Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 1 | /* |
| 2 | * HTMLtree.c : implemetation of access function for an HTML tree. |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * Daniel.Veillard@w3.org |
| 7 | */ |
| 8 | |
| 9 | #include "config.h" |
| 10 | #include <stdio.h> |
| 11 | #include <ctype.h> |
| 12 | #include <stdlib.h> |
| 13 | #include <string.h> /* for memset() only ! */ |
| 14 | |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 15 | #include "HTMLparser.h" |
| 16 | #include "HTMLtree.h" |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 17 | #include "entities.h" |
| 18 | #include "valid.h" |
| 19 | |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 20 | /** |
| 21 | * htmlDtdDump: |
| 22 | * @buf: the HTML buffer output |
| 23 | * @doc: the document |
| 24 | * |
| 25 | * Dump the HTML document DTD, if any. |
| 26 | */ |
| 27 | static void |
| 28 | htmlDtdDump(xmlBufferPtr buf, xmlDocPtr doc) { |
| 29 | xmlDtdPtr cur = doc->intSubset; |
| 30 | |
| 31 | if (cur == NULL) { |
| 32 | fprintf(stderr, "htmlDtdDump : no internal subset\n"); |
| 33 | return; |
| 34 | } |
| 35 | xmlBufferWriteChar(buf, "<!DOCTYPE "); |
| 36 | xmlBufferWriteCHAR(buf, cur->name); |
| 37 | if (cur->ExternalID != NULL) { |
| 38 | xmlBufferWriteChar(buf, " PUBLIC "); |
| 39 | xmlBufferWriteQuotedString(buf, cur->ExternalID); |
Daniel Veillard | 1566d3a | 1999-07-15 14:24:29 +0000 | [diff] [blame^] | 40 | if (cur->SystemID != NULL) { |
| 41 | xmlBufferWriteChar(buf, " "); |
| 42 | xmlBufferWriteQuotedString(buf, cur->SystemID); |
| 43 | } |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 44 | } else if (cur->SystemID != NULL) { |
| 45 | xmlBufferWriteChar(buf, " SYSTEM "); |
| 46 | xmlBufferWriteQuotedString(buf, cur->SystemID); |
| 47 | } |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 48 | xmlBufferWriteChar(buf, ">\n"); |
| 49 | } |
| 50 | |
| 51 | /** |
| 52 | * htmlAttrDump: |
| 53 | * @buf: the HTML buffer output |
| 54 | * @doc: the document |
| 55 | * @cur: the attribute pointer |
| 56 | * |
| 57 | * Dump an HTML attribute |
| 58 | */ |
| 59 | static void |
| 60 | htmlAttrDump(xmlBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur) { |
| 61 | CHAR *value; |
| 62 | |
| 63 | if (cur == NULL) { |
| 64 | fprintf(stderr, "htmlAttrDump : property == NULL\n"); |
| 65 | return; |
| 66 | } |
| 67 | xmlBufferWriteChar(buf, " "); |
| 68 | xmlBufferWriteCHAR(buf, cur->name); |
| 69 | value = xmlNodeListGetString(doc, cur->val, 0); |
| 70 | if (value) { |
| 71 | xmlBufferWriteChar(buf, "="); |
| 72 | xmlBufferWriteQuotedString(buf, value); |
| 73 | free(value); |
| 74 | } else { |
| 75 | xmlBufferWriteChar(buf, "=\"\""); |
| 76 | } |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * htmlAttrListDump: |
| 81 | * @buf: the HTML buffer output |
| 82 | * @doc: the document |
| 83 | * @cur: the first attribute pointer |
| 84 | * |
| 85 | * Dump a list of HTML attributes |
| 86 | */ |
| 87 | static void |
| 88 | htmlAttrListDump(xmlBufferPtr buf, xmlDocPtr doc, xmlAttrPtr cur) { |
| 89 | if (cur == NULL) { |
| 90 | fprintf(stderr, "htmlAttrListDump : property == NULL\n"); |
| 91 | return; |
| 92 | } |
| 93 | while (cur != NULL) { |
| 94 | htmlAttrDump(buf, doc, cur); |
| 95 | cur = cur->next; |
| 96 | } |
| 97 | } |
| 98 | |
| 99 | |
| 100 | static void |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 101 | htmlNodeDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 102 | /** |
| 103 | * htmlNodeListDump: |
| 104 | * @buf: the HTML buffer output |
| 105 | * @doc: the document |
| 106 | * @cur: the first node |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 107 | * |
| 108 | * Dump an HTML node list, recursive behaviour,children are printed too. |
| 109 | */ |
| 110 | static void |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 111 | htmlNodeListDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur) { |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 112 | if (cur == NULL) { |
| 113 | fprintf(stderr, "htmlNodeListDump : node == NULL\n"); |
| 114 | return; |
| 115 | } |
| 116 | while (cur != NULL) { |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 117 | htmlNodeDump(buf, doc, cur); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 118 | cur = cur->next; |
| 119 | } |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * htmlNodeDump: |
| 124 | * @buf: the HTML buffer output |
| 125 | * @doc: the document |
| 126 | * @cur: the current node |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 127 | * |
| 128 | * Dump an HTML node, recursive behaviour,children are printed too. |
| 129 | */ |
| 130 | static void |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 131 | htmlNodeDump(xmlBufferPtr buf, xmlDocPtr doc, xmlNodePtr cur) { |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 132 | htmlElemDescPtr info; |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 133 | |
| 134 | if (cur == NULL) { |
| 135 | fprintf(stderr, "htmlNodeDump : node == NULL\n"); |
| 136 | return; |
| 137 | } |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 138 | /* |
| 139 | * Special cases. |
| 140 | */ |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 141 | if (cur->type == HTML_TEXT_NODE) { |
| 142 | if (cur->content != NULL) { |
| 143 | CHAR *buffer; |
| 144 | |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 145 | /* uses the HTML encoding routine !!!!!!!!!! */ |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 146 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
| 147 | if (buffer != NULL) { |
| 148 | xmlBufferWriteCHAR(buf, buffer); |
| 149 | free(buffer); |
| 150 | } |
| 151 | } |
| 152 | return; |
| 153 | } |
| 154 | if (cur->type == HTML_COMMENT_NODE) { |
| 155 | if (cur->content != NULL) { |
| 156 | xmlBufferWriteChar(buf, "<!--"); |
| 157 | xmlBufferWriteCHAR(buf, cur->content); |
| 158 | xmlBufferWriteChar(buf, "-->"); |
| 159 | } |
| 160 | return; |
| 161 | } |
| 162 | if (cur->type == HTML_ENTITY_REF_NODE) { |
| 163 | xmlBufferWriteChar(buf, "&"); |
| 164 | xmlBufferWriteCHAR(buf, cur->name); |
| 165 | xmlBufferWriteChar(buf, ";"); |
| 166 | return; |
| 167 | } |
| 168 | |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 169 | /* |
| 170 | * Get specific HTmL info for taht node. |
| 171 | */ |
| 172 | info = htmlTagLookup(cur->name); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 173 | |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 174 | xmlBufferWriteChar(buf, "<"); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 175 | xmlBufferWriteCHAR(buf, cur->name); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 176 | if (cur->properties != NULL) |
| 177 | htmlAttrListDump(buf, doc, cur->properties); |
| 178 | |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 179 | if (info->empty) { |
| 180 | xmlBufferWriteChar(buf, ">"); |
| 181 | if (cur->next != NULL) { |
| 182 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 183 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 184 | xmlBufferWriteChar(buf, "\n"); |
| 185 | } |
| 186 | return; |
| 187 | } |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 188 | if ((cur->content == NULL) && (cur->childs == NULL)) { |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 189 | if (info->endTag != 0) |
| 190 | xmlBufferWriteChar(buf, ">"); |
| 191 | else { |
| 192 | xmlBufferWriteChar(buf, "></"); |
| 193 | xmlBufferWriteCHAR(buf, cur->name); |
| 194 | xmlBufferWriteChar(buf, ">"); |
| 195 | } |
| 196 | if (cur->next != NULL) { |
| 197 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 198 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 199 | xmlBufferWriteChar(buf, "\n"); |
| 200 | } |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 201 | return; |
| 202 | } |
| 203 | xmlBufferWriteChar(buf, ">"); |
| 204 | if (cur->content != NULL) { |
| 205 | CHAR *buffer; |
| 206 | |
| 207 | buffer = xmlEncodeEntitiesReentrant(doc, cur->content); |
| 208 | if (buffer != NULL) { |
| 209 | xmlBufferWriteCHAR(buf, buffer); |
| 210 | free(buffer); |
| 211 | } |
| 212 | } |
| 213 | if (cur->childs != NULL) { |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 214 | if ((cur->childs->type != HTML_TEXT_NODE) && |
| 215 | (cur->childs->type != HTML_ENTITY_REF_NODE)) |
| 216 | xmlBufferWriteChar(buf, "\n"); |
| 217 | htmlNodeListDump(buf, doc, cur->childs); |
| 218 | if ((cur->last->type != HTML_TEXT_NODE) && |
| 219 | (cur->last->type != HTML_ENTITY_REF_NODE)) |
| 220 | xmlBufferWriteChar(buf, "\n"); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 221 | } |
| 222 | xmlBufferWriteChar(buf, "</"); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 223 | xmlBufferWriteCHAR(buf, cur->name); |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 224 | xmlBufferWriteChar(buf, ">"); |
| 225 | if (cur->next != NULL) { |
| 226 | if ((cur->next->type != HTML_TEXT_NODE) && |
| 227 | (cur->next->type != HTML_ENTITY_REF_NODE)) |
| 228 | xmlBufferWriteChar(buf, "\n"); |
| 229 | } |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 230 | } |
| 231 | |
| 232 | /** |
| 233 | * htmlDocContentDump: |
| 234 | * @buf: the HTML buffer output |
| 235 | * @cur: the document |
| 236 | * |
| 237 | * Dump an HTML document. |
| 238 | */ |
| 239 | static void |
| 240 | htmlDocContentDump(xmlBufferPtr buf, xmlDocPtr cur) { |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 241 | if (cur->intSubset != NULL) |
| 242 | htmlDtdDump(buf, cur); |
| 243 | if (cur->root != NULL) { |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 244 | htmlNodeDump(buf, cur, cur->root); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 245 | } |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 246 | xmlBufferWriteChar(buf, "\n"); |
Daniel Veillard | 167b509 | 1999-07-07 04:19:20 +0000 | [diff] [blame] | 247 | } |
| 248 | |
| 249 | /** |
| 250 | * htmlDocDumpMemory: |
| 251 | * @cur: the document |
| 252 | * @mem: OUT: the memory pointer |
| 253 | * @size: OUT: the memory lenght |
| 254 | * |
| 255 | * Dump an HTML document in memory and return the CHAR * and it's size. |
| 256 | * It's up to the caller to free the memory. |
| 257 | */ |
| 258 | void |
| 259 | htmlDocDumpMemory(xmlDocPtr cur, CHAR**mem, int *size) { |
| 260 | xmlBufferPtr buf; |
| 261 | |
| 262 | if (cur == NULL) { |
| 263 | #ifdef DEBUG_TREE |
| 264 | fprintf(stderr, "xmlDocDumpMemory : document == NULL\n"); |
| 265 | #endif |
| 266 | *mem = NULL; |
| 267 | *size = 0; |
| 268 | return; |
| 269 | } |
| 270 | buf = xmlBufferCreate(); |
| 271 | if (buf == NULL) { |
| 272 | *mem = NULL; |
| 273 | *size = 0; |
| 274 | return; |
| 275 | } |
| 276 | htmlDocContentDump(buf, cur); |
| 277 | *mem = buf->content; |
| 278 | *size = buf->use; |
| 279 | memset(buf, -1, sizeof(xmlBuffer)); |
| 280 | free(buf); |
| 281 | } |
| 282 | |
| 283 | |
| 284 | /** |
| 285 | * htmlDocDump: |
| 286 | * @f: the FILE* |
| 287 | * @cur: the document |
| 288 | * |
| 289 | * Dump an HTML document to an open FILE. |
| 290 | */ |
| 291 | void |
| 292 | htmlDocDump(FILE *f, xmlDocPtr cur) { |
| 293 | xmlBufferPtr buf; |
| 294 | |
| 295 | if (cur == NULL) { |
| 296 | #ifdef DEBUG_TREE |
| 297 | fprintf(stderr, "xmlDocDump : document == NULL\n"); |
| 298 | #endif |
| 299 | return; |
| 300 | } |
| 301 | buf = xmlBufferCreate(); |
| 302 | if (buf == NULL) return; |
| 303 | htmlDocContentDump(buf, cur); |
| 304 | xmlBufferDump(f, buf); |
| 305 | xmlBufferFree(buf); |
| 306 | } |
| 307 | |
| 308 | /** |
| 309 | * htmlSaveFile: |
| 310 | * @filename: the filename |
| 311 | * @cur: the document |
| 312 | * |
| 313 | * Dump an HTML document to a file. |
| 314 | * |
| 315 | * returns: the number of byte written or -1 in case of failure. |
| 316 | */ |
| 317 | int |
| 318 | htmlSaveFile(const char *filename, xmlDocPtr cur) { |
| 319 | xmlBufferPtr buf; |
| 320 | FILE *output = NULL; |
| 321 | int ret; |
| 322 | |
| 323 | /* |
| 324 | * save the content to a temp buffer. |
| 325 | */ |
| 326 | buf = xmlBufferCreate(); |
| 327 | if (buf == NULL) return(0); |
| 328 | htmlDocContentDump(buf, cur); |
| 329 | |
| 330 | output = fopen(filename, "w"); |
| 331 | if (output == NULL) return(-1); |
| 332 | ret = xmlBufferDump(output, buf); |
| 333 | fclose(output); |
| 334 | |
| 335 | xmlBufferFree(buf); |
| 336 | return(ret * sizeof(CHAR)); |
| 337 | } |
| 338 | |