RSSyl: Use g_mkstemp() instead of mkstemp()
[claws.git] / src / plugins / rssyl / parse822.c
1 /*
2  * Claws-Mail-- a GTK+ based, lightweight, and fast e-mail client
3  * Copyright (C) 1999-2004 Hiroyuki Yamamoto
4  * This file (C) 2005 Andrej Kacian <andrej@kacian.sk>
5  *
6  * This program is free software; you can redistribute it and/or modify
7  * it under the terms of the GNU General Public License as published by
8  * the Free Software Foundation; either version 2 of the License, or
9  * (at your option) any later version.
10  *
11  * This program is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14  * GNU General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public License
17  * along with this program; if not, write to the Free Software
18  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
19  */
20
21 #ifdef HAVE_CONFIG_H
22 #  include "config.h"
23 #endif
24
25 /* Global includes */
26 #include <sys/stat.h>
27 #include <glib.h>
28 #include <pthread.h>
29
30 /* Claws Mail includes */
31 #include <common/claws.h>
32 #include <procheader.h>
33 #include <common/utils.h>
34 #include <main.h>
35
36 /* Local includes */
37 #include "libfeed/feed.h"
38 #include "libfeed/feeditem.h"
39 #include "libfeed/date.h"
40 #include "parse822.h"
41 #include "rssyl_feed.h"
42 #include "rssyl_parse_feed.h"
43 #include "strutils.h"
44
45 /* rssyl_parse_folder_item_file()
46  *
47  * Parse a RFC822-formatted feed item given by "path", and returns a
48  * pointer to a newly-allocated FeedItem struct, which contains all required data.
49  *
50  */
51 FeedItem *rssyl_parse_folder_item_file(gchar *path)
52 {
53         gchar *contents, **lines, **line, **splid, *tmp, *tmp2;
54         GError *error = NULL;
55         FeedItem *item;
56         RFeedCtx *ctx;
57         gint i = 0;
58         gboolean parsing_headers = TRUE, past_html_tag = FALSE, past_endhtml_tag = FALSE;
59         gboolean started_author = FALSE, started_subject = FALSE;
60         gboolean started_link = FALSE, started_clink = FALSE, got_original_title = FALSE;
61
62         debug_print("RSSyl: parsing '%s'\n", path);
63
64         g_file_get_contents(path, &contents, NULL, &error);
65
66         if( error )
67                 g_warning("GError: '%s'\n", error->message);
68
69         if( contents != NULL ) {
70                 lines = strsplit_no_copy(contents, '\n');
71         } else {
72                 g_warning("Badly formatted file found, ignoring: '%s'\n", path);
73                 return NULL;
74         }
75
76         ctx = g_new0(RFeedCtx, 1);
77         ctx->path = g_strdup(path); /* store filesystem path to source file */
78         ctx->last_seen = 0;
79
80         item = feed_item_new(NULL);
81         item->data = ctx;
82
83         while( lines[i] ) {
84                 if( parsing_headers && lines[i] && !strlen(lines[i]) ) {
85                         parsing_headers = FALSE;
86                         debug_print("RSSyl: finished parsing headers\n");
87                 }
88
89                 if( parsing_headers ) {
90                         line = g_strsplit(lines[i], ": ", 2);
91                         if( line[0] && line[1] && strlen(line[0]) && lines[i][0] != ' ') {
92                                 started_author = FALSE;
93                                 started_subject = FALSE;
94                                 started_link = FALSE;
95                                 started_clink = FALSE;
96
97                                 /* Author */
98                                 if( !strcmp(line[0], "From") ) {
99                                         feed_item_set_author(item, line[1]);
100                                         debug_print("RSSyl: got author '%s'\n", feed_item_get_author(item));
101                                         started_author = TRUE;
102                                 }
103
104                                 /* Date */
105                                 if( !strcmp(line[0], "Date") ) {
106                                         feed_item_set_date_modified(item,
107                                                         procheader_date_parse(NULL, line[1], 0));
108                                         debug_print("RSSyl: got date \n" );
109                                 }
110
111                                 /* Title */
112                                 if( !strcmp(line[0], "Subject") && !got_original_title ) {
113                                         feed_item_set_title(item,line[1]);
114                                         debug_print("RSSyl: got title '%s'\n", feed_item_get_title(item));
115                                         started_subject = TRUE;
116                                 }
117
118                                 /* Original (including HTML) title - Atom feeds */
119                                 if( !strcmp(line[0], "X-RSSyl-OrigTitle") ) {
120                                         feed_item_set_title(item, line[1]);
121                                         debug_print("RSSyl: got original title '%s'\n",
122                                                         feed_item_get_title(item));
123                                         got_original_title = TRUE;
124                                 }
125
126                                 /* URL */
127                                 if( !strcmp(line[0], "X-RSSyl-URL") ) {
128                                         feed_item_set_url(item, line[1]);
129                                         debug_print("RSSyl: got link '%s'\n", feed_item_get_url(item));
130                                         started_link = TRUE;
131                                 }
132
133                                 /* Last-Seen timestamp */
134                                 if( !strcmp(line[0], "X-RSSyl-Last-Seen") ) {
135                                         ctx->last_seen = atol(line[1]);
136                                         debug_print("RSSyl: got last_seen timestamp %ld\n", ctx->last_seen);
137                                 }
138
139                                 /* ID */
140                                 if( !strcmp(line[0], "Message-ID") ) {
141                                         if (line[1][0] != '<' || line[1][strlen(line[1])-1] != '>') {
142                                                 debug_print("RSSyl: malformed Message-ID, ignoring...\n");
143                                         } else {
144                                                 /* Get the ID from within < and >. */
145                                                 tmp = line[1] + 1;
146                                                 tmp2 = g_strndup(tmp, strlen(tmp) - 1);
147                                                 feed_item_set_id(item, tmp2);
148                                                 g_free(tmp2);
149                                         }
150                                 }
151
152                                 /* Feed comments */
153                                 if( !strcmp(line[0], "X-RSSyl-Comments") ) {
154                                         feed_item_set_comments_url(item, line[1]);
155                                         debug_print("RSSyl: got clink '%s'\n", feed_item_get_comments_url(item));
156                                         started_clink = TRUE;
157                                 }
158
159                                 /* References */
160                                 if( !strcmp(line[0], "References") ) {
161                                         splid = g_strsplit_set(line[1], "<>", 3);
162                                         if( strlen(splid[1]) != 0 )
163                                                 feed_item_set_parent_id(item, line[1]);
164                                         g_strfreev(splid);
165                                 }
166
167                         } else if (lines[i][0] == ' ') {
168                                 gchar *tmp = NULL;
169                                 /* continuation line */
170                                 if (started_author) {
171                                         tmp = g_strdup_printf("%s %s", feed_item_get_author(item), lines[i]+1);
172                                         feed_item_set_author(item, tmp);
173                                         debug_print("RSSyl: updated author to '%s'\n", tmp);
174                                         g_free(tmp);
175                                 } else if (started_subject) {
176                                         tmp = g_strdup_printf("%s %s", feed_item_get_title(item), lines[i]+1);
177                                         feed_item_set_title(item, tmp);
178                                         debug_print("RSSyl: updated title to '%s'\n", tmp);
179                                         g_free(tmp);
180                                 } else if (started_link) {
181                                         tmp = g_strdup_printf("%s%s", feed_item_get_url(item), lines[i]+1);
182                                         feed_item_set_url(item, tmp);
183                                         debug_print("RSSyl: updated link to '%s'\n", tmp);
184                                         g_free(tmp);
185                                 } else if (started_clink) {
186                                         tmp = g_strdup_printf("%s%s", feed_item_get_comments_url(item), lines[i]+1);
187                                         feed_item_set_comments_url(item, tmp);
188                                         debug_print("RSSyl: updated comments_link to '%s'\n", tmp);
189                                 }
190                         }
191                         g_strfreev(line);
192                 } else {
193                         if( !strcmp(lines[i], RSSYL_TEXT_START) ) {
194                                 debug_print("RSSyl: Leading html tag found at line %d\n", i);
195                                 past_html_tag = TRUE;
196                                 i++;
197                                 continue;
198                         }
199                         while( past_html_tag && !past_endhtml_tag && lines[i] ) {
200                                 if( !strcmp(lines[i], RSSYL_TEXT_END) ) {
201                                         debug_print("RSSyl: Trailing html tag found at line %d\n", i);
202                                         past_endhtml_tag = TRUE;
203                                         i++;
204                                         continue;
205                                 }
206                                 if( feed_item_get_text(item) != NULL ) {
207                                         gint e_len, n_len;
208                                         e_len = strlen(item->text);
209                                         n_len = strlen(lines[i]);
210                                         item->text = g_realloc(item->text, e_len + n_len + 2);
211                                         *(item->text+e_len) = '\n';
212                                         strcpy(item->text+e_len+1, lines[i]);
213                                         *(item->text+e_len+n_len+1) = '\0';
214                                 } else {
215                                         item->text = g_strdup(lines[i]);
216                                 }
217                                 i++;
218                         }
219
220                         if( lines[i] == NULL )
221                                 return item;
222                 }
223
224                 i++;
225         }
226         g_free(lines);
227         g_free(contents);
228         return item;
229 }
230
231 static void rssyl_flush_folder_func(gpointer data, gpointer user_data)
232 {
233         FeedItem *item = (FeedItem *)data;
234         RFeedCtx *ctx = (RFeedCtx *)item->data;
235
236         if( ctx != NULL && ctx->path != NULL)
237                 g_free(ctx->path);
238         feed_item_free(item);
239 }
240
241 static void rssyl_folder_read_existing_real(RFolderItem *ritem)
242 {
243         gchar *path = NULL, *fname = NULL;
244         GDir *dp;
245         const gchar *d;
246         GError *error;
247         gint num;
248         FeedItem *item = NULL;
249         RFeedCtx *ctx;
250
251         g_return_if_fail(ritem != NULL);
252
253         path = folder_item_get_path(&ritem->item);
254         g_return_if_fail(path != NULL);
255
256         debug_print("RSSyl: reading existing items from '%s'\n", path);
257
258         /* Flush contents if any, so we can add new */
259         if( g_slist_length(ritem->items) > 0 ) {
260                 g_slist_foreach(ritem->items, (GFunc)rssyl_flush_folder_func, NULL);
261                 g_slist_free(ritem->items);
262         }
263         ritem->items = NULL;
264         ritem->last_update = 0;
265
266         if( (dp = g_dir_open(path, 0, &error)) == NULL ) {
267                 FILE_OP_ERROR(path, "g_dir_open");
268                 debug_print("g_dir_open on \"%s\" failed with error %d (%s)\n",
269                                 path, error->code, error->message);
270                 g_free(path);
271                 return;
272         }
273
274         while( (d = g_dir_read_name(dp)) != NULL ) {
275                 if( claws_is_exiting() ) {
276                         g_dir_close(dp);
277                         g_free(path);
278                         return;
279                 }
280
281                 if( d[0] != '.' && (num = to_number(d)) > 0 ) {
282                         fname = g_strdup_printf("%s%c%s", path, G_DIR_SEPARATOR, d);
283                         if (!g_file_test(fname, G_FILE_TEST_IS_REGULAR)) {
284                                 debug_print("RSSyl: not a regular file: '%s', ignoring it\n", fname);
285                                 g_free(fname);
286                                 continue;
287                         }
288
289                         debug_print("RSSyl: starting to parse '%s'\n", d);
290                         if( (item = rssyl_parse_folder_item_file(fname)) != NULL ) {
291                                 /* Find latest timestamp */
292                                 ctx = (RFeedCtx *)item->data;
293                                 if( ritem->last_update < ctx->last_seen )
294                                         ritem->last_update = ctx->last_seen;
295                                 debug_print("RSSyl: Appending '%s'\n", feed_item_get_title(item));
296                                 ritem->items = g_slist_prepend(ritem->items, item);
297                         }
298                         g_free(fname);
299                 }
300         }
301
302         g_dir_close(dp);
303         g_free(path);
304
305         ritem->items = g_slist_reverse(ritem->items);
306 }
307
308 #ifdef USE_PTHREAD
309 static void *rssyl_read_existing_thr(void *arg)
310 {
311         RParseCtx *ctx = (RParseCtx *)arg;
312
313         rssyl_folder_read_existing_real(ctx->ritem);
314         ctx->ready = TRUE;
315         return NULL;
316 }
317 #endif
318
319 void rssyl_folder_read_existing(RFolderItem *ritem)
320 {
321 #ifdef USE_PTHREAD
322         RParseCtx *ctx;
323         pthread_t pt;
324 #endif
325
326         g_return_if_fail(ritem != NULL);
327
328
329 #ifdef USE_PTHREAD
330         ctx = g_new0(RParseCtx, 1);
331         ctx->ritem = ritem;
332         ctx->ready = FALSE;
333
334         if( pthread_create(&pt, PTHREAD_CREATE_JOINABLE, rssyl_read_existing_thr,
335                                 (void *)ctx) != 0 ) {
336                 /* Couldn't create thread, let's continue non-threaded. */
337                 rssyl_folder_read_existing_real(ritem);
338         } else {
339                 /* Thread started, wait until it is done. */
340                 debug_print("RSSyl: waiting for thread to finish\n");
341                 while( !ctx->ready ) {
342                         claws_do_idle();
343                 }
344
345                 debug_print("RSSyl: thread finished\n");
346                 pthread_join(pt, NULL);
347         }
348
349         g_free(ctx);
350 #else
351         rssyl_folder_read_existing_real(ritem);
352 #endif
353 }