src/plugins/rssyl/libfeed/parser_atom10.c

   1 /*
   2  * Copyright (C) 2006 Andrej Kacian <andrej@kacian.sk>
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License as
   6  * published by the Free Software Foundation; either version 2 of the
   7  * License, or (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public
  15  * License along with this program; if not, write to the
  16  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  17  * Boston, MA 02111-1307, USA.
  18  */
  19 #define __USE_GNU
  20
  21 #include <glib.h>
  22 #include <expat.h>
  23 #include <string.h>
  24 #include <stdio.h>
  25
  26 #include "feed.h"
  27 #include "feeditem.h"
  28 #include "date.h"
  29 #include "parser.h"
  30 #include "parser_atom10.h"
  31
  32 void feed_parser_atom10_start(void *data, const gchar *el, const gchar **attr)
  33 {
  34         FeedParserCtx *ctx = (FeedParserCtx *)data;
  35         gchar *a = NULL;
  36
  37         if( ctx->depth == 1 ) {
  38
  39                 if( !strcmp(el, "entry") ) {
  40                         /* Start of new feed item found.
  41                          * Create a new FeedItem, freeing the one we already have, if any. */
  42                         if( ctx->curitem != NULL )
  43                                 feed_item_free(ctx->curitem);
  44                         ctx->curitem = feed_item_new(ctx->feed);
  45                         ctx->location = FEED_LOC_ATOM10_ENTRY;
  46                 } else if( !strcmp(el, "author") ) {
  47                         /* Start of author info for the feed found.
  48                          * Set correct location. */
  49                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  50                 } else ctx->location = FEED_LOC_ATOM10_NONE;
  51
  52         } else if( ctx->depth == 2 ) {
  53
  54                 if( !strcmp(el, "author") ) {
  55                         /* Start of author info for current feed item.
  56                          * Set correct location. */
  57                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  58                 } else if( !strcmp(el, "link") ) {
  59                         /* Capture item URL, from the "url" XML attribute. */
  60                         if (ctx->curitem && ctx->location == FEED_LOC_ATOM10_ENTRY)
  61                   ctx->curitem->url = g_strdup(feed_parser_get_attribute_value(attr, "href"));
  62                 } else if( !strcmp(el, "source") ) {
  63                         ctx->location = FEED_LOC_ATOM10_SOURCE;
  64                 } else ctx->location = FEED_LOC_ATOM10_ENTRY;
  65
  66                 if( !strcmp(el, "title") ) {
  67                         a = feed_parser_get_attribute_value(attr, "type");
  68                         if( !a || !strcmp(a, "text") )
  69                                 ctx->curitem->title_format = FEED_ITEM_TITLE_TEXT;
  70                         else if( !strcmp(a, "html") )
  71                                 ctx->curitem->title_format = FEED_ITEM_TITLE_HTML;
  72                         else if( !strcmp(a, "xhtml") )
  73                                 ctx->curitem->title_format = FEED_ITEM_TITLE_XHTML;
  74                         else
  75                                 ctx->curitem->title_format = FEED_ITEM_TITLE_UNKNOWN;
  76                 } else if (!strcmp(el, "content") ) {
  77                         a = feed_parser_get_attribute_value(attr, "type");
  78                         if (a && !strcmp(a, "xhtml")) {
  79                                 ctx->curitem->xhtml_content = TRUE;
  80                                 ctx->location = FEED_LOC_ATOM10_CONTENT;
  81                         }
  82                 }
  83         }
  84
  85         ctx->depth++;
  86 }
  87
  88 void feed_parser_atom10_end(void *data, const gchar *el)
  89 {
  90         FeedParserCtx *ctx = (FeedParserCtx *)data;
  91         Feed *feed = ctx->feed;
  92         gchar *text = NULL;
  93
  94         if( ctx->str != NULL )
  95                 text = ctx->str->str;
  96
  97         ctx->depth--;
  98
  99         switch( ctx->depth ) {
 100
 101                 case 0:
 102
 103                         if( !strcmp(el, "feed") ) {
 104                                 /* We have finished parsing the feed, reverse the list
 105                                  * so it's not upside down. */
 106                                 feed->items = g_slist_reverse(ctx->feed->items);
 107                         }
 108
 109                         break;
 110
 111                 case 1:
 112
 113                         /* decide if we just received </entry>, so we can
 114                          * add a complete item to feed */
 115                         if( !strcmp(el, "entry") ) {
 116
 117                                 /* append the complete feed item */
 118                                 if( ctx->curitem->id && ctx->curitem->title
 119                                                 && ctx->curitem->date_modified ) {
 120                                         feed->items =
 121                                                 g_slist_prepend(feed->items, (gpointer)ctx->curitem);
 122                                 }
 123
 124                                 /* since it's in the linked list, lose this pointer */
 125                                 ctx->curitem = NULL;
 126
 127                         } else if( !strcmp(el, "title") ) {     /* so it wasn't end of item */
 128                                 FILL(feed->title)
 129                         } else if( !strcmp(el, "summary" ) ) {
 130                                 FILL(feed->description)
 131                         } else if( !strcmp(el, "updated" ) ) {
 132                                 feed->date = parseISO8601Date(text);
 133                         }
 134                         /* FIXME: add more later */
 135
 136                         break;
 137
 138                 case 2:
 139
 140                         if( ctx->curitem == NULL )
 141                                 break;
 142
 143                         switch(ctx->location) {
 144
 145                                 /* We're in feed/entry */
 146                                 case FEED_LOC_ATOM10_ENTRY:
 147                                         if( !strcmp(el, "title") ) {
 148                                                 FILL(ctx->curitem->title)
 149                                         } else if( !strcmp(el, "summary") ) {
 150                                                 FILL(ctx->curitem->summary)
 151                                         } else if( !strcmp(el, "content") ) {
 152                                                 if (!ctx->curitem->xhtml_content)
 153                                                         FILL(ctx->curitem->text);
 154                                         } else if( !strcmp(el, "id") ) {
 155                                                 FILL(ctx->curitem->id);
 156                                                 feed_item_set_id_permalink(ctx->curitem, TRUE);
 157                                         } else if( !strcmp(el, "published") ) {
 158                                                 ctx->curitem->date_published = parseISO8601Date(text);
 159                                         } else if( !strcmp(el, "updated") ) {
 160                                                 ctx->curitem->date_modified = parseISO8601Date(text);
 161                                         }
 162
 163                                         break;
 164
 165                                 /* We're in feed/author or about to leave feed/entry/author */
 166                                 case FEED_LOC_ATOM10_AUTHOR:
 167                                         if( !strcmp(el, "author" ) ) {
 168                                                 /* We just finished parsing <author> */
 169                                                 ctx->curitem->author = g_strdup_printf("%s%s%s%s%s",
 170                                                                 ctx->name ? ctx->name : "",
 171                                                                 ctx->name && ctx->mail ? " <" : ctx->mail ? "<" : "",
 172                                                                 ctx->mail ? ctx->mail : "",
 173                                                                 ctx->mail ? ">" : "",
 174                                                                 !ctx->name && !ctx->mail ? "N/A" : "");
 175                                                 ctx->location = FEED_LOC_ATOM10_ENTRY;
 176                                         } else if( !strcmp(el, "name") ) {
 177                                                 FILL(feed->author);
 178                                         }
 179
 180                                         break;
 181                         }
 182
 183                         break;
 184
 185                 case 3:
 186
 187                         if( ctx->curitem == NULL )
 188                                 break;
 189
 190                         switch(ctx->location) {
 191
 192                                 /* We're in feed/entry/author */
 193                                 case FEED_LOC_ATOM10_AUTHOR:
 194                                         if( !strcmp(el, "name") ) {
 195                                                 FILL(ctx->name);
 196                                         } else if( !strcmp(el, "email") ) {
 197                                                 FILL(ctx->mail);
 198                                         }
 199
 200                                         break;
 201
 202                                 /* We're in feed/entry/source */
 203                                 case FEED_LOC_ATOM10_SOURCE:
 204                                         if( !strcmp(el, "title" ) ) {
 205                                                 FILL(ctx->curitem->sourcetitle);
 206                                         } else if( !strcmp(el, "id" ) ) {
 207                                                 FILL(ctx->curitem->sourceid);
 208                                         } else if( !strcmp(el, "updated" ) ) {
 209                                                 ctx->curitem->sourcedate = parseISO8601Date(text);
 210                                         }
 211
 212                                         break;
 213
 214                                 case FEED_LOC_ATOM10_CONTENT:
 215                                         if (!strcmp(el, "div") && ctx->curitem->xhtml_content)
 216                                                 FILL(ctx->curitem->text);
 217                                         break;
 218
 219                                 }
 220
 221
 222                         break;
 223         }
 224
 225         if( ctx->str != NULL ) {
 226                 g_string_free(ctx->str, TRUE);
 227                 ctx->str = NULL;
 228         }
 229         ctx->str = NULL;
 230 }