src/plugins/bogofilter/bogofilter.c

   1 /*
   2  * Claws Mail -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 1999-2012 Colin Leroy <colin@colino.net> and
   4  * the Claws Mail team
   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 3 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, see <http://www.gnu.org/licenses/>.
  18  *
  19  */
  20
  21 #ifdef HAVE_CONFIG_H
  22 #  include "config.h"
  23 #include "claws-features.h"
  24 #endif
  25
  26 #include "defs.h"
  27
  28 #include <sys/types.h>
  29 #include <sys/wait.h>
  30 #include <errno.h>
  31
  32 #include <glib.h>
  33 #include <glib/gi18n.h>
  34
  35 #if HAVE_LOCALE_H
  36 #  include <locale.h>
  37 #endif
  38
  39 #include "common/claws.h"
  40 #include "common/version.h"
  41 #include "plugin.h"
  42 #include "common/utils.h"
  43 #include "hooks.h"
  44 #include "procmsg.h"
  45 #include "folder.h"
  46 #include "prefs.h"
  47 #include "prefs_gtk.h"
  48
  49 #include "bogofilter.h"
  50 #include "inc.h"
  51 #include "log.h"
  52 #include "prefs_common.h"
  53 #include "alertpanel.h"
  54 #include "addr_compl.h"
  55
  56 #ifdef HAVE_SYSEXITS_H
  57 #include <sysexits.h>
  58 #endif
  59 #ifdef HAVE_ERRNO_H
  60 #include <errno.h>
  61 #endif
  62 #ifdef HAVE_SYS_ERRNO_H
  63 #include <sys/errno.h>
  64 #endif
  65 #ifdef HAVE_TIME_H
  66 #include <time.h>
  67 #endif
  68 #ifdef HAVE_SYS_TIME_H
  69 #include <sys/time.h>
  70 #endif
  71 #ifdef HAVE_SIGNAL_H
  72 #include <signal.h>
  73 #endif
  74 #ifdef HAVE_PWD_H
  75 #include <pwd.h>
  76 #endif
  77 #ifdef USE_PTHREAD
  78 #include <pthread.h>
  79 #endif
  80
  81 #define PLUGIN_NAME (_("Bogofilter"))
  82
  83 static guint hook_id = -1;
  84 static MessageCallback message_callback;
  85
  86 static BogofilterConfig config;
  87
  88 static PrefParam param[] = {
  89         {"process_emails", "TRUE", &config.process_emails, P_BOOL,
  90          NULL, NULL, NULL},
  91         {"receive_spam", "1", &config.receive_spam, P_INT,
  92          NULL, NULL, NULL},
  93         {"save_folder", NULL, &config.save_folder, P_STRING,
  94          NULL, NULL, NULL},
  95         {"save_unsure", "FALSE", &config.save_unsure, P_BOOL,
  96          NULL, NULL, NULL},
  97         {"save_unsure_folder", NULL, &config.save_unsure_folder, P_STRING,
  98          NULL, NULL, NULL},
  99         {"max_size", "250", &config.max_size, P_INT,
 100          NULL, NULL, NULL},
 101         {"bogopath", "bogofilter", &config.bogopath, P_STRING,
 102          NULL, NULL, NULL},
 103         {"insert_header", "FALSE", &config.insert_header, P_BOOL,
 104          NULL, NULL, NULL},
 105         {"whitelist_ab", "FALSE", &config.whitelist_ab, P_BOOL,
 106          NULL, NULL, NULL},
 107         {"whitelist_ab_folder", N_("Any"), &config.whitelist_ab_folder, P_STRING,
 108          NULL, NULL, NULL},
 109         {"learn_from_whitelist", "FALSE", &config.learn_from_whitelist, P_BOOL,
 110          NULL, NULL, NULL},
 111         {"mark_as_read", "TRUE", &config.mark_as_read, P_BOOL,
 112          NULL, NULL, NULL},
 113
 114         {NULL, NULL, NULL, P_OTHER, NULL, NULL, NULL}
 115 };
 116
 117 /*
 118  * Helper function for spawn_with_input() - write an entire
 119  * string to a fd.
 120  */
 121 static gboolean
 122 write_all (int         fd,
 123            const char *buf,
 124            gsize       to_write)
 125 {
 126   while (to_write > 0)
 127     {
 128       gssize count = write (fd, buf, to_write);
 129       if (count < 0)
 130         {
 131           if (errno != EINTR)
 132             return FALSE;
 133         }
 134       else
 135         {
 136           to_write -= count;
 137           buf += count;
 138         }
 139     }
 140
 141   return TRUE;
 142 }
 143
 144 typedef struct _BogoFilterData {
 145         MailFilteringData *mail_filtering_data;
 146         gchar **bogo_args;
 147         GSList *msglist;
 148         GSList *new_hams;
 149         GSList *new_unsure;
 150         GSList *new_spams;
 151         GSList *whitelisted_new_spams;
 152         gboolean done;
 153         int status;
 154         gboolean in_thread;
 155 } BogoFilterData;
 156
 157 static BogoFilterData *to_filter_data = NULL;
 158 #ifdef USE_PTHREAD
 159 static gboolean filter_th_done = FALSE;
 160 static pthread_mutex_t list_mutex = PTHREAD_MUTEX_INITIALIZER;
 161 static pthread_mutex_t wait_mutex = PTHREAD_MUTEX_INITIALIZER;
 162 static pthread_cond_t wait_cond = PTHREAD_COND_INITIALIZER;
 163 #endif
 164
 165 static void bogofilter_do_filter(BogoFilterData *data)
 166 {
 167         GPid bogo_pid;
 168         gint bogo_stdin, bogo_stdout;
 169         GError *error = NULL;
 170         gboolean bogo_forked;
 171         int status = 0;
 172         MsgInfo *msginfo;
 173         GSList *cur = NULL;
 174         int total = 0, curnum = 1;
 175         gchar *file = NULL;
 176         gchar buf[BUFSIZ];
 177
 178         total = g_slist_length(data->msglist);
 179
 180         bogo_forked = g_spawn_async_with_pipes(
 181                         NULL, data->bogo_args,NULL, G_SPAWN_SEARCH_PATH|G_SPAWN_DO_NOT_REAP_CHILD,
 182                         NULL, NULL, &bogo_pid, &bogo_stdin,
 183                         &bogo_stdout, NULL, &error);
 184
 185         if (bogo_forked == FALSE) {
 186                 g_warning("%s", error ? error->message:"ERROR???");
 187                 g_error_free(error);
 188                 error = NULL;
 189                 status = -1;
 190         } else {
 191
 192                 if (config.whitelist_ab) {
 193                         gchar *ab_folderpath;
 194
 195                         if (*config.whitelist_ab_folder == '\0' ||
 196                                 strcasecmp(config.whitelist_ab_folder, "Any") == 0) {
 197                                 /* match the whole addressbook */
 198                                 ab_folderpath = NULL;
 199                         } else {
 200                                 /* match the specific book/folder of the addressbook */
 201                                 ab_folderpath = config.whitelist_ab_folder;
 202                         }
 203
 204                         start_address_completion(ab_folderpath);
 205                 }
 206
 207                 for (cur = data->msglist; cur; cur = cur->next) {
 208                         gboolean whitelisted = FALSE;
 209                         msginfo = (MsgInfo *)cur->data;
 210                         debug_print("Filtering message %d (%d/%d)\n", msginfo->msgnum, curnum, total);
 211
 212                         if (message_callback != NULL)
 213                                 message_callback(NULL, total, curnum++, data->in_thread);
 214
 215                         if (config.whitelist_ab && msginfo->from &&
 216                             found_in_addressbook(msginfo->from))
 217                                 whitelisted = TRUE;
 218
 219                         /* can set flags (SCANNED, ATTACHMENT) but that's ok
 220                          * as GUI updates are hooked not direct */
 221
 222                         file = procmsg_get_message_file(msginfo);
 223
 224                         if (file) {
 225                                 gchar *tmp = g_strdup_printf("%s\n",file);
 226                                 /* send filename to bogofilter */
 227                                 write_all(bogo_stdin, tmp, strlen(tmp));
 228                                 g_free(tmp);
 229                                 memset(buf, 0, sizeof(buf));
 230                                 /* get the result */
 231                                 if (read(bogo_stdout, buf, sizeof(buf)-1) < 0) {
 232                                         g_warning("bogofilter short read");
 233                                         debug_print("message %d is ham\n", msginfo->msgnum);
 234                                         data->mail_filtering_data->unfiltered = g_slist_prepend(
 235                                                 data->mail_filtering_data->unfiltered, msginfo);
 236                                         data->new_hams = g_slist_prepend(data->new_hams, msginfo);
 237                                 } else {
 238                                         gchar **parts = NULL;
 239
 240                                         buf[sizeof(buf) - 1] = '\0';
 241                                         if (strchr(buf, '/')) {
 242                                                 tmp = strrchr(buf, '/')+1;
 243                                         } else {
 244                                                 tmp = buf;
 245                                         }
 246                                         parts = g_strsplit(tmp, " ", 0);
 247                                         debug_print("read %s\n", buf);
 248
 249                                         /* note the result if the header if needed */
 250                                         if (parts && parts[0] && parts[1] && parts[2] &&
 251                                             FOLDER_TYPE(msginfo->folder->folder) == F_MH &&
 252                                             config.insert_header) {
 253                                                 gchar *tmpfile = get_tmp_file();
 254                                                 FILE *input = g_fopen(file, "r");
 255                                                 FILE *output = g_fopen(tmpfile, "w");
 256                                                 if (strstr(parts[2], "\n"))
 257                                                         *(strstr(parts[2], "\n")) = '\0';
 258                                                 if (input && !output)
 259                                                         fclose (input);
 260                                                 else if (!input && output)
 261                                                         fclose (output);
 262                                                 else if (input && output) {
 263                                                         gchar tmpbuf[BUFFSIZE];
 264                                                         gboolean err = FALSE;
 265                                                         const gchar *bogosity = *parts[1] == 'S' ? "Spam":
 266                                                                                  (*parts[1] == 'H' ? "Ham":"Unsure");
 267                                                         gchar *tmpstr = g_strdup_printf(
 268                                                                         "X-Bogosity: %s, spamicity=%s%s\n",
 269                                                                         bogosity, parts[2],
 270                                                                         whitelisted?" [whitelisted]":"");
 271                                                         if (fwrite(tmpstr, 1, strlen(tmpstr), output) < strlen(tmpstr)) {
 272                                                                 err = TRUE;
 273                                                         } else {
 274                                                                 while (fgets(tmpbuf, sizeof(buf), input)) {
 275                                                                         if (fputs(tmpbuf, output) == EOF) {
 276                                                                                 err = TRUE;
 277                                                                                 break;
 278                                                                         }
 279                                                                 }
 280                                                         }
 281                                                         fclose(input);
 282                                                         if (fclose(output) == EOF)
 283                                                                 err = TRUE;
 284                                                         if (!err)
 285                                                                 move_file(tmpfile, file, TRUE);
 286                                                         g_free(tmpstr);
 287                                                 }
 288                                                 g_free(tmpfile);
 289                                         }
 290
 291                                         /* file the mail */
 292                                         if (!whitelisted && parts && parts[0] && parts[1] && *parts[1] == 'S') {
 293
 294                                                 debug_print("message %d is spam\n", msginfo->msgnum);
 295                                                 /* Spam will be filtered away, unless we want "mark only".
 296                                                  * In that case, we want it among unfiltered messages, so
 297                                                  * it gets processed further. */
 298                                                 if (config.receive_spam == SPAM_MARK_ONLY) {
 299                                                         data->mail_filtering_data->unfiltered = g_slist_prepend(
 300                                                                 data->mail_filtering_data->unfiltered, msginfo);
 301                                                 } else {
 302                                                         data->mail_filtering_data->filtered = g_slist_prepend(
 303                                                                 data->mail_filtering_data->filtered, msginfo);
 304                                                 }
 305                                                 data->new_spams = g_slist_prepend(data->new_spams, msginfo);
 306
 307                                         } else if (whitelisted && parts && parts[0] && parts[1] &&
 308                                                         (*parts[1] == 'S' || *parts[1] == 'U')) {
 309
 310                                                 debug_print("message %d is whitelisted %s\n", msginfo->msgnum,
 311                                                         *parts[1] == 'S' ? "spam":"unsure");
 312                                                 /* Whitelisted spam will *not* be filtered away, but continue
 313                                                  * their trip through filtering as if it was ham. */
 314                                                 data->mail_filtering_data->unfiltered = g_slist_prepend(
 315                                                         data->mail_filtering_data->unfiltered, msginfo);
 316                                                 /* But it gets put in a different list, so that we
 317                                                  * can still flag it and inform the user that it is
 318                                                  * considered a spam (so that he can teach bogo that
 319                                                  * it was not). */
 320                                                 data->whitelisted_new_spams = g_slist_prepend(data->whitelisted_new_spams, msginfo);
 321
 322                                         } else if (config.save_unsure && parts && parts[0] && parts[1] && *parts[1] == 'U') {
 323
 324                                                 debug_print("message %d is unsure\n", msginfo->msgnum);
 325                                                 /* Spam will be filtered away */
 326                                                 data->mail_filtering_data->filtered = g_slist_prepend(
 327                                                         data->mail_filtering_data->filtered, msginfo);
 328                                                 data->new_unsure = g_slist_prepend(data->new_unsure, msginfo);
 329
 330                                         } else {
 331
 332                                                 debug_print("message %d is ham\n", msginfo->msgnum);
 333                                                 data->mail_filtering_data->unfiltered = g_slist_prepend(
 334                                                         data->mail_filtering_data->unfiltered, msginfo);
 335                                                 data->new_hams = g_slist_prepend(data->new_hams, msginfo);
 336
 337                                         }
 338                                         g_strfreev(parts);
 339                                 }
 340                                 g_free(file);
 341                         } else {
 342                                 data->mail_filtering_data->unfiltered = g_slist_prepend(
 343                                         data->mail_filtering_data->unfiltered, msginfo);
 344                                 data->new_hams = g_slist_prepend(data->new_hams, msginfo);
 345                         }
 346                 }
 347                 if (config.whitelist_ab)
 348                         end_address_completion();
 349         }
 350         if (status != -1) {
 351                 close(bogo_stdout);
 352                 close(bogo_stdin);
 353                 waitpid(bogo_pid, &status, 0);
 354                 if (!WIFEXITED(status))
 355                         status = -1;
 356                 else
 357                         status = WEXITSTATUS(status);
 358         }
 359
 360         to_filter_data->status = status;
 361 }
 362
 363 #ifdef USE_PTHREAD
 364 static void *bogofilter_filtering_thread(void *data)
 365 {
 366         while (!filter_th_done) {
 367                 pthread_mutex_lock(&list_mutex);
 368                 if (to_filter_data == NULL || to_filter_data->done == TRUE) {
 369                         pthread_mutex_unlock(&list_mutex);
 370                         debug_print("thread is waiting for something to filter\n");
 371                         pthread_mutex_lock(&wait_mutex);
 372                         pthread_cond_wait(&wait_cond, &wait_mutex);
 373                         pthread_mutex_unlock(&wait_mutex);
 374                 } else {
 375                         debug_print("thread awaken with something to filter\n");
 376                         to_filter_data->done = FALSE;
 377                         bogofilter_do_filter(to_filter_data);
 378                         pthread_mutex_unlock(&list_mutex);
 379                         to_filter_data->done = TRUE;
 380                         usleep(100);
 381                 }
 382         }
 383         return NULL;
 384 }
 385
 386 static pthread_t filter_th = 0;
 387
 388 static void bogofilter_start_thread(void)
 389 {
 390         filter_th_done = FALSE;
 391         if (filter_th != 0 || 1)
 392                 return;
 393         if (pthread_create(&filter_th, NULL,
 394                         bogofilter_filtering_thread,
 395                         NULL) != 0) {
 396                 filter_th = 0;
 397                 return;
 398         }
 399         debug_print("thread created\n");
 400 }
 401
 402 static void bogofilter_stop_thread(void)
 403 {
 404         void *res;
 405         while (pthread_mutex_trylock(&list_mutex) != 0) {
 406                 GTK_EVENTS_FLUSH();
 407                 usleep(100);
 408         }
 409         if (filter_th != 0) {
 410                 filter_th_done = TRUE;
 411                 debug_print("waking thread up\n");
 412                 pthread_mutex_lock(&wait_mutex);
 413                 pthread_cond_broadcast(&wait_cond);
 414                 pthread_mutex_unlock(&wait_mutex);
 415                 pthread_join(filter_th, &res);
 416                 filter_th = 0;
 417         }
 418         pthread_mutex_unlock(&list_mutex);
 419         debug_print("thread done\n");
 420 }
 421 #endif
 422
 423 static gboolean mail_filtering_hook(gpointer source, gpointer data)
 424 {
 425         MailFilteringData *mail_filtering_data = (MailFilteringData *) source;
 426         MsgInfo *msginfo = mail_filtering_data->msginfo;
 427         GSList *msglist = mail_filtering_data->msglist;
 428         GSList *cur = NULL;
 429         static gboolean warned_error = FALSE;
 430         int status = 0;
 431         int total = 0, curnum = 0;
 432         GSList *new_hams = NULL, *new_spams = NULL;
 433         GSList *new_unsure, *whitelisted_new_spams = NULL;
 434         gchar *bogo_exec = (config.bogopath && *config.bogopath) ? config.bogopath:"bogofilter";
 435         gchar *bogo_args[4];
 436         gboolean ok_to_thread = TRUE;
 437
 438         bogo_args[0] = bogo_exec;
 439         bogo_args[1] = "-T";
 440         bogo_args[2] = "-b";
 441         bogo_args[3] = NULL;
 442
 443         if (!config.process_emails) {
 444                 return FALSE;
 445         }
 446
 447         if (msglist == NULL && msginfo != NULL) {
 448                 g_warning("wrong call to bogofilter mail_filtering_hook");
 449                 return FALSE;
 450         }
 451
 452         total = g_slist_length(msglist);
 453
 454         /* we have to make sure the mails are cached - or it'll break on IMAP */
 455         if (message_callback != NULL)
 456                 message_callback(_("Bogofilter: fetching bodies..."), total, 0, FALSE);
 457         for (cur = msglist; cur; cur = cur->next) {
 458                 gchar *file = procmsg_get_message_file((MsgInfo *)cur->data);
 459                 if (file == NULL)
 460                         ok_to_thread = FALSE;
 461                 if (message_callback != NULL)
 462                         message_callback(NULL, total, curnum++, FALSE);
 463                 g_free(file);
 464         }
 465         if (message_callback != NULL)
 466                 message_callback(NULL, 0, 0, FALSE);
 467
 468         if (message_callback != NULL)
 469                 message_callback(_("Bogofilter: filtering messages..."), total, 0, FALSE);
 470
 471 #ifdef USE_PTHREAD
 472         while (pthread_mutex_trylock(&list_mutex) != 0) {
 473                 GTK_EVENTS_FLUSH();
 474                 usleep(100);
 475         }
 476 #endif
 477         to_filter_data = g_new0(BogoFilterData, 1);
 478         to_filter_data->msglist = msglist;
 479         to_filter_data->mail_filtering_data = mail_filtering_data;
 480         to_filter_data->new_hams = NULL;
 481         to_filter_data->new_unsure = NULL;
 482         to_filter_data->new_spams = NULL;
 483         to_filter_data->whitelisted_new_spams = NULL;
 484         to_filter_data->done = FALSE;
 485         to_filter_data->status = -1;
 486         to_filter_data->bogo_args = bogo_args;
 487 #ifdef USE_PTHREAD
 488         to_filter_data->in_thread = (filter_th != 0 && ok_to_thread);
 489 #else
 490         to_filter_data->in_thread = FALSE;
 491 #endif
 492
 493 #ifdef USE_PTHREAD
 494         pthread_mutex_unlock(&list_mutex);
 495
 496         if (filter_th != 0 && ok_to_thread) {
 497                 debug_print("waking thread to let it filter things\n");
 498                 pthread_mutex_lock(&wait_mutex);
 499                 pthread_cond_broadcast(&wait_cond);
 500                 pthread_mutex_unlock(&wait_mutex);
 501
 502                 while (!to_filter_data->done) {
 503                         GTK_EVENTS_FLUSH();
 504                         usleep(100);
 505                 }
 506         }
 507
 508         while (pthread_mutex_trylock(&list_mutex) != 0) {
 509                 GTK_EVENTS_FLUSH();
 510                 usleep(100);
 511
 512         }
 513         if (filter_th == 0 || !ok_to_thread)
 514                 bogofilter_do_filter(to_filter_data);
 515 #else
 516         bogofilter_do_filter(to_filter_data);
 517 #endif
 518
 519         new_hams = to_filter_data->new_hams;
 520         new_unsure = to_filter_data->new_unsure;
 521         new_spams = to_filter_data->new_spams;
 522         whitelisted_new_spams = to_filter_data->whitelisted_new_spams;
 523         status = to_filter_data->status;
 524         g_free(to_filter_data);
 525         to_filter_data = NULL;
 526 #ifdef USE_PTHREAD
 527         pthread_mutex_unlock(&list_mutex);
 528 #endif
 529
 530
 531         /* unflag hams */
 532         for (cur = new_hams; cur; cur = cur->next) {
 533                 MsgInfo *msginfo = (MsgInfo *)cur->data;
 534                 procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 535                 debug_print("unflagging ham: %d\n", msginfo->msgnum);
 536         }
 537         /* unflag unsure */
 538         for (cur = new_unsure; cur; cur = cur->next) {
 539                 MsgInfo *msginfo = (MsgInfo *)cur->data;
 540                 procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 541                 debug_print("unflagging unsure: %d\n", msginfo->msgnum);
 542         }
 543         if (config.learn_from_whitelist && whitelisted_new_spams) {
 544                 /* flag whitelisted spams */
 545                 for (cur = whitelisted_new_spams; cur; cur = cur->next) {
 546                         MsgInfo *msginfo = (MsgInfo *)cur->data;
 547                         procmsg_msginfo_set_flags(msginfo, MSG_SPAM, 0);
 548                         debug_print("flagging whitelisted non-ham: %d\n", msginfo->msgnum);
 549                 }
 550                 /* correct bogo */
 551                 bogofilter_learn(NULL, whitelisted_new_spams, FALSE);
 552
 553                 /* unflag them */
 554                 for (cur = whitelisted_new_spams; cur; cur = cur->next) {
 555                         MsgInfo *msginfo = (MsgInfo *)cur->data;
 556                         procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 557                         debug_print("unflagging whitelisted non-ham: %d\n", msginfo->msgnum);
 558                 }
 559         } else {
 560                 for (cur = whitelisted_new_spams; cur; cur = cur->next) {
 561                         MsgInfo *msginfo = (MsgInfo *)cur->data;
 562                         procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 563                         debug_print("not flagging whitelisted non-ham: %d\n", msginfo->msgnum);
 564                 }
 565         }
 566
 567         /* flag spams and delete them if config.receive_spam == 0
 568          * (if config.receive_spam is set to 1, we'll move them later,
 569          * mark as spam only if set to 2) */
 570         for (cur = new_spams; cur; cur = cur->next) {
 571                 MsgInfo *msginfo = (MsgInfo *)cur->data;
 572                 if (config.receive_spam != SPAM_DELETE) {
 573                         if (config.mark_as_read)
 574                                 procmsg_msginfo_unset_flags(msginfo, ~0, 0);
 575                         procmsg_msginfo_set_flags(msginfo, MSG_SPAM, 0);
 576                 } else {
 577                         folder_item_remove_msg(msginfo->folder, msginfo->msgnum);
 578                 }
 579         }
 580
 581         if (status < 0 || status > 2) { /* I/O or other errors */
 582                 gchar *msg = NULL;
 583
 584                 if (status == 3)
 585                         msg =  g_strdup_printf(_("The Bogofilter plugin couldn't filter "
 586                                            "a message. The probable cause of the "
 587                                            "error is that it didn't learn from any mail.\n"
 588                                            "Use \"/Mark/Mark as spam\" and \"/Mark/Mark as "
 589                                            "ham\" to train Bogofilter with a few hundred "
 590                                            "spam and ham messages."));
 591                 else
 592                         msg =  g_strdup_printf(_("The Bogofilter plugin couldn't filter "
 593                                            "a message. The command `%s %s %s` couldn't be run."),
 594                                            bogo_args[0], bogo_args[1], bogo_args[2]);
 595                 if (!prefs_common_get_prefs()->no_recv_err_panel) {
 596                         if (!warned_error) {
 597                                 alertpanel_error("%s", msg);
 598                         }
 599                         warned_error = TRUE;
 600                 } else {
 601                         log_error(LOG_PROTOCOL, "%s\n", msg);
 602                 }
 603                 g_free(msg);
 604         }
 605         if (status < 0 || status > 2) {
 606                 g_slist_free(mail_filtering_data->filtered);
 607                 g_slist_free(mail_filtering_data->unfiltered);
 608                 mail_filtering_data->filtered = NULL;
 609                 mail_filtering_data->unfiltered = NULL;
 610         } else {
 611                 if (config.receive_spam == SPAM_MARK_AND_SAVE && new_spams) {
 612                         FolderItem *save_folder = NULL;
 613
 614                         if ((!config.save_folder) ||
 615                             (config.save_folder[0] == '\0') ||
 616                             ((save_folder = folder_find_item_from_identifier(config.save_folder)) == NULL)) {
 617                                 if (mail_filtering_data->account && mail_filtering_data->account->set_trash_folder) {
 618                                         save_folder = folder_find_item_from_identifier(
 619                                                 mail_filtering_data->account->trash_folder);
 620                                         if (save_folder)
 621                                                 debug_print("found trash folder from account's advanced settings\n");
 622                                 }
 623                                 if (save_folder == NULL && mail_filtering_data->account &&
 624                                     mail_filtering_data->account->folder) {
 625                                         save_folder = mail_filtering_data->account->folder->trash;
 626                                         if (save_folder)
 627                                                 debug_print("found trash folder from account's trash\n");
 628                                 }
 629                                 if (save_folder == NULL && mail_filtering_data->account &&
 630                                     !mail_filtering_data->account->folder)  {
 631                                         if (mail_filtering_data->account->inbox) {
 632                                                 FolderItem *item = folder_find_item_from_identifier(
 633                                                         mail_filtering_data->account->inbox);
 634                                                 if (item && item->folder->trash) {
 635                                                         save_folder = item->folder->trash;
 636                                                         debug_print("found trash folder from account's inbox\n");
 637                                                 }
 638                                         }
 639                                         if (!save_folder && mail_filtering_data->account->local_inbox) {
 640                                                 FolderItem *item = folder_find_item_from_identifier(
 641                                                         mail_filtering_data->account->local_inbox);
 642                                                 if (item && item->folder->trash) {
 643                                                         save_folder = item->folder->trash;
 644                                                         debug_print("found trash folder from account's local_inbox\n");
 645                                                 }
 646                                         }
 647                                 }
 648                                 if (save_folder == NULL) {
 649                                         debug_print("using default trash folder\n");
 650                                         save_folder = folder_get_default_trash();
 651                                 }
 652                         }
 653                         if (save_folder) {
 654                                 for (cur = new_spams; cur; cur = cur->next) {
 655                                         msginfo = (MsgInfo *)cur->data;
 656                                         msginfo->filter_op = IS_MOVE;
 657                                         msginfo->to_filter_folder = save_folder;
 658                                 }
 659                         }
 660                 }
 661                 if (config.save_unsure && new_unsure) {
 662                         FolderItem *save_unsure_folder = NULL;
 663
 664                         if ((!config.save_unsure_folder) ||
 665                             (config.save_unsure_folder[0] == '\0') ||
 666                             ((save_unsure_folder = folder_find_item_from_identifier(config.save_unsure_folder)) == NULL)) {
 667                                 if (mail_filtering_data->account)
 668                                         save_unsure_folder = folder_find_item_from_identifier(
 669                                                 mail_filtering_data->account->inbox);
 670                                 if (save_unsure_folder == NULL && mail_filtering_data->account &&
 671                                     mail_filtering_data->account->folder)
 672                                         save_unsure_folder = mail_filtering_data->account->folder->inbox;
 673                                 if (save_unsure_folder == NULL && mail_filtering_data->account &&
 674                                     !mail_filtering_data->account->folder)  {
 675                                         if (mail_filtering_data->account->inbox) {
 676                                                 FolderItem *item = folder_find_item_from_identifier(
 677                                                         mail_filtering_data->account->inbox);
 678                                                 if (item) {
 679                                                         save_unsure_folder = item;
 680                                                 }
 681                                         }
 682                                         if (!save_unsure_folder && mail_filtering_data->account->local_inbox) {
 683                                                 FolderItem *item = folder_find_item_from_identifier(
 684                                                         mail_filtering_data->account->local_inbox);
 685                                                 if (item) {
 686                                                         save_unsure_folder = item;
 687                                                 }
 688                                         }
 689                                 }
 690                                 if (save_unsure_folder == NULL)
 691                                         save_unsure_folder = folder_get_default_inbox();
 692                         }
 693                         if (save_unsure_folder) {
 694                                 for (cur = new_unsure; cur; cur = cur->next) {
 695                                         msginfo = (MsgInfo *)cur->data;
 696                                         msginfo->filter_op = IS_MOVE;
 697                                         msginfo->to_filter_folder = save_unsure_folder;
 698                                 }
 699                         }
 700                 }
 701         }
 702         g_slist_free(new_hams);
 703         g_slist_free(new_unsure);
 704         g_slist_free(new_spams);
 705         g_slist_free(whitelisted_new_spams);
 706
 707         if (message_callback != NULL)
 708                 message_callback(NULL, 0, 0, FALSE);
 709         mail_filtering_data->filtered   = g_slist_reverse(
 710                 mail_filtering_data->filtered);
 711         mail_filtering_data->unfiltered = g_slist_reverse(
 712                 mail_filtering_data->unfiltered);
 713
 714         return FALSE;
 715 }
 716
 717 BogofilterConfig *bogofilter_get_config(void)
 718 {
 719         return &config;
 720 }
 721
 722 int bogofilter_learn(MsgInfo *msginfo, GSList *msglist, gboolean spam)
 723 {
 724         gchar *cmd = NULL;
 725         gchar *file = NULL;
 726         const gchar *bogo_exec = (config.bogopath && *config.bogopath) ? config.bogopath:"bogofilter";
 727         gint status = 0;
 728
 729         if (msginfo == NULL && msglist == NULL) {
 730                 return -1;
 731         }
 732
 733         if (msginfo) {
 734                 file = procmsg_get_message_file(msginfo);
 735                 if (file == NULL) {
 736                         return -1;
 737                 } else {
 738                         if (message_callback != NULL)
 739                                 message_callback(_("Bogofilter: learning from message..."), 0, 0, FALSE);
 740                         if (spam)
 741                                 /* learn as spam */
 742                                 cmd = g_strdup_printf("%s -s -I '%s'", bogo_exec, file);
 743                         else if (MSG_IS_SPAM(msginfo->flags))
 744                                 /* correct bogofilter, this wasn't spam */
 745                                 cmd = g_strdup_printf("%s -Sn -I '%s'", bogo_exec, file);
 746                         else
 747                                 /* learn as ham */
 748                                 cmd = g_strdup_printf("%s -n -I '%s'", bogo_exec, file);
 749
 750                         debug_print("%s\n", cmd);
 751                         if ((status = execute_command_line(cmd, FALSE, NULL)) != 0)
 752                                 log_error(LOG_PROTOCOL, _("Learning failed; `%s` returned with status %d."),
 753                                                 cmd, status);
 754                         g_free(cmd);
 755                         g_free(file);
 756                         if (message_callback != NULL)
 757                                 message_callback(NULL, 0, 0, FALSE);
 758                 }
 759         } else if (msglist) {
 760                 GSList *cur = msglist;
 761                 MsgInfo *info;
 762                 int total = g_slist_length(msglist);
 763                 int done = 0;
 764                 gboolean some_correction = FALSE, some_no_correction = FALSE;
 765
 766                 if (message_callback != NULL)
 767                         message_callback(_("Bogofilter: learning from messages..."), total, 0, FALSE);
 768
 769                 for (cur = msglist; cur && status == 0; cur = cur->next) {
 770                         info = (MsgInfo *)cur->data;
 771                         if (spam)
 772                                 some_no_correction = TRUE;
 773                         else if (MSG_IS_SPAM(info->flags))
 774                                 /* correct bogofilter, this wasn't spam */
 775                                 some_correction = TRUE;
 776                         else
 777                                 some_no_correction = TRUE;
 778
 779                 }
 780
 781                 if (some_correction && some_no_correction) {
 782                         /* we potentially have to do different stuff for every mail */
 783                         for (cur = msglist; cur && status == 0; cur = cur->next) {
 784                                 info = (MsgInfo *)cur->data;
 785                                 file = procmsg_get_message_file(info);
 786
 787                                 if (spam)
 788                                         /* learn as spam */
 789                                         cmd = g_strdup_printf("%s -s -I '%s'", bogo_exec, file);
 790                                 else if (MSG_IS_SPAM(info->flags))
 791                                         /* correct bogofilter, this wasn't spam */
 792                                         cmd = g_strdup_printf("%s -Sn -I '%s'", bogo_exec, file);
 793                                 else
 794                                         /* learn as ham */
 795                                         cmd = g_strdup_printf("%s -n -I '%s'", bogo_exec, file);
 796
 797                                 debug_print("%s\n", cmd);
 798                                 if ((status = execute_command_line(cmd, FALSE, NULL)) != 0)
 799                                         log_error(LOG_PROTOCOL, _("Learning failed; `%s` returned with status %d."),
 800                                                         cmd, status);
 801
 802                                 g_free(cmd);
 803                                 g_free(file);
 804                                 done++;
 805                                 if (message_callback != NULL)
 806                                         message_callback(NULL, total, done, FALSE);
 807                         }
 808                 } else if (some_correction || some_no_correction) {
 809                         cur = msglist;
 810
 811                         gchar *bogo_args[4];
 812                         GPid bogo_pid;
 813                         gint bogo_stdin;
 814                         GError *error = NULL;
 815                         gboolean bogo_forked;
 816
 817                         bogo_args[0] = (gchar *)bogo_exec;
 818                         if (some_correction && !some_no_correction)
 819                                 bogo_args[1] = "-Sn";
 820                         else if (some_no_correction && !some_correction)
 821                                 bogo_args[1] = spam ? "-s":"-n";
 822                         bogo_args[2] = "-b";
 823                         bogo_args[3] = NULL;
 824                         debug_print("|%s %s %s ...\n", bogo_args[0], bogo_args[1], bogo_args[2]);
 825                         bogo_forked = g_spawn_async_with_pipes(
 826                                         NULL, bogo_args,NULL, G_SPAWN_SEARCH_PATH|G_SPAWN_DO_NOT_REAP_CHILD,
 827                                         NULL, NULL, &bogo_pid, &bogo_stdin,
 828                                         NULL, NULL, &error);
 829
 830                         while (bogo_forked && cur) {
 831                                 gchar *tmp = NULL;
 832                                 info = (MsgInfo *)cur->data;
 833                                 file = procmsg_get_message_file(info);
 834                                 if (file) {
 835                                         tmp = g_strdup_printf("%s\n",
 836                                                 file);
 837                                         write_all(bogo_stdin, tmp, strlen(tmp));
 838                                         g_free(tmp);
 839                                 }
 840                                 g_free(file);
 841                                 done++;
 842                                 if (message_callback != NULL)
 843                                         message_callback(NULL, total, done, FALSE);
 844                                 cur = cur->next;
 845                         }
 846                         if (bogo_forked) {
 847                                 close(bogo_stdin);
 848                                 waitpid(bogo_pid, &status, 0);
 849                                 if (!WIFEXITED(status))
 850                                         status = -1;
 851                                 else
 852                                         status = WEXITSTATUS(status);
 853                         }
 854                         if (!bogo_forked || status != 0) {
 855                                 log_error(LOG_PROTOCOL, _("Learning failed; `%s %s %s` returned with error:\n%s"),
 856                                                 bogo_args[0], bogo_args[1], bogo_args[2],
 857                                                 error ? error->message:_("Unknown error"));
 858                                 if (error)
 859                                         g_error_free(error);
 860                         }
 861
 862                 }
 863
 864                 if (message_callback != NULL)
 865                         message_callback(NULL, 0, 0, FALSE);
 866         }
 867         return 0;
 868 }
 869
 870 void bogofilter_save_config(void)
 871 {
 872         PrefFile *pfile;
 873         gchar *rcpath;
 874
 875         debug_print("Saving Bogofilter Page\n");
 876
 877         rcpath = g_strconcat(get_rc_dir(), G_DIR_SEPARATOR_S, COMMON_RC, NULL);
 878         pfile = prefs_write_open(rcpath);
 879         g_free(rcpath);
 880         if (!pfile || (prefs_set_block_label(pfile, "Bogofilter") < 0))
 881                 return;
 882
 883         if (prefs_write_param(param, pfile->fp) < 0) {
 884                 g_warning("Failed to write Bogofilter configuration to file");
 885                 prefs_file_close_revert(pfile);
 886                 return;
 887         }
 888         if (fprintf(pfile->fp, "\n") < 0) {
 889                 FILE_OP_ERROR(rcpath, "fprintf");
 890                 prefs_file_close_revert(pfile);
 891         } else
 892                 prefs_file_close(pfile);
 893 }
 894
 895 void bogofilter_set_message_callback(MessageCallback callback)
 896 {
 897         message_callback = callback;
 898 }
 899
 900 gint plugin_init(gchar **error)
 901 {
 902         gchar *rcpath;
 903
 904         hook_id = -1;
 905
 906         if (!check_plugin_version(MAKE_NUMERIC_VERSION(2,9,2,72),
 907                                 VERSION_NUMERIC, PLUGIN_NAME, error))
 908                 return -1;
 909
 910         prefs_set_default(param);
 911         rcpath = g_strconcat(get_rc_dir(), G_DIR_SEPARATOR_S, COMMON_RC, NULL);
 912         prefs_read_config(param, "Bogofilter", rcpath, NULL);
 913         g_free(rcpath);
 914
 915         bogofilter_gtk_init();
 916
 917         debug_print("Bogofilter plugin loaded\n");
 918
 919 #ifdef USE_PTHREAD
 920         bogofilter_start_thread();
 921 #endif
 922
 923         if (config.process_emails) {
 924                 bogofilter_register_hook();
 925         }
 926
 927         procmsg_register_spam_learner(bogofilter_learn);
 928         procmsg_spam_set_folder(config.save_folder, bogofilter_get_spam_folder);
 929
 930         return 0;
 931
 932 }
 933
 934 FolderItem *bogofilter_get_spam_folder(MsgInfo *msginfo)
 935 {
 936         FolderItem *item = folder_find_item_from_identifier(config.save_folder);
 937
 938         if (item || msginfo == NULL || msginfo->folder == NULL)
 939                 return item;
 940
 941         if (msginfo->folder->folder &&
 942             msginfo->folder->folder->account &&
 943             msginfo->folder->folder->account->set_trash_folder) {
 944                 item = folder_find_item_from_identifier(
 945                         msginfo->folder->folder->account->trash_folder);
 946         }
 947
 948         if (item == NULL &&
 949             msginfo->folder->folder &&
 950             msginfo->folder->folder->trash)
 951                 item = msginfo->folder->folder->trash;
 952
 953         if (item == NULL)
 954                 item = folder_get_default_trash();
 955
 956         debug_print("bogo spam dir: %s\n", folder_item_get_path(item));
 957         return item;
 958 }
 959
 960 gboolean plugin_done(void)
 961 {
 962         if (hook_id != (guint) -1) {
 963                 bogofilter_unregister_hook();
 964         }
 965 #ifdef USE_PTHREAD
 966         bogofilter_stop_thread();
 967 #endif
 968         g_free(config.save_folder);
 969         bogofilter_gtk_done();
 970         procmsg_unregister_spam_learner(bogofilter_learn);
 971         procmsg_spam_set_folder(NULL, NULL);
 972         debug_print("Bogofilter plugin unloaded\n");
 973         return TRUE;
 974 }
 975
 976 const gchar *plugin_name(void)
 977 {
 978         return PLUGIN_NAME;
 979 }
 980
 981 const gchar *plugin_desc(void)
 982 {
 983         return _("This plugin can check all messages that are received from an "
 984                  "IMAP, LOCAL or POP account for spam using Bogofilter. "
 985                  "You will need Bogofilter installed locally.\n"
 986                  "\n"
 987                  "Before Bogofilter can recognize spam messages, you have to "
 988                  "train it by marking a few hundred spam and ham messages "
 989                  "with the use of \"/Mark/Mark as spam\" and \"/Mark/Mark as "
 990                  "ham\".\n"
 991                  "\n"
 992                  "When a message is identified as spam it can be deleted or "
 993                  "saved in a specially designated folder.\n"
 994                  "\n"
 995                  "Options can be found in /Configuration/Preferences/Plugins/Bogofilter");
 996 }
 997
 998 const gchar *plugin_type(void)
 999 {
1000         return "GTK2";
1001 }
1002
1003 const gchar *plugin_licence(void)
1004 {
1005         return "GPL3+";
1006 }
1007
1008 const gchar *plugin_version(void)
1009 {
1010         return VERSION;
1011 }
1012
1013 struct PluginFeature *plugin_provides(void)
1014 {
1015         static struct PluginFeature features[] =
1016                 { {PLUGIN_FILTERING, N_("Spam detection")},
1017                   {PLUGIN_FILTERING, N_("Spam learning")},
1018                   {PLUGIN_NOTHING, NULL}};
1019         return features;
1020 }
1021
1022 void bogofilter_register_hook(void)
1023 {
1024         if (hook_id == (guint) -1)
1025                 hook_id = hooks_register_hook(MAIL_LISTFILTERING_HOOKLIST, mail_filtering_hook, NULL);
1026         if (hook_id == (guint) -1) {
1027                 g_warning("Failed to register mail filtering hook");
1028                 config.process_emails = FALSE;
1029         }
1030 }
1031
1032 void bogofilter_unregister_hook(void)
1033 {
1034         if (hook_id != (guint) -1) {
1035                 hooks_unregister_hook(MAIL_LISTFILTERING_HOOKLIST, hook_id);
1036         }
1037         hook_id = -1;
1038 }