProp(meta, BAD_CAST"http-equiv", BAD_CAST"Content-Type"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 219 | return(0); |
| 220 | } |
| 221 | cur = cur->children; |
| 222 | |
| 223 | found_meta: |
| 224 | if (encoding != NULL) { |
| 225 | /* |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 226 | * Create a new Meta element with the right attributes |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 227 | */ |
| 228 | |
| 229 | meta = xmlNewDocNode(doc, NULL, BAD_CAST"meta", NULL); |
| 230 | xmlAddPrevSibling(cur, meta); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 231 | xmlNewProp(meta, BAD_CAST"content", BAD_CAST newcontent); |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 232 | xmlNewProp(meta, BAD_CAST"http-equiv", BAD_CAST"Content-Type"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 233 | } |
| 234 | |
| 235 | /* |
| 236 | * Search and destroy all the remaining the meta elements carrying |
| 237 | * encoding informations |
| 238 | */ |
| 239 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 240 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 241 | if (xmlStrcasecmp(cur->name, BAD_CAST"meta") == 0) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 242 | xmlAttrPtr attr = cur->properties; |
| 243 | int http; |
| 244 | const xmlChar *value; |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 245 | int same_charset; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 246 | |
| 247 | content = NULL; |
| 248 | http = 0; |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 249 | same_charset = 0; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 250 | while (attr != NULL) { |
| 251 | if ((attr->children != NULL) && |
| 252 | (attr->children->type == XML_TEXT_NODE) && |
| 253 | (attr->children->next == NULL)) { |
| 254 | #ifndef XML_USE_BUFFER_CONTENT |
| 255 | value = attr->children->content; |
| 256 | #else |
| 257 | value = xmlBufferContent(attr->children->content); |
| 258 | #endif |
| 259 | if ((!xmlStrcasecmp(attr->name, BAD_CAST"http-equiv")) |
| 260 | && (!xmlStrcasecmp(value, BAD_CAST"Content-Type"))) |
| 261 | http = 1; |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 262 | else |
| 263 | { |
| 264 | if ((value != NULL) && |
| 265 | (!xmlStrcasecmp(attr->name, BAD_CAST"content"))) |
| 266 | content = value; |
| 267 | else |
| 268 | if ((!xmlStrcasecmp(attr->name, BAD_CAST"charset")) |
| 269 | && (!xmlStrcasecmp(value, encoding))) |
| 270 | same_charset = 1; |
| 271 | } |
| 272 | if ((http != 0) && (content != NULL) && (same_charset != 0)) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 273 | break; |
| 274 | } |
| 275 | attr = attr->next; |
| 276 | } |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 277 | if ((http != 0) && (content != NULL) && (same_charset != 0)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 278 | meta = cur; |
| 279 | cur = cur->next; |
| 280 | xmlUnlinkNode(meta); |
| 281 | xmlFreeNode(meta); |
| 282 | continue; |
| 283 | } |
| 284 | |
| 285 | } |
| 286 | } |
| 287 | cur = cur->next; |
| 288 | } |
| 289 | return(0); |
| 290 | } |
| 291 | |
| 292 | /************************************************************************ |
| 293 | * * |
| 294 | * Dumping HTML tree content to a simple buffer * |
| 295 | * * |
| 296 | ************************************************************************/ |
| 297 | |
| 298 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 299 | htmlDocContentDump(xmlBufferPtr buf, xmlDocPtr cur, int format); |
Daniel Veillard | 86fd5a7 | 2001-12-13 14:55:21 +0000 | [diff] [blame] | 300 | static void |
| 301 | htmlNodeDumpFormat(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur, |
| 302 | int format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 303 | |
| 304 | /** |
| 305 | * htmlDtdDump: |
| 306 | * @buf: the HTML buffer output |
| 307 | * @doc: the document |
| 308 | * |
| 309 | * Dump the HTML document DTD, if any. |
| 310 | */ |
| 311 | static void |
| 312 | htmlDtdDump(xmlBufferPtr buf, xmlDocPtr doc) { |
| 313 | xmlDtdPtr cur = doc->intSubset; |
| 314 | |
| 315 | if (cur == NULL) { |
| 316 | xmlGenericError(xmlGenericErrorContext, |
| 317 | "htmlDtdDump : no internal subset\n"); |
| 318 | return; |
| 319 | } |
| 320 | xmlBufferWriteChar(buf, "<!DOCTYPE "); |
| 321 | xmlBufferWriteCHAR(buf, cur->name); |
| 322 | if (cur->ExternalID != NULL) { |
| 323 | xmlBufferWriteChar(buf, " PUBLIC "); |
| 324 | xmlBufferWriteQuotedString(buf, cur->ExternalID); |
| 325 | if (cur->SystemID != NULL) { |
| 326 | xmlBufferWriteChar(buf, " "); |
| 327 | xmlBufferWriteQuotedString(buf, cur->SystemID); |
| 328 | } |
| 329 | } else if (cur->SystemID != NULL) { |
| 330 | xmlBufferWriteChar(buf, " SYSTEM "); |
| 331 | xmlBufferWriteQuotedString(buf, cur->SystemID); |
| 332 | } |
| 333 | xmlBufferWriteChar(buf, ">\n"); |
| 334 | } |
| 335 | |
| 336 | /** |
| 337 | * htmlAttrDump: |
| 338 | * @buf: the HTML buffer output |
| 339 | * @doc: the document |
| 340 | * @cur: the attribute pointer |
| 341 | * |
| 342 | * Dump an HTML attribute |
| 343 | */ |
| 344 | static void |
| 345 | htmlAttrDump(xmlBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur) { |
| 346 | xmlChar *value; |
| 347 | |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 348 | /* |
| 349 | * TODO: The html output method should not escape a & character |
| 350 | * occurring in an attribute value immediately followed by |
| 351 | * a { character (see Section B.7.1 of the HTML 4.0 Recommendation). |
| 352 | */ |
| 353 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 354 | if (cur == NULL) { |
| 355 | xmlGenericError(xmlGenericErrorContext, |
| 356 | "htmlAttrDump : property == NULL\n"); |
| 357 | return; |
| 358 | } |
| 359 | xmlBufferWriteChar(buf, " "); |
| 360 | xmlBufferWriteCHAR(buf, cur->name); |
| 361 | if (cur->children != NULL) { |
| 362 | value = xmlNodeListGetString(doc, cur->children, 0); |
| 363 | if (value) { |
| 364 | xmlBufferWriteChar(buf, "="); |
| 365 | xmlBufferWriteQuotedString(buf, value); |
| 366 | xmlFree(value); |
| 367 | } else { |
| 368 | xmlBufferWriteChar(buf, "=\"\""); |
| 369 | } |
| 370 | } |
| 371 | } |
| 372 | |
| 373 | /** |
| 374 | * htmlAttrListDump: |
| 375 | * @buf: the HTML buffer output |
| 376 | * @doc: the document |
| 377 | * @cur: the first attribute pointer |
| 378 | * |
| 379 | * Dump a list of HTML attributes |
| 380 | */ |
| 381 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 382 | htmlAttrListDump(xmlBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur, int format) { |
| 383 | int i = 0; |
| 384 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 385 | if (cur == NULL) { |
| 386 | xmlGenericError(xmlGenericErrorContext, |
| 387 | "htmlAttrListDump : property == NULL\n"); |
| 388 | return; |
| 389 | } |
| 390 | while (cur != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 391 | i++; |
| 392 | if ((format) && (i >= 5)) { |
| 393 | i = 0; |
| 394 | xmlBufferWriteChar(buf, "\n"); |
| 395 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 396 | htmlAttrDump(buf, doc, cur); |
| 397 | cur = cur->next; |
| 398 | } |
| 399 | } |
| 400 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 401 | /** |
| 402 | * htmlNodeListDump: |
| 403 | * @buf: the HTML buffer output |
| 404 | * @doc: the document |
| 405 | * @cur: the first node |
| 406 | * |
| 407 | * Dump an HTML node list, recursive behaviour,children are printed too. |
| 408 | */ |
| 409 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 410 | htmlNodeListDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 411 | if (cur == NULL) { |
| 412 | xmlGenericError(xmlGenericErrorContext, |
| 413 | "htmlNodeListDump : node == NULL\n"); |
| 414 | return; |
| 415 | } |
| 416 | while (cur != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 417 | htmlNodeDumpFormat(buf, doc, cur, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 418 | cur = cur->next; |
| 419 | } |
| 420 | } |
| 421 | |
| 422 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 423 | * htmlNodeDumpFormat: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 424 | * @buf: the HTML buffer output |
| 425 | * @doc: the document |
| 426 | * @cur: the current node |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 427 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 428 | * |
| 429 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 430 | */ |
Daniel Veillard | 86fd5a7 | 2001-12-13 14:55:21 +0000 | [diff] [blame] | 431 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 432 | htmlNodeDumpFormat(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur, |
| 433 | int format) { |
Daniel Veillard | bb37129 | 2001-08-16 23:26:59 +0000 | [diff] [blame] | 434 | const htmlElemDesc * info; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 435 | |
| 436 | if (cur == NULL) { |
| 437 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 438 | "htmlNodeDumpFormat : node == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 439 | return; |
| 440 | } |
| 441 | /* |
| 442 | * Special cases. |
| 443 | */ |
| 444 | if (cur->type == XML_DTD_NODE) |
| 445 | return; |
| 446 | if (cur->type == XML_HTML_DOCUMENT_NODE) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 447 | htmlDocContentDump(buf, (xmlDocPtr) cur, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 448 | return; |
| 449 | } |
| 450 | if (cur->type == HTML_TEXT_NODE) { |
| 451 | if (cur->content != NULL) { |
Daniel Veillard | b44025c | 2001-10-11 22:55:55 +0000 | [diff] [blame] | 452 | if (((cur->name == (const xmlChar *)xmlStringText) || |
| 453 | (cur->name != (const xmlChar *)xmlStringTextNoenc)) && |
Daniel Veillard | 6e93c4a | 2001-06-05 20:57:42 +0000 | [diff] [blame] | 454 | ((cur->parent == NULL) || |
| 455 | (!xmlStrEqual(cur->parent->name, BAD_CAST "script")))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 456 | xmlChar *buffer; |
| 457 | |
| 458 | #ifndef XML_USE_BUFFER_CONTENT |
| 459 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
| 460 | #else |
| 461 | buffer = xmlEncodeEntitiesReentrant(doc, |
| 462 | xmlBufferContent(cur->content)); |
| 463 | #endif |
| 464 | if (buffer != NULL) { |
| 465 | xmlBufferWriteCHAR(buf, buffer); |
| 466 | xmlFree(buffer); |
| 467 | } |
| 468 | } else { |
| 469 | xmlBufferWriteCHAR(buf, cur->content); |
| 470 | } |
| 471 | } |
| 472 | return; |
| 473 | } |
| 474 | if (cur->type == HTML_COMMENT_NODE) { |
| 475 | if (cur->content != NULL) { |
| 476 | xmlBufferWriteChar(buf, "<!--"); |
| 477 | #ifndef XML_USE_BUFFER_CONTENT |
| 478 | xmlBufferWriteCHAR(buf, cur->content); |
| 479 | #else |
| 480 | xmlBufferWriteCHAR(buf, xmlBufferContent(cur->content)); |
| 481 | #endif |
| 482 | xmlBufferWriteChar(buf, "-->"); |
| 483 | } |
| 484 | return; |
| 485 | } |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 486 | if (cur->type == HTML_PI_NODE) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 487 | if (cur->name == NULL) |
| 488 | return; |
| 489 | xmlBufferWriteChar(buf, "<?"); |
| 490 | xmlBufferWriteCHAR(buf, cur->name); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 491 | if (cur->content != NULL) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 492 | xmlBufferWriteChar(buf, " "); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 493 | #ifndef XML_USE_BUFFER_CONTENT |
| 494 | xmlBufferWriteCHAR(buf, cur->content); |
| 495 | #else |
| 496 | xmlBufferWriteCHAR(buf, xmlBufferContent(cur->content)); |
| 497 | #endif |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 498 | } |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 499 | xmlBufferWriteChar(buf, ">"); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 500 | return; |
| 501 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 502 | if (cur->type == HTML_ENTITY_REF_NODE) { |
| 503 | xmlBufferWriteChar(buf, "&"); |
| 504 | xmlBufferWriteCHAR(buf, cur->name); |
| 505 | xmlBufferWriteChar(buf, ";"); |
| 506 | return; |
| 507 | } |
Daniel Veillard | 083c266 | 2001-05-08 08:27:14 +0000 | [diff] [blame] | 508 | if (cur->type == HTML_PRESERVE_NODE) { |
| 509 | if (cur->content != NULL) { |
| 510 | #ifndef XML_USE_BUFFER_CONTENT |
| 511 | xmlBufferWriteCHAR(buf, cur->content); |
| 512 | #else |
| 513 | xmlBufferWriteCHAR(buf, xmlBufferContent(cur->content)); |
| 514 | #endif |
| 515 | } |
| 516 | return; |
| 517 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 518 | |
| 519 | /* |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 520 | * Get specific HTML info for that node. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 521 | */ |
| 522 | info = htmlTagLookup(cur->name); |
| 523 | |
| 524 | xmlBufferWriteChar(buf, "<"); |
| 525 | xmlBufferWriteCHAR(buf, cur->name); |
| 526 | if (cur->properties != NULL) |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 527 | htmlAttrListDump(buf, doc, cur->properties, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 528 | |
| 529 | if ((info != NULL) && (info->empty)) { |
| 530 | xmlBufferWriteChar(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 531 | if ((format) && (info != NULL) && (!info->isinline) && |
| 532 | (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 533 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 534 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 535 | xmlBufferWriteChar(buf, "\n"); |
| 536 | } |
| 537 | return; |
| 538 | } |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 539 | if (((cur->type == XML_ELEMENT_NODE) || (cur->content == NULL)) && |
| 540 | (cur->children == NULL)) { |
Daniel Veillard | 083c266 | 2001-05-08 08:27:14 +0000 | [diff] [blame] | 541 | if ((info != NULL) && (info->saveEndTag != 0) && |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 542 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "html")) && |
| 543 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "body"))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 544 | xmlBufferWriteChar(buf, ">"); |
Daniel Veillard | 083c266 | 2001-05-08 08:27:14 +0000 | [diff] [blame] | 545 | } else { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 546 | xmlBufferWriteChar(buf, "></"); |
| 547 | xmlBufferWriteCHAR(buf, cur->name); |
| 548 | xmlBufferWriteChar(buf, ">"); |
| 549 | } |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 550 | if ((format) && (info != NULL) && (!info->isinline) && |
| 551 | (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 552 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 553 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 554 | xmlBufferWriteChar(buf, "\n"); |
| 555 | } |
| 556 | return; |
| 557 | } |
| 558 | xmlBufferWriteChar(buf, ">"); |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 559 | if ((cur->type != XML_ELEMENT_NODE) && (cur->content != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 560 | xmlChar *buffer; |
| 561 | |
| 562 | #ifndef XML_USE_BUFFER_CONTENT |
Daniel Veillard | 083c266 | 2001-05-08 08:27:14 +0000 | [diff] [blame] | 563 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 564 | #else |
Daniel Veillard | 083c266 | 2001-05-08 08:27:14 +0000 | [diff] [blame] | 565 | buffer = xmlEncodeEntitiesReentrant(doc, |
| 566 | xmlBufferContent(cur->content)); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 567 | #endif |
| 568 | if (buffer != NULL) { |
| 569 | xmlBufferWriteCHAR(buf, buffer); |
| 570 | xmlFree(buffer); |
| 571 | } |
| 572 | } |
| 573 | if (cur->children != NULL) { |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 574 | if ((format) && (info != NULL) && (!info->isinline) && |
| 575 | (cur->children->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 576 | (cur->children->type != HTML_ENTITY_REF_NODE) && |
| 577 | (cur->children != cur->last)) |
| 578 | xmlBufferWriteChar(buf, "\n"); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 579 | htmlNodeListDump(buf, doc, cur->children, format); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 580 | if ((format) && (info != NULL) && (!info->isinline) && |
| 581 | (cur->last->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 582 | (cur->last->type != HTML_ENTITY_REF_NODE) && |
| 583 | (cur->children != cur->last)) |
| 584 | xmlBufferWriteChar(buf, "\n"); |
| 585 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 586 | xmlBufferWriteChar(buf, "</"); |
| 587 | xmlBufferWriteCHAR(buf, cur->name); |
| 588 | xmlBufferWriteChar(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 589 | if ((format) && (info != NULL) && (!info->isinline) && |
| 590 | (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 591 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 592 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 593 | xmlBufferWriteChar(buf, "\n"); |
| 594 | } |
| 595 | } |
| 596 | |
| 597 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 598 | * htmlNodeDump: |
| 599 | * @buf: the HTML buffer output |
| 600 | * @doc: the document |
| 601 | * @cur: the current node |
| 602 | * |
| 603 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 604 | * and formatting returns are added. |
| 605 | */ |
| 606 | void |
| 607 | htmlNodeDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur) { |
| 608 | htmlNodeDumpFormat(buf, doc, cur, 1); |
| 609 | } |
| 610 | |
| 611 | /** |
| 612 | * htmlNodeDumpFileFormat: |
| 613 | * @out: the FILE pointer |
| 614 | * @doc: the document |
| 615 | * @cur: the current node |
| 616 | * @encoding: the document encoding |
| 617 | * @format: should formatting spaces been added |
| 618 | * |
| 619 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 620 | * |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 621 | * TODO: if encoding == NULL try to save in the doc encoding |
| 622 | * |
| 623 | * returns: the number of byte written or -1 in case of failure. |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 624 | */ |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 625 | int |
| 626 | htmlNodeDumpFileFormat(FILE *out, xmlDocPtr doc, |
| 627 | xmlNodePtr cur, const char *encoding, int format) { |
| 628 | xmlOutputBufferPtr buf; |
| 629 | xmlCharEncodingHandlerPtr handler = NULL; |
| 630 | int ret; |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 631 | |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 632 | if (encoding != NULL) { |
| 633 | xmlCharEncoding enc; |
| 634 | |
| 635 | enc = xmlParseCharEncoding(encoding); |
| 636 | if (enc != XML_CHAR_ENCODING_UTF8) { |
| 637 | handler = xmlFindCharEncodingHandler(encoding); |
| 638 | if (handler == NULL) |
| 639 | return(-1); |
| 640 | } |
| 641 | } |
| 642 | |
| 643 | /* |
| 644 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 645 | */ |
| 646 | if (handler == NULL) |
| 647 | handler = xmlFindCharEncodingHandler("HTML"); |
| 648 | if (handler == NULL) |
| 649 | handler = xmlFindCharEncodingHandler("ascii"); |
| 650 | |
| 651 | /* |
| 652 | * save the content to a temp buffer. |
| 653 | */ |
| 654 | buf = xmlOutputBufferCreateFile(out, handler); |
| 655 | if (buf == NULL) return(0); |
| 656 | |
| 657 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, format); |
| 658 | |
| 659 | ret = xmlOutputBufferClose(buf); |
| 660 | return(ret); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 661 | } |
| 662 | |
| 663 | /** |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 664 | * htmlNodeDumpFile: |
| 665 | * @out: the FILE pointer |
| 666 | * @doc: the document |
| 667 | * @cur: the current node |
| 668 | * |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 669 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 670 | * and formatting returns are added. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 671 | */ |
| 672 | void |
| 673 | htmlNodeDumpFile(FILE *out, xmlDocPtr doc, xmlNodePtr cur) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 674 | htmlNodeDumpFileFormat(out, doc, cur, NULL, 1); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 675 | } |
| 676 | |
| 677 | /** |
| 678 | * htmlDocContentDump: |
| 679 | * @buf: the HTML buffer output |
| 680 | * @cur: the document |
| 681 | * |
| 682 | * Dump an HTML document. |
| 683 | */ |
| 684 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 685 | htmlDocContentDump(xmlBufferPtr buf, xmlDocPtr cur, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 686 | int type; |
| 687 | |
| 688 | /* |
| 689 | * force to output the stuff as HTML, especially for entities |
| 690 | */ |
| 691 | type = cur->type; |
| 692 | cur->type = XML_HTML_DOCUMENT_NODE; |
| 693 | if (cur->intSubset != NULL) |
| 694 | htmlDtdDump(buf, cur); |
| 695 | else { |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 696 | /* Default to HTML-4.0 transitional @@@@ */ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 697 | xmlBufferWriteChar(buf, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\" \"http://www.w3.org/TR/REC-html40/loose.dtd\">"); |
| 698 | |
| 699 | } |
| 700 | if (cur->children != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 701 | htmlNodeListDump(buf, cur, cur->children, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 702 | } |
| 703 | xmlBufferWriteChar(buf, "\n"); |
| 704 | cur->type = (xmlElementType) type; |
| 705 | } |
| 706 | |
| 707 | /** |
| 708 | * htmlDocDumpMemory: |
| 709 | * @cur: the document |
| 710 | * @mem: OUT: the memory pointer |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 711 | * @size: OUT: the memory length |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 712 | * |
| 713 | * Dump an HTML document in memory and return the xmlChar * and it's size. |
| 714 | * It's up to the caller to free the memory. |
| 715 | */ |
| 716 | void |
| 717 | htmlDocDumpMemory(xmlDocPtr cur, xmlChar**mem, int *size) { |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 718 | xmlOutputBufferPtr buf; |
| 719 | xmlCharEncodingHandlerPtr handler = NULL; |
| 720 | const char *encoding; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 721 | |
| 722 | if (cur == NULL) { |
| 723 | #ifdef DEBUG_TREE |
| 724 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 725 | "htmlDocDumpMemory : document == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 726 | #endif |
| 727 | *mem = NULL; |
| 728 | *size = 0; |
| 729 | return; |
| 730 | } |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 731 | |
| 732 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 733 | |
| 734 | if (encoding != NULL) { |
| 735 | xmlCharEncoding enc; |
| 736 | |
| 737 | enc = xmlParseCharEncoding(encoding); |
| 738 | if (enc != cur->charset) { |
| 739 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 740 | /* |
| 741 | * Not supported yet |
| 742 | */ |
| 743 | *mem = NULL; |
| 744 | *size = 0; |
| 745 | return; |
| 746 | } |
| 747 | |
| 748 | handler = xmlFindCharEncodingHandler(encoding); |
| 749 | if (handler == NULL) { |
| 750 | *mem = NULL; |
| 751 | *size = 0; |
| 752 | return; |
| 753 | } |
| 754 | } |
| 755 | } |
| 756 | |
| 757 | /* |
| 758 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 759 | */ |
| 760 | if (handler == NULL) |
| 761 | handler = xmlFindCharEncodingHandler("HTML"); |
| 762 | if (handler == NULL) |
| 763 | handler = xmlFindCharEncodingHandler("ascii"); |
| 764 | |
| 765 | buf = xmlAllocOutputBuffer(handler); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 766 | if (buf == NULL) { |
| 767 | *mem = NULL; |
| 768 | *size = 0; |
| 769 | return; |
| 770 | } |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 771 | |
| 772 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 773 | xmlOutputBufferFlush(buf); |
| 774 | if (buf->conv != NULL) { |
| 775 | *size = buf->conv->use; |
| 776 | *mem = xmlStrndup(buf->conv->content, *size); |
| 777 | } else { |
| 778 | *size = buf->buffer->use; |
| 779 | *mem = xmlStrndup(buf->buffer->content, *size); |
| 780 | } |
| 781 | (void)xmlOutputBufferClose(buf); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 782 | } |
| 783 | |
| 784 | |
| 785 | /************************************************************************ |
| 786 | * * |
| 787 | * Dumping HTML tree content to an I/O output buffer * |
| 788 | * * |
| 789 | ************************************************************************/ |
| 790 | |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 791 | void |
| 792 | htmlDocContentDumpFormatOutput(xmlOutputBufferPtr buf, xmlDocPtr cur, |
| 793 | const char *encoding, int format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 794 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 795 | * htmlDtdDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 796 | * @buf: the HTML buffer output |
| 797 | * @doc: the document |
| 798 | * @encoding: the encoding string |
| 799 | * |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 800 | * TODO: check whether encoding is needed |
| 801 | * |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 802 | * Dump the HTML document DTD, if any. |
| 803 | */ |
| 804 | static void |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 805 | htmlDtdDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
Daniel Veillard | c86a4fa | 2001-03-26 16:28:29 +0000 | [diff] [blame] | 806 | const char *encoding ATTRIBUTE_UNUSED) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 807 | xmlDtdPtr cur = doc->intSubset; |
| 808 | |
| 809 | if (cur == NULL) { |
| 810 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 811 | "htmlDtdDumpOutput : no internal subset\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 812 | return; |
| 813 | } |
| 814 | xmlOutputBufferWriteString(buf, "<!DOCTYPE "); |
| 815 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 816 | if (cur->ExternalID != NULL) { |
| 817 | xmlOutputBufferWriteString(buf, " PUBLIC "); |
| 818 | xmlBufferWriteQuotedString(buf->buffer, cur->ExternalID); |
| 819 | if (cur->SystemID != NULL) { |
| 820 | xmlOutputBufferWriteString(buf, " "); |
| 821 | xmlBufferWriteQuotedString(buf->buffer, cur->SystemID); |
| 822 | } |
| 823 | } else if (cur->SystemID != NULL) { |
| 824 | xmlOutputBufferWriteString(buf, " SYSTEM "); |
| 825 | xmlBufferWriteQuotedString(buf->buffer, cur->SystemID); |
| 826 | } |
| 827 | xmlOutputBufferWriteString(buf, ">\n"); |
| 828 | } |
| 829 | |
| 830 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 831 | * htmlAttrDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 832 | * @buf: the HTML buffer output |
| 833 | * @doc: the document |
| 834 | * @cur: the attribute pointer |
| 835 | * @encoding: the encoding string |
| 836 | * |
| 837 | * Dump an HTML attribute |
| 838 | */ |
| 839 | static void |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 840 | htmlAttrDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur, |
Daniel Veillard | c86a4fa | 2001-03-26 16:28:29 +0000 | [diff] [blame] | 841 | const char *encoding ATTRIBUTE_UNUSED) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 842 | xmlChar *value; |
| 843 | |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 844 | /* |
| 845 | * TODO: The html output method should not escape a & character |
| 846 | * occurring in an attribute value immediately followed by |
| 847 | * a { character (see Section B.7.1 of the HTML 4.0 Recommendation). |
| 848 | */ |
| 849 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 850 | if (cur == NULL) { |
| 851 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 852 | "htmlAttrDumpOutput : property == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 853 | return; |
| 854 | } |
| 855 | xmlOutputBufferWriteString(buf, " "); |
| 856 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 857 | if (cur->children != NULL) { |
| 858 | value = xmlNodeListGetString(doc, cur->children, 0); |
| 859 | if (value) { |
| 860 | xmlOutputBufferWriteString(buf, "="); |
| 861 | xmlBufferWriteQuotedString(buf->buffer, value); |
| 862 | xmlFree(value); |
| 863 | } else { |
| 864 | xmlOutputBufferWriteString(buf, "=\"\""); |
| 865 | } |
| 866 | } |
| 867 | } |
| 868 | |
| 869 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 870 | * htmlAttrListDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 871 | * @buf: the HTML buffer output |
| 872 | * @doc: the document |
| 873 | * @cur: the first attribute pointer |
| 874 | * @encoding: the encoding string |
| 875 | * |
| 876 | * Dump a list of HTML attributes |
| 877 | */ |
| 878 | static void |
| 879 | htmlAttrListDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur, const char *encoding) { |
| 880 | if (cur == NULL) { |
| 881 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 882 | "htmlAttrListDumpOutput : property == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 883 | return; |
| 884 | } |
| 885 | while (cur != NULL) { |
| 886 | htmlAttrDumpOutput(buf, doc, cur, encoding); |
| 887 | cur = cur->next; |
| 888 | } |
| 889 | } |
| 890 | |
| 891 | |
| 892 | void htmlNodeDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 893 | xmlNodePtr cur, const char *encoding); |
| 894 | |
| 895 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 896 | * htmlNodeListDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 897 | * @buf: the HTML buffer output |
| 898 | * @doc: the document |
| 899 | * @cur: the first node |
| 900 | * @encoding: the encoding string |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 901 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 902 | * |
| 903 | * Dump an HTML node list, recursive behaviour,children are printed too. |
| 904 | */ |
| 905 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 906 | htmlNodeListDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 907 | xmlNodePtr cur, const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 908 | if (cur == NULL) { |
| 909 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 910 | "htmlNodeListDumpOutput : node == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 911 | return; |
| 912 | } |
| 913 | while (cur != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 914 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 915 | cur = cur->next; |
| 916 | } |
| 917 | } |
| 918 | |
| 919 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 920 | * htmlNodeDumpFormatOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 921 | * @buf: the HTML buffer output |
| 922 | * @doc: the document |
| 923 | * @cur: the current node |
| 924 | * @encoding: the encoding string |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 925 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 926 | * |
| 927 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 928 | */ |
| 929 | void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 930 | htmlNodeDumpFormatOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 931 | xmlNodePtr cur, const char *encoding, int format) { |
Daniel Veillard | bb37129 | 2001-08-16 23:26:59 +0000 | [diff] [blame] | 932 | const htmlElemDesc * info; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 933 | |
| 934 | if (cur == NULL) { |
| 935 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 936 | "htmlNodeDumpFormatOutput : node == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 937 | return; |
| 938 | } |
| 939 | /* |
| 940 | * Special cases. |
| 941 | */ |
| 942 | if (cur->type == XML_DTD_NODE) |
| 943 | return; |
| 944 | if (cur->type == XML_HTML_DOCUMENT_NODE) { |
| 945 | htmlDocContentDumpOutput(buf, (xmlDocPtr) cur, encoding); |
| 946 | return; |
| 947 | } |
| 948 | if (cur->type == HTML_TEXT_NODE) { |
| 949 | if (cur->content != NULL) { |
Daniel Veillard | b44025c | 2001-10-11 22:55:55 +0000 | [diff] [blame] | 950 | if (((cur->name == (const xmlChar *)xmlStringText) || |
| 951 | (cur->name != (const xmlChar *)xmlStringTextNoenc)) && |
Daniel Veillard | 6e93c4a | 2001-06-05 20:57:42 +0000 | [diff] [blame] | 952 | ((cur->parent == NULL) || |
| 953 | (!xmlStrEqual(cur->parent->name, BAD_CAST "script")))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 954 | xmlChar *buffer; |
| 955 | |
| 956 | #ifndef XML_USE_BUFFER_CONTENT |
| 957 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
| 958 | #else |
| 959 | buffer = xmlEncodeEntitiesReentrant(doc, |
| 960 | xmlBufferContent(cur->content)); |
| 961 | #endif |
| 962 | if (buffer != NULL) { |
| 963 | xmlOutputBufferWriteString(buf, (const char *)buffer); |
| 964 | xmlFree(buffer); |
| 965 | } |
| 966 | } else { |
| 967 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
| 968 | } |
| 969 | } |
| 970 | return; |
| 971 | } |
| 972 | if (cur->type == HTML_COMMENT_NODE) { |
| 973 | if (cur->content != NULL) { |
| 974 | xmlOutputBufferWriteString(buf, "<!--"); |
| 975 | #ifndef XML_USE_BUFFER_CONTENT |
| 976 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
| 977 | #else |
| 978 | xmlOutputBufferWriteString(buf, (const char *) |
| 979 | xmlBufferContent(cur->content)); |
| 980 | #endif |
| 981 | xmlOutputBufferWriteString(buf, "-->"); |
| 982 | } |
| 983 | return; |
| 984 | } |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 985 | if (cur->type == HTML_PI_NODE) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 986 | if (cur->name == NULL) |
| 987 | return; |
| 988 | xmlOutputBufferWriteString(buf, "<?"); |
| 989 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 990 | if (cur->content != NULL) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 991 | xmlOutputBufferWriteString(buf, " "); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 992 | #ifndef XML_USE_BUFFER_CONTENT |
| 993 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
| 994 | #else |
| 995 | xmlOutputBufferWriteString(buf, (const char *) |
| 996 | xmlBufferContent(cur->content)); |
| 997 | #endif |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 998 | } |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 999 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 1000 | return; |
| 1001 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1002 | if (cur->type == HTML_ENTITY_REF_NODE) { |
| 1003 | xmlOutputBufferWriteString(buf, "&"); |
| 1004 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 1005 | xmlOutputBufferWriteString(buf, ";"); |
| 1006 | return; |
| 1007 | } |
| 1008 | if (cur->type == HTML_PRESERVE_NODE) { |
| 1009 | if (cur->content != NULL) { |
| 1010 | #ifndef XML_USE_BUFFER_CONTENT |
| 1011 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
| 1012 | #else |
| 1013 | xmlOutputBufferWriteString(buf, (const char *) |
| 1014 | xmlBufferContent(cur->content)); |
| 1015 | #endif |
| 1016 | } |
| 1017 | return; |
| 1018 | } |
| 1019 | |
| 1020 | /* |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 1021 | * Get specific HTML info for that node. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1022 | */ |
| 1023 | info = htmlTagLookup(cur->name); |
| 1024 | |
| 1025 | xmlOutputBufferWriteString(buf, "<"); |
| 1026 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 1027 | if (cur->properties != NULL) |
| 1028 | htmlAttrListDumpOutput(buf, doc, cur->properties, encoding); |
| 1029 | |
| 1030 | if ((info != NULL) && (info->empty)) { |
| 1031 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 1032 | if ((format) && (!info->isinline) && (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1033 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | 8a92629 | 2001-06-07 11:20:20 +0000 | [diff] [blame] | 1034 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 1035 | (cur->parent != NULL) && |
| 1036 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1037 | xmlOutputBufferWriteString(buf, "\n"); |
| 1038 | } |
| 1039 | return; |
| 1040 | } |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 1041 | if (((cur->type == XML_ELEMENT_NODE) || (cur->content == NULL)) && |
| 1042 | (cur->children == NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1043 | if ((info != NULL) && (info->saveEndTag != 0) && |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 1044 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "html")) && |
| 1045 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "body"))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1046 | xmlOutputBufferWriteString(buf, ">"); |
| 1047 | } else { |
| 1048 | xmlOutputBufferWriteString(buf, "></"); |
| 1049 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 1050 | xmlOutputBufferWriteString(buf, ">"); |
| 1051 | } |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 1052 | if ((format) && (cur->next != NULL) && |
| 1053 | (info != NULL) && (!info->isinline)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1054 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | 8a92629 | 2001-06-07 11:20:20 +0000 | [diff] [blame] | 1055 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 1056 | (cur->parent != NULL) && |
| 1057 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1058 | xmlOutputBufferWriteString(buf, "\n"); |
| 1059 | } |
| 1060 | return; |
| 1061 | } |
| 1062 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 1063 | if ((cur->type != XML_ELEMENT_NODE) && |
| 1064 | (cur->content != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1065 | /* |
| 1066 | * Uses the OutputBuffer property to automatically convert |
| 1067 | * invalids to charrefs |
| 1068 | */ |
| 1069 | |
| 1070 | #ifndef XML_USE_BUFFER_CONTENT |
| 1071 | xmlOutputBufferWriteString(buf, (const char *) cur->content); |
| 1072 | #else |
| 1073 | xmlOutputBufferWriteString(buf, |
| 1074 | (const char *) xmlBufferContent(cur->content)); |
| 1075 | #endif |
| 1076 | } |
| 1077 | if (cur->children != NULL) { |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 1078 | if ((format) && (info != NULL) && (!info->isinline) && |
| 1079 | (cur->children->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1080 | (cur->children->type != HTML_ENTITY_REF_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 1081 | (cur->children != cur->last) && |
| 1082 | (!xmlStrEqual(cur->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1083 | xmlOutputBufferWriteString(buf, "\n"); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1084 | htmlNodeListDumpOutput(buf, doc, cur->children, encoding, format); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 1085 | if ((format) && (info != NULL) && (!info->isinline) && |
| 1086 | (cur->last->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1087 | (cur->last->type != HTML_ENTITY_REF_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 1088 | (cur->children != cur->last) && |
| 1089 | (!xmlStrEqual(cur->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1090 | xmlOutputBufferWriteString(buf, "\n"); |
| 1091 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1092 | xmlOutputBufferWriteString(buf, "</"); |
| 1093 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 1094 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 1095 | if ((format) && (info != NULL) && (!info->isinline) && |
| 1096 | (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1097 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 1098 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 1099 | (cur->parent != NULL) && |
| 1100 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1101 | xmlOutputBufferWriteString(buf, "\n"); |
| 1102 | } |
| 1103 | } |
| 1104 | |
| 1105 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1106 | * htmlNodeDumpOutput: |
| 1107 | * @buf: the HTML buffer output |
| 1108 | * @doc: the document |
| 1109 | * @cur: the current node |
| 1110 | * @encoding: the encoding string |
| 1111 | * |
| 1112 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 1113 | * and formatting returns/spaces are added. |
| 1114 | */ |
| 1115 | void |
| 1116 | htmlNodeDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 1117 | xmlNodePtr cur, const char *encoding) { |
| 1118 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, 1); |
| 1119 | } |
| 1120 | |
| 1121 | /** |
| 1122 | * htmlDocContentDumpFormatOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1123 | * @buf: the HTML buffer output |
| 1124 | * @cur: the document |
| 1125 | * @encoding: the encoding string |
Daniel Veillard | 9d06d30 | 2002-01-22 18:15:52 +0000 | [diff] [blame] | 1126 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1127 | * |
| 1128 | * Dump an HTML document. |
| 1129 | */ |
| 1130 | void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1131 | htmlDocContentDumpFormatOutput(xmlOutputBufferPtr buf, xmlDocPtr cur, |
| 1132 | const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1133 | int type; |
| 1134 | |
| 1135 | /* |
| 1136 | * force to output the stuff as HTML, especially for entities |
| 1137 | */ |
| 1138 | type = cur->type; |
| 1139 | cur->type = XML_HTML_DOCUMENT_NODE; |
Daniel Veillard | 4dd9346 | 2001-04-02 15:16:19 +0000 | [diff] [blame] | 1140 | if (cur->intSubset != NULL) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1141 | htmlDtdDumpOutput(buf, cur, NULL); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1142 | } |
| 1143 | if (cur->children != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1144 | htmlNodeListDumpOutput(buf, cur, cur->children, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1145 | } |
| 1146 | xmlOutputBufferWriteString(buf, "\n"); |
| 1147 | cur->type = (xmlElementType) type; |
| 1148 | } |
| 1149 | |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1150 | /** |
| 1151 | * htmlDocContentDumpOutput: |
| 1152 | * @buf: the HTML buffer output |
| 1153 | * @cur: the document |
| 1154 | * @encoding: the encoding string |
| 1155 | * |
| 1156 | * Dump an HTML document. Formating return/spaces are added. |
| 1157 | */ |
| 1158 | void |
| 1159 | htmlDocContentDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr cur, |
| 1160 | const char *encoding) { |
| 1161 | htmlDocContentDumpFormatOutput(buf, cur, encoding, 1); |
| 1162 | } |
| 1163 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1164 | /************************************************************************ |
| 1165 | * * |
| 1166 | * Saving functions front-ends * |
| 1167 | * * |
| 1168 | ************************************************************************/ |
| 1169 | |
| 1170 | /** |
| 1171 | * htmlDocDump: |
| 1172 | * @f: the FILE* |
| 1173 | * @cur: the document |
| 1174 | * |
| 1175 | * Dump an HTML document to an open FILE. |
| 1176 | * |
| 1177 | * returns: the number of byte written or -1 in case of failure. |
| 1178 | */ |
| 1179 | int |
| 1180 | htmlDocDump(FILE *f, xmlDocPtr cur) { |
| 1181 | xmlOutputBufferPtr buf; |
| 1182 | xmlCharEncodingHandlerPtr handler = NULL; |
| 1183 | const char *encoding; |
| 1184 | int ret; |
| 1185 | |
| 1186 | if (cur == NULL) { |
| 1187 | #ifdef DEBUG_TREE |
| 1188 | xmlGenericError(xmlGenericErrorContext, |
| 1189 | "htmlDocDump : document == NULL\n"); |
| 1190 | #endif |
| 1191 | return(-1); |
| 1192 | } |
| 1193 | |
| 1194 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 1195 | |
| 1196 | if (encoding != NULL) { |
| 1197 | xmlCharEncoding enc; |
| 1198 | |
| 1199 | enc = xmlParseCharEncoding(encoding); |
| 1200 | if (enc != cur->charset) { |
| 1201 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 1202 | /* |
| 1203 | * Not supported yet |
| 1204 | */ |
| 1205 | return(-1); |
| 1206 | } |
| 1207 | |
| 1208 | handler = xmlFindCharEncodingHandler(encoding); |
| 1209 | if (handler == NULL) |
| 1210 | return(-1); |
| 1211 | } |
| 1212 | } |
| 1213 | |
| 1214 | /* |
| 1215 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 1216 | */ |
| 1217 | if (handler == NULL) |
| 1218 | handler = xmlFindCharEncodingHandler("HTML"); |
| 1219 | if (handler == NULL) |
| 1220 | handler = xmlFindCharEncodingHandler("ascii"); |
| 1221 | |
| 1222 | buf = xmlOutputBufferCreateFile(f, handler); |
| 1223 | if (buf == NULL) return(-1); |
| 1224 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 1225 | |
| 1226 | ret = xmlOutputBufferClose(buf); |
| 1227 | return(ret); |
| 1228 | } |
| 1229 | |
| 1230 | /** |
| 1231 | * htmlSaveFile: |
| 1232 | * @filename: the filename (or URL) |
| 1233 | * @cur: the document |
| 1234 | * |
| 1235 | * Dump an HTML document to a file. If @filename is "-" the stdout file is |
| 1236 | * used. |
| 1237 | * returns: the number of byte written or -1 in case of failure. |
| 1238 | */ |
| 1239 | int |
| 1240 | htmlSaveFile(const char *filename, xmlDocPtr cur) { |
| 1241 | xmlOutputBufferPtr buf; |
| 1242 | xmlCharEncodingHandlerPtr handler = NULL; |
| 1243 | const char *encoding; |
| 1244 | int ret; |
| 1245 | |
| 1246 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 1247 | |
| 1248 | if (encoding != NULL) { |
| 1249 | xmlCharEncoding enc; |
| 1250 | |
| 1251 | enc = xmlParseCharEncoding(encoding); |
| 1252 | if (enc != cur->charset) { |
| 1253 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 1254 | /* |
| 1255 | * Not supported yet |
| 1256 | */ |
| 1257 | return(-1); |
| 1258 | } |
| 1259 | |
| 1260 | handler = xmlFindCharEncodingHandler(encoding); |
| 1261 | if (handler == NULL) |
| 1262 | return(-1); |
| 1263 | } |
| 1264 | } |
| 1265 | |
| 1266 | /* |
| 1267 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 1268 | */ |
| 1269 | if (handler == NULL) |
| 1270 | handler = xmlFindCharEncodingHandler("HTML"); |
| 1271 | if (handler == NULL) |
| 1272 | handler = xmlFindCharEncodingHandler("ascii"); |
| 1273 | |
| 1274 | /* |
| 1275 | * save the content to a temp buffer. |
| 1276 | */ |
| 1277 | buf = xmlOutputBufferCreateFilename(filename, handler, cur->compression); |
| 1278 | if (buf == NULL) return(0); |
| 1279 | |
| 1280 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 1281 | |
| 1282 | ret = xmlOutputBufferClose(buf); |
| 1283 | return(ret); |
| 1284 | } |
| 1285 | |
| 1286 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1287 | * htmlSaveFileFormat: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1288 | * @filename: the filename |
| 1289 | * @cur: the document |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1290 | * @format: should formatting spaces been added |
| 1291 | * @encoding: the document encoding |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1292 | * |
| 1293 | * Dump an HTML document to a file using a given encoding. |
| 1294 | * |
| 1295 | * returns: the number of byte written or -1 in case of failure. |
| 1296 | */ |
| 1297 | int |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1298 | htmlSaveFileFormat(const char *filename, xmlDocPtr cur, |
| 1299 | const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1300 | xmlOutputBufferPtr buf; |
| 1301 | xmlCharEncodingHandlerPtr handler = NULL; |
| 1302 | int ret; |
| 1303 | |
| 1304 | if (encoding != NULL) { |
| 1305 | xmlCharEncoding enc; |
| 1306 | |
| 1307 | enc = xmlParseCharEncoding(encoding); |
| 1308 | if (enc != cur->charset) { |
| 1309 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 1310 | /* |
| 1311 | * Not supported yet |
| 1312 | */ |
| 1313 | return(-1); |
| 1314 | } |
| 1315 | |
| 1316 | handler = xmlFindCharEncodingHandler(encoding); |
| 1317 | if (handler == NULL) |
| 1318 | return(-1); |
| 1319 | htmlSetMetaEncoding(cur, (const xmlChar *) encoding); |
| 1320 | } |
Daniel Veillard | 4dd9346 | 2001-04-02 15:16:19 +0000 | [diff] [blame] | 1321 | } else { |
| 1322 | htmlSetMetaEncoding(cur, (const xmlChar *) "UTF-8"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1323 | } |
| 1324 | |
| 1325 | /* |
| 1326 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 1327 | */ |
| 1328 | if (handler == NULL) |
| 1329 | handler = xmlFindCharEncodingHandler("HTML"); |
| 1330 | if (handler == NULL) |
| 1331 | handler = xmlFindCharEncodingHandler("ascii"); |
| 1332 | |
| 1333 | /* |
| 1334 | * save the content to a temp buffer. |
| 1335 | */ |
| 1336 | buf = xmlOutputBufferCreateFilename(filename, handler, 0); |
| 1337 | if (buf == NULL) return(0); |
| 1338 | |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1339 | htmlDocContentDumpFormatOutput(buf, cur, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1340 | |
| 1341 | ret = xmlOutputBufferClose(buf); |
| 1342 | return(ret); |
| 1343 | } |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1344 | |
| 1345 | /** |
| 1346 | * htmlSaveFileEnc: |
| 1347 | * @filename: the filename |
| 1348 | * @cur: the document |
| 1349 | * @encoding: the document encoding |
| 1350 | * |
| 1351 | * Dump an HTML document to a file using a given encoding |
| 1352 | * and formatting returns/spaces are added. |
| 1353 | * |
| 1354 | * returns: the number of byte written or -1 in case of failure. |
| 1355 | */ |
| 1356 | int |
| 1357 | htmlSaveFileEnc(const char *filename, xmlDocPtr cur, const char *encoding) { |
| 1358 | return(htmlSaveFileFormat(filename, cur, encoding, 1)); |
| 1359 | } |
| 1360 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1361 | #endif /* LIBXML_HTML_ENABLED */ |