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                 /* Make sure we are in one of known locations within the XML structure.
  55                  * This condition should never be true on a valid Atom feed. */
  56                 if (ctx->location != FEED_LOC_ATOM10_AUTHOR &&
  57                                 ctx->location != FEED_LOC_ATOM10_ENTRY) {
  58                         ctx->depth++;
  59                         return;
  60                 }
  61
  62                 if( !strcmp(el, "author") ) {
  63                         /* Start of author info for current feed item.
  64                          * Set correct location. */
  65                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  66                 } else if( !strcmp(el, "link") ) {
  67                         /* Capture item URL, from the "url" XML attribute. */
  68                         if (ctx->curitem && ctx->location == FEED_LOC_ATOM10_ENTRY)
  69                   ctx->curitem->url = g_strdup(feed_parser_get_attribute_value(attr, "href"));
  70                 } else if( !strcmp(el, "source") ) {
  71                         ctx->location = FEED_LOC_ATOM10_SOURCE;
  72                 } else ctx->location = FEED_LOC_ATOM10_ENTRY;
  73
  74                 if( !strcmp(el, "title") ) {
  75                         a = feed_parser_get_attribute_value(attr, "type");
  76                         if( !a || !strcmp(a, "text") )
  77                                 ctx->curitem->title_format = FEED_ITEM_TITLE_TEXT;
  78                         else if( !strcmp(a, "html") )
  79                                 ctx->curitem->title_format = FEED_ITEM_TITLE_HTML;
  80                         else if( !strcmp(a, "xhtml") )
  81                                 ctx->curitem->title_format = FEED_ITEM_TITLE_XHTML;
  82                         else
  83                                 ctx->curitem->title_format = FEED_ITEM_TITLE_UNKNOWN;
  84                 } else if (!strcmp(el, "content") ) {
  85                         a = feed_parser_get_attribute_value(attr, "type");
  86                         if (a && !strcmp(a, "xhtml")) {
  87                                 ctx->curitem->xhtml_content = TRUE;
  88                                 ctx->location = FEED_LOC_ATOM10_CONTENT;
  89                         }
  90                 }
  91         }
  92
  93         ctx->depth++;
  94 }
  95
  96 void feed_parser_atom10_end(void *data, const gchar *el)
  97 {
  98         FeedParserCtx *ctx = (FeedParserCtx *)data;
  99         Feed *feed = ctx->feed;
 100         gchar *text = NULL;
 101
 102         if( ctx->str != NULL )
 103                 text = ctx->str->str;
 104         else
 105                 text = "";
 106
 107         switch( ctx->depth ) {
 108
 109                 case 0:
 110                         /* Just in case. */
 111                         break;
 112
 113                 case 1:
 114
 115                         if( !strcmp(el, "feed") ) {
 116                                 /* We have finished parsing the feed, reverse the list
 117                                  * so it's not upside down. */
 118                                 feed->items = g_slist_reverse(ctx->feed->items);
 119                         }
 120
 121                         break;
 122
 123                 case 2:
 124
 125                         /* decide if we just received </entry>, so we can
 126                          * add a complete item to feed */
 127                         if( !strcmp(el, "entry") ) {
 128
 129                                 /* append the complete feed item */
 130                                 if( ctx->curitem->id && ctx->curitem->title
 131                                                 && ctx->curitem->date_modified ) {
 132                                         feed->items =
 133                                                 g_slist_prepend(feed->items, (gpointer)ctx->curitem);
 134                                 }
 135
 136                                 /* since it's in the linked list, lose this pointer */
 137                                 ctx->curitem = NULL;
 138
 139                         } else if( !strcmp(el, "title") ) {     /* so it wasn't end of item */
 140                                 FILL(feed->title)
 141                         } else if( !strcmp(el, "summary" ) ) {
 142                                 FILL(feed->description)
 143                         } else if( !strcmp(el, "updated" ) ) {
 144                                 feed->date = parseISO8601Date(text);
 145                         }
 146                         /* FIXME: add more later */
 147
 148                         break;
 149
 150                 case 3:
 151
 152                         if( ctx->curitem == NULL )
 153                                 break;
 154
 155                         switch(ctx->location) {
 156
 157                                 /* We're in feed/entry */
 158                                 case FEED_LOC_ATOM10_ENTRY:
 159                                         if( !strcmp(el, "title") ) {
 160                                                 FILL(ctx->curitem->title)
 161                                         } else if( !strcmp(el, "summary") ) {
 162                                                 FILL(ctx->curitem->summary)
 163                                         } else if( !strcmp(el, "content") ) {
 164                                                 if (!ctx->curitem->xhtml_content)
 165                                                         FILL(ctx->curitem->text)
 166                                         } else if( !strcmp(el, "id") ) {
 167                                                 FILL(ctx->curitem->id)
 168                                                 feed_item_set_id_permalink(ctx->curitem, TRUE);
 169                                         } else if( !strcmp(el, "published") ) {
 170                                                 ctx->curitem->date_published = parseISO8601Date(text);
 171                                         } else if( !strcmp(el, "updated") ) {
 172                                                 ctx->curitem->date_modified = parseISO8601Date(text);
 173                                         }
 174
 175                                         break;
 176
 177                                 /* We're in feed/author or about to leave feed/entry/author */
 178                                 case FEED_LOC_ATOM10_AUTHOR:
 179                                         if( !strcmp(el, "author" ) ) {
 180                                                 /* We just finished parsing <author> */
 181                                                 ctx->curitem->author = g_strdup_printf("%s%s%s%s%s",
 182                                                                 ctx->name ? ctx->name : "",
 183                                                                 ctx->name && ctx->mail ? " <" : ctx->mail ? "<" : "",
 184                                                                 ctx->mail ? ctx->mail : "",
 185                                                                 ctx->mail ? ">" : "",
 186                                                                 !ctx->name && !ctx->mail ? "N/A" : "");
 187                                                 ctx->location = FEED_LOC_ATOM10_ENTRY;
 188                                         } else if( !strcmp(el, "name") ) {
 189                                                 FILL(feed->author)
 190                                         }
 191
 192                                         break;
 193                         }
 194
 195                         break;
 196
 197                 case 4:
 198
 199                         if( ctx->curitem == NULL )
 200                                 break;
 201
 202                         switch(ctx->location) {
 203
 204                                 /* We're in feed/entry/author */
 205                                 case FEED_LOC_ATOM10_AUTHOR:
 206                                         if( !strcmp(el, "name") ) {
 207                                                 FILL(ctx->name)
 208                                         } else if( !strcmp(el, "email") ) {
 209                                                 FILL(ctx->mail)
 210                                         }
 211
 212                                         break;
 213
 214                                 /* We're in feed/entry/source */
 215                                 case FEED_LOC_ATOM10_SOURCE:
 216                                         if( !strcmp(el, "title" ) ) {
 217                                                 FILL(ctx->curitem->sourcetitle)
 218                                         } else if( !strcmp(el, "id" ) ) {
 219                                                 FILL(ctx->curitem->sourceid)
 220                                         } else if( !strcmp(el, "updated" ) ) {
 221                                                 ctx->curitem->sourcedate = parseISO8601Date(text);
 222                                         }
 223
 224                                         break;
 225
 226                                 case FEED_LOC_ATOM10_CONTENT:
 227                                         if (!strcmp(el, "div") && ctx->curitem->xhtml_content)
 228                                                 FILL(ctx->curitem->text)
 229                                         break;
 230
 231                                 }
 232
 233
 234                         break;
 235         }
 236
 237         if( ctx->str != NULL ) {
 238                 g_string_free(ctx->str, TRUE);
 239                 ctx->str = NULL;
 240         }
 241         ctx->str = NULL;
 242
 243         ctx->depth--;
 244 }