vcs.maemo.org Git - monky/blob - src/prss.c

   1 /*
   2  * Copyright (c) 2007 Mikko Sysikaski <mikko.sysikaski@gmail.com>
   3  *                    Toni Spets <toni.spets@gmail.com>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17
  18 #ifdef RSS
  19
  20 #include <libxml/parser.h>
  21 #include <libxml/tree.h>
  22 #include <string.h>
  23 #include <stdlib.h>
  24 #include <stdio.h>
  25 #include "prss.h"
  26
  27 #ifndef PARSE_OPTIONS
  28 #define PARSE_OPTIONS 0
  29 #endif
  30
  31 PRSS* prss_parse_doc(xmlDocPtr doc);
  32
  33 PRSS* prss_parse_data(const char* xml_data)
  34 {
  35         xmlDocPtr doc = xmlReadMemory(xml_data, strlen(xml_data), "", NULL, PARSE_OPTIONS);
  36         if (!doc)
  37                 return NULL;
  38
  39         return prss_parse_doc(doc);
  40 }
  41 PRSS* prss_parse_file(const char* xml_file)
  42 {
  43         xmlDocPtr doc = xmlReadFile(xml_file, NULL, PARSE_OPTIONS);
  44         if (!doc)
  45                 return NULL;
  46
  47         return prss_parse_doc(doc);
  48 }
  49 void prss_free(PRSS* data)
  50 {
  51         if (!data)
  52                 return;
  53         xmlFreeDoc(data->_data);
  54         free(data->items);
  55         free(data);
  56 }
  57
  58 static inline void prss_null(PRSS* p)
  59 {
  60         memset(p, 0, sizeof(PRSS));
  61 }
  62 static inline void prss_null_item(PRSS_Item* i)
  63 {
  64         memset(i, 0, sizeof(PRSS_Item));
  65 }
  66
  67 static inline void read_item(PRSS_Item* res, xmlNodePtr data)
  68 {
  69         prss_null_item(res);
  70
  71         res->title = res->link = res->description = NULL;
  72         for(; data; data = data->next) {
  73                 if (data->type != XML_ELEMENT_NODE)
  74                         continue;
  75                 xmlNodePtr child = data->children;
  76                 if (!child)
  77                         continue;
  78
  79                 if (!strcmp((char*)data->name, "title")) {
  80                         res->title = (char*)child->content;
  81                 } else if (!strcmp((char*)data->name, "link")) {
  82                         res->link = (char*)child->content;
  83                 } else if (!strcmp((char*)data->name, "description")) {
  84                         res->description = (char*)child->content;
  85                 } else if (!strcmp((char*)data->name, "category")) {
  86                         res->category = (char*)child->content;
  87                 } else if (!strcmp((char*)data->name, "pubDate")) {
  88                         res->pubdate = (char*)child->content;
  89                 } else if (!strcmp((char*)data->name, "guid")) {
  90                         res->guid = (char*)child->content;
  91                 }
  92         }
  93 }
  94 static inline void read_element(PRSS* res, xmlNodePtr n)
  95 {
  96         if (n->type != XML_ELEMENT_NODE)
  97                 return;
  98         xmlNodePtr child = n->children;
  99         if (!child)
 100                 return;
 101
 102         if (!strcmp((char*)n->name, "title")) {
 103                 res->title = (char*)child->content;
 104         } else if (!strcmp((char*)n->name, "link")) {
 105                 res->link = (char*)child->content;
 106         } else if (!strcmp((char*)n->name, "description")) {
 107                 res->description = (char*)child->content;
 108         } else if (!strcmp((char*)n->name, "language")) {
 109                 res->language = (char*)child->content;
 110         } else if (!strcmp((char*)n->name, "pubDate")) {
 111                 res->pubdate = (char*)child->content;
 112         } else if (!strcmp((char*)n->name, "lastBuildDate")) {
 113                 res->lastbuilddate = (char*)child->content;
 114         } else if (!strcmp((char*)n->name, "generator")) {
 115                 res->generator = (char*)child->content;
 116         } else if (!strcmp((char*)n->name, "docs")) {
 117                 res->docs = (char*)child->content;
 118         } else if (!strcmp((char*)n->name, "managingEditor")) {
 119                 res->managingeditor = (char*)child->content;
 120         } else if (!strcmp((char*)n->name, "webMaster")) {
 121                 res->webmaster = (char*)child->content;
 122         } else if (!strcmp((char*)n->name, "item")) {
 123                 read_item(&res->items[res->item_count++], n->children);
 124         }
 125 }
 126
 127 static inline int parse_rss_2_0(PRSS* res, xmlNodePtr root)
 128 {
 129         xmlNodePtr channel = root->children;
 130         while(channel
 131                         && (channel->type!=XML_ELEMENT_NODE
 132                          || strcmp((char*)channel->name, "channel")))
 133                 channel = channel->next;
 134         if (!channel)
 135                 return 0;
 136
 137         int items = 0;
 138         xmlNodePtr n;
 139         for(n = channel->children; n; n = n->next)
 140                 if (n->type==XML_ELEMENT_NODE && !strcmp((char*)n->name, "item"))
 141                         ++items;
 142
 143         res->items = malloc(items*sizeof(PRSS_Item));
 144         res->item_count = 0;
 145
 146         for(n = channel->children; n; n = n->next) {
 147                 read_element(res, n);
 148         }
 149
 150         return 1;
 151 }
 152 static inline int parse_rss_1_0(PRSS* res, xmlNodePtr root)
 153 {
 154         int items = 0;
 155         xmlNodePtr n;
 156         for(n = root->children; n; n = n->next) {
 157                 if (n->type==XML_ELEMENT_NODE) {
 158                         if (!strcmp((char*)n->name, "item"))
 159                                 ++items;
 160                         else if (!strcmp((char*)n->name, "channel")) {
 161                                 xmlNodePtr i;
 162                                 for(i = n->children; i; i = i->next) {
 163                                         read_element(res, i);
 164                                 }
 165                         }
 166                 }
 167         }
 168
 169         res->items = malloc(items*sizeof(PRSS_Item));
 170         res->item_count = 0;
 171
 172         for(n = root->children; n; n = n->next) {
 173                 if (n->type==XML_ELEMENT_NODE && !strcmp((char*)n->name, "item"))
 174                         read_item(&res->items[res->item_count++], n->children);
 175         }
 176
 177         return 1;
 178 }
 179 static inline int parse_rss_0_9x(PRSS* res, xmlNodePtr root)
 180 {
 181         // almost same...
 182         return parse_rss_2_0(res, root);
 183 }
 184
 185 PRSS* prss_parse_doc(xmlDocPtr doc)
 186 {
 187         // FIXME: doc shouldn't be freed after failure when called explicitly from program!
 188
 189         xmlNodePtr root = xmlDocGetRootElement(doc);
 190         PRSS* result = malloc(sizeof(PRSS));
 191         prss_null(result);
 192         result->_data = doc;
 193         do {
 194                 if (root->type == XML_ELEMENT_NODE) {
 195                         if (!strcmp((char*)root->name, "RDF")) {
 196                                 // RSS 1.0 document
 197                                 if (!parse_rss_1_0(result, root)) {
 198                                         free(result);
 199                                         xmlFreeDoc(doc);
 200                                         return NULL;
 201                                 }
 202                                 return result;
 203                         } else if (!strcmp((char*)root->name, "rss")) {
 204                                 // RSS 2.0 or <1.0 document
 205                                 if (!parse_rss_2_0(result, root)) {
 206                                         free(result);
 207                                         xmlFreeDoc(doc);
 208                                         return NULL;
 209                                 }
 210                                 return result;
 211                         }
 212                 }
 213                 root = root->next;
 214         } while(root);
 215         free(result);
 216         return NULL;
 217 }
 218
 219 #endif