src/advsearch.c

   1 /*
   2  * Claws Mail -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 2012-2014 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 3 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, see <http://www.gnu.org/licenses/>.
  17  *
  18  */
  19
  20 #ifdef HAVE_CONFIG_H
  21 # include "config.h"
  22 # include "claws-features.h"
  23 #endif
  24
  25 #include "advsearch.h"
  26
  27 #include <glib.h>
  28 #include <ctype.h>
  29
  30 #include "matcher.h"
  31 #include "matcher_parser.h"
  32 #include "utils.h"
  33 #include "prefs_common.h"
  34
  35 struct _AdvancedSearch {
  36         struct {
  37                 AdvancedSearchType       type;
  38                 gchar                   *matchstring;
  39         } request;
  40
  41         MatcherList                     *predicate;
  42         gboolean                         is_fast;
  43         gboolean                         search_aborted;
  44
  45         struct {
  46                 gboolean (*cb)(gpointer data, guint at, guint matched, guint total);
  47                 gpointer data;
  48         } on_progress_cb;
  49         struct {
  50                 void (*cb)(gpointer data);
  51                 gpointer data;
  52         } on_error_cb;
  53 };
  54
  55 void advsearch_set_on_progress_cb(AdvancedSearch *search, gboolean (*cb)(gpointer, guint, guint, guint), gpointer data)
  56 {
  57         search->on_progress_cb.cb = cb;
  58         search->on_progress_cb.data = data;
  59 }
  60
  61 void advsearch_set_on_error_cb(AdvancedSearch* search, void (*cb)(gpointer data), gpointer data)
  62 {
  63         search->on_error_cb.cb = cb;
  64         search->on_error_cb.data = data;
  65 }
  66
  67 static void prepare_matcher(AdvancedSearch *search);
  68 static gboolean search_impl(MsgInfoList **messages, AdvancedSearch* search,
  69                             FolderItem* folderItem, gboolean recursive);
  70
  71 // --------------------------
  72
  73 AdvancedSearch* advsearch_new()
  74 {
  75         AdvancedSearch *result;
  76
  77         result = g_new0(AdvancedSearch, 1);
  78
  79         return result;
  80 }
  81
  82 void advsearch_free(AdvancedSearch *search)
  83 {
  84         if (search->predicate != NULL)
  85                 matcherlist_free(search->predicate);
  86
  87         g_free(search->request.matchstring);
  88         g_free(search);
  89 }
  90
  91 void advsearch_set(AdvancedSearch *search, AdvancedSearchType type, const gchar *matchstring)
  92 {
  93         cm_return_if_fail(search != NULL);
  94
  95         search->request.type = type;
  96
  97         g_free(search->request.matchstring);
  98         search->request.matchstring = g_strdup(matchstring);
  99
 100         prepare_matcher(search);
 101 }
 102
 103 gboolean advsearch_is_fast(AdvancedSearch *search)
 104 {
 105         cm_return_val_if_fail(search != NULL, FALSE);
 106
 107         return search->is_fast;
 108 }
 109
 110 gboolean advsearch_has_proper_predicate(AdvancedSearch *search)
 111 {
 112         cm_return_val_if_fail(search != NULL, FALSE);
 113
 114         return search->predicate != NULL;
 115 }
 116
 117 gboolean advsearch_search_msgs_in_folders(AdvancedSearch* search, MsgInfoList **messages,
 118                                           FolderItem* folderItem, gboolean recursive)
 119 {
 120         if (search == NULL || search->predicate == NULL)
 121                 return FALSE;
 122
 123         search->search_aborted = FALSE;
 124         return search_impl(messages, search, folderItem, recursive);
 125 }
 126
 127 void advsearch_abort(AdvancedSearch *search)
 128 {
 129         search->search_aborted = TRUE;
 130 }
 131
 132 gchar *advsearch_expand_search_string(const gchar *search_string)
 133 {
 134         int i = 0;
 135         gchar term_char, save_char;
 136         gchar *cmd_start, *cmd_end;
 137         GString *matcherstr;
 138         gchar *returnstr = NULL;
 139         gchar *copy_str;
 140         gboolean casesens, dontmatch, regex;
 141         /* list of allowed pattern abbreviations */
 142         struct {
 143                 gchar           *abbreviated;   /* abbreviation */
 144                 gchar           *command;       /* actual matcher command */
 145                 gint            numparams;      /* number of params for cmd */
 146                 gboolean        qualifier;      /* do we append stringmatch operations */
 147                 gboolean        quotes;         /* do we need quotes */
 148         }
 149         cmds[] = {
 150                 { "a",  "all",                          0,      FALSE,  FALSE },
 151                 { "ag", "age_greater",                  1,      FALSE,  FALSE },
 152                 { "al", "age_lower",                    1,      FALSE,  FALSE },
 153                 { "agh","age_greater_hours",            1,      FALSE,  FALSE },
 154                 { "alh","age_lower_hours",              1,      FALSE,  FALSE },
 155                 { "b",  "body_part",                    1,      TRUE,   TRUE  },
 156                 { "B",  "message",                      1,      TRUE,   TRUE  },
 157                 { "c",  "cc",                           1,      TRUE,   TRUE  },
 158                 { "C",  "to_or_cc",                     1,      TRUE,   TRUE  },
 159                 { "D",  "deleted",                      0,      FALSE,  FALSE },
 160                 { "da", "date_after",                   1,      FALSE,  TRUE  },
 161                 { "db", "date_before",                  1,      FALSE,  TRUE  },
 162                 { "e",  "header \"Sender\"",            1,      TRUE,   TRUE  },
 163                 { "E",  "execute",                      1,      FALSE,  TRUE  },
 164                 { "f",  "from",                         1,      TRUE,   TRUE  },
 165                 { "F",  "forwarded",                    0,      FALSE,  FALSE },
 166                 { "h",  "headers_part",                 1,      TRUE,   TRUE  },
 167                 { "H",  "headers_cont",                 1,      TRUE,   TRUE  },
 168                 { "ha", "has_attachments",              0,      FALSE,  FALSE },
 169                 { "i",  "messageid",                    1,      TRUE,   TRUE  },
 170                 { "I",  "inreplyto",                    1,      TRUE,   TRUE  },
 171                 { "k",  "colorlabel",                   1,      FALSE,  FALSE },
 172                 { "L",  "locked",                       0,      FALSE,  FALSE },
 173                 { "n",  "newsgroups",                   1,      TRUE,   TRUE  },
 174                 { "N",  "new",                          0,      FALSE,  FALSE },
 175                 { "O",  "~new",                         0,      FALSE,  FALSE },
 176                 { "r",  "replied",                      0,      FALSE,  FALSE },
 177                 { "R",  "~unread",                      0,      FALSE,  FALSE },
 178                 { "s",  "subject",                      1,      TRUE,   TRUE  },
 179                 { "se", "score_equal",                  1,      FALSE,  FALSE },
 180                 { "sg", "score_greater",                1,      FALSE,  FALSE },
 181                 { "sl", "score_lower",                  1,      FALSE,  FALSE },
 182                 { "Se", "size_equal",                   1,      FALSE,  FALSE },
 183                 { "Sg", "size_greater",                 1,      FALSE,  FALSE },
 184                 { "Ss", "size_smaller",                 1,      FALSE,  FALSE },
 185                 { "t",  "to",                           1,      TRUE,   TRUE  },
 186                 { "tg", "tag",                          1,      TRUE,   TRUE  },
 187                 { "T",  "marked",                       0,      FALSE,  FALSE },
 188                 { "U",  "unread",                       0,      FALSE,  FALSE },
 189                 { "x",  "references",                   1,      TRUE,   TRUE  },
 190                 { "X",  "test",                         1,      FALSE,  FALSE },
 191                 { "y",  "header \"X-Label\"",           1,      TRUE,   TRUE  },
 192                 { "&",  "&",                            0,      FALSE,  FALSE },
 193                 { "|",  "|",                            0,      FALSE,  FALSE },
 194                 { "p",  "partial",                      0,      FALSE,  FALSE },
 195                 { NULL, NULL,                           0,      FALSE,  FALSE }
 196         };
 197
 198         if (search_string == NULL)
 199                 return NULL;
 200
 201         copy_str = g_strdup(search_string);
 202
 203         matcherstr = g_string_sized_new(16);
 204         cmd_start = copy_str;
 205         while (cmd_start && *cmd_start) {
 206                 /* skip all white spaces */
 207                 while (*cmd_start && isspace((guchar)*cmd_start))
 208                         cmd_start++;
 209                 cmd_end = cmd_start;
 210
 211                 /* extract a command */
 212                 while (*cmd_end && !isspace((guchar)*cmd_end))
 213                         cmd_end++;
 214
 215                 /* save character */
 216                 save_char = *cmd_end;
 217                 *cmd_end = '\0';
 218
 219                 dontmatch = FALSE;
 220                 casesens = FALSE;
 221                 regex = FALSE;
 222
 223                 /* ~ and ! mean logical NOT */
 224                 if (*cmd_start == '~' || *cmd_start == '!')
 225                 {
 226                         dontmatch = TRUE;
 227                         cmd_start++;
 228                 }
 229                 /* % means case sensitive match */
 230                 if (*cmd_start == '%')
 231                 {
 232                         casesens = TRUE;
 233                         cmd_start++;
 234                 }
 235                 /* # means regex match */
 236                 if (*cmd_start == '#') {
 237                         regex = TRUE;
 238                         cmd_start++;
 239                 }
 240
 241                 /* find matching abbreviation */
 242                 for (i = 0; cmds[i].command; i++) {
 243                         if (!strcmp(cmd_start, cmds[i].abbreviated)) {
 244                                 /* restore character */
 245                                 *cmd_end = save_char;
 246
 247                                 /* copy command */
 248                                 if (matcherstr->len > 0) {
 249                                         g_string_append(matcherstr, " ");
 250                                 }
 251                                 if (dontmatch)
 252                                         g_string_append(matcherstr, "~");
 253                                 g_string_append(matcherstr, cmds[i].command);
 254                                 g_string_append(matcherstr, " ");
 255
 256                                 /* stop if no params required */
 257                                 if (cmds[i].numparams == 0)
 258                                         break;
 259
 260                                 /* extract a parameter, allow quotes */
 261                                 while (*cmd_end && isspace((guchar)*cmd_end))
 262                                         cmd_end++;
 263
 264                                 cmd_start = cmd_end;
 265                                 if (*cmd_start == '"') {
 266                                         term_char = '"';
 267                                         cmd_end++;
 268                                 }
 269                                 else
 270                                         term_char = ' ';
 271
 272                                 /* extract actual parameter */
 273                                 while ((*cmd_end) && (*cmd_end != term_char))
 274                                         cmd_end++;
 275
 276                                 if (*cmd_end == '"')
 277                                         cmd_end++;
 278
 279                                 save_char = *cmd_end;
 280                                 *cmd_end = '\0';
 281
 282                                 if (cmds[i].qualifier) {
 283                                         if (casesens)
 284                                                 g_string_append(matcherstr, regex ? "regexp " : "match ");
 285                                         else
 286                                                 g_string_append(matcherstr, regex ? "regexpcase " : "matchcase ");
 287                                 }
 288
 289                                 /* do we need to add quotes ? */
 290                                 if (cmds[i].quotes && term_char != '"')
 291                                         g_string_append(matcherstr, "\"");
 292
 293                                 /* copy actual parameter */
 294                                 g_string_append(matcherstr, cmd_start);
 295
 296                                 /* do we need to add quotes ? */
 297                                 if (cmds[i].quotes && term_char != '"')
 298                                         g_string_append(matcherstr, "\"");
 299
 300                                 /* restore original character */
 301                                 *cmd_end = save_char;
 302
 303                                 break;
 304                         }
 305                 }
 306
 307                 if (*cmd_end)
 308                         cmd_end++;
 309                 cmd_start = cmd_end;
 310         }
 311
 312         g_free(copy_str);
 313
 314         /* return search string if no match is found to allow
 315            all available filtering expressions in advanced search */
 316         if (matcherstr->len > 0) {
 317                 returnstr = g_string_free(matcherstr, FALSE);
 318         } else {
 319                 returnstr = g_strdup(search_string);
 320                 g_string_free(matcherstr, TRUE);
 321         }
 322         return returnstr;
 323 }
 324
 325 static void prepare_matcher_extended(AdvancedSearch *search)
 326 {
 327         gchar *newstr = advsearch_expand_search_string(search->request.matchstring);
 328
 329         if (newstr && newstr[0] != '\0') {
 330                 search->predicate = matcher_parser_get_cond(newstr, &search->is_fast);
 331                 g_free(newstr);
 332         }
 333 }
 334
 335 #define debug_matcher_list(prefix, list)                                        \
 336 do {                                                                            \
 337         gchar *str = list ? matcherlist_to_string(list) : g_strdup("(NULL)");   \
 338                                                                                 \
 339         debug_print("%s: %s\n", prefix, str);                                   \
 340                                                                                 \
 341         g_free(str);                                                            \
 342 } while(0)
 343
 344 static void prepare_matcher_tag(AdvancedSearch *search)
 345 {
 346         gchar **words = search->request.matchstring
 347                         ? g_strsplit(search->request.matchstring, " ", -1)
 348                         : NULL;
 349         gint i = 0;
 350
 351         if (search->predicate == NULL) {
 352                 search->predicate = g_new0(MatcherList, 1);
 353                 search->predicate->bool_and = FALSE;
 354                 search->is_fast = TRUE;
 355         }
 356
 357         while (words && words[i] && *words[i]) {
 358                 MatcherProp *matcher;
 359
 360                 g_strstrip(words[i]);
 361
 362                 matcher = matcherprop_new(MATCHCRITERIA_TAG, NULL,
 363                                           MATCHTYPE_MATCHCASE, words[i], 0);
 364
 365                 search->predicate->matchers = g_slist_prepend(search->predicate->matchers, matcher);
 366
 367                 i++;
 368         }
 369         g_strfreev(words);
 370 }
 371
 372 static void prepare_matcher_header(AdvancedSearch *search, gint match_header)
 373 {
 374         MatcherProp *matcher;
 375
 376         if (search->predicate == NULL) {
 377                 search->predicate = g_new0(MatcherList, 1);
 378                 search->predicate->bool_and = FALSE;
 379                 search->is_fast = TRUE;
 380         }
 381
 382         matcher = matcherprop_new(match_header, NULL, MATCHTYPE_MATCHCASE,
 383                         search->request.matchstring, 0);
 384
 385         search->predicate->matchers = g_slist_prepend(search->predicate->matchers, matcher);
 386 }
 387
 388 static void prepare_matcher_mixed(AdvancedSearch *search)
 389 {
 390         prepare_matcher_tag(search);
 391         debug_matcher_list("tag matcher list", search->predicate);
 392
 393         /* we want an OR search */
 394         if (search->predicate)
 395                 search->predicate->bool_and = FALSE;
 396
 397         prepare_matcher_header(search, MATCHCRITERIA_SUBJECT);
 398         debug_matcher_list("tag + subject matcher list", search->predicate);
 399         prepare_matcher_header(search, MATCHCRITERIA_FROM);
 400         debug_matcher_list("tag + subject + from matcher list", search->predicate);
 401         prepare_matcher_header(search, MATCHCRITERIA_TO);
 402         debug_matcher_list("tag + subject + from + to matcher list", search->predicate);
 403         prepare_matcher_header(search, MATCHCRITERIA_CC);
 404         debug_matcher_list("tag + subject + from + to + cc matcher list", search->predicate);
 405 }
 406
 407 static void prepare_matcher(AdvancedSearch *search)
 408 {
 409         const gchar *search_string;
 410
 411         cm_return_if_fail(search != NULL);
 412
 413         if (search->predicate) {
 414                 matcherlist_free(search->predicate);
 415                 search->predicate = NULL;
 416         }
 417
 418         search_string = search->request.matchstring;
 419
 420         if (search_string == NULL || search_string[0] == '\0')
 421                 return;
 422
 423         switch (search->request.type) {
 424                 case ADVANCED_SEARCH_SUBJECT:
 425                         prepare_matcher_header(search, MATCHCRITERIA_SUBJECT);
 426                         debug_matcher_list("subject search", search->predicate);
 427                         break;
 428
 429                 case ADVANCED_SEARCH_FROM:
 430                         prepare_matcher_header(search, MATCHCRITERIA_FROM);
 431                         debug_matcher_list("from search", search->predicate);
 432                         break;
 433
 434                 case ADVANCED_SEARCH_TO:
 435                         prepare_matcher_header(search, MATCHCRITERIA_TO);
 436                         debug_matcher_list("to search", search->predicate);
 437                         break;
 438
 439                 case ADVANCED_SEARCH_TAG:
 440                         prepare_matcher_tag(search);
 441                         debug_matcher_list("tag search", search->predicate);
 442                         break;
 443
 444                 case ADVANCED_SEARCH_MIXED:
 445                         prepare_matcher_mixed(search);
 446                         debug_matcher_list("mixed search", search->predicate);
 447                         break;
 448
 449                 case ADVANCED_SEARCH_EXTENDED:
 450                         prepare_matcher_extended(search);
 451                         debug_matcher_list("extended search", search->predicate);
 452                         break;
 453
 454                 default:
 455                         debug_print("unknown search type (%d)\n", search->request.type);
 456                         break;
 457         }
 458 }
 459
 460 static gboolean search_progress_notify_cb(gpointer data, gboolean on_server, guint at,
 461                 guint matched, guint total)
 462 {
 463         AdvancedSearch *search = (AdvancedSearch*) data;
 464
 465         if (search->search_aborted)
 466                 return FALSE;
 467
 468         if (on_server || search->on_progress_cb.cb == NULL)
 469                 return TRUE;
 470
 471         return search->on_progress_cb.cb(search->on_progress_cb.data, at, matched, total);
 472 }
 473
 474 static gboolean search_filter_folder(MsgNumberList **msgnums, AdvancedSearch *search,
 475                                           FolderItem *folderItem, gboolean onServer)
 476 {
 477         gint matched;
 478         gboolean tried_server = onServer;
 479
 480         matched = folder_item_search_msgs(folderItem->folder,
 481                 folderItem,
 482                 msgnums,
 483                 &onServer,
 484                 search->predicate,
 485                 search_progress_notify_cb,
 486                 search);
 487
 488         if (matched < 0) {
 489                 if (search->on_error_cb.cb != NULL)
 490                         search->on_error_cb.cb(search->on_error_cb.data);
 491                 return FALSE;
 492         }
 493
 494         if (folderItem->folder->klass->supports_server_search && tried_server && !onServer) {
 495                 return search_filter_folder(msgnums, search, folderItem, onServer);
 496         } else {
 497                 return TRUE;
 498         }
 499 }
 500
 501 static gboolean search_impl(MsgInfoList **messages, AdvancedSearch* search,
 502                             FolderItem* folderItem, gboolean recursive)
 503 {
 504         if (recursive) {
 505                 if (!search_impl(messages, search, folderItem, FALSE))
 506                         return FALSE;
 507
 508                 if (folderItem->node->children != NULL && !search->search_aborted) {
 509                         GNode *node;
 510                         for (node = folderItem->node->children; node != NULL; node = node->next) {
 511                                 FolderItem *cur = FOLDER_ITEM(node->data);
 512                                 debug_print("in: %s\n", cur->path);
 513                                 if (!search_impl(messages, search, cur, TRUE))
 514                                         return FALSE;
 515                         }
 516                 }
 517         } else if (!folderItem->no_select) {
 518                 MsgNumberList *msgnums = NULL;
 519                 MsgNumberList *cur;
 520                 MsgInfoList *msgs = NULL;
 521                 gboolean can_search_on_server = folderItem->folder->klass->supports_server_search;
 522
 523                 if (!search_filter_folder(&msgnums, search, folderItem,
 524                                           can_search_on_server)) {
 525                         g_slist_free(msgnums);
 526                         return FALSE;
 527                 }
 528
 529                 for (cur = msgnums; cur != NULL; cur = cur->next) {
 530                         MsgInfo *msg = folder_item_get_msginfo(folderItem, GPOINTER_TO_UINT(cur->data));
 531
 532                         msgs = g_slist_prepend(msgs, msg);
 533                 }
 534
 535                 while (msgs != NULL) {
 536                         MsgInfoList *front = msgs;
 537
 538                         msgs = msgs->next;
 539
 540                         front->next = *messages;
 541                         *messages = front;
 542                 }
 543
 544                 g_slist_free(msgnums);
 545         }
 546
 547         return TRUE;
 548 }