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 if( !strcmp(el, "link") ) {
  51                         /* Link tag for the feed */
  52                         g_free(ctx->feed->link);
  53                         ctx->feed->link = g_strdup(feed_parser_get_attribute_value(attr, "href"));
  54                 } else ctx->location = FEED_LOC_ATOM10_NONE;
  55
  56         } else if( ctx->depth == 2 ) {
  57
  58                 /* Make sure we are in one of known locations within the XML structure.
  59                  * This condition should never be true on a valid Atom feed. */
  60                 if (ctx->location != FEED_LOC_ATOM10_AUTHOR &&
  61                                 ctx->location != FEED_LOC_ATOM10_ENTRY) {
  62                         ctx->depth++;
  63                         return;
  64                 }
  65
  66                 if( !strcmp(el, "author") ) {
  67                         /* Start of author info for current feed item.
  68                          * Set correct location. */
  69                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  70                 } else if( !strcmp(el, "link") ) {
  71                         /* Capture item URL, from the "url" XML attribute. */
  72                         if (ctx->curitem && ctx->location == FEED_LOC_ATOM10_ENTRY)
  73                   ctx->curitem->url = g_strdup(feed_parser_get_attribute_value(attr, "href"));
  74                 } else if( !strcmp(el, "source") ) {
  75                         ctx->location = FEED_LOC_ATOM10_SOURCE;
  76                 } else ctx->location = FEED_LOC_ATOM10_ENTRY;
  77
  78                 if( !strcmp(el, "title") ) {
  79                         a = feed_parser_get_attribute_value(attr, "type");
  80                         if( !a || !strcmp(a, "text") )
  81                                 ctx->curitem->title_format = FEED_ITEM_TITLE_TEXT;
  82                         else if( !strcmp(a, "html") )
  83                                 ctx->curitem->title_format = FEED_ITEM_TITLE_HTML;
  84                         else if( !strcmp(a, "xhtml") )
  85                                 ctx->curitem->title_format = FEED_ITEM_TITLE_XHTML;
  86                         else
  87                                 ctx->curitem->title_format = FEED_ITEM_TITLE_UNKNOWN;
  88                 } else if (!strcmp(el, "content") ) {
  89                         a = feed_parser_get_attribute_value(attr, "type");
  90                         if (a && !strcmp(a, "xhtml")) {
  91                                 ctx->curitem->xhtml_content = TRUE;
  92                                 ctx->location = FEED_LOC_ATOM10_CONTENT;
  93                         }
  94                 }
  95         }
  96
  97         ctx->depth++;
  98 }
  99
 100 void feed_parser_atom10_end(void *data, const gchar *el)
 101 {
 102         FeedParserCtx *ctx = (FeedParserCtx *)data;
 103         Feed *feed = ctx->feed;
 104         gchar *text = NULL, *tmp;
 105
 106         if( ctx->str != NULL )
 107                 text = ctx->str->str;
 108         else
 109                 text = "";
 110
 111         switch( ctx->depth ) {
 112
 113                 case 0:
 114                         /* Just in case. */
 115                         break;
 116
 117                 case 1:
 118
 119                         if( !strcmp(el, "feed") ) {
 120                                 /* We have finished parsing the feed, reverse the list
 121                                  * so it's not upside down. */
 122                                 feed->items = g_slist_reverse(ctx->feed->items);
 123                         }
 124
 125                         break;
 126
 127                 case 2:
 128
 129                         /* decide if we just received </entry>, so we can
 130                          * add a complete item to feed */
 131                         if( !strcmp(el, "entry") ) {
 132
 133                                 /* Fix up URL, if it is relative */
 134                                 if (ctx->curitem->url != NULL &&
 135                                                 !strstr("://", ctx->curitem->url) &&
 136                                                 ctx->feed->link != NULL) {
 137                                         tmp = g_strconcat(ctx->feed->link,
 138                                                         (ctx->curitem->url[0] == '/' ? "" : "/"),
 139                                                         ctx->curitem->url, NULL);
 140                                         feed_item_set_url(ctx->curitem, tmp);
 141                                         g_free(tmp);
 142                                 }
 143
 144                                 /* append the complete feed item */
 145                                 if( ctx->curitem->id && ctx->curitem->title
 146                                                 && ctx->curitem->date_modified ) {
 147                                         feed->items =
 148                                                 g_slist_prepend(feed->items, (gpointer)ctx->curitem);
 149                                 }
 150
 151                                 /* since it's in the linked list, lose this pointer */
 152                                 ctx->curitem = NULL;
 153
 154                         } else if( !strcmp(el, "title") ) {     /* so it wasn't end of item */
 155                                 FILL(feed->title)
 156                         } else if( !strcmp(el, "summary" ) ) {
 157                                 FILL(feed->description)
 158                         } else if( !strcmp(el, "updated" ) ) {
 159                                 feed->date = parseISO8601Date(text);
 160                         }
 161                         /* FIXME: add more later */
 162
 163                         break;
 164
 165                 case 3:
 166
 167                         if( ctx->curitem == NULL )
 168                                 break;
 169
 170                         switch(ctx->location) {
 171
 172                                 /* We're in feed/entry */
 173                                 case FEED_LOC_ATOM10_ENTRY:
 174                                         if( !strcmp(el, "title") ) {
 175                                                 FILL(ctx->curitem->title)
 176                                         } else if( !strcmp(el, "summary") ) {
 177                                                 FILL(ctx->curitem->summary)
 178                                         } else if( !strcmp(el, "content") ) {
 179                                                 if (!ctx->curitem->xhtml_content)
 180                                                         FILL(ctx->curitem->text)
 181                                         } else if( !strcmp(el, "id") ) {
 182                                                 FILL(ctx->curitem->id)
 183                                                 feed_item_set_id_permalink(ctx->curitem, TRUE);
 184                                         } else if( !strcmp(el, "published") ) {
 185                                                 ctx->curitem->date_published = parseISO8601Date(text);
 186                                         } else if( !strcmp(el, "updated") ) {
 187                                                 ctx->curitem->date_modified = parseISO8601Date(text);
 188                                         }
 189
 190                                         break;
 191
 192                                 /* We're in feed/author or about to leave feed/entry/author */
 193                                 case FEED_LOC_ATOM10_AUTHOR:
 194                                         if( !strcmp(el, "author" ) ) {
 195                                                 /* We just finished parsing <author> */
 196                                                 ctx->curitem->author = g_strdup_printf("%s%s%s%s%s",
 197                                                                 ctx->name ? ctx->name : "",
 198                                                                 ctx->name && ctx->mail ? " <" : ctx->mail ? "<" : "",
 199                                                                 ctx->mail ? ctx->mail : "",
 200                                                                 ctx->mail ? ">" : "",
 201                                                                 !ctx->name && !ctx->mail ? "N/A" : "");
 202                                                 ctx->location = FEED_LOC_ATOM10_ENTRY;
 203                                         } else if( !strcmp(el, "name") ) {
 204                                                 FILL(feed->author)
 205                                         }
 206
 207                                         break;
 208                         }
 209
 210                         break;
 211
 212                 case 4:
 213
 214                         if( ctx->curitem == NULL )
 215                                 break;
 216
 217                         switch(ctx->location) {
 218
 219                                 /* We're in feed/entry/author */
 220                                 case FEED_LOC_ATOM10_AUTHOR:
 221                                         if( !strcmp(el, "name") ) {
 222                                                 FILL(ctx->name)
 223                                         } else if( !strcmp(el, "email") ) {
 224                                                 FILL(ctx->mail)
 225                                         }
 226
 227                                         break;
 228
 229                                 /* We're in feed/entry/source */
 230                                 case FEED_LOC_ATOM10_SOURCE:
 231                                         if( !strcmp(el, "title" ) ) {
 232                                                 FILL(ctx->curitem->sourcetitle)
 233                                         } else if( !strcmp(el, "id" ) ) {
 234                                                 FILL(ctx->curitem->sourceid)
 235                                         } else if( !strcmp(el, "updated" ) ) {
 236                                                 ctx->curitem->sourcedate = parseISO8601Date(text);
 237                                         }
 238
 239                                         break;
 240
 241                                 case FEED_LOC_ATOM10_CONTENT:
 242                                         if (!strcmp(el, "div") && ctx->curitem->xhtml_content)
 243                                                 FILL(ctx->curitem->text)
 244                                         break;
 245
 246                                 }
 247
 248
 249                         break;
 250         }
 251
 252         if( ctx->str != NULL ) {
 253                 g_string_free(ctx->str, TRUE);
 254                 ctx->str = NULL;
 255         }
 256         ctx->str = NULL;
 257
 258         ctx->depth--;
 259 }