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'", error->message);
  68                 g_error_free(error);
  69         }
  70
  71         if( contents != NULL ) {
  72                 lines = strsplit_no_copy(contents, '\n');
  73         } else {
  74                 g_warning("Badly formatted file found, ignoring: '%s'", path);
  75                 return NULL;
  76         }
  77
  78         ctx = g_new0(RFeedCtx, 1);
  79         ctx->path = g_strdup(path); /* store filesystem path to source file */
  80         ctx->last_seen = 0;
  81
  82         item = feed_item_new(NULL);
  83         item->data = ctx;
  84
  85         while( lines[i] ) {
  86                 if( parsing_headers && lines[i] && !strlen(lines[i]) ) {
  87                         parsing_headers = FALSE;
  88                         debug_print("RSSyl: finished parsing headers\n");
  89                 }
  90
  91                 if( parsing_headers ) {
  92                         line = g_strsplit(lines[i], ": ", 2);
  93                         if( line[0] && line[1] && strlen(line[0]) && lines[i][0] != ' ') {
  94                                 started_author = FALSE;
  95                                 started_subject = FALSE;
  96                                 started_link = FALSE;
  97                                 started_clink = FALSE;
  98
  99                                 /* Author */
 100                                 if( !strcmp(line[0], "From") ) {
 101                                         feed_item_set_author(item, line[1]);
 102                                         debug_print("RSSyl: got author '%s'\n", feed_item_get_author(item));
 103                                         started_author = TRUE;
 104                                 }
 105
 106                                 /* Date */
 107                                 if( !strcmp(line[0], "Date") ) {
 108                                         feed_item_set_date_modified(item,
 109                                                         procheader_date_parse(NULL, line[1], 0));
 110                                         debug_print("RSSyl: got date \n" );
 111                                 }
 112
 113                                 /* Title */
 114                                 if( !strcmp(line[0], "Subject") && !got_original_title ) {
 115                                         feed_item_set_title(item,line[1]);
 116                                         debug_print("RSSyl: got title '%s'\n", feed_item_get_title(item));
 117                                         started_subject = TRUE;
 118                                 }
 119
 120                                 /* Original (including HTML) title - Atom feeds */
 121                                 if( !strcmp(line[0], "X-RSSyl-OrigTitle") ) {
 122                                         feed_item_set_title(item, line[1]);
 123                                         debug_print("RSSyl: got original title '%s'\n",
 124                                                         feed_item_get_title(item));
 125                                         got_original_title = TRUE;
 126                                 }
 127
 128                                 /* URL */
 129                                 if( !strcmp(line[0], "X-RSSyl-URL") ) {
 130                                         feed_item_set_url(item, line[1]);
 131                                         debug_print("RSSyl: got link '%s'\n", feed_item_get_url(item));
 132                                         started_link = TRUE;
 133                                 }
 134
 135                                 /* Last-Seen timestamp */
 136                                 if( !strcmp(line[0], "X-RSSyl-Last-Seen") ) {
 137                                         ctx->last_seen = atol(line[1]);
 138                                         debug_print("RSSyl: got last_seen timestamp %lld\n", (long long)ctx->last_seen);
 139                                 }
 140
 141                                 /* ID */
 142                                 if( !strcmp(line[0], "Message-ID") ) {
 143                                         if (line[1][0] != '<' || line[1][strlen(line[1])-1] != '>') {
 144                                                 debug_print("RSSyl: malformed Message-ID, ignoring...\n");
 145                                         } else {
 146                                                 /* Get the ID from within < and >. */
 147                                                 tmp = line[1] + 1;
 148                                                 tmp2 = g_strndup(tmp, strlen(tmp) - 1);
 149                                                 feed_item_set_id(item, tmp2);
 150                                                 g_free(tmp2);
 151                                         }
 152                                 }
 153
 154                                 /* Feed comments */
 155                                 if( !strcmp(line[0], "X-RSSyl-Comments") ) {
 156                                         feed_item_set_comments_url(item, line[1]);
 157                                         debug_print("RSSyl: got clink '%s'\n", feed_item_get_comments_url(item));
 158                                         started_clink = TRUE;
 159                                 }
 160
 161                                 /* References */
 162                                 if( !strcmp(line[0], "References") ) {
 163                                         splid = g_strsplit_set(line[1], "<>", 3);
 164                                         if( strlen(splid[1]) != 0 )
 165                                                 feed_item_set_parent_id(item, line[1]);
 166                                         g_strfreev(splid);
 167                                 }
 168
 169                         } else if (lines[i][0] == ' ') {
 170                                 gchar *tmp = NULL;
 171                                 /* continuation line */
 172                                 if (started_author) {
 173                                         tmp = g_strdup_printf("%s %s", feed_item_get_author(item), lines[i]+1);
 174                                         feed_item_set_author(item, tmp);
 175                                         debug_print("RSSyl: updated author to '%s'\n", tmp);
 176                                         g_free(tmp);
 177                                 } else if (started_subject) {
 178                                         tmp = g_strdup_printf("%s %s", feed_item_get_title(item), lines[i]+1);
 179                                         feed_item_set_title(item, tmp);
 180                                         debug_print("RSSyl: updated title to '%s'\n", tmp);
 181                                         g_free(tmp);
 182                                 } else if (started_link) {
 183                                         tmp = g_strdup_printf("%s%s", feed_item_get_url(item), lines[i]+1);
 184                                         feed_item_set_url(item, tmp);
 185                                         debug_print("RSSyl: updated link to '%s'\n", tmp);
 186                                         g_free(tmp);
 187                                 } else if (started_clink) {
 188                                         tmp = g_strdup_printf("%s%s", feed_item_get_comments_url(item), lines[i]+1);
 189                                         feed_item_set_comments_url(item, tmp);
 190                                         debug_print("RSSyl: updated comments_link to '%s'\n", tmp);
 191                                 }
 192                         }
 193                         g_strfreev(line);
 194                 } else {
 195                         if( !strcmp(lines[i], RSSYL_TEXT_START) ) {
 196                                 debug_print("RSSyl: Leading html tag found at line %d\n", i);
 197                                 past_html_tag = TRUE;
 198                                 i++;
 199                                 continue;
 200                         }
 201                         while( past_html_tag && !past_endhtml_tag && lines[i] ) {
 202                                 if( !strcmp(lines[i], RSSYL_TEXT_END) ) {
 203                                         debug_print("RSSyl: Trailing html tag found at line %d\n", i);
 204                                         past_endhtml_tag = TRUE;
 205                                         i++;
 206                                         continue;
 207                                 }
 208                                 if( feed_item_get_text(item) != NULL ) {
 209                                         gint e_len, n_len;
 210                                         e_len = strlen(item->text);
 211                                         n_len = strlen(lines[i]);
 212                                         item->text = g_realloc(item->text, e_len + n_len + 2);
 213                                         *(item->text+e_len) = '\n';
 214                                         strcpy(item->text+e_len+1, lines[i]);
 215                                         *(item->text+e_len+n_len+1) = '\0';
 216                                 } else {
 217                                         item->text = g_strdup(lines[i]);
 218                                 }
 219                                 i++;
 220                         }
 221
 222                         if( lines[i] == NULL )
 223                                 return item;
 224                 }
 225
 226                 i++;
 227         }
 228         g_free(lines);
 229         g_free(contents);
 230         return item;
 231 }
 232
 233 static void rssyl_flush_folder_func(gpointer data, gpointer user_data)
 234 {
 235         FeedItem *item = (FeedItem *)data;
 236         RFeedCtx *ctx = (RFeedCtx *)item->data;
 237
 238         if( ctx != NULL && ctx->path != NULL) {
 239                 g_free(ctx->path);
 240         }
 241         feed_item_free(item);
 242 }
 243
 244 static void rssyl_folder_read_existing_real(RFolderItem *ritem)
 245 {
 246         gchar *path = NULL, *fname = NULL;
 247         GDir *dp;
 248         const gchar *d;
 249         GError *error = NULL;
 250         gint num;
 251         FeedItem *item = NULL;
 252         RFeedCtx *ctx;
 253
 254         g_return_if_fail(ritem != NULL);
 255
 256         path = folder_item_get_path(&ritem->item);
 257         g_return_if_fail(path != NULL);
 258
 259         debug_print("RSSyl: reading existing items from '%s'\n", path);
 260
 261         /* Flush contents if any, so we can add new */
 262         if( g_slist_length(ritem->items) > 0 ) {
 263                 g_slist_foreach(ritem->items, (GFunc)rssyl_flush_folder_func, NULL);
 264                 g_slist_free(ritem->items);
 265         }
 266         ritem->items = NULL;
 267         ritem->last_update = 0;
 268
 269         if( (dp = g_dir_open(path, 0, &error)) == NULL ) {
 270                 debug_print("g_dir_open on \"%s\" failed with error %d (%s)\n",
 271                                 path, error->code, error->message);
 272                 g_error_free(error);
 273                 g_free(path);
 274                 return;
 275         }
 276
 277         while( (d = g_dir_read_name(dp)) != NULL ) {
 278                 if( claws_is_exiting() ) {
 279                         g_dir_close(dp);
 280                         g_free(path);
 281                         return;
 282                 }
 283
 284                 if( d[0] != '.' && (num = to_number(d)) > 0 ) {
 285                         fname = g_strdup_printf("%s%c%s", path, G_DIR_SEPARATOR, d);
 286                         if (!g_file_test(fname, G_FILE_TEST_IS_REGULAR)) {
 287                                 debug_print("RSSyl: not a regular file: '%s', ignoring it\n", fname);
 288                                 g_free(fname);
 289                                 continue;
 290                         }
 291
 292                         debug_print("RSSyl: starting to parse '%s'\n", d);
 293                         if( (item = rssyl_parse_folder_item_file(fname)) != NULL ) {
 294                                 /* Find latest timestamp */
 295                                 ctx = (RFeedCtx *)item->data;
 296                                 if( ritem->last_update < ctx->last_seen )
 297                                         ritem->last_update = ctx->last_seen;
 298                                 debug_print("RSSyl: Appending '%s'\n", feed_item_get_title(item));
 299                                 ritem->items = g_slist_prepend(ritem->items, item);
 300                         }
 301                         g_free(fname);
 302                 }
 303         }
 304
 305         g_dir_close(dp);
 306         g_free(path);
 307
 308         ritem->items = g_slist_reverse(ritem->items);
 309 }
 310
 311 #ifdef USE_PTHREAD
 312 static void *rssyl_read_existing_thr(void *arg)
 313 {
 314         RParseCtx *ctx = (RParseCtx *)arg;
 315
 316         rssyl_folder_read_existing_real(ctx->ritem);
 317         ctx->ready = TRUE;
 318         return NULL;
 319 }
 320 #endif
 321
 322 void rssyl_folder_read_existing(RFolderItem *ritem)
 323 {
 324 #ifdef USE_PTHREAD
 325         RParseCtx *ctx;
 326         pthread_t pt;
 327 #endif
 328
 329         g_return_if_fail(ritem != NULL);
 330
 331
 332 #ifdef USE_PTHREAD
 333         ctx = g_new0(RParseCtx, 1);
 334         ctx->ritem = ritem;
 335         ctx->ready = FALSE;
 336
 337         if( pthread_create(&pt, NULL, rssyl_read_existing_thr,
 338                                 (void *)ctx) != 0 ) {
 339                 /* Couldn't create thread, let's continue non-threaded. */
 340                 rssyl_folder_read_existing_real(ritem);
 341         } else {
 342                 /* Thread started, wait until it is done. */
 343                 debug_print("RSSyl: waiting for thread to finish\n");
 344                 while( !ctx->ready ) {
 345                         claws_do_idle();
 346                 }
 347
 348                 debug_print("RSSyl: thread finished\n");
 349                 pthread_join(pt, NULL);
 350         }
 351
 352         g_free(ctx);
 353 #else
 354         rssyl_folder_read_existing_real(ritem);
 355 #endif
 356 }