src/plugins/rssyl/rssyl_add_item.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  * - DESCRIPTION HERE
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License
  19  * along with this program; if not, write to the Free Software
  20  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  21  */
  22
  23 #ifdef HAVE_CONFIG_H
  24 #  include "config.h"
  25 #endif
  26
  27 /* Global includes */
  28 #include <glib.h>
  29 #include <glib/gi18n.h>
  30 #include <string.h>
  31
  32 /* Claws Mail includes */
  33 #include <codeconv.h>
  34 #include <procmsg.h>
  35 #include <common/utils.h>
  36
  37 /* Local includes */
  38 #include "libfeed/date.h"
  39 #include "libfeed/feeditem.h"
  40 #include "parse822.h"
  41 #include "rssyl.h"
  42 #include "rssyl_deleted.h"
  43 #include "rssyl_feed.h"
  44 #include "rssyl_parse_feed.h"
  45 #include "strutils.h"
  46
  47 /* rssyl_cb_feed_compare()
  48  *
  49  * GCompareFunc function called by glib2's g_slist_find_custom().
  50  */
  51
  52 static gint rssyl_cb_feed_compare(const FeedItem *a, const FeedItem *b)
  53 {
  54         gboolean date_eq = FALSE, url_eq = FALSE, title_eq = FALSE;
  55         gboolean no_url = FALSE, no_date = FALSE, no_title = FALSE;
  56         gchar *atit = NULL, *btit = NULL;
  57
  58         g_return_val_if_fail(a != NULL && b != NULL, 1);
  59
  60         /* ID should be unique. If it matches, we've found what we came for. */
  61         if( (a->id != NULL) && (b->id != NULL) ) {
  62                         if( !strcmp(a->id, b->id) ) {
  63                                 return 0;
  64                         }
  65
  66                         /* If both IDs are present, but they do not match, these are not the
  67                          * droids we're looking for. */
  68                         return 1;
  69         }
  70
  71         /* Ok, we have no ID to aid us. Let's have a look at item timestamps
  72          * and item title & url. */
  73         if( (a->url != NULL) && (b->url != NULL) ) {
  74                 if( !strcmp(a->url, b->url) )
  75                         url_eq = TRUE;
  76         } else
  77                 no_url = TRUE;
  78
  79         if( (a->title != NULL) && (b->title != NULL) ) {
  80                 atit = conv_unmime_header(a->title, CS_UTF_8, FALSE);
  81                 btit = conv_unmime_header(b->title, CS_UTF_8, FALSE);
  82                 if( !strcmp(atit, btit) )
  83                         title_eq = TRUE;
  84                 g_free(atit);
  85                 g_free(btit);
  86         } else
  87                 no_title = TRUE;
  88
  89         /* If there's no 'published' timestamp for the item, we can only judge
  90          * by item url - 'modified' timestamp can have changed if the item was
  91          * updated recently. */
  92         if( b->date_published <= 0 ) {
  93                 if( b->date_modified > 0 ) {
  94                         /* If the item has 'modified' timestamp, we can only rely on url
  95                          * and title at this point. */
  96                         if( (url_eq || no_url) && title_eq
  97                                 && (a->date_modified >= b->date_modified) )
  98                                 return 0;
  99                         else
 100                                 return 1;
 101                 } else {
 102                         /* No timestamp of any kind, we'll just assume if both title and url
 103                          * match, we found the right item. Items in such feeds rarely change,
 104                          * and if they do, there's no way we can really */
 105                         if( (url_eq || no_url) && title_eq )
 106                                 return 0;
 107                         else
 108                                 return 1;
 109                 }
 110         }
 111
 112         /* Check if 'published' or at least 'modified' timestamps match */
 113         if( ((a->date_published > 0) && (b->date_published > 0) &&
 114                         (a->date_published == b->date_published))
 115                         || ((a->date_modified > 0) && (b->date_modified > 0) &&
 116                         (a->date_modified == b->date_modified))) {
 117                 date_eq = TRUE;
 118         } else
 119                 no_date = TRUE;
 120
 121         /* If 'published' time and item url match, it is reasonable to assume
 122          * we found our item. */
 123         if( (no_url || url_eq) && date_eq )
 124                 return 0;
 125
 126         /* There is no timestamp and the url matches (or there is none),
 127          * we need to compare titles, ... */
 128         if( (no_url || url_eq) && no_date ) {
 129                 if( title_eq )
 130                         return 0;
 131                 else
 132                         return 1;
 133         }
 134
 135         /* ... and as a last resort, if there is no title, item texts. */
 136         if( no_title && a->text && b->text ) {
 137                 if( !strcmp(a->text, b->text) )
 138                         return 0;
 139                 else
 140                         return 1;
 141         }
 142
 143         /* We don't know this item. */
 144         return 1;
 145 }
 146
 147 enum {
 148         ITEM_UNCHANGED,
 149         ITEM_CHANGED_TEXTONLY,
 150         ITEM_CHANGED
 151 };
 152
 153 static gint rssyl_feed_item_changed(FeedItem *new_item, FeedItem *old_item )
 154 {
 155         debug_print("RSSyl: comparing '%s' and '%s'\n",
 156                         new_item->title, old_item->title);
 157
 158         /* if both have title ... */
 159         if( old_item->title && new_item->title ) {
 160                 gchar *old = conv_unmime_header(old_item->title, CS_UTF_8, FALSE);
 161                 gchar *new = conv_unmime_header(new_item->title, CS_UTF_8, FALSE);
 162                 if( strcmp(old, new) != 0 ) { /* ... compare "unmimed" titles */
 163                         debug_print("RSSyl:\t\titem titles differ:\nOLD: '%s'\nNEW: '%s'\n",
 164                                         old, new);
 165                         g_free(old);
 166                         g_free(new);
 167                         return ITEM_CHANGED;
 168                 }
 169                 g_free(old);
 170                 g_free(new);
 171         } else {
 172                 /* if atleast one has a title, they differ */
 173                 if( old_item->title || new_item->title ) {
 174                         debug_print("RSSyl:\t\t+/- title\n");
 175                         return ITEM_CHANGED;
 176                 }
 177         }
 178
 179         if( old_item->author && new_item->author ) {
 180                 gchar *old = conv_unmime_header(old_item->author, CS_UTF_8, TRUE);
 181                 gchar *new = conv_unmime_header(new_item->author, CS_UTF_8, TRUE);
 182                 if( strcmp(old, new) ) {  /* ... compare "unmimed" authors */
 183                         g_free(old);
 184                         g_free(new);
 185                         debug_print("RSSyl:\t\titem authors differ\n");
 186                         return ITEM_CHANGED;
 187                 }
 188                 g_free(old);
 189                 g_free(new);
 190         } else {
 191                 /* if atleast one has author, they differ */
 192                 if( old_item->author || new_item->author ) {
 193                         debug_print("RSSyl:\t\t+/- author\n");
 194                         return ITEM_CHANGED;
 195                 }
 196         }
 197
 198         /* if both have text ... */
 199         if( old_item->text && new_item->text ) {
 200                 if( strcmp(old_item->text, new_item->text) ) { /* ... compare them */
 201                         debug_print("RSSyl:\t\titem texts differ\n");
 202                         debug_print("\nOLD: '%s'\n", old_item->text);
 203                         debug_print("\nNEW: '%s'\n", new_item->text);
 204
 205                         return ITEM_CHANGED_TEXTONLY;
 206                 }
 207         } else {
 208                 /* if at least one has some text, they differ */
 209                 if( old_item->text || new_item->text ) {
 210                         debug_print("RSSyl:\t\t+/- text\n");
 211                         return ITEM_CHANGED_TEXTONLY;
 212                 }
 213         }
 214
 215         /* they don't seem to differ */
 216         return ITEM_UNCHANGED;
 217 }
 218
 219 enum {
 220         EXISTS_NEW,
 221         EXISTS_UNCHANGED,
 222         EXISTS_CHANGED,
 223         EXISTS_CHANGED_TEXTONLY
 224 };
 225
 226 /* rssyl_feed_item_exists()
 227  *
 228  * Returns 1 if a feed item already exists locally, 2 if there's a changed
 229  * item with link that already belongs to existing item, 3 if only item's
 230  * text has changed, 0 if item is new.
 231  */
 232
 233 static guint rssyl_feed_item_exists(RFolderItem *ritem, FeedItem *fitem,
 234                 FeedItem **oldfitem)
 235 {
 236         GSList *item = NULL;
 237         FeedItem *efitem = NULL;
 238         gint changed;
 239
 240         g_return_val_if_fail(ritem != NULL, FALSE);
 241         g_return_val_if_fail(fitem != NULL, FALSE);
 242
 243         if( ritem->items == NULL || g_slist_length(ritem->items) == 0 )
 244                 return EXISTS_NEW;
 245
 246         if( (item = g_slist_find_custom(ritem->items,
 247                                         (gconstpointer)fitem, (GCompareFunc)rssyl_cb_feed_compare)) ) {
 248                 efitem = (FeedItem *)item->data;
 249                 if( (changed = rssyl_feed_item_changed(fitem, efitem)) > ITEM_UNCHANGED ) {
 250                         *oldfitem = efitem;
 251                         if (changed == ITEM_CHANGED_TEXTONLY)
 252                                 return EXISTS_CHANGED_TEXTONLY;
 253                         else
 254                                 return EXISTS_CHANGED;
 255                 }
 256
 257                 return EXISTS_UNCHANGED;
 258         }
 259
 260         return EXISTS_NEW;
 261 }
 262
 263 /* =============================================================== */
 264
 265 void rssyl_add_item(RFolderItem *ritem, FeedItem *feed_item)
 266 {
 267         FeedItem *old_item = NULL;
 268         MsgFlags *flags;
 269         MsgPermFlags oldperm_flags = 0;
 270         MsgInfo *msginfo;
 271         FILE *f;
 272         gint fd, d, dif, errno = 0;
 273         time_t tmpd;
 274         gchar *meta_charset = NULL;
 275         gchar *baseurl = NULL;
 276         gchar *template = NULL;
 277         gchar *tmp = NULL, *tmpurl = NULL, *tmpid = NULL;
 278         gchar *dirname = NULL;
 279         gchar *text = NULL;
 280         gchar *heading = NULL;
 281         gchar hdr[1024];
 282         FeedItemEnclosure *enc = NULL;
 283         RFeedCtx *ctx;
 284
 285         g_return_if_fail(ritem != NULL);
 286
 287         /* If item title is empty, try to fill it from source title (Atom only). */
 288         tmp = feed_item_get_sourcetitle(feed_item);
 289         if( feed_item_get_title(feed_item) == NULL ||
 290                         strlen(feed_item->title) == 0 ) {
 291                 if( tmp != NULL && strlen(tmp) > 0 )
 292                         feed_item_set_title(feed_item, tmp);
 293                 else
 294                         feed_item_set_title(feed_item, C_("Empty RSS feed title placeholder", "(empty)"));
 295         }
 296
 297 /*
 298         if (feed_item_get_id(feed_item) == NULL) {
 299                 debug_print("RSSyl: item ID empty, using its URL as ID.\n");
 300                 feed_item_set_id(feed_item, feed_item_get_url(feed_item));
 301         }
 302 */
 303
 304         /* If neither item date is set, use date from source (Atom only). */
 305         if( feed_item_get_date_modified(feed_item) == -1 &&
 306                         feed_item_get_date_published(feed_item) == -1 )
 307                 feed_item_set_date_published(feed_item,
 308                                 feed_item_get_sourcedate(feed_item));
 309
 310         /* Fix up subject, url and ID (rssyl_format_string()) so that
 311          * comparing doesn't break. */
 312         debug_print("RSSyl: fixing up subject '%s'\n", feed_item_get_title(feed_item));
 313         feed_item_set_title(feed_item, rssyl_format_string(feed_item_get_title(feed_item), TRUE, TRUE));
 314         debug_print("RSSyl: fixing up URL\n");
 315         feed_item_set_url(feed_item, rssyl_format_string(feed_item_get_url(feed_item),
 316                                 TRUE, TRUE));
 317         if( feed_item_get_id(feed_item) != NULL ) {
 318                 debug_print("RSSyl: fixing up ID\n");
 319                 feed_item_set_id(feed_item, rssyl_format_string(feed_item_get_id(feed_item),
 320                                         TRUE, TRUE));
 321         }
 322
 323         /* If there's a summary, but no text, use summary as text. */
 324         if( feed_item_get_text(feed_item) == NULL &&
 325                         (tmp = feed_item_get_summary(feed_item)) != NULL ) {
 326                 feed_item_set_text(feed_item, tmp);
 327                 g_free(feed_item->summary);     /* We do not need summary in rssyl now. */
 328                 feed_item->summary = NULL;
 329         }
 330
 331         /* Do not add if the item already exists, update if it does exist, but
 332          * has changed. */
 333         dif = rssyl_feed_item_exists(ritem, feed_item, &old_item);
 334         debug_print("RSSyl: rssyl_feed_item_exists returned %d\n", dif);
 335
 336         if( dif == EXISTS_UNCHANGED ) {
 337                 debug_print("RSSyl: This item already exists, skipping...\n");
 338                 return;
 339         }
 340
 341         /* Item is already in the list, but has changed */
 342         if( dif >= EXISTS_CHANGED && old_item != NULL ) {
 343                 debug_print("RSSyl: Item changed, removing old one and adding new.\n");
 344
 345                 /* Store permflags of the old item. */
 346                 ctx = (RFeedCtx *)old_item->data;
 347                 msginfo = folder_item_get_msginfo((FolderItem *)ritem,
 348                                 atoi(g_path_get_basename(ctx->path)));
 349                 oldperm_flags = msginfo->flags.perm_flags;
 350
 351                 ritem->items = g_slist_remove(ritem->items, old_item);
 352                 if (g_unlink(ctx->path) != 0) {
 353                         debug_print("RSSyl: Error, could not delete file '%s': %s\n",
 354                                         ctx->path, g_strerror(errno));
 355                 }
 356
 357                 g_free(ctx->path);
 358                 feed_item_free(old_item);
 359                 old_item = NULL;
 360         }
 361
 362         /* Check against list of deleted items. */
 363         if (rssyl_deleted_check(ritem->deleted_items, feed_item)) {
 364                 debug_print("RSSyl: Item '%s' found among deleted items, NOT adding it.\n",
 365                                 feed_item_get_title(feed_item));
 366                 return;
 367         }
 368
 369         /* Add a new item, formatting its title along the way */
 370         debug_print("RSSyl: Adding item '%s'\n", feed_item_get_title(feed_item));
 371         ritem->items = g_slist_prepend(ritem->items, feed_item_copy(feed_item));
 372
 373         dirname = folder_item_get_path(&ritem->item);
 374         template = g_strconcat(dirname, G_DIR_SEPARATOR_S,
 375                         RSSYL_TMP_TEMPLATE, NULL);
 376         fd = mkstemp(template);
 377
 378         f = fdopen(fd, "w");
 379         if(f == NULL) {
 380                 g_warning("Couldn't open file '%s', not adding msg!\n", template);
 381                 g_free(template);
 382                 return;
 383         }
 384
 385         /* From */
 386         if( (tmp = feed_item_get_author(feed_item)) != NULL ) {
 387                 if( g_utf8_validate(tmp, -1, NULL)) {
 388                         conv_encode_header_full(hdr, 1023, tmp, strlen("From: "),
 389                                         TRUE, CS_UTF_8);
 390                         fprintf(f, "From: %s\n", hdr);
 391                 } else
 392                         fprintf(f, "From: %s\n", tmp);
 393         }
 394
 395         /* Date */
 396         if( (tmpd = feed_item_get_date_modified(feed_item)) != -1 ) {
 397                 tmp = createRFC822Date(&tmpd);
 398                 debug_print("RSSyl: using date_modified: '%s'\n", tmp);
 399         } else if( (tmpd = feed_item_get_date_published(feed_item)) != -1 ) {
 400                 tmp = createRFC822Date(&tmpd);
 401                 debug_print("RSSyl: using date_published: '%s'\n", tmp);
 402         } else {
 403                 tmpd = time(NULL);
 404                 tmp = createRFC822Date(&tmpd);
 405         }
 406
 407         if( tmp != NULL ) {
 408                 fprintf(f, "Date: %s\n", tmp);
 409                 g_free(tmp);
 410         }
 411
 412         if( (tmp = feed_item_get_title(feed_item)) != NULL ) {
 413
 414                 /* (Atom only) Strip HTML markup from title for the Subject line. */
 415                 if( feed_item_get_title_format(feed_item) == FEED_ITEM_TITLE_HTML
 416                                 || feed_item_get_title_format(feed_item) == FEED_ITEM_TITLE_XHTML) {
 417                         debug_print("RSSyl: item title is HTML/XHTML, stripping tags for Subject line\n");
 418                         tmp = g_strdup(tmp);
 419                         strip_html(tmp);
 420                 }
 421
 422                 if( g_utf8_validate(tmp, -1, NULL) ) {
 423                         conv_encode_header_full(hdr, 1023, tmp, strlen("Subject: "),
 424                                         FALSE, CS_UTF_8);
 425                         debug_print("RSSyl: Subject: %s\n", hdr);
 426                         fprintf(f, "Subject: %s\n", hdr);
 427                 } else
 428                         fprintf(f, "Subject: %s\n", tmp);
 429
 430                 if( feed_item_get_title_format(feed_item) == FEED_ITEM_TITLE_HTML
 431                                 || feed_item_get_title_format(feed_item) == FEED_ITEM_TITLE_XHTML) {
 432                         g_free(tmp);
 433                         fprintf(f, "X-RSSyl-OrigTitle: %s\n", feed_item_get_title(feed_item));
 434                 }
 435         } else {
 436                 debug_print("RSSyl: No feed title, it seems\n");
 437                 fprintf(f, "Subject: (empty)\n");
 438         }
 439
 440         /* X-RSSyl-URL */
 441         if( (tmpurl = feed_item_get_url(feed_item)) == NULL ) {
 442                 if( feed_item_get_id(feed_item) != NULL &&
 443                                 feed_item_id_is_permalink(feed_item) ) {
 444                         tmpurl = feed_item_get_id(feed_item);
 445                 }
 446         }
 447
 448         if( tmpurl != NULL )
 449                 fprintf(f, "X-RSSyl-URL: %s\n", tmpurl);
 450
 451         if( ritem->last_update > 0) {
 452                 fprintf(f, "X-RSSyl-Last-Seen: %ld\n", ritem->last_update);
 453         }
 454
 455         /* Message-ID */
 456         if( (tmpid = feed_item_get_id(feed_item)) == NULL )
 457                 tmpid = feed_item_get_url(feed_item);
 458         if( tmpid != NULL )
 459                 fprintf(f, "Message-ID: <%s>\n", tmpid);
 460
 461         /* X-RSSyl-Comments */
 462         if( (text = feed_item_get_comments_url(feed_item)) != NULL )
 463                 fprintf(f, "X-RSSyl-Comments: %s\n", text);
 464
 465         /* References */
 466         if( (text = feed_item_get_parent_id(feed_item)) != NULL )
 467                 fprintf(f, "References: <%s>\n", text);
 468
 469         /* Content-Type */
 470         text = feed_item_get_text(feed_item);
 471         if( text && g_utf8_validate(text, -1, NULL) ) {
 472                 fprintf(f, "Content-Type: text/html; charset=UTF-8\n\n");
 473                 meta_charset = g_strdup("<meta http-equiv=\"Content-Type\" "
 474                                 "content=\"text/html; charset=UTF-8\">");
 475         } else {
 476                 fprintf(f, "Content-Type: text/html\n\n");
 477         }
 478
 479         /* construct base href */
 480         if( feed_item_get_url(feed_item) != NULL )
 481                 baseurl = g_strdup_printf("<base href=\"%s\">\n",
 482                         feed_item_get_url(feed_item) );
 483
 484         if( ritem->write_heading )
 485                 heading = g_strdup_printf("<h2>%s</h2>\n<br><br>\n",
 486                                 feed_item_get_title(feed_item));
 487
 488         /* Message body */
 489         fprintf(f, "<html><head>"
 490                         "%s\n"
 491                         "%s"
 492                         "</head>\n<body>\n"
 493                         "%s\n"
 494                         "URL: <a href=\"%s\">%s</a>\n\n<br><br>\n"
 495                         RSSYL_TEXT_START"\n"
 496                         "%s%s"
 497                         RSSYL_TEXT_END"\n\n",
 498                         (meta_charset ? meta_charset : ""),
 499                         (baseurl ? baseurl : ""),
 500                         (heading ? heading : ""),
 501                         (tmpurl ? tmpurl : ""),
 502                         (tmpurl ? tmpurl : "n/a"),
 503                         (text ? text : ""), (text ? "\n" : "") );
 504
 505         g_free(meta_charset);
 506         g_free(baseurl);
 507         g_free(heading);
 508
 509         if( (enc = feed_item_get_enclosure(feed_item)) != NULL )
 510                 fprintf(f, "<p><a href=\"%s\">Attached media file</a> [%s] (%ld bytes)</p>\n",
 511                                 feed_item_enclosure_get_url(enc),
 512                                 feed_item_enclosure_get_type(enc),
 513                                 feed_item_enclosure_get_size(enc) );
 514
 515         fprintf(f, "</body></html>\n");
 516         fclose(f);
 517
 518         g_return_if_fail(template != NULL);
 519
 520         flags = g_new(MsgFlags, 1);
 521         flags->perm_flags = MSG_NEW | MSG_UNREAD;
 522         flags->tmp_flags = 0;
 523
 524         d = folder_item_add_msg(&ritem->item, template, flags, TRUE);
 525         g_free(template);
 526
 527         ctx = g_new0(RFeedCtx, 1);
 528         ctx->path = (gpointer)g_strdup_printf("%s%c%d", dirname,
 529                         G_DIR_SEPARATOR, d);
 530         ctx->last_seen = ritem->last_update;
 531         ((FeedItem *)ritem->items->data)->data = (gpointer)ctx;
 532
 533         /* Unset unread+new if the changed item wasn't set unread and user
 534          * doesn't want to see it unread because of the change. */
 535         if (!(oldperm_flags & MSG_UNREAD) && (ritem->silent_update == 2
 536                         || (ritem->silent_update == 1 && dif == EXISTS_CHANGED_TEXTONLY)))
 537                 procmsg_msginfo_unset_flags(
 538                                 folder_item_get_msginfo((FolderItem *)ritem, d), MSG_NEW | MSG_UNREAD, 0);
 539
 540         debug_print("RSSyl: folder_item_add_msg(): %d\n", d);
 541 }