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 enum {
  33         FEED_LOC_ATOM10_NONE,
  34         FEED_LOC_ATOM10_ENTRY,
  35         FEED_LOC_ATOM10_AUTHOR,
  36         FEED_LOC_ATOM10_SOURCE,
  37         FEED_LOC_ATOM10_CONTENT
  38 } FeedAtom10Locations;
  39
  40 void feed_parser_atom10_start(void *data, const gchar *el, const gchar **attr)
  41 {
  42         FeedParserCtx *ctx = (FeedParserCtx *)data;
  43         gchar *a = NULL;
  44
  45         if( ctx->depth == 1 ) {
  46
  47                 if( !strcmp(el, "entry") ) {
  48                         /* Start of new feed item found.
  49                          * Create a new FeedItem, freeing the one we already have, if any. */
  50                         if( ctx->curitem != NULL )
  51                                 feed_item_free(ctx->curitem);
  52                         ctx->curitem = feed_item_new(ctx->feed);
  53                         ctx->location = FEED_LOC_ATOM10_ENTRY;
  54                 } else if( !strcmp(el, "author") ) {
  55                         /* Start of author info for the feed found.
  56                          * Set correct location. */
  57                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  58                 } else if( !strcmp(el, "link") ) {
  59                         if (!feed_parser_get_attribute_value(attr, "rel")) {
  60                                 /* Link tag for the feed */
  61                                 g_free(ctx->feed->link);
  62                                 ctx->feed->link =
  63                                         g_strdup(feed_parser_get_attribute_value(attr, "href"));
  64                         }
  65                 } else ctx->location = FEED_LOC_ATOM10_NONE;
  66
  67         } else if( ctx->depth == 2 ) {
  68
  69                 /* Make sure we are in one of known locations within the XML structure.
  70                  * This condition should never be true on a valid Atom feed. */
  71                 if (ctx->location != FEED_LOC_ATOM10_AUTHOR &&
  72                                 ctx->location != FEED_LOC_ATOM10_ENTRY) {
  73                         ctx->depth++;
  74                         return;
  75                 }
  76
  77                 if( !strcmp(el, "author") ) {
  78                         /* Start of author info for current feed item.
  79                          * Set correct location. */
  80                         ctx->location = FEED_LOC_ATOM10_AUTHOR;
  81                 } else if( !strcmp(el, "link") ) {
  82                         /* Capture item URL, from the "url" XML attribute. */
  83                         if (ctx->curitem && ctx->location == FEED_LOC_ATOM10_ENTRY)
  84                   ctx->curitem->url = g_strdup(feed_parser_get_attribute_value(attr, "href"));
  85                 } else if( !strcmp(el, "source") ) {
  86                         ctx->location = FEED_LOC_ATOM10_SOURCE;
  87                 } else ctx->location = FEED_LOC_ATOM10_ENTRY;
  88
  89                 if( !strcmp(el, "title") ) {
  90                         a = feed_parser_get_attribute_value(attr, "type");
  91                         if( !a || !strcmp(a, "text") )
  92                                 ctx->curitem->title_format = FEED_ITEM_TITLE_TEXT;
  93                         else if( !strcmp(a, "html") )
  94                                 ctx->curitem->title_format = FEED_ITEM_TITLE_HTML;
  95                         else if( !strcmp(a, "xhtml") )
  96                                 ctx->curitem->title_format = FEED_ITEM_TITLE_XHTML;
  97                         else
  98                                 ctx->curitem->title_format = FEED_ITEM_TITLE_UNKNOWN;
  99                 } else if (!strcmp(el, "content") ) {
 100                         a = feed_parser_get_attribute_value(attr, "type");
 101                         if (a && !strcmp(a, "xhtml")) {
 102                                 ctx->curitem->xhtml_content = TRUE;
 103                                 ctx->location = FEED_LOC_ATOM10_CONTENT;
 104                         }
 105                 }
 106         }
 107
 108         ctx->depth++;
 109 }
 110
 111 void feed_parser_atom10_end(void *data, const gchar *el)
 112 {
 113         FeedParserCtx *ctx = (FeedParserCtx *)data;
 114         Feed *feed = ctx->feed;
 115         gchar *text = NULL, *tmp;
 116
 117         if( ctx->str != NULL )
 118                 text = g_strstrip(g_strdup(ctx->str->str));
 119         else
 120                 text = "";
 121
 122         switch( ctx->depth ) {
 123
 124                 case 0:
 125                         /* Just in case. */
 126                         break;
 127
 128                 case 1:
 129
 130                         if( !strcmp(el, "feed") ) {
 131                                 /* We have finished parsing the feed, reverse the list
 132                                  * so it's not upside down. */
 133                                 feed->items = g_slist_reverse(ctx->feed->items);
 134                         }
 135
 136                         break;
 137
 138                 case 2:
 139
 140                         /* decide if we just received </entry>, so we can
 141                          * add a complete item to feed */
 142                         if( !strcmp(el, "entry") ) {
 143
 144                                 /* Fix up URL, if it is relative */
 145                                 if (ctx->curitem->url != NULL &&
 146                                                 !strstr(ctx->curitem->url, "://") &&
 147                                                 ctx->feed->link != NULL) {
 148                                         tmp = g_strconcat(ctx->feed->link,
 149                                                         (ctx->curitem->url[0] == '/' ? "" : "/"),
 150                                                         ctx->curitem->url, NULL);
 151                                         feed_item_set_url(ctx->curitem, tmp);
 152                                         g_free(tmp);
 153                                 }
 154
 155                                 /* append the complete feed item */
 156                                 if( ctx->curitem->id && ctx->curitem->title
 157                                                 && ctx->curitem->date_modified ) {
 158                                         feed->items =
 159                                                 g_slist_prepend(feed->items, (gpointer)ctx->curitem);
 160                                 }
 161
 162                                 /* since it's in the linked list, lose this pointer */
 163                                 ctx->curitem = NULL;
 164
 165                         } else if( !strcmp(el, "title") ) {     /* so it wasn't end of item */
 166                                 FILL(feed->title)
 167                         } else if( !strcmp(el, "summary" ) ) {
 168                                 FILL(feed->description)
 169                         } else if( !strcmp(el, "updated" ) ) {
 170                                 feed->date = parseISO8601Date(text);
 171                         }
 172                         /* FIXME: add more later */
 173
 174                         break;
 175
 176                 case 3:
 177
 178                         if( ctx->curitem == NULL )
 179                                 break;
 180
 181                         switch(ctx->location) {
 182
 183                                 /* We're in feed/entry */
 184                                 case FEED_LOC_ATOM10_ENTRY:
 185                                         if( !strcmp(el, "title") ) {
 186                                                 FILL(ctx->curitem->title)
 187                                         } else if( !strcmp(el, "summary") ) {
 188                                                 FILL(ctx->curitem->summary)
 189                                         } else if( !strcmp(el, "content") ) {
 190                                                 if (!ctx->curitem->xhtml_content)
 191                                                         FILL(ctx->curitem->text)
 192                                         } else if( !strcmp(el, "id") ) {
 193                                                 FILL(ctx->curitem->id)
 194                                                 feed_item_set_id_permalink(ctx->curitem, TRUE);
 195                                         } else if( !strcmp(el, "published") ) {
 196                                                 ctx->curitem->date_published = parseISO8601Date(text);
 197                                         } else if( !strcmp(el, "updated") ) {
 198                                                 ctx->curitem->date_modified = parseISO8601Date(text);
 199                                         }
 200
 201                                         break;
 202
 203                                 /* We're in feed/author or about to leave feed/entry/author */
 204                                 case FEED_LOC_ATOM10_AUTHOR:
 205                                         if( !strcmp(el, "author" ) ) {
 206                                                 /* We just finished parsing <author> */
 207                                                 ctx->curitem->author = g_strdup_printf("%s%s%s%s%s",
 208                                                                 ctx->name ? ctx->name : "",
 209                                                                 ctx->name && ctx->mail ? " <" : ctx->mail ? "<" : "",
 210                                                                 ctx->mail ? ctx->mail : "",
 211                                                                 ctx->mail ? ">" : "",
 212                                                                 !ctx->name && !ctx->mail ? "N/A" : "");
 213                                                 ctx->location = FEED_LOC_ATOM10_ENTRY;
 214                                         } else if( !strcmp(el, "name") ) {
 215                                                 FILL(feed->author)
 216                                         }
 217
 218                                         break;
 219                         }
 220
 221                         break;
 222
 223                 case 4:
 224
 225                         if( ctx->curitem == NULL )
 226                                 break;
 227
 228                         switch(ctx->location) {
 229
 230                                 /* We're in feed/entry/author */
 231                                 case FEED_LOC_ATOM10_AUTHOR:
 232                                         if( !strcmp(el, "name") ) {
 233                                                 FILL(ctx->name)
 234                                         } else if( !strcmp(el, "email") ) {
 235                                                 FILL(ctx->mail)
 236                                         }
 237
 238                                         break;
 239
 240                                 /* We're in feed/entry/source */
 241                                 case FEED_LOC_ATOM10_SOURCE:
 242                                         if( !strcmp(el, "title" ) ) {
 243                                                 FILL(ctx->curitem->sourcetitle)
 244                                         } else if( !strcmp(el, "id" ) ) {
 245                                                 FILL(ctx->curitem->sourceid)
 246                                         } else if( !strcmp(el, "updated" ) ) {
 247                                                 ctx->curitem->sourcedate = parseISO8601Date(text);
 248                                         }
 249
 250                                         break;
 251
 252                                 case FEED_LOC_ATOM10_CONTENT:
 253                                         if (!strcmp(el, "div") && ctx->curitem->xhtml_content)
 254                                                 FILL(ctx->curitem->text)
 255                                         break;
 256
 257                                 }
 258
 259
 260                         break;
 261         }
 262
 263         if( ctx->str != NULL ) {
 264                 g_free(text);
 265                 g_string_free(ctx->str, TRUE);
 266                 ctx->str = NULL;
 267         }
 268         ctx->str = NULL;
 269
 270         ctx->depth--;
 271 }