src/advsearch.c

   1 /*
   2  * Claws Mail -- a GTK based, lightweight, and fast e-mail client
   3  * Copyright (C) 2012-2023 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 #ifdef HAVE_CONFIG_H
  20 # include "config.h"
  21 # include "claws-features.h"
  22 #endif
  23
  24 #include "advsearch.h"
  25
  26 #include <glib.h>
  27 #include <ctype.h>
  28
  29 #include "matcher.h"
  30 #include "matcher_parser.h"
  31 #include "utils.h"
  32 #include "prefs_common.h"
  33 #include "timing.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
 133 static void advsearch_extract_param(GString *matcherstr, gchar **cmd_start_, gchar **cmd_end_, gboolean quotes, gboolean qualifier, gboolean casesens, gboolean regex)
 134 {
 135         gchar *cmd_start, *cmd_end;
 136         gchar term_char, save_char;
 137
 138         cmd_start = *cmd_start_;
 139         cmd_end   = *cmd_end_;
 140
 141         /* extract a parameter, allow quotes */
 142         while (*cmd_end && isspace((guchar)*cmd_end))
 143                 cmd_end++;
 144
 145         cmd_start = cmd_end;
 146         if (*cmd_start == '"') {
 147                 term_char = '"';
 148                 cmd_end++;
 149         }
 150         else
 151                 term_char = ' ';
 152
 153         /* extract actual parameter */
 154         while ((*cmd_end) && (*cmd_end != term_char))
 155                 cmd_end++;
 156
 157         if (*cmd_end == '"')
 158                 cmd_end++;
 159
 160         save_char = *cmd_end;
 161         *cmd_end = '\0';
 162
 163         if (qualifier) {
 164                 if (casesens)
 165                         g_string_append(matcherstr, regex ? "regexp " : "match ");
 166                 else
 167                         g_string_append(matcherstr, regex ? "regexpcase " : "matchcase ");
 168         }
 169
 170         /* do we need to add quotes ? */
 171         if (quotes && term_char != '"')
 172                 g_string_append(matcherstr, "\"");
 173
 174         /* copy actual parameter */
 175         g_string_append(matcherstr, cmd_start);
 176
 177         /* do we need to add quotes ? */
 178         if (quotes && term_char != '"')
 179                 g_string_append(matcherstr, "\"");
 180
 181         /* restore original character */
 182         *cmd_end = save_char;
 183
 184         *cmd_end_   = cmd_end;
 185         *cmd_start_ = cmd_start;
 186         return;
 187 }
 188
 189 gchar *advsearch_expand_search_string(const gchar *search_string)
 190 {
 191         int i = 0;
 192         gchar *cmd_start, *cmd_end;
 193         gchar save_char;
 194         GString *matcherstr;
 195         gchar *returnstr = NULL;
 196         gchar *copy_str;
 197         gboolean casesens, dontmatch, regex;
 198         /* list of allowed pattern abbreviations */
 199         struct {
 200                 gchar           *abbreviated;   /* abbreviation */
 201                 gchar           *command;       /* actual matcher command */
 202                 gint            numparams;      /* number of params for cmd */
 203                 gboolean        qualifier;      /* do we append stringmatch operations */
 204                 gboolean        quotes;         /* do we need quotes */
 205         }
 206         cmds[] = {
 207                 { "a",  "all",                          0,      FALSE,  FALSE },
 208                 { "ag", "age_greater",                  1,      FALSE,  FALSE },
 209                 { "al", "age_lower",                    1,      FALSE,  FALSE },
 210                 { "agh","age_greater_hours",            1,      FALSE,  FALSE },
 211                 { "alh","age_lower_hours",              1,      FALSE,  FALSE },
 212                 { "b",  "body_part",                    1,      TRUE,   TRUE  },
 213                 { "B",  "message",                      1,      TRUE,   TRUE  },
 214                 { "c",  "cc",                           1,      TRUE,   TRUE  },
 215                 { "C",  "to_or_cc",                     1,      TRUE,   TRUE  },
 216                 { "D",  "deleted",                      0,      FALSE,  FALSE },
 217                 { "da", "date_after",                   1,      FALSE,  TRUE  },
 218                 { "db", "date_before",                  1,      FALSE,  TRUE  },
 219                 { "e",  "header \"Sender\"",            1,      TRUE,   TRUE  },
 220                 { "E",  "execute",                      1,      FALSE,  TRUE  },
 221                 { "f",  "from",                         1,      TRUE,   TRUE  },
 222                 { "F",  "forwarded",                    0,      FALSE,  FALSE },
 223                 { "h",  "headers_part",                 1,      TRUE,   TRUE  },
 224                 { "H",  "headers_cont",                 1,      TRUE,   TRUE  },
 225                 { "ha", "has_attachments",              0,      FALSE,  FALSE },
 226                 { "i",  "messageid",                    1,      TRUE,   TRUE  },
 227                 { "I",  "inreplyto",                    1,      TRUE,   TRUE  },
 228                 { "k",  "colorlabel",                   1,      FALSE,  FALSE },
 229                 { "L",  "locked",                       0,      FALSE,  FALSE },
 230                 { "n",  "newsgroups",                   1,      TRUE,   TRUE  },
 231                 { "N",  "new",                          0,      FALSE,  FALSE },
 232                 { "O",  "~new",                         0,      FALSE,  FALSE },
 233                 { "r",  "replied",                      0,      FALSE,  FALSE },
 234                 { "R",  "~unread",                      0,      FALSE,  FALSE },
 235                 { "s",  "subject",                      1,      TRUE,   TRUE  },
 236                 { "se", "score_equal",                  1,      FALSE,  FALSE },
 237                 { "sg", "score_greater",                1,      FALSE,  FALSE },
 238                 { "sl", "score_lower",                  1,      FALSE,  FALSE },
 239                 { "Se", "size_equal",                   1,      FALSE,  FALSE },
 240                 { "Sg", "size_greater",                 1,      FALSE,  FALSE },
 241                 { "Ss", "size_smaller",                 1,      FALSE,  FALSE },
 242                 { "t",  "to",                           1,      TRUE,   TRUE  },
 243                 { "tg", "tag",                          1,      TRUE,   TRUE  },
 244                 { "T",  "marked",                       0,      FALSE,  FALSE },
 245                 { "U",  "unread",                       0,      FALSE,  FALSE },
 246                 { "x",  "references",                   1,      TRUE,   TRUE  },
 247                 { "X",  "test",                         1,      FALSE,  FALSE },
 248                 { "v",  "header",                       2,      TRUE,   TRUE  },
 249                 { "&",  "&",                            0,      FALSE,  FALSE },
 250                 { "|",  "|",                            0,      FALSE,  FALSE },
 251                 { "p",  "partial",                      0,      FALSE,  FALSE },
 252                 { NULL, NULL,                           0,      FALSE,  FALSE }
 253         };
 254
 255         if (search_string == NULL)
 256                 return NULL;
 257
 258         copy_str = g_strdup(search_string);
 259
 260         matcherstr = g_string_sized_new(16);
 261         cmd_start = copy_str;
 262         while (cmd_start && *cmd_start) {
 263                 /* skip all white spaces */
 264                 while (*cmd_start && isspace((guchar)*cmd_start))
 265                         cmd_start++;
 266                 cmd_end = cmd_start;
 267
 268                 /* extract a command */
 269                 while (*cmd_end && !isspace((guchar)*cmd_end))
 270                         cmd_end++;
 271
 272                 /* save character */
 273                 save_char = *cmd_end;
 274                 *cmd_end = '\0';
 275
 276                 dontmatch = FALSE;
 277                 casesens = FALSE;
 278                 regex = FALSE;
 279
 280                 /* ~ and ! mean logical NOT */
 281                 if (*cmd_start == '~' || *cmd_start == '!')
 282                 {
 283                         dontmatch = TRUE;
 284                         cmd_start++;
 285                 }
 286                 /* % means case sensitive match */
 287                 if (*cmd_start == '%')
 288                 {
 289                         casesens = TRUE;
 290                         cmd_start++;
 291                 }
 292                 /* # means regex match */
 293                 if (*cmd_start == '#') {
 294                         regex = TRUE;
 295                         cmd_start++;
 296                 }
 297
 298                 /* find matching abbreviation */
 299                 for (i = 0; cmds[i].command; i++) {
 300                         if (!strcmp(cmd_start, cmds[i].abbreviated)) {
 301                                 /* restore character */
 302                                 *cmd_end = save_char;
 303
 304                                 /* copy command */
 305                                 if (matcherstr->len > 0) {
 306                                         g_string_append(matcherstr, " ");
 307                                 }
 308                                 if (dontmatch)
 309                                         g_string_append(matcherstr, "~");
 310                                 g_string_append(matcherstr, cmds[i].command);
 311                                 g_string_append(matcherstr, " ");
 312
 313                                 /* stop if no params required */
 314                                 if (cmds[i].numparams == 0)
 315                                         break;
 316
 317                                 /* extract a first parameter before the final matched one */
 318                                 if (cmds[i].numparams == 2)
 319                                 {
 320                                         advsearch_extract_param(matcherstr, &cmd_start, &cmd_end, cmds[i].quotes, FALSE, casesens, regex);
 321                                         g_string_append(matcherstr, " ");
 322                                 }
 323                                 advsearch_extract_param(matcherstr, &cmd_start, &cmd_end, cmds[i].quotes, cmds[i].qualifier, casesens, regex);
 324                                 break;
 325                         }
 326                 }
 327
 328                 if (*cmd_end)
 329                         cmd_end++;
 330                 cmd_start = cmd_end;
 331         }
 332
 333         g_free(copy_str);
 334
 335         /* return search string if no match is found to allow
 336            all available filtering expressions in advanced search */
 337         if (matcherstr->len > 0) {
 338                 returnstr = g_string_free(matcherstr, FALSE);
 339         } else {
 340                 returnstr = g_strdup(search_string);
 341                 g_string_free(matcherstr, TRUE);
 342         }
 343         return returnstr;
 344 }
 345
 346 static void prepare_matcher_extended(AdvancedSearch *search)
 347 {
 348         gchar *newstr = advsearch_expand_search_string(search->request.matchstring);
 349
 350         if (newstr && newstr[0] != '\0') {
 351                 search->predicate = matcher_parser_get_cond(newstr, &search->is_fast);
 352                 g_free(newstr);
 353         }
 354 }
 355
 356 #define debug_matcher_list(prefix, list)                                        \
 357 do {                                                                            \
 358         gchar *str = list ? matcherlist_to_string(list) : g_strdup("(NULL)");   \
 359                                                                                 \
 360         debug_print("%s: %s\n", prefix, str);                                   \
 361                                                                                 \
 362         g_free(str);                                                            \
 363 } while(0)
 364
 365 static void prepare_matcher_tag(AdvancedSearch *search)
 366 {
 367         gchar **words = search->request.matchstring
 368                         ? g_strsplit(search->request.matchstring, " ", -1)
 369                         : NULL;
 370         gint i = 0;
 371
 372         if (search->predicate == NULL) {
 373                 search->predicate = g_new0(MatcherList, 1);
 374                 search->predicate->bool_and = FALSE;
 375                 search->is_fast = TRUE;
 376         }
 377
 378         while (words && words[i] && *words[i]) {
 379                 MatcherProp *matcher;
 380
 381                 g_strstrip(words[i]);
 382
 383                 matcher = matcherprop_new(MATCHCRITERIA_TAG, NULL,
 384                                           MATCHTYPE_MATCHCASE, words[i], 0);
 385
 386                 search->predicate->matchers = g_slist_prepend(search->predicate->matchers, matcher);
 387
 388                 i++;
 389         }
 390         g_strfreev(words);
 391 }
 392
 393 static void prepare_matcher_header(AdvancedSearch *search, gint match_header)
 394 {
 395         MatcherProp *matcher;
 396
 397         if (search->predicate == NULL) {
 398                 search->predicate = g_new0(MatcherList, 1);
 399                 search->predicate->bool_and = FALSE;
 400                 search->is_fast = TRUE;
 401         }
 402
 403         matcher = matcherprop_new(match_header, NULL, MATCHTYPE_MATCHCASE,
 404                         search->request.matchstring, 0);
 405
 406         search->predicate->matchers = g_slist_prepend(search->predicate->matchers, matcher);
 407 }
 408
 409 static void prepare_matcher_mixed(AdvancedSearch *search)
 410 {
 411         prepare_matcher_tag(search);
 412         debug_matcher_list("tag matcher list", search->predicate);
 413
 414         /* we want an OR search */
 415         if (search->predicate)
 416                 search->predicate->bool_and = FALSE;
 417
 418         prepare_matcher_header(search, MATCHCRITERIA_SUBJECT);
 419         debug_matcher_list("tag + subject matcher list", search->predicate);
 420         prepare_matcher_header(search, MATCHCRITERIA_FROM);
 421         debug_matcher_list("tag + subject + from matcher list", search->predicate);
 422         prepare_matcher_header(search, MATCHCRITERIA_TO);
 423         debug_matcher_list("tag + subject + from + to matcher list", search->predicate);
 424         prepare_matcher_header(search, MATCHCRITERIA_CC);
 425         debug_matcher_list("tag + subject + from + to + cc matcher list", search->predicate);
 426 }
 427
 428 static void prepare_matcher(AdvancedSearch *search)
 429 {
 430         const gchar *search_string;
 431
 432         cm_return_if_fail(search != NULL);
 433
 434         if (search->predicate) {
 435                 matcherlist_free(search->predicate);
 436                 search->predicate = NULL;
 437         }
 438
 439         search_string = search->request.matchstring;
 440
 441         if (search_string == NULL || search_string[0] == '\0')
 442                 return;
 443
 444         switch (search->request.type) {
 445                 case ADVANCED_SEARCH_SUBJECT:
 446                         prepare_matcher_header(search, MATCHCRITERIA_SUBJECT);
 447                         debug_matcher_list("subject search", search->predicate);
 448                         break;
 449
 450                 case ADVANCED_SEARCH_FROM:
 451                         prepare_matcher_header(search, MATCHCRITERIA_FROM);
 452                         debug_matcher_list("from search", search->predicate);
 453                         break;
 454
 455                 case ADVANCED_SEARCH_TO:
 456                         prepare_matcher_header(search, MATCHCRITERIA_TO);
 457                         debug_matcher_list("to search", search->predicate);
 458                         break;
 459
 460                 case ADVANCED_SEARCH_TAG:
 461                         prepare_matcher_tag(search);
 462                         debug_matcher_list("tag search", search->predicate);
 463                         break;
 464
 465                 case ADVANCED_SEARCH_MIXED:
 466                         prepare_matcher_mixed(search);
 467                         debug_matcher_list("mixed search", search->predicate);
 468                         break;
 469
 470                 case ADVANCED_SEARCH_EXTENDED:
 471                         prepare_matcher_extended(search);
 472                         debug_matcher_list("extended search", search->predicate);
 473                         break;
 474
 475                 default:
 476                         debug_print("unknown search type (%d)\n", search->request.type);
 477                         break;
 478         }
 479 }
 480
 481 static gboolean search_progress_notify_cb(gpointer data, gboolean on_server, guint at,
 482                 guint matched, guint total)
 483 {
 484         AdvancedSearch *search = (AdvancedSearch*) data;
 485
 486         if (search->search_aborted)
 487                 return FALSE;
 488
 489         if (on_server || search->on_progress_cb.cb == NULL)
 490                 return TRUE;
 491
 492         return search->on_progress_cb.cb(search->on_progress_cb.data, at, matched, total);
 493 }
 494
 495 static gboolean search_filter_folder(MsgNumberList **msgnums, AdvancedSearch *search,
 496                                           FolderItem *folderItem, gboolean onServer)
 497 {
 498         gint matched;
 499         gboolean tried_server = onServer;
 500
 501         matched = folder_item_search_msgs(folderItem->folder,
 502                 folderItem,
 503                 msgnums,
 504                 &onServer,
 505                 search->predicate,
 506                 search_progress_notify_cb,
 507                 search);
 508
 509         if (matched < 0) {
 510                 if (search->on_error_cb.cb != NULL)
 511                         search->on_error_cb.cb(search->on_error_cb.data);
 512                 return FALSE;
 513         }
 514
 515         if (folderItem->folder->klass->supports_server_search && tried_server && !onServer) {
 516                 return search_filter_folder(msgnums, search, folderItem, onServer);
 517         } else {
 518                 return TRUE;
 519         }
 520 }
 521
 522 static gboolean search_impl(MsgInfoList **messages, AdvancedSearch* search,
 523                             FolderItem* folderItem, gboolean recursive)
 524 {
 525         if (recursive) {
 526                 START_TIMING("recursive");
 527                 if (!search_impl(messages, search, folderItem, FALSE)) {
 528                         END_TIMING();
 529                         return FALSE;
 530                 }
 531                 if (folderItem->node->children != NULL && !search->search_aborted) {
 532                         GNode *node;
 533                         for (node = folderItem->node->children; node != NULL; node = node->next) {
 534                                 FolderItem *cur = FOLDER_ITEM(node->data);
 535                                 debug_print("in: %s\n", cur->path);
 536                                 if (!search_impl(messages, search, cur, TRUE)) {
 537                                         END_TIMING();
 538                                         return FALSE;
 539                                 }
 540                         }
 541                 }
 542                 END_TIMING();
 543         } else if (!folderItem->no_select) {
 544                 MsgNumberList *msgnums = NULL;
 545                 MsgNumberList *cur;
 546                 MsgInfoList *msgs = NULL;
 547                 gboolean can_search_on_server = folderItem->folder->klass->supports_server_search;
 548                 START_TIMING("folder");
 549                 if (!search_filter_folder(&msgnums, search, folderItem,
 550                                           can_search_on_server)) {
 551                         g_slist_free(msgnums);
 552                         END_TIMING();
 553                         return FALSE;
 554                 }
 555
 556                 for (cur = msgnums; cur != NULL; cur = cur->next) {
 557                         MsgInfo *msg = folder_item_get_msginfo(folderItem, GPOINTER_TO_UINT(cur->data));
 558
 559                         msgs = g_slist_prepend(msgs, msg);
 560                 }
 561
 562                 while (msgs != NULL) {
 563                         MsgInfoList *front = msgs;
 564
 565                         msgs = msgs->next;
 566
 567                         front->next = *messages;
 568                         *messages = front;
 569                 }
 570
 571                 g_slist_free(msgnums);
 572                 END_TIMING();
 573         }
 574
 575         return TRUE;
 576 }