src/plugins/bogofilter/bogofilter.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 1999-2006 Hiroyuki Yamamoto and the Claws Mail Team
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  */
  19
  20 #ifdef HAVE_CONFIG_H
  21 #  include "config.h"
  22 #endif
  23
  24 #include "defs.h"
  25
  26 #include <sys/types.h>
  27 #include <sys/wait.h>
  28 #include <errno.h>
  29
  30 #include <glib.h>
  31 #include <glib/gi18n.h>
  32
  33 #if HAVE_LOCALE_H
  34 #  include <locale.h>
  35 #endif
  36
  37 #include "common/sylpheed.h"
  38 #include "common/version.h"
  39 #include "plugin.h"
  40 #include "common/utils.h"
  41 #include "hooks.h"
  42 #include "procmsg.h"
  43 #include "folder.h"
  44 #include "prefs.h"
  45 #include "prefs_gtk.h"
  46
  47 #include "bogofilter.h"
  48 #include "inc.h"
  49 #include "log.h"
  50 #include "prefs_common.h"
  51 #include "alertpanel.h"
  52
  53 #ifdef HAVE_SYSEXITS_H
  54 #include <sysexits.h>
  55 #endif
  56 #ifdef HAVE_ERRNO_H
  57 #include <errno.h>
  58 #endif
  59 #ifdef HAVE_SYS_ERRNO_H
  60 #include <sys/errno.h>
  61 #endif
  62 #ifdef HAVE_TIME_H
  63 #include <time.h>
  64 #endif
  65 #ifdef HAVE_SYS_TIME_H
  66 #include <sys/time.h>
  67 #endif
  68 #ifdef HAVE_SIGNAL_H
  69 #include <signal.h>
  70 #endif
  71 #ifdef HAVE_PWD_H
  72 #include <pwd.h>
  73 #endif
  74
  75 #define MAILS_PER_BATCH 20
  76
  77 static guint hook_id = -1;
  78 static MessageCallback message_callback;
  79
  80 static BogofilterConfig config;
  81
  82 static PrefParam param[] = {
  83         {"process_emails", "TRUE", &config.process_emails, P_BOOL,
  84          NULL, NULL, NULL},
  85         {"receive_spam", "TRUE", &config.receive_spam, P_BOOL,
  86          NULL, NULL, NULL},
  87         {"save_folder", NULL, &config.save_folder, P_STRING,
  88          NULL, NULL, NULL},
  89         {"max_size", "250", &config.max_size, P_INT,
  90          NULL, NULL, NULL},
  91         {"bogopath", "bogofilter", &config.bogopath, P_STRING,
  92          NULL, NULL, NULL},
  93
  94         {NULL, NULL, NULL, P_OTHER, NULL, NULL, NULL}
  95 };
  96
  97 /*
  98  * Helper function for spawn_with_input() - write an entire
  99  * string to a fd.
 100  */
 101 static gboolean
 102 write_all (int         fd,
 103            const char *buf,
 104            gsize       to_write)
 105 {
 106   while (to_write > 0)
 107     {
 108       gssize count = write (fd, buf, to_write);
 109       if (count < 0)
 110         {
 111           if (errno != EINTR)
 112             return FALSE;
 113         }
 114       else
 115         {
 116           to_write -= count;
 117           buf += count;
 118         }
 119     }
 120
 121   return TRUE;
 122 }
 123
 124 static gboolean mail_filtering_hook(gpointer source, gpointer data)
 125 {
 126         MailFilteringData *mail_filtering_data = (MailFilteringData *) source;
 127         MsgInfo *msginfo = mail_filtering_data->msginfo;
 128         GSList *msglist = mail_filtering_data->msglist;
 129         GSList *cur = NULL;
 130         static gboolean warned_error = FALSE;
 131         gchar *file = NULL;
 132         int status = 0;
 133         gchar *bogo_exec = (config.bogopath && *config.bogopath) ? config.bogopath:"bogofilter";
 134         int total = 0, curnum = 0;
 135         GSList *spams = NULL;
 136         gchar buf[BUFSIZ];
 137
 138         gchar *bogo_args[4];
 139         GPid bogo_pid;
 140         gint bogo_stdin, bogo_stdout;
 141         GError *error = NULL;
 142         gboolean bogo_forked;
 143
 144         if (!config.process_emails) {
 145                 return FALSE;
 146         }
 147
 148         if (msglist == NULL && msginfo != NULL) {
 149                 g_warning("wrong call to bogofilter mail_filtering_hook");
 150                 return FALSE;
 151         }
 152
 153         total = g_slist_length(msglist);
 154         if (message_callback != NULL)
 155                 message_callback(_("Bogofilter: filtering messages..."), total, 0);
 156
 157         bogo_args[0] = bogo_exec;
 158         bogo_args[1] = "-T";
 159         bogo_args[2] = "-b";
 160         bogo_args[3] = NULL;
 161
 162         bogo_forked = g_spawn_async_with_pipes(
 163                         NULL, bogo_args,NULL, G_SPAWN_SEARCH_PATH|G_SPAWN_DO_NOT_REAP_CHILD,
 164                         NULL, NULL, &bogo_pid, &bogo_stdin,
 165                         &bogo_stdout, NULL, &error);
 166
 167         if (bogo_forked == FALSE) {
 168                 g_warning("%s\n", error ? error->message:"ERROR???");
 169                 g_error_free(error);
 170                 error = NULL;
 171                 status = -1;
 172         } else {
 173                 for (cur = msglist; cur; cur = cur->next) {
 174                         msginfo = (MsgInfo *)cur->data;
 175                         debug_print("Filtering message %d (%d/%d)\n", msginfo->msgnum, curnum, total);
 176
 177                         if (message_callback != NULL)
 178                                 message_callback(NULL, total, curnum++);
 179
 180                         file = procmsg_get_message_file(msginfo);
 181
 182                         if (file) {
 183                                 gchar *tmp = g_strdup_printf("%s\n",file);
 184                                 write_all(bogo_stdin, tmp, strlen(tmp));
 185                                 g_free(tmp);
 186                                 memset(buf, 0, sizeof(buf));
 187                                 if (read(bogo_stdout, buf, sizeof(buf)-1) < 0) {
 188                                         g_warning("bogofilter short read\n");
 189                                         debug_print("message %d is ham\n", msginfo->msgnum);
 190                                         procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 191                                         mail_filtering_data->unfiltered = g_slist_prepend(
 192                                                 mail_filtering_data->unfiltered, msginfo);
 193                                 } else {
 194                                         gchar **parts = NULL;
 195                                         if (strchr(buf, '/')) {
 196                                                 tmp = strrchr(buf, '/')+1;
 197                                         } else {
 198                                                 tmp = buf;
 199                                         }
 200                                         parts = g_strsplit(tmp, " ", 0);
 201                                         debug_print("read %s\n", buf);
 202                                         if (parts && parts[0] && parts[1] && *parts[1] == 'S') {
 203                                                 debug_print("message %d is spam\n", msginfo->msgnum);
 204                                                 procmsg_msginfo_set_flags(msginfo, MSG_SPAM, 0);
 205
 206                                                 if (config.receive_spam) {
 207                                                         procmsg_msginfo_unset_flags(msginfo, ~0, 0);
 208                                                         procmsg_msginfo_set_flags(msginfo, MSG_SPAM, 0);
 209                                                         spams = g_slist_prepend(spams, msginfo);
 210                                                 } else {
 211                                                         folder_item_remove_msg(msginfo->folder, msginfo->msgnum);
 212                                                 }
 213
 214                                                 mail_filtering_data->filtered = g_slist_prepend(
 215                                                         mail_filtering_data->filtered, msginfo);
 216                                         } else {
 217                                                 debug_print("message %d is ham\n", msginfo->msgnum);
 218                                                 procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 219                                                 mail_filtering_data->unfiltered = g_slist_prepend(
 220                                                         mail_filtering_data->unfiltered, msginfo);
 221                                         }
 222                                         g_strfreev(parts);
 223                                 }
 224                                 g_free(file);
 225                         } else {
 226                                 procmsg_msginfo_unset_flags(msginfo, MSG_SPAM, 0);
 227                                 mail_filtering_data->unfiltered = g_slist_prepend(
 228                                         mail_filtering_data->unfiltered, msginfo);
 229                         }
 230                 }
 231         }
 232
 233         if (status != -1) {
 234                 close(bogo_stdout);
 235                 close(bogo_stdin);
 236                 waitpid(bogo_pid, &status, 0);
 237                 if (!WIFEXITED(status))
 238                         status = -1;
 239                 else
 240                         status = WEXITSTATUS(status);
 241         }
 242
 243         if (status < 0 || status > 2) { /* I/O or other errors */
 244                 gchar *msg = NULL;
 245
 246                 if (status == 3)
 247                         msg =  g_strdup_printf(_("The Bogofilter plugin couldn't filter "
 248                                            "a message. The probable cause of the "
 249                                            "error is that it didn't learn from any mail.\n"
 250                                            "Use \"/Mark/Mark as spam\" and \"/Mark/Mark as "
 251                                            "ham\" to train Bogofilter with a few hundred "
 252                                            "spam and ham messages."));
 253                 else
 254                         msg =  g_strdup_printf(_("The Bogofilter plugin couldn't filter "
 255                                            "a message. the command `%s %s %s` couldn't be run."),
 256                                            bogo_args[0], bogo_args[1], bogo_args[2]);
 257                 if (!prefs_common.no_recv_err_panel) {
 258                         if (!warned_error) {
 259                                 alertpanel_error(msg);
 260                         }
 261                         warned_error = TRUE;
 262                 } else {
 263                         gchar *tmp = g_strdup_printf("%s\n", msg);
 264                         log_error(tmp);
 265                         g_free(tmp);
 266                 }
 267                 g_free(msg);
 268         }
 269         if (status < 0 || status > 2) {
 270                 g_slist_free(mail_filtering_data->filtered);
 271                 g_slist_free(mail_filtering_data->unfiltered);
 272                 g_slist_free(spams);
 273                 mail_filtering_data->filtered = NULL;
 274                 mail_filtering_data->unfiltered = NULL;
 275         } else if (config.receive_spam && spams) {
 276                 FolderItem *save_folder;
 277
 278                 if ((!config.save_folder) ||
 279                     (config.save_folder[0] == '\0') ||
 280                     ((save_folder = folder_find_item_from_identifier(config.save_folder)) == NULL))
 281                         save_folder = folder_get_default_trash();
 282                 if (save_folder) {
 283                         for (cur = spams; cur; cur = cur->next) {
 284                                 msginfo = (MsgInfo *)cur->data;
 285                                 msginfo->is_move = TRUE;
 286                                 msginfo->to_filter_folder = save_folder;
 287                         }
 288                 }
 289         }
 290
 291         if (message_callback != NULL)
 292                 message_callback(NULL, 0, 0);
 293         mail_filtering_data->filtered = g_slist_reverse(
 294                 mail_filtering_data->filtered);
 295         mail_filtering_data->unfiltered = g_slist_reverse(
 296                 mail_filtering_data->unfiltered);
 297
 298         return FALSE;
 299 }
 300
 301 BogofilterConfig *bogofilter_get_config(void)
 302 {
 303         return &config;
 304 }
 305
 306 int bogofilter_learn(MsgInfo *msginfo, GSList *msglist, gboolean spam)
 307 {
 308         gchar *cmd = NULL;
 309         gchar *file = NULL;
 310         const gchar *bogo_exec = (config.bogopath && *config.bogopath) ? config.bogopath:"bogofilter";
 311         gint status = 0;
 312         if (msginfo == NULL && msglist == NULL) {
 313                 return -1;
 314         }
 315
 316         if (msginfo) {
 317                 file = procmsg_get_message_file(msginfo);
 318                 if (file == NULL) {
 319                         return -1;
 320                 } else {
 321                         if (message_callback != NULL)
 322                                 message_callback(_("Bogofilter: learning from message..."), 0, 0);
 323                         if (spam)
 324                                 /* learn as spam */
 325                                 cmd = g_strdup_printf("%s -s -I '%s'", bogo_exec, file);
 326                         else if (MSG_IS_SPAM(msginfo->flags))
 327                                 /* correct bogofilter, this wasn't spam */
 328                                 cmd = g_strdup_printf("%s -Sn -I '%s'", bogo_exec, file);
 329                         else
 330                                 /* learn as ham */
 331                                 cmd = g_strdup_printf("%s -n -I '%s'", bogo_exec, file);
 332                         if ((status = execute_command_line(cmd, FALSE)) != 0)
 333                                 log_error(_("Learning failed; `%s` returned with status %d."),
 334                                                 cmd, status);
 335                         g_free(cmd);
 336                         g_free(file);
 337                         if (message_callback != NULL)
 338                                 message_callback(NULL, 0, 0);
 339                         return 0;
 340                 }
 341         }
 342         if (msglist) {
 343                 GSList *cur = msglist;
 344                 MsgInfo *info;
 345                 int total = g_slist_length(msglist);
 346                 int done = 0;
 347                 gboolean some_correction = FALSE, some_no_correction = FALSE;
 348
 349                 if (message_callback != NULL)
 350                         message_callback(_("Bogofilter: learning from messages..."), total, 0);
 351
 352                 for (cur = msglist; cur && status == 0; cur = cur->next) {
 353                         info = (MsgInfo *)cur->data;
 354                         if (spam)
 355                                 some_no_correction = TRUE;
 356                         else if (MSG_IS_SPAM(info->flags))
 357                                 /* correct bogofilter, this wasn't spam */
 358                                 some_correction = TRUE;
 359                         else
 360                                 some_no_correction = TRUE;
 361
 362                 }
 363
 364                 if (some_correction && some_no_correction) {
 365                         /* we potentially have to do different stuff for every mail */
 366                         for (cur = msglist; cur && status == 0; cur = cur->next) {
 367                                 info = (MsgInfo *)cur->data;
 368                                 file = procmsg_get_message_file(info);
 369
 370                                 if (spam)
 371                                         /* learn as spam */
 372                                         cmd = g_strdup_printf("%s -s -I '%s'", bogo_exec, file);
 373                                 else if (MSG_IS_SPAM(info->flags))
 374                                         /* correct bogofilter, this wasn't spam */
 375                                         cmd = g_strdup_printf("%s -Sn -I '%s'", bogo_exec, file);
 376                                 else
 377                                         /* learn as ham */
 378                                         cmd = g_strdup_printf("%s -n -I '%s'", bogo_exec, file);
 379
 380                                 if ((status = execute_command_line(cmd, FALSE)) != 0)
 381                                         log_error(_("Learning failed; `%s` returned with status %d."),
 382                                                         cmd, status);
 383
 384                                 g_free(cmd);
 385                                 g_free(file);
 386                                 done++;
 387                                 if (message_callback != NULL)
 388                                         message_callback(NULL, total, done);
 389                         }
 390                 } else if (some_correction || some_no_correction) {
 391                         cur = msglist;
 392
 393                         gchar *bogo_args[4];
 394                         GPid bogo_pid;
 395                         gint bogo_stdin;
 396                         GError *error = NULL;
 397                         gboolean bogo_forked;
 398
 399                         bogo_args[0] = (gchar *)bogo_exec;
 400                         if (some_correction && !some_no_correction)
 401                                 bogo_args[1] = "-Sn";
 402                         else if (some_no_correction && !some_correction)
 403                                 bogo_args[1] = spam ? "-s":"-n";
 404                         bogo_args[2] = "-b";
 405                         bogo_args[3] = NULL;
 406
 407                         bogo_forked = g_spawn_async_with_pipes(
 408                                         NULL, bogo_args,NULL, G_SPAWN_SEARCH_PATH|G_SPAWN_DO_NOT_REAP_CHILD,
 409                                         NULL, NULL, &bogo_pid, &bogo_stdin,
 410                                         NULL, NULL, &error);
 411
 412                         while (bogo_forked && cur) {
 413                                 gchar *tmp = NULL;
 414                                 info = (MsgInfo *)cur->data;
 415                                 file = procmsg_get_message_file(info);
 416                                 if (file) {
 417                                         tmp = g_strdup_printf("%s\n",
 418                                                 file);
 419                                         write_all(bogo_stdin, tmp, strlen(tmp));
 420                                         g_free(tmp);
 421                                 }
 422                                 g_free(file);
 423                                 done++;
 424                                 if (message_callback != NULL)
 425                                         message_callback(NULL, total, done);
 426                                 cur = cur->next;
 427                         }
 428                         if (bogo_forked) {
 429                                 close(bogo_stdin);
 430                                 waitpid(bogo_pid, &status, 0);
 431                                 if (!WIFEXITED(status))
 432                                         status = -1;
 433                                 else
 434                                         status = WEXITSTATUS(status);
 435                         }
 436                         if (!bogo_forked || status != 0) {
 437                                 log_error(_("Learning failed; `%s %s %s` returned with error:\n%s"),
 438                                                 bogo_args[0], bogo_args[1], bogo_args[2],
 439                                                 error ? error->message:_("Unknown error"));
 440                                 if (error)
 441                                         g_error_free(error);
 442                         }
 443
 444                 }
 445
 446                 if (message_callback != NULL)
 447                         message_callback(NULL, 0, 0);
 448                 return 0;
 449         }
 450         return -1;
 451 }
 452
 453 void bogofilter_save_config(void)
 454 {
 455         PrefFile *pfile;
 456         gchar *rcpath;
 457
 458         debug_print("Saving Bogofilter Page\n");
 459
 460         rcpath = g_strconcat(get_rc_dir(), G_DIR_SEPARATOR_S, COMMON_RC, NULL);
 461         pfile = prefs_write_open(rcpath);
 462         g_free(rcpath);
 463         if (!pfile || (prefs_set_block_label(pfile, "Bogofilter") < 0))
 464                 return;
 465
 466         if (prefs_write_param(param, pfile->fp) < 0) {
 467                 g_warning("Failed to write Bogofilter configuration to file\n");
 468                 prefs_file_close_revert(pfile);
 469                 return;
 470         }
 471         fprintf(pfile->fp, "\n");
 472
 473         prefs_file_close(pfile);
 474 }
 475
 476 void bogofilter_set_message_callback(MessageCallback callback)
 477 {
 478         message_callback = callback;
 479 }
 480
 481 gint plugin_init(gchar **error)
 482 {
 483         gchar *rcpath;
 484
 485         hook_id = -1;
 486
 487         if ((sylpheed_get_version() > VERSION_NUMERIC)) {
 488                 *error = g_strdup(_("Your version of Claws Mail is newer than the version the Bogofilter plugin was built with"));
 489                 return -1;
 490         }
 491
 492         if ((sylpheed_get_version() < MAKE_NUMERIC_VERSION(0, 9, 3, 86))) {
 493                 *error = g_strdup(_("Your version of Claws Mail is too old for the Bogofilter plugin"));
 494                 return -1;
 495         }
 496
 497         prefs_set_default(param);
 498         rcpath = g_strconcat(get_rc_dir(), G_DIR_SEPARATOR_S, COMMON_RC, NULL);
 499         prefs_read_config(param, "Bogofilter", rcpath, NULL);
 500         g_free(rcpath);
 501
 502         bogofilter_gtk_init();
 503
 504         debug_print("Bogofilter plugin loaded\n");
 505
 506         if (config.process_emails) {
 507                 bogofilter_register_hook();
 508         }
 509
 510         procmsg_register_spam_learner(bogofilter_learn);
 511         procmsg_spam_set_folder(config.save_folder);
 512
 513         return 0;
 514
 515 }
 516
 517 void plugin_done(void)
 518 {
 519         if (hook_id != -1) {
 520                 bogofilter_unregister_hook();
 521         }
 522         g_free(config.save_folder);
 523         bogofilter_gtk_done();
 524         procmsg_unregister_spam_learner(bogofilter_learn);
 525         procmsg_spam_set_folder(NULL);
 526         debug_print("Bogofilter plugin unloaded\n");
 527 }
 528
 529 const gchar *plugin_name(void)
 530 {
 531         return _("Bogofilter");
 532 }
 533
 534 const gchar *plugin_desc(void)
 535 {
 536         return _("This plugin can check all messages that are received from an "
 537                  "IMAP, LOCAL or POP account for spam using Bogofilter. "
 538                  "You will need Bogofilter installed locally.\n "
 539                  "\n"
 540                  "Before Bogofilter can recognize spam messages, you have to "
 541                  "train it by marking a few hundred spam and ham messages. "
 542                  "Use \"/Mark/Mark as Spam\" and \"/Mark/Mark as ham\" to "
 543                  "train Bogofilter.\n"
 544                  "\n"
 545                  "When a message is identified as spam it can be deleted or "
 546                  "saved in a specially designated folder.\n"
 547                  "\n"
 548                  "Options can be found in /Configuration/Preferences/Plugins/Bogofilter");
 549 }
 550
 551 const gchar *plugin_type(void)
 552 {
 553         return "GTK2";
 554 }
 555
 556 const gchar *plugin_licence(void)
 557 {
 558         return "GPL";
 559 }
 560
 561 const gchar *plugin_version(void)
 562 {
 563         return VERSION;
 564 }
 565
 566 struct PluginFeature *plugin_provides(void)
 567 {
 568         static struct PluginFeature features[] =
 569                 { {PLUGIN_FILTERING, N_("Spam detection")},
 570                   {PLUGIN_FILTERING, N_("Spam learning")},
 571                   {PLUGIN_NOTHING, NULL}};
 572         return features;
 573 }
 574
 575 void bogofilter_register_hook(void)
 576 {
 577         if (hook_id == -1)
 578                 hook_id = hooks_register_hook(MAIL_LISTFILTERING_HOOKLIST, mail_filtering_hook, NULL);
 579         if (hook_id == -1) {
 580                 g_warning("Failed to register mail filtering hook");
 581                 config.process_emails = FALSE;
 582         }
 583 }
 584
 585 void bogofilter_unregister_hook(void)
 586 {
 587         if (hook_id != -1) {
 588                 hooks_unregister_hook(MAIL_LISTFILTERING_HOOKLIST, hook_id);
 589         }
 590         hook_id = -1;
 591 }