git.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 #include <libxml/parser.h>
  19 #include <libxml/tree.h>
  20 #include <string.h>
  21 #include <stdlib.h>
  22 #include <stdio.h>
  23 #include "prss.h"
  24
  25 #ifndef PARSE_OPTIONS
  26 #define PARSE_OPTIONS 0
  27 #endif
  28
  29 PRSS* prss_parse_doc(xmlDocPtr doc);
  30
  31 PRSS* prss_parse_data(const char* xml_data)
  32 {
  33         xmlDocPtr doc = xmlReadMemory(xml_data, strlen(xml_data), "", NULL, PARSE_OPTIONS);
  34         if (!doc)
  35                 return NULL;
  36
  37         return prss_parse_doc(doc);
  38 }
  39 PRSS* prss_parse_file(const char* xml_file)
  40 {
  41         xmlDocPtr doc = xmlReadFile(xml_file, NULL, PARSE_OPTIONS);
  42         if (!doc)
  43                 return NULL;
  44
  45         return prss_parse_doc(doc);
  46 }
  47 void prss_free(PRSS* data)
  48 {
  49         if (!data)
  50                 return;
  51         xmlFreeDoc(data->_data);
  52         free(data->items);
  53         free(data);
  54 }
  55
  56 static inline void prss_null(PRSS* p)
  57 {
  58         memset(p, 0, sizeof(PRSS));
  59 }
  60 static inline void prss_null_item(PRSS_Item* i)
  61 {
  62         memset(i, 0, sizeof(PRSS_Item));
  63 }
  64
  65 static inline void read_item(PRSS_Item* res, xmlNodePtr data)
  66 {
  67         prss_null_item(res);
  68
  69         res->title = res->link = res->description = NULL;
  70         for(; data; data = data->next) {
  71                 if (data->type != XML_ELEMENT_NODE)
  72                         continue;
  73                 xmlNodePtr child = data->children;
  74                 if (!child)
  75                         continue;
  76
  77                 if (!strcmp((char*)data->name, "title")) {
  78                         res->title = (char*)child->content;
  79                 } else if (!strcmp((char*)data->name, "link")) {
  80                         res->link = (char*)child->content;
  81                 } else if (!strcmp((char*)data->name, "description")) {
  82                         res->description = (char*)child->content;
  83                 } else if (!strcmp((char*)data->name, "category")) {
  84                         res->category = (char*)child->content;
  85                 } else if (!strcmp((char*)data->name, "pubDate")) {
  86                         res->pubdate = (char*)child->content;
  87                 } else if (!strcmp((char*)data->name, "guid")) {
  88                         res->guid = (char*)child->content;
  89                 }
  90         }
  91 }
  92 static inline void read_element(PRSS* res, xmlNodePtr n)
  93 {
  94         if (n->type != XML_ELEMENT_NODE)
  95                 return;
  96         xmlNodePtr child = n->children;
  97         if (!child)
  98                 return;
  99
 100         if (!strcmp((char*)n->name, "title")) {
 101                 res->title = (char*)child->content;
 102         } else if (!strcmp((char*)n->name, "link")) {
 103                 res->link = (char*)child->content;
 104         } else if (!strcmp((char*)n->name, "description")) {
 105                 res->description = (char*)child->content;
 106         } else if (!strcmp((char*)n->name, "language")) {
 107                 res->language = (char*)child->content;
 108         } else if (!strcmp((char*)n->name, "pubDate")) {
 109                 res->pubdate = (char*)child->content;
 110         } else if (!strcmp((char*)n->name, "lastBuildDate")) {
 111                 res->lastbuilddate = (char*)child->content;
 112         } else if (!strcmp((char*)n->name, "generator")) {
 113                 res->generator = (char*)child->content;
 114         } else if (!strcmp((char*)n->name, "docs")) {
 115                 res->docs = (char*)child->content;
 116         } else if (!strcmp((char*)n->name, "managingEditor")) {
 117                 res->managingeditor = (char*)child->content;
 118         } else if (!strcmp((char*)n->name, "webMaster")) {
 119                 res->webmaster = (char*)child->content;
 120         } else if (!strcmp((char*)n->name, "item")) {
 121                 read_item(&res->items[res->item_count++], n->children);
 122         }
 123 }
 124
 125 static inline int parse_rss_2_0(PRSS* res, xmlNodePtr root)
 126 {
 127         xmlNodePtr channel = root->children;
 128         while(channel
 129                         && (channel->type!=XML_ELEMENT_NODE
 130                          || strcmp((char*)channel->name, "channel")))
 131                 channel = channel->next;
 132         if (!channel)
 133                 return 0;
 134
 135         int items = 0;
 136         xmlNodePtr n;
 137         for(n = channel->children; n; n = n->next)
 138                 if (n->type==XML_ELEMENT_NODE && !strcmp((char*)n->name, "item"))
 139                         ++items;
 140
 141         res->items = malloc(items*sizeof(PRSS_Item));
 142         res->item_count = 0;
 143
 144         for(n = channel->children; n; n = n->next) {
 145                 read_element(res, n);
 146         }
 147
 148         return 1;
 149 }
 150 static inline int parse_rss_1_0(PRSS* res, xmlNodePtr root)
 151 {
 152         int items = 0;
 153         xmlNodePtr n;
 154         for(n = root->children; n; n = n->next) {
 155                 if (n->type==XML_ELEMENT_NODE) {
 156                         if (!strcmp((char*)n->name, "item"))
 157                                 ++items;
 158                         else if (!strcmp((char*)n->name, "channel")) {
 159                                 xmlNodePtr i;
 160                                 for(i = n->children; i; i = i->next) {
 161                                         read_element(res, i);
 162                                 }
 163                         }
 164                 }
 165         }
 166
 167         res->items = malloc(items*sizeof(PRSS_Item));
 168         res->item_count = 0;
 169
 170         for(n = root->children; n; n = n->next) {
 171                 if (n->type==XML_ELEMENT_NODE && !strcmp((char*)n->name, "item"))
 172                         read_item(&res->items[res->item_count++], n->children);
 173         }
 174
 175         return 1;
 176 }
 177 static inline int parse_rss_0_9x(PRSS* res, xmlNodePtr root)
 178 {
 179         // almost same...
 180         return parse_rss_2_0(res, root);
 181 }
 182
 183 PRSS* prss_parse_doc(xmlDocPtr doc)
 184 {
 185         // FIXME: doc shouldn't be freed after failure when called explicitly from program!
 186
 187         xmlNodePtr root = xmlDocGetRootElement(doc);
 188         PRSS* result = malloc(sizeof(PRSS));
 189         prss_null(result);
 190         result->_data = doc;
 191         do {
 192                 if (root->type == XML_ELEMENT_NODE) {
 193                         if (!strcmp((char*)root->name, "RDF")) {
 194                                 // RSS 1.0 document
 195                                 if (!parse_rss_1_0(result, root)) {
 196                                         free(result);
 197                                         xmlFreeDoc(doc);
 198                                         return NULL;
 199                                 }
 200                                 return result;
 201                         } else if (!strcmp((char*)root->name, "rss")) {
 202                                 // RSS 2.0 or <1.0 document
 203                                 if (!parse_rss_2_0(result, root)) {
 204                                         free(result);
 205                                         xmlFreeDoc(doc);
 206                                         return NULL;
 207                                 }
 208                                 return result;
 209                         }
 210                 }
 211                 root = root->next;
 212         } while(root);
 213         free(result);
 214         return NULL;
 215 }