RSSyl: handle expiring comments after their parent is gone
[claws.git] / src / plugins / rssyl / rssyl_parse_feed.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
20 #ifdef HAVE_CONFIG_H
21 #  include "config.h"
22 #endif
23
24 /* Global includes */
25 #include <glib.h>
26 #include <glib/gi18n.h>
27
28 /* Claws Mail includes */
29 #include <log.h>
30 #include <codeconv.h>
31 #include <main.h>
32 #include <procmsg.h>
33
34 /* Local includes */
35 #include "libfeed/feed.h"
36 #include "libfeed/feeditem.h"
37 #include "libfeed/date.h"
38 #include "parse822.h"
39 #include "rssyl.h"
40 #include "rssyl_add_item.h"
41 #include "rssyl_deleted.h"
42 #include "rssyl_feed.h"
43 #include "rssyl_parse_feed.h"
44 #include "rssyl_prefs.h"
45 #include "strutils.h"
46
47 static void rssyl_foreach_parse_func(gpointer data, gpointer user_data)
48 {
49         FeedItem *feed_item = (FeedItem *)data;
50         RFolderItem *ritem = (RFolderItem *)user_data;
51
52         rssyl_add_item(ritem, feed_item);
53 }
54
55 struct _RSSylExpireItemsCtx {
56         gboolean exists;
57         FeedItem *item;
58         GSList *expired_ids;
59 };
60
61 typedef struct _RSSylExpireItemsCtx RSSylExpireItemsCtx;
62
63 static void expire_items_func(gpointer data, gpointer user_data)
64 {
65         RSSylExpireItemsCtx *ctx = (RSSylExpireItemsCtx *)user_data;
66         FeedItem *item = (FeedItem *)data;
67         gchar *id = NULL, *id2 = NULL;
68
69         if( (id = feed_item_get_id(item)) == NULL )
70                 id = feed_item_get_url(item);
71
72         if( id == NULL )
73                 return;
74
75         if( (id2 = feed_item_get_id(ctx->item)) == NULL )
76                 id2 = feed_item_get_url(ctx->item);
77
78         if( id2 == NULL )
79                 return;
80
81         /* Simply check ID, as we should have up-to-date items right now. */
82         if( !strcmp(id, id2) )
83                 ctx->exists = TRUE;
84 }
85
86 static void rssyl_expire_items(RFolderItem *ritem, Feed *feed)
87 {
88         FeedItem *item = NULL;
89         GSList *i = NULL;
90         RSSylExpireItemsCtx *ctx = NULL;
91         RFeedCtx *fctx;
92
93         debug_print("RSSyl: rssyl_expire_items()\n");
94
95         g_return_if_fail(ritem != NULL);
96         g_return_if_fail(ritem->items != NULL);
97         g_return_if_fail(feed != NULL);
98
99         ctx = malloc( sizeof(RSSylExpireItemsCtx) );
100         ctx->expired_ids = NULL;
101
102         /* Check each locally stored item, if it is still in the upstream
103          * feed - xnay it if not. */
104         for( i = ritem->items; i != NULL; i = i->next ) {
105                 item = (FeedItem *)i->data;
106
107                 /* Comments will be expired later, once we know which parent items
108                  * have been expired. */
109                 if (feed_item_get_parent_id(item) != NULL)
110                         continue;
111
112                 /* Find matching item in the fresh feed. */
113                 ctx->exists = FALSE;
114                 ctx->item = item;
115                 feed_foreach_item(feed, expire_items_func, ctx);
116
117                 if( !ctx->exists ) {
118                         /* No match, add item ids to the list and get rid of it. */
119                         debug_print("RSSyl: expiring '%s'\n", feed_item_get_id(item));
120                         ctx->expired_ids = g_slist_prepend(ctx->expired_ids,
121                                         g_strdup(feed_item_get_id(item)));
122                         fctx = (RFeedCtx *)item->data;
123                         g_remove(fctx->path);
124                 }
125         }
126
127         /* Now do one more pass over folder contents, and expire comments
128          * whose parents are gone. */
129         for( i = ritem->items; i != NULL; i = i->next ) {
130                 item = (FeedItem *)i->data;
131
132                 /* Handle comments expiration. */
133                 if (feed_item_get_parent_id(item) != NULL) {
134                         /* If its parent's id is on list of expired ids, this comment
135                          * can go as well. */
136                         if (g_slist_find_custom(ctx->expired_ids,
137                                         feed_item_get_parent_id(item), (GCompareFunc)g_strcmp0)) {
138                                 debug_print("RSSyl: expiring comment '%s'\n", feed_item_get_id(item));
139                                 fctx = (RFeedCtx *)item->data;
140                                 g_remove(fctx->path);
141                         }
142                 }
143         }
144
145         debug_print("RSSyl: expired %d items\n", g_slist_length(ctx->expired_ids));
146
147         g_slist_free_full(ctx->expired_ids, g_free);
148         g_free(ctx);
149 }
150
151 /* -------------------------------------------------------------------------
152  * rssyl_parse_feed() */
153
154 gboolean rssyl_parse_feed(RFolderItem *ritem, Feed *feed)
155 {
156         gchar *tmp = NULL, *tmp2 = NULL;
157         gint i = 1;
158
159         g_return_val_if_fail(ritem != NULL, FALSE);
160         g_return_val_if_fail(feed != NULL, FALSE);
161         g_return_val_if_fail(feed->title != NULL, FALSE);
162
163         debug_print("RSSyl: parse_feed\n");
164
165         /* Set the last_update timestamp here, so it is the same for all items */
166         ritem->last_update = time(NULL);
167
168         /* If the upstream feed changed its title, change name of our folder
169          * accordingly even if user has renamed it before. This makes sure that
170          * user will be aware of the upstream title change. */
171         if( !ritem->ignore_title_rename &&
172                         (ritem->official_title == NULL ||
173                         strcmp(feed->title, ritem->official_title)) ) {
174                 g_free(ritem->official_title);
175                 ritem->official_title = g_strdup(feed->title);
176
177                 tmp = rssyl_format_string(feed->title, TRUE, TRUE);
178
179                 tmp2 = g_strdup(tmp);
180                 while (folder_item_rename(&ritem->item, tmp2) != 0 && i < 20) {
181                         g_free(tmp2);
182                         tmp2 = g_strdup_printf("%s__%d", tmp, ++i);
183                         debug_print("RSSyl: couldn't rename, trying '%s'\n", tmp2);
184                 }
185                 /* TODO: handle case when i reaches 20 */
186         
187                 g_free(tmp);
188                 g_free(tmp2);
189
190                 /* FIXME: update name in properties */
191                 /* FIXME: store feed properties */
192         }
193
194         folder_item_update_freeze();
195
196         /* Read contents of folder, so we can check for duplicates/updates */
197         rssyl_folder_read_existing(ritem);
198
199         if( claws_is_exiting() ) {
200                 debug_print("RSSyl: Claws-Mail is exiting, bailing out\n");
201                 log_print(LOG_PROTOCOL, RSSYL_LOG_ABORTED_EXITING, ritem->url);
202                 folder_item_update_thaw();
203                 return TRUE;
204         }
205
206         /* Populate the ->deleted_items list so that we can check it when
207          * adding each item. */
208         ritem->deleted_items = rssyl_deleted_update(ritem);
209
210         /* Parse each item in the feed, adding or updating existing items if
211          * necessary */
212         if( feed_n_items(feed) > 0 )
213                 feed_foreach_item(feed, rssyl_foreach_parse_func, (gpointer)ritem);
214
215         if( !ritem->keep_old && !ritem->fetching_comments ) {
216                 rssyl_folder_read_existing(ritem);
217                 rssyl_expire_items(ritem, feed);
218         }
219
220         rssyl_deleted_free(ritem->deleted_items);
221
222         folder_item_scan(&ritem->item);
223         folder_item_update_thaw();
224
225         if( !ritem->fetching_comments )
226                 log_print(LOG_PROTOCOL, RSSYL_LOG_UPDATED, ritem->url);
227
228         return TRUE;
229 }