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 | |
Daniel Veillard | 34ce8be | 2002-03-18 19:37:11 +0000 | [diff] [blame] | 10 | #define IN_LIBXML |
Bjorn Reese | 70a9da5 | 2001-04-21 16:57:29 +0000 | [diff] [blame] | 11 | #include "libxml.h" |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 12 | #ifdef LIBXML_HTML_ENABLED |
| 13 | |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 14 | #include <string.h> /* for memset() only ! */ |
| 15 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 16 | #ifdef HAVE_CTYPE_H |
| 17 | #include <ctype.h> |
| 18 | #endif |
| 19 | #ifdef HAVE_STDLIB_H |
| 20 | #include <stdlib.h> |
| 21 | #endif |
| 22 | |
| 23 | #include <libxml/xmlmemory.h> |
| 24 | #include <libxml/HTMLparser.h> |
| 25 | #include <libxml/HTMLtree.h> |
| 26 | #include <libxml/entities.h> |
| 27 | #include <libxml/valid.h> |
| 28 | #include <libxml/xmlerror.h> |
| 29 | #include <libxml/parserInternals.h> |
Daniel Veillard | 3c01b1d | 2001-10-17 15:58:35 +0000 | [diff] [blame] | 30 | #include <libxml/globals.h> |
Daniel Veillard | eb475a3 | 2002-04-14 22:00:22 +0000 | [diff] [blame] | 31 | #include <libxml/uri.h> |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 32 | |
| 33 | /************************************************************************ |
| 34 | * * |
| 35 | * Getting/Setting encoding meta tags * |
| 36 | * * |
| 37 | ************************************************************************/ |
| 38 | |
| 39 | /** |
| 40 | * htmlGetMetaEncoding: |
| 41 | * @doc: the document |
| 42 | * |
| 43 | * Encoding definition lookup in the Meta tags |
| 44 | * |
| 45 | * Returns the current encoding as flagged in the HTML source |
| 46 | */ |
| 47 | const xmlChar * |
| 48 | htmlGetMetaEncoding(htmlDocPtr doc) { |
| 49 | htmlNodePtr cur; |
| 50 | const xmlChar *content; |
| 51 | const xmlChar *encoding; |
| 52 | |
| 53 | if (doc == NULL) |
| 54 | return(NULL); |
| 55 | cur = doc->children; |
| 56 | |
| 57 | /* |
| 58 | * Search the html |
| 59 | */ |
| 60 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 61 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 62 | if (xmlStrEqual(cur->name, BAD_CAST"html")) |
| 63 | break; |
| 64 | if (xmlStrEqual(cur->name, BAD_CAST"head")) |
| 65 | goto found_head; |
| 66 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) |
| 67 | goto found_meta; |
| 68 | } |
| 69 | cur = cur->next; |
| 70 | } |
| 71 | if (cur == NULL) |
| 72 | return(NULL); |
| 73 | cur = cur->children; |
| 74 | |
| 75 | /* |
| 76 | * Search the head |
| 77 | */ |
| 78 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 79 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 80 | if (xmlStrEqual(cur->name, BAD_CAST"head")) |
| 81 | break; |
| 82 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) |
| 83 | goto found_meta; |
| 84 | } |
| 85 | cur = cur->next; |
| 86 | } |
| 87 | if (cur == NULL) |
| 88 | return(NULL); |
| 89 | found_head: |
| 90 | cur = cur->children; |
| 91 | |
| 92 | /* |
| 93 | * Search the meta elements |
| 94 | */ |
| 95 | found_meta: |
| 96 | while (cur != NULL) { |
Daniel Veillard | 5151c06 | 2001-10-23 13:10:19 +0000 | [diff] [blame] | 97 | if ((cur->type == XML_ELEMENT_NODE) && (cur->name != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 98 | if (xmlStrEqual(cur->name, BAD_CAST"meta")) { |
| 99 | xmlAttrPtr attr = cur->properties; |
| 100 | int http; |
| 101 | const xmlChar *value; |
| 102 | |
| 103 | content = NULL; |
| 104 | http = 0; |
| 105 | while (attr != NULL) { |
| 106 | if ((attr->children != NULL) && |
| 107 | (attr->children->type == XML_TEXT_NODE) && |
| 108 | (attr->children->next == NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 109 | value = attr->children->content; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 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); |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 217 | xmlNewProp(meta, BAD_CAST"http-equiv", BAD_CAST"Content-Type"); |
Daniel Veillard | 3a42f3f | 2002-07-17 17:57:34 +0000 | [diff] [blame] | 218 | xmlNewProp(meta, BAD_CAST"content", BAD_CAST newcontent); |
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); |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 231 | xmlNewProp(meta, BAD_CAST"http-equiv", BAD_CAST"Content-Type"); |
Daniel Veillard | 3a42f3f | 2002-07-17 17:57:34 +0000 | [diff] [blame] | 232 | xmlNewProp(meta, BAD_CAST"content", BAD_CAST newcontent); |
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; |
| 245 | |
| 246 | content = NULL; |
| 247 | http = 0; |
| 248 | while (attr != NULL) { |
| 249 | if ((attr->children != NULL) && |
| 250 | (attr->children->type == XML_TEXT_NODE) && |
| 251 | (attr->children->next == NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 252 | value = attr->children->content; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 253 | if ((!xmlStrcasecmp(attr->name, BAD_CAST"http-equiv")) |
| 254 | && (!xmlStrcasecmp(value, BAD_CAST"Content-Type"))) |
| 255 | http = 1; |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 256 | else |
| 257 | { |
| 258 | if ((value != NULL) && |
| 259 | (!xmlStrcasecmp(attr->name, BAD_CAST"content"))) |
| 260 | content = value; |
Daniel Veillard | 1ed3f88 | 2001-04-18 09:45:35 +0000 | [diff] [blame] | 261 | } |
Daniel Veillard | 4e0e297 | 2002-03-06 21:39:42 +0000 | [diff] [blame] | 262 | if ((http != 0) && (content != NULL)) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 263 | break; |
| 264 | } |
| 265 | attr = attr->next; |
| 266 | } |
Daniel Veillard | 4e0e297 | 2002-03-06 21:39:42 +0000 | [diff] [blame] | 267 | if ((http != 0) && (content != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 268 | meta = cur; |
| 269 | cur = cur->next; |
| 270 | xmlUnlinkNode(meta); |
| 271 | xmlFreeNode(meta); |
| 272 | continue; |
| 273 | } |
| 274 | |
| 275 | } |
| 276 | } |
| 277 | cur = cur->next; |
| 278 | } |
| 279 | return(0); |
| 280 | } |
| 281 | |
Daniel Veillard | c084e47 | 2002-08-12 13:27:28 +0000 | [diff] [blame] | 282 | /** |
| 283 | * booleanHTMLAttrs: |
| 284 | * |
| 285 | * These are the HTML attributes which will be output |
| 286 | * in minimized form, i.e. <option selected="selected"> will be |
| 287 | * output as <option selected>, as per XSLT 1.0 16.2 "HTML Output Method" |
| 288 | * |
| 289 | */ |
| 290 | static const char* htmlBooleanAttrs[] = { |
| 291 | "checked", "compact", "declare", "defer", "disabled", "ismap", |
| 292 | "multiple", "nohref", "noresize", "noshade", "nowrap", "readonly", |
| 293 | "selected", NULL |
| 294 | }; |
| 295 | |
| 296 | |
| 297 | /** |
| 298 | * htmlIsBooleanAttr: |
| 299 | * @name: the name of the attribute to check |
| 300 | * |
| 301 | * Determine if a given attribute is a boolean attribute. |
| 302 | * |
| 303 | * returns: false if the attribute is not boolean, true otherwise. |
| 304 | */ |
| 305 | int |
| 306 | htmlIsBooleanAttr(const xmlChar *name) |
| 307 | { |
| 308 | int i = 0; |
| 309 | |
| 310 | while (htmlBooleanAttrs[i] != NULL) { |
Daniel Veillard | abe0174 | 2002-09-26 12:40:03 +0000 | [diff] [blame] | 311 | if (xmlStrcasecmp((const xmlChar *)htmlBooleanAttrs[i], name) == 0) |
Daniel Veillard | c084e47 | 2002-08-12 13:27:28 +0000 | [diff] [blame] | 312 | return 1; |
| 313 | i++; |
| 314 | } |
| 315 | return 0; |
| 316 | } |
| 317 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 318 | /************************************************************************ |
| 319 | * * |
| 320 | * Dumping HTML tree content to a simple buffer * |
| 321 | * * |
| 322 | ************************************************************************/ |
| 323 | |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 324 | static int |
Daniel Veillard | 86fd5a7 | 2001-12-13 14:55:21 +0000 | [diff] [blame] | 325 | htmlNodeDumpFormat(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur, |
| 326 | int format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 327 | |
| 328 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 329 | * htmlNodeDumpFormat: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 330 | * @buf: the HTML buffer output |
| 331 | * @doc: the document |
| 332 | * @cur: the current node |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 333 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 334 | * |
| 335 | * Dump an HTML node, recursive behaviour,children are printed too. |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 336 | * |
| 337 | * Returns the number of byte written or -1 in case of error |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 338 | */ |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 339 | static int |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 340 | htmlNodeDumpFormat(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur, |
| 341 | int format) { |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 342 | unsigned int use; |
| 343 | int ret; |
| 344 | xmlOutputBufferPtr outbuf; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 345 | |
| 346 | if (cur == NULL) { |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 347 | return (-1); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 348 | } |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 349 | if (buf == NULL) { |
| 350 | return (-1); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 351 | } |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 352 | outbuf = (xmlOutputBufferPtr) xmlMalloc(sizeof(xmlOutputBuffer)); |
| 353 | if (outbuf == NULL) { |
| 354 | xmlGenericError(xmlGenericErrorContext, |
| 355 | "htmlNodeDumpFormat: out of memory!\n"); |
| 356 | return (-1); |
| 357 | } |
| 358 | memset(outbuf, 0, (size_t) sizeof(xmlOutputBuffer)); |
| 359 | outbuf->buffer = buf; |
| 360 | outbuf->encoder = NULL; |
| 361 | outbuf->writecallback = NULL; |
| 362 | outbuf->closecallback = NULL; |
| 363 | outbuf->context = NULL; |
| 364 | outbuf->written = 0; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 365 | |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 366 | use = buf->use; |
| 367 | htmlNodeDumpFormatOutput(outbuf, doc, cur, NULL, format); |
| 368 | xmlFree(outbuf); |
| 369 | ret = buf->use - use; |
| 370 | return (ret); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 371 | } |
| 372 | |
| 373 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 374 | * htmlNodeDump: |
| 375 | * @buf: the HTML buffer output |
| 376 | * @doc: the document |
| 377 | * @cur: the current node |
| 378 | * |
| 379 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 380 | * and formatting returns are added. |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 381 | * |
| 382 | * Returns the number of byte written or -1 in case of error |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 383 | */ |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 384 | int |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 385 | htmlNodeDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur) { |
Daniel Veillard | 8db67d2 | 2002-11-27 19:39:27 +0000 | [diff] [blame] | 386 | return(htmlNodeDumpFormat(buf, doc, cur, 1)); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 387 | } |
| 388 | |
| 389 | /** |
| 390 | * htmlNodeDumpFileFormat: |
| 391 | * @out: the FILE pointer |
| 392 | * @doc: the document |
| 393 | * @cur: the current node |
| 394 | * @encoding: the document encoding |
| 395 | * @format: should formatting spaces been added |
| 396 | * |
| 397 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 398 | * |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 399 | * TODO: if encoding == NULL try to save in the doc encoding |
| 400 | * |
| 401 | * returns: the number of byte written or -1 in case of failure. |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 402 | */ |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 403 | int |
| 404 | htmlNodeDumpFileFormat(FILE *out, xmlDocPtr doc, |
| 405 | xmlNodePtr cur, const char *encoding, int format) { |
| 406 | xmlOutputBufferPtr buf; |
| 407 | xmlCharEncodingHandlerPtr handler = NULL; |
| 408 | int ret; |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 409 | |
Daniel Veillard | c4f631d | 2001-06-14 11:11:59 +0000 | [diff] [blame] | 410 | if (encoding != NULL) { |
| 411 | xmlCharEncoding enc; |
| 412 | |
| 413 | enc = xmlParseCharEncoding(encoding); |
| 414 | if (enc != XML_CHAR_ENCODING_UTF8) { |
| 415 | handler = xmlFindCharEncodingHandler(encoding); |
| 416 | if (handler == NULL) |
| 417 | return(-1); |
| 418 | } |
| 419 | } |
| 420 | |
| 421 | /* |
| 422 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 423 | */ |
| 424 | if (handler == NULL) |
| 425 | handler = xmlFindCharEncodingHandler("HTML"); |
| 426 | if (handler == NULL) |
| 427 | handler = xmlFindCharEncodingHandler("ascii"); |
| 428 | |
| 429 | /* |
| 430 | * save the content to a temp buffer. |
| 431 | */ |
| 432 | buf = xmlOutputBufferCreateFile(out, handler); |
| 433 | if (buf == NULL) return(0); |
| 434 | |
| 435 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, format); |
| 436 | |
| 437 | ret = xmlOutputBufferClose(buf); |
| 438 | return(ret); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 439 | } |
| 440 | |
| 441 | /** |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 442 | * htmlNodeDumpFile: |
| 443 | * @out: the FILE pointer |
| 444 | * @doc: the document |
| 445 | * @cur: the current node |
| 446 | * |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 447 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 448 | * and formatting returns are added. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 449 | */ |
| 450 | void |
| 451 | htmlNodeDumpFile(FILE *out, xmlDocPtr doc, xmlNodePtr cur) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 452 | htmlNodeDumpFileFormat(out, doc, cur, NULL, 1); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 453 | } |
| 454 | |
| 455 | /** |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 456 | * htmlDocDumpMemory: |
| 457 | * @cur: the document |
| 458 | * @mem: OUT: the memory pointer |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 459 | * @size: OUT: the memory length |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 460 | * |
| 461 | * Dump an HTML document in memory and return the xmlChar * and it's size. |
| 462 | * It's up to the caller to free the memory. |
| 463 | */ |
| 464 | void |
| 465 | htmlDocDumpMemory(xmlDocPtr cur, xmlChar**mem, int *size) { |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 466 | xmlOutputBufferPtr buf; |
| 467 | xmlCharEncodingHandlerPtr handler = NULL; |
| 468 | const char *encoding; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 469 | |
| 470 | if (cur == NULL) { |
| 471 | #ifdef DEBUG_TREE |
| 472 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 473 | "htmlDocDumpMemory : document == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 474 | #endif |
| 475 | *mem = NULL; |
| 476 | *size = 0; |
| 477 | return; |
| 478 | } |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 479 | |
| 480 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 481 | |
| 482 | if (encoding != NULL) { |
| 483 | xmlCharEncoding enc; |
| 484 | |
| 485 | enc = xmlParseCharEncoding(encoding); |
| 486 | if (enc != cur->charset) { |
| 487 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 488 | /* |
| 489 | * Not supported yet |
| 490 | */ |
| 491 | *mem = NULL; |
| 492 | *size = 0; |
| 493 | return; |
| 494 | } |
| 495 | |
| 496 | handler = xmlFindCharEncodingHandler(encoding); |
| 497 | if (handler == NULL) { |
| 498 | *mem = NULL; |
| 499 | *size = 0; |
| 500 | return; |
| 501 | } |
| 502 | } |
| 503 | } |
| 504 | |
| 505 | /* |
| 506 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 507 | */ |
| 508 | if (handler == NULL) |
| 509 | handler = xmlFindCharEncodingHandler("HTML"); |
| 510 | if (handler == NULL) |
| 511 | handler = xmlFindCharEncodingHandler("ascii"); |
| 512 | |
| 513 | buf = xmlAllocOutputBuffer(handler); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 514 | if (buf == NULL) { |
| 515 | *mem = NULL; |
| 516 | *size = 0; |
| 517 | return; |
| 518 | } |
Daniel Veillard | 2d70372 | 2001-05-30 18:32:34 +0000 | [diff] [blame] | 519 | |
| 520 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 521 | xmlOutputBufferFlush(buf); |
| 522 | if (buf->conv != NULL) { |
| 523 | *size = buf->conv->use; |
| 524 | *mem = xmlStrndup(buf->conv->content, *size); |
| 525 | } else { |
| 526 | *size = buf->buffer->use; |
| 527 | *mem = xmlStrndup(buf->buffer->content, *size); |
| 528 | } |
| 529 | (void)xmlOutputBufferClose(buf); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 530 | } |
| 531 | |
| 532 | |
| 533 | /************************************************************************ |
| 534 | * * |
| 535 | * Dumping HTML tree content to an I/O output buffer * |
| 536 | * * |
| 537 | ************************************************************************/ |
| 538 | |
Daniel Veillard | c084e47 | 2002-08-12 13:27:28 +0000 | [diff] [blame] | 539 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 540 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 541 | * htmlDtdDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 542 | * @buf: the HTML buffer output |
| 543 | * @doc: the document |
| 544 | * @encoding: the encoding string |
| 545 | * |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 546 | * TODO: check whether encoding is needed |
| 547 | * |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 548 | * Dump the HTML document DTD, if any. |
| 549 | */ |
| 550 | static void |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 551 | htmlDtdDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
Daniel Veillard | c86a4fa | 2001-03-26 16:28:29 +0000 | [diff] [blame] | 552 | const char *encoding ATTRIBUTE_UNUSED) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 553 | xmlDtdPtr cur = doc->intSubset; |
| 554 | |
| 555 | if (cur == NULL) { |
| 556 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 557 | "htmlDtdDumpOutput : no internal subset\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 558 | return; |
| 559 | } |
| 560 | xmlOutputBufferWriteString(buf, "<!DOCTYPE "); |
| 561 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 562 | if (cur->ExternalID != NULL) { |
| 563 | xmlOutputBufferWriteString(buf, " PUBLIC "); |
| 564 | xmlBufferWriteQuotedString(buf->buffer, cur->ExternalID); |
| 565 | if (cur->SystemID != NULL) { |
| 566 | xmlOutputBufferWriteString(buf, " "); |
| 567 | xmlBufferWriteQuotedString(buf->buffer, cur->SystemID); |
| 568 | } |
| 569 | } else if (cur->SystemID != NULL) { |
| 570 | xmlOutputBufferWriteString(buf, " SYSTEM "); |
| 571 | xmlBufferWriteQuotedString(buf->buffer, cur->SystemID); |
| 572 | } |
| 573 | xmlOutputBufferWriteString(buf, ">\n"); |
| 574 | } |
| 575 | |
| 576 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 577 | * htmlAttrDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 578 | * @buf: the HTML buffer output |
| 579 | * @doc: the document |
| 580 | * @cur: the attribute pointer |
| 581 | * @encoding: the encoding string |
| 582 | * |
| 583 | * Dump an HTML attribute |
| 584 | */ |
| 585 | static void |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 586 | htmlAttrDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur, |
Daniel Veillard | c86a4fa | 2001-03-26 16:28:29 +0000 | [diff] [blame] | 587 | const char *encoding ATTRIBUTE_UNUSED) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 588 | xmlChar *value; |
| 589 | |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 590 | /* |
| 591 | * TODO: The html output method should not escape a & character |
| 592 | * occurring in an attribute value immediately followed by |
| 593 | * a { character (see Section B.7.1 of the HTML 4.0 Recommendation). |
| 594 | */ |
| 595 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 596 | if (cur == NULL) { |
| 597 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 598 | "htmlAttrDumpOutput : property == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 599 | return; |
| 600 | } |
| 601 | xmlOutputBufferWriteString(buf, " "); |
| 602 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
Daniel Veillard | c084e47 | 2002-08-12 13:27:28 +0000 | [diff] [blame] | 603 | if ((cur->children != NULL) && (!htmlIsBooleanAttr(cur->name))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 604 | value = xmlNodeListGetString(doc, cur->children, 0); |
| 605 | if (value) { |
| 606 | xmlOutputBufferWriteString(buf, "="); |
Daniel Veillard | ad11b30 | 2002-08-12 14:53:41 +0000 | [diff] [blame] | 607 | if ((!xmlStrcasecmp(cur->name, BAD_CAST "href")) || |
| 608 | (!xmlStrcasecmp(cur->name, BAD_CAST "src"))) { |
Daniel Veillard | eb475a3 | 2002-04-14 22:00:22 +0000 | [diff] [blame] | 609 | xmlChar *escaped; |
| 610 | xmlChar *tmp = value; |
| 611 | |
| 612 | while (IS_BLANK(*tmp)) tmp++; |
| 613 | |
Daniel Veillard | 6231e84 | 2002-04-18 11:54:04 +0000 | [diff] [blame] | 614 | escaped = xmlURIEscapeStr(tmp, BAD_CAST"@/:=?;#%&"); |
Daniel Veillard | eb475a3 | 2002-04-14 22:00:22 +0000 | [diff] [blame] | 615 | if (escaped != NULL) { |
| 616 | xmlBufferWriteQuotedString(buf->buffer, escaped); |
| 617 | xmlFree(escaped); |
| 618 | } else { |
| 619 | xmlBufferWriteQuotedString(buf->buffer, value); |
| 620 | } |
| 621 | } else { |
| 622 | xmlBufferWriteQuotedString(buf->buffer, value); |
| 623 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 624 | xmlFree(value); |
| 625 | } else { |
| 626 | xmlOutputBufferWriteString(buf, "=\"\""); |
| 627 | } |
| 628 | } |
| 629 | } |
| 630 | |
| 631 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 632 | * htmlAttrListDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 633 | * @buf: the HTML buffer output |
| 634 | * @doc: the document |
| 635 | * @cur: the first attribute pointer |
| 636 | * @encoding: the encoding string |
| 637 | * |
| 638 | * Dump a list of HTML attributes |
| 639 | */ |
| 640 | static void |
| 641 | htmlAttrListDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur, const char *encoding) { |
| 642 | if (cur == NULL) { |
| 643 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 644 | "htmlAttrListDumpOutput : property == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 645 | return; |
| 646 | } |
| 647 | while (cur != NULL) { |
| 648 | htmlAttrDumpOutput(buf, doc, cur, encoding); |
| 649 | cur = cur->next; |
| 650 | } |
| 651 | } |
| 652 | |
| 653 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 654 | |
| 655 | /** |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 656 | * htmlNodeListDumpOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 657 | * @buf: the HTML buffer output |
| 658 | * @doc: the document |
| 659 | * @cur: the first node |
| 660 | * @encoding: the encoding string |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 661 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 662 | * |
| 663 | * Dump an HTML node list, recursive behaviour,children are printed too. |
| 664 | */ |
| 665 | static void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 666 | htmlNodeListDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 667 | xmlNodePtr cur, const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 668 | if (cur == NULL) { |
| 669 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 670 | "htmlNodeListDumpOutput : node == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 671 | return; |
| 672 | } |
| 673 | while (cur != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 674 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 675 | cur = cur->next; |
| 676 | } |
| 677 | } |
| 678 | |
| 679 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 680 | * htmlNodeDumpFormatOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 681 | * @buf: the HTML buffer output |
| 682 | * @doc: the document |
| 683 | * @cur: the current node |
| 684 | * @encoding: the encoding string |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 685 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 686 | * |
| 687 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 688 | */ |
| 689 | void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 690 | htmlNodeDumpFormatOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 691 | xmlNodePtr cur, const char *encoding, int format) { |
Daniel Veillard | bb37129 | 2001-08-16 23:26:59 +0000 | [diff] [blame] | 692 | const htmlElemDesc * info; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 693 | |
| 694 | if (cur == NULL) { |
| 695 | xmlGenericError(xmlGenericErrorContext, |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 696 | "htmlNodeDumpFormatOutput : node == NULL\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 697 | return; |
| 698 | } |
| 699 | /* |
| 700 | * Special cases. |
| 701 | */ |
| 702 | if (cur->type == XML_DTD_NODE) |
| 703 | return; |
| 704 | if (cur->type == XML_HTML_DOCUMENT_NODE) { |
| 705 | htmlDocContentDumpOutput(buf, (xmlDocPtr) cur, encoding); |
| 706 | return; |
| 707 | } |
| 708 | if (cur->type == HTML_TEXT_NODE) { |
| 709 | if (cur->content != NULL) { |
Daniel Veillard | b44025c | 2001-10-11 22:55:55 +0000 | [diff] [blame] | 710 | if (((cur->name == (const xmlChar *)xmlStringText) || |
| 711 | (cur->name != (const xmlChar *)xmlStringTextNoenc)) && |
Daniel Veillard | 6e93c4a | 2001-06-05 20:57:42 +0000 | [diff] [blame] | 712 | ((cur->parent == NULL) || |
Daniel Veillard | 44892f7 | 2002-10-16 15:23:26 +0000 | [diff] [blame] | 713 | ((xmlStrcasecmp(cur->parent->name, BAD_CAST "script")) && |
| 714 | (xmlStrcasecmp(cur->parent->name, BAD_CAST "style"))))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 715 | xmlChar *buffer; |
| 716 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 717 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 718 | if (buffer != NULL) { |
| 719 | xmlOutputBufferWriteString(buf, (const char *)buffer); |
| 720 | xmlFree(buffer); |
| 721 | } |
| 722 | } else { |
| 723 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
| 724 | } |
| 725 | } |
| 726 | return; |
| 727 | } |
| 728 | if (cur->type == HTML_COMMENT_NODE) { |
| 729 | if (cur->content != NULL) { |
| 730 | xmlOutputBufferWriteString(buf, "<!--"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 731 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 732 | xmlOutputBufferWriteString(buf, "-->"); |
| 733 | } |
| 734 | return; |
| 735 | } |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 736 | if (cur->type == HTML_PI_NODE) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 737 | if (cur->name == NULL) |
| 738 | return; |
| 739 | xmlOutputBufferWriteString(buf, "<?"); |
| 740 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 741 | if (cur->content != NULL) { |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 742 | xmlOutputBufferWriteString(buf, " "); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 743 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 744 | } |
Daniel Veillard | 5146f20 | 2001-04-25 10:29:44 +0000 | [diff] [blame] | 745 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 7533cc8 | 2001-04-24 15:52:00 +0000 | [diff] [blame] | 746 | return; |
| 747 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 748 | if (cur->type == HTML_ENTITY_REF_NODE) { |
| 749 | xmlOutputBufferWriteString(buf, "&"); |
| 750 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 751 | xmlOutputBufferWriteString(buf, ";"); |
| 752 | return; |
| 753 | } |
| 754 | if (cur->type == HTML_PRESERVE_NODE) { |
| 755 | if (cur->content != NULL) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 756 | xmlOutputBufferWriteString(buf, (const char *)cur->content); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 757 | } |
| 758 | return; |
| 759 | } |
| 760 | |
| 761 | /* |
Daniel Veillard | cbaf399 | 2001-12-31 16:16:02 +0000 | [diff] [blame] | 762 | * Get specific HTML info for that node. |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 763 | */ |
| 764 | info = htmlTagLookup(cur->name); |
| 765 | |
| 766 | xmlOutputBufferWriteString(buf, "<"); |
| 767 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 768 | if (cur->properties != NULL) |
| 769 | htmlAttrListDumpOutput(buf, doc, cur->properties, encoding); |
| 770 | |
| 771 | if ((info != NULL) && (info->empty)) { |
| 772 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 773 | if ((format) && (!info->isinline) && (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 774 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | 8a92629 | 2001-06-07 11:20:20 +0000 | [diff] [blame] | 775 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 776 | (cur->parent != NULL) && |
| 777 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 778 | xmlOutputBufferWriteString(buf, "\n"); |
| 779 | } |
| 780 | return; |
| 781 | } |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 782 | if (((cur->type == XML_ELEMENT_NODE) || (cur->content == NULL)) && |
| 783 | (cur->children == NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 784 | if ((info != NULL) && (info->saveEndTag != 0) && |
Daniel Veillard | eca60d0 | 2001-06-13 07:45:41 +0000 | [diff] [blame] | 785 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "html")) && |
| 786 | (xmlStrcmp(BAD_CAST info->name, BAD_CAST "body"))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 787 | xmlOutputBufferWriteString(buf, ">"); |
| 788 | } else { |
| 789 | xmlOutputBufferWriteString(buf, "></"); |
| 790 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 791 | xmlOutputBufferWriteString(buf, ">"); |
| 792 | } |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 793 | if ((format) && (cur->next != NULL) && |
| 794 | (info != NULL) && (!info->isinline)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 795 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | 8a92629 | 2001-06-07 11:20:20 +0000 | [diff] [blame] | 796 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 797 | (cur->parent != NULL) && |
| 798 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 799 | xmlOutputBufferWriteString(buf, "\n"); |
| 800 | } |
| 801 | return; |
| 802 | } |
| 803 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 7db3773 | 2001-07-12 01:20:08 +0000 | [diff] [blame] | 804 | if ((cur->type != XML_ELEMENT_NODE) && |
| 805 | (cur->content != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 806 | /* |
| 807 | * Uses the OutputBuffer property to automatically convert |
| 808 | * invalids to charrefs |
| 809 | */ |
| 810 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 811 | xmlOutputBufferWriteString(buf, (const char *) cur->content); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 812 | } |
| 813 | if (cur->children != NULL) { |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 814 | if ((format) && (info != NULL) && (!info->isinline) && |
| 815 | (cur->children->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 816 | (cur->children->type != HTML_ENTITY_REF_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 817 | (cur->children != cur->last) && |
| 818 | (!xmlStrEqual(cur->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 819 | xmlOutputBufferWriteString(buf, "\n"); |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 820 | htmlNodeListDumpOutput(buf, doc, cur->children, encoding, format); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 821 | if ((format) && (info != NULL) && (!info->isinline) && |
| 822 | (cur->last->type != HTML_TEXT_NODE) && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 823 | (cur->last->type != HTML_ENTITY_REF_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 824 | (cur->children != cur->last) && |
| 825 | (!xmlStrEqual(cur->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 826 | xmlOutputBufferWriteString(buf, "\n"); |
| 827 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 828 | xmlOutputBufferWriteString(buf, "</"); |
| 829 | xmlOutputBufferWriteString(buf, (const char *)cur->name); |
| 830 | xmlOutputBufferWriteString(buf, ">"); |
Daniel Veillard | 02bb170 | 2001-06-13 21:11:59 +0000 | [diff] [blame] | 831 | if ((format) && (info != NULL) && (!info->isinline) && |
| 832 | (cur->next != NULL)) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 833 | if ((cur->next->type != HTML_TEXT_NODE) && |
Daniel Veillard | f0c5376 | 2001-06-07 16:07:07 +0000 | [diff] [blame] | 834 | (cur->next->type != HTML_ENTITY_REF_NODE) && |
| 835 | (cur->parent != NULL) && |
| 836 | (!xmlStrEqual(cur->parent->name, BAD_CAST "pre"))) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 837 | xmlOutputBufferWriteString(buf, "\n"); |
| 838 | } |
| 839 | } |
| 840 | |
| 841 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 842 | * htmlNodeDumpOutput: |
| 843 | * @buf: the HTML buffer output |
| 844 | * @doc: the document |
| 845 | * @cur: the current node |
| 846 | * @encoding: the encoding string |
| 847 | * |
| 848 | * Dump an HTML node, recursive behaviour,children are printed too, |
| 849 | * and formatting returns/spaces are added. |
| 850 | */ |
| 851 | void |
| 852 | htmlNodeDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr doc, |
| 853 | xmlNodePtr cur, const char *encoding) { |
| 854 | htmlNodeDumpFormatOutput(buf, doc, cur, encoding, 1); |
| 855 | } |
| 856 | |
| 857 | /** |
| 858 | * htmlDocContentDumpFormatOutput: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 859 | * @buf: the HTML buffer output |
| 860 | * @cur: the document |
| 861 | * @encoding: the encoding string |
Daniel Veillard | 9d06d30 | 2002-01-22 18:15:52 +0000 | [diff] [blame] | 862 | * @format: should formatting spaces been added |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 863 | * |
| 864 | * Dump an HTML document. |
| 865 | */ |
| 866 | void |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 867 | htmlDocContentDumpFormatOutput(xmlOutputBufferPtr buf, xmlDocPtr cur, |
| 868 | const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 869 | int type; |
| 870 | |
| 871 | /* |
| 872 | * force to output the stuff as HTML, especially for entities |
| 873 | */ |
| 874 | type = cur->type; |
| 875 | cur->type = XML_HTML_DOCUMENT_NODE; |
Daniel Veillard | 4dd9346 | 2001-04-02 15:16:19 +0000 | [diff] [blame] | 876 | if (cur->intSubset != NULL) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 877 | htmlDtdDumpOutput(buf, cur, NULL); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 878 | } |
| 879 | if (cur->children != NULL) { |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 880 | htmlNodeListDumpOutput(buf, cur, cur->children, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 881 | } |
| 882 | xmlOutputBufferWriteString(buf, "\n"); |
| 883 | cur->type = (xmlElementType) type; |
| 884 | } |
| 885 | |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 886 | /** |
| 887 | * htmlDocContentDumpOutput: |
| 888 | * @buf: the HTML buffer output |
| 889 | * @cur: the document |
| 890 | * @encoding: the encoding string |
| 891 | * |
| 892 | * Dump an HTML document. Formating return/spaces are added. |
| 893 | */ |
| 894 | void |
| 895 | htmlDocContentDumpOutput(xmlOutputBufferPtr buf, xmlDocPtr cur, |
| 896 | const char *encoding) { |
| 897 | htmlDocContentDumpFormatOutput(buf, cur, encoding, 1); |
| 898 | } |
| 899 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 900 | /************************************************************************ |
| 901 | * * |
| 902 | * Saving functions front-ends * |
| 903 | * * |
| 904 | ************************************************************************/ |
| 905 | |
| 906 | /** |
| 907 | * htmlDocDump: |
| 908 | * @f: the FILE* |
| 909 | * @cur: the document |
| 910 | * |
| 911 | * Dump an HTML document to an open FILE. |
| 912 | * |
| 913 | * returns: the number of byte written or -1 in case of failure. |
| 914 | */ |
| 915 | int |
| 916 | htmlDocDump(FILE *f, xmlDocPtr cur) { |
| 917 | xmlOutputBufferPtr buf; |
| 918 | xmlCharEncodingHandlerPtr handler = NULL; |
| 919 | const char *encoding; |
| 920 | int ret; |
| 921 | |
| 922 | if (cur == NULL) { |
| 923 | #ifdef DEBUG_TREE |
| 924 | xmlGenericError(xmlGenericErrorContext, |
| 925 | "htmlDocDump : document == NULL\n"); |
| 926 | #endif |
| 927 | return(-1); |
| 928 | } |
| 929 | |
| 930 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 931 | |
| 932 | if (encoding != NULL) { |
| 933 | xmlCharEncoding enc; |
| 934 | |
| 935 | enc = xmlParseCharEncoding(encoding); |
| 936 | if (enc != cur->charset) { |
| 937 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 938 | /* |
| 939 | * Not supported yet |
| 940 | */ |
| 941 | return(-1); |
| 942 | } |
| 943 | |
| 944 | handler = xmlFindCharEncodingHandler(encoding); |
| 945 | if (handler == NULL) |
| 946 | return(-1); |
| 947 | } |
| 948 | } |
| 949 | |
| 950 | /* |
| 951 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 952 | */ |
| 953 | if (handler == NULL) |
| 954 | handler = xmlFindCharEncodingHandler("HTML"); |
| 955 | if (handler == NULL) |
| 956 | handler = xmlFindCharEncodingHandler("ascii"); |
| 957 | |
| 958 | buf = xmlOutputBufferCreateFile(f, handler); |
| 959 | if (buf == NULL) return(-1); |
| 960 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 961 | |
| 962 | ret = xmlOutputBufferClose(buf); |
| 963 | return(ret); |
| 964 | } |
| 965 | |
| 966 | /** |
| 967 | * htmlSaveFile: |
| 968 | * @filename: the filename (or URL) |
| 969 | * @cur: the document |
| 970 | * |
| 971 | * Dump an HTML document to a file. If @filename is "-" the stdout file is |
| 972 | * used. |
| 973 | * returns: the number of byte written or -1 in case of failure. |
| 974 | */ |
| 975 | int |
| 976 | htmlSaveFile(const char *filename, xmlDocPtr cur) { |
| 977 | xmlOutputBufferPtr buf; |
| 978 | xmlCharEncodingHandlerPtr handler = NULL; |
| 979 | const char *encoding; |
| 980 | int ret; |
| 981 | |
| 982 | encoding = (const char *) htmlGetMetaEncoding(cur); |
| 983 | |
| 984 | if (encoding != NULL) { |
| 985 | xmlCharEncoding enc; |
| 986 | |
| 987 | enc = xmlParseCharEncoding(encoding); |
| 988 | if (enc != cur->charset) { |
| 989 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 990 | /* |
| 991 | * Not supported yet |
| 992 | */ |
| 993 | return(-1); |
| 994 | } |
| 995 | |
| 996 | handler = xmlFindCharEncodingHandler(encoding); |
| 997 | if (handler == NULL) |
| 998 | return(-1); |
| 999 | } |
| 1000 | } |
| 1001 | |
| 1002 | /* |
| 1003 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 1004 | */ |
| 1005 | if (handler == NULL) |
| 1006 | handler = xmlFindCharEncodingHandler("HTML"); |
| 1007 | if (handler == NULL) |
| 1008 | handler = xmlFindCharEncodingHandler("ascii"); |
| 1009 | |
| 1010 | /* |
| 1011 | * save the content to a temp buffer. |
| 1012 | */ |
| 1013 | buf = xmlOutputBufferCreateFilename(filename, handler, cur->compression); |
| 1014 | if (buf == NULL) return(0); |
| 1015 | |
| 1016 | htmlDocContentDumpOutput(buf, cur, NULL); |
| 1017 | |
| 1018 | ret = xmlOutputBufferClose(buf); |
| 1019 | return(ret); |
| 1020 | } |
| 1021 | |
| 1022 | /** |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1023 | * htmlSaveFileFormat: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1024 | * @filename: the filename |
| 1025 | * @cur: the document |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1026 | * @format: should formatting spaces been added |
| 1027 | * @encoding: the document encoding |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1028 | * |
| 1029 | * Dump an HTML document to a file using a given encoding. |
| 1030 | * |
| 1031 | * returns: the number of byte written or -1 in case of failure. |
| 1032 | */ |
| 1033 | int |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1034 | htmlSaveFileFormat(const char *filename, xmlDocPtr cur, |
| 1035 | const char *encoding, int format) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1036 | xmlOutputBufferPtr buf; |
| 1037 | xmlCharEncodingHandlerPtr handler = NULL; |
| 1038 | int ret; |
| 1039 | |
| 1040 | if (encoding != NULL) { |
| 1041 | xmlCharEncoding enc; |
| 1042 | |
| 1043 | enc = xmlParseCharEncoding(encoding); |
| 1044 | if (enc != cur->charset) { |
| 1045 | if (cur->charset != XML_CHAR_ENCODING_UTF8) { |
| 1046 | /* |
| 1047 | * Not supported yet |
| 1048 | */ |
| 1049 | return(-1); |
| 1050 | } |
| 1051 | |
| 1052 | handler = xmlFindCharEncodingHandler(encoding); |
| 1053 | if (handler == NULL) |
| 1054 | return(-1); |
| 1055 | htmlSetMetaEncoding(cur, (const xmlChar *) encoding); |
| 1056 | } |
Daniel Veillard | 4dd9346 | 2001-04-02 15:16:19 +0000 | [diff] [blame] | 1057 | } else { |
| 1058 | htmlSetMetaEncoding(cur, (const xmlChar *) "UTF-8"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1059 | } |
| 1060 | |
| 1061 | /* |
| 1062 | * Fallback to HTML or ASCII when the encoding is unspecified |
| 1063 | */ |
| 1064 | if (handler == NULL) |
| 1065 | handler = xmlFindCharEncodingHandler("HTML"); |
| 1066 | if (handler == NULL) |
| 1067 | handler = xmlFindCharEncodingHandler("ascii"); |
| 1068 | |
| 1069 | /* |
| 1070 | * save the content to a temp buffer. |
| 1071 | */ |
| 1072 | buf = xmlOutputBufferCreateFilename(filename, handler, 0); |
| 1073 | if (buf == NULL) return(0); |
| 1074 | |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1075 | htmlDocContentDumpFormatOutput(buf, cur, encoding, format); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1076 | |
| 1077 | ret = xmlOutputBufferClose(buf); |
| 1078 | return(ret); |
| 1079 | } |
Daniel Veillard | 95d845f | 2001-06-13 13:48:46 +0000 | [diff] [blame] | 1080 | |
| 1081 | /** |
| 1082 | * htmlSaveFileEnc: |
| 1083 | * @filename: the filename |
| 1084 | * @cur: the document |
| 1085 | * @encoding: the document encoding |
| 1086 | * |
| 1087 | * Dump an HTML document to a file using a given encoding |
| 1088 | * and formatting returns/spaces are added. |
| 1089 | * |
| 1090 | * returns: the number of byte written or -1 in case of failure. |
| 1091 | */ |
| 1092 | int |
| 1093 | htmlSaveFileEnc(const char *filename, xmlDocPtr cur, const char *encoding) { |
| 1094 | return(htmlSaveFileFormat(filename, cur, encoding, 1)); |
| 1095 | } |
| 1096 | |
Daniel Veillard | c084e47 | 2002-08-12 13:27:28 +0000 | [diff] [blame] | 1097 | |
| 1098 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1099 | #endif /* LIBXML_HTML_ENABLED */ |