git.notmuchmail.org Git - notmuch/blob - lib/query.cc

   1 /* query.cc - Support for searching a notmuch database
   2  *
   3  * Copyright © 2009 Carl Worth
   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  * Author: Carl Worth <cworth@cworth.org>
  19  */
  20
  21 #include "notmuch-private.h"
  22 #include "database-private.h"
  23
  24 #include <glib.h> /* GHashTable, GPtrArray */
  25
  26 struct _notmuch_query {
  27     notmuch_database_t *notmuch;
  28     const char *query_string;
  29     notmuch_sort_t sort;
  30     notmuch_string_list_t *exclude_terms;
  31     notmuch_exclude_t omit_excluded;
  32 };
  33
  34 typedef struct _notmuch_mset_messages {
  35     notmuch_messages_t base;
  36     notmuch_database_t *notmuch;
  37     Xapian::MSetIterator iterator;
  38     Xapian::MSetIterator iterator_end;
  39 } notmuch_mset_messages_t;
  40
  41 struct _notmuch_doc_id_set {
  42     unsigned char *bitmap;
  43     unsigned int bound;
  44 };
  45
  46 #define DOCIDSET_WORD(bit) ((bit) / CHAR_BIT)
  47 #define DOCIDSET_BIT(bit) ((bit) % CHAR_BIT)
  48
  49 struct visible _notmuch_threads {
  50     notmuch_query_t *query;
  51
  52     /* The ordered list of doc ids matched by the query. */
  53     GArray *doc_ids;
  54     /* Our iterator's current position in doc_ids. */
  55     unsigned int doc_id_pos;
  56     /* The set of matched docid's that have not been assigned to a
  57      * thread. Initially, this contains every docid in doc_ids. */
  58     notmuch_doc_id_set_t match_set;
  59 };
  60
  61 /* We need this in the message functions so forward declare. */
  62 static notmuch_bool_t
  63 _notmuch_doc_id_set_init (void *ctx,
  64                           notmuch_doc_id_set_t *doc_ids,
  65                           GArray *arr);
  66
  67 static notmuch_bool_t
  68 _debug_query (void)
  69 {
  70     char *env = getenv ("NOTMUCH_DEBUG_QUERY");
  71     return (env && strcmp (env, "") != 0);
  72 }
  73
  74 notmuch_query_t *
  75 notmuch_query_create (notmuch_database_t *notmuch,
  76                       const char *query_string)
  77 {
  78     notmuch_query_t *query;
  79
  80     if (_debug_query ())
  81         fprintf (stderr, "Query string is:\n%s\n", query_string);
  82
  83     query = talloc (notmuch, notmuch_query_t);
  84     if (unlikely (query == NULL))
  85         return NULL;
  86
  87     query->notmuch = notmuch;
  88
  89     query->query_string = talloc_strdup (query, query_string);
  90
  91     query->sort = NOTMUCH_SORT_NEWEST_FIRST;
  92
  93     query->exclude_terms = _notmuch_string_list_create (query);
  94
  95     query->omit_excluded = NOTMUCH_EXCLUDE_TRUE;
  96
  97     return query;
  98 }
  99
 100 const char *
 101 notmuch_query_get_query_string (notmuch_query_t *query)
 102 {
 103     return query->query_string;
 104 }
 105
 106 void
 107 notmuch_query_set_omit_excluded (notmuch_query_t *query,
 108                                  notmuch_exclude_t omit_excluded)
 109 {
 110     query->omit_excluded = omit_excluded;
 111 }
 112
 113 void
 114 notmuch_query_set_sort (notmuch_query_t *query, notmuch_sort_t sort)
 115 {
 116     query->sort = sort;
 117 }
 118
 119 notmuch_sort_t
 120 notmuch_query_get_sort (notmuch_query_t *query)
 121 {
 122     return query->sort;
 123 }
 124
 125 void
 126 notmuch_query_add_tag_exclude (notmuch_query_t *query, const char *tag)
 127 {
 128     char *term = talloc_asprintf (query, "%s%s", _find_prefix ("tag"), tag);
 129     _notmuch_string_list_append (query->exclude_terms, term);
 130 }
 131
 132 /* We end up having to call the destructors explicitly because we had
 133  * to use "placement new" in order to initialize C++ objects within a
 134  * block that we allocated with talloc. So C++ is making talloc
 135  * slightly less simple to use, (we wouldn't need
 136  * talloc_set_destructor at all otherwise).
 137  */
 138 static int
 139 _notmuch_messages_destructor (notmuch_mset_messages_t *messages)
 140 {
 141     messages->iterator.~MSetIterator ();
 142     messages->iterator_end.~MSetIterator ();
 143
 144     return 0;
 145 }
 146
 147 /* Return a query that matches messages with the excluded tags
 148  * registered with query.  Any tags that explicitly appear in xquery
 149  * will not be excluded, and will be removed from the list of exclude
 150  * tags.  The caller of this function has to combine the returned
 151  * query appropriately.*/
 152 static Xapian::Query
 153 _notmuch_exclude_tags (notmuch_query_t *query, Xapian::Query xquery)
 154 {
 155     Xapian::Query exclude_query = Xapian::Query::MatchNothing;
 156
 157     for (notmuch_string_node_t *term = query->exclude_terms->head; term;
 158          term = term->next) {
 159         Xapian::TermIterator it = xquery.get_terms_begin ();
 160         Xapian::TermIterator end = xquery.get_terms_end ();
 161         for (; it != end; it++) {
 162             if ((*it).compare (term->string) == 0)
 163                 break;
 164         }
 165         if (it == end)
 166             exclude_query = Xapian::Query (Xapian::Query::OP_OR,
 167                                     exclude_query, Xapian::Query (term->string));
 168         else
 169             term->string = talloc_strdup (query, "");
 170     }
 171     return exclude_query;
 172 }
 173
 174 notmuch_messages_t *
 175 notmuch_query_search_messages (notmuch_query_t *query)
 176 {
 177     notmuch_status_t status;
 178     notmuch_messages_t *messages;
 179     status = notmuch_query_search_messages_st (query, &messages);
 180     if (status)
 181         return NULL;
 182     else
 183         return messages;
 184 }
 185
 186 notmuch_status_t
 187 notmuch_query_search_messages_st (notmuch_query_t *query,
 188                                   notmuch_messages_t **out)
 189 {
 190     notmuch_database_t *notmuch = query->notmuch;
 191     const char *query_string = query->query_string;
 192     notmuch_mset_messages_t *messages;
 193
 194     messages = talloc (query, notmuch_mset_messages_t);
 195     if (unlikely (messages == NULL))
 196         return NOTMUCH_STATUS_OUT_OF_MEMORY;
 197
 198     try {
 199
 200         messages->base.is_of_list_type = FALSE;
 201         messages->base.iterator = NULL;
 202         messages->notmuch = notmuch;
 203         new (&messages->iterator) Xapian::MSetIterator ();
 204         new (&messages->iterator_end) Xapian::MSetIterator ();
 205
 206         talloc_set_destructor (messages, _notmuch_messages_destructor);
 207
 208         Xapian::Enquire enquire (*notmuch->xapian_db);
 209         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
 210                                                    _find_prefix ("type"),
 211                                                    "mail"));
 212         Xapian::Query string_query, final_query, exclude_query;
 213         Xapian::MSet mset;
 214         Xapian::MSetIterator iterator;
 215         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
 216                               Xapian::QueryParser::FLAG_PHRASE |
 217                               Xapian::QueryParser::FLAG_LOVEHATE |
 218                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
 219                               Xapian::QueryParser::FLAG_WILDCARD |
 220                               Xapian::QueryParser::FLAG_PURE_NOT);
 221
 222         if (strcmp (query_string, "") == 0 ||
 223             strcmp (query_string, "*") == 0)
 224         {
 225             final_query = mail_query;
 226         } else {
 227             string_query = notmuch->query_parser->
 228                 parse_query (query_string, flags);
 229             final_query = Xapian::Query (Xapian::Query::OP_AND,
 230                                          mail_query, string_query);
 231         }
 232         messages->base.excluded_doc_ids = NULL;
 233
 234         if ((query->omit_excluded != NOTMUCH_EXCLUDE_FALSE) && (query->exclude_terms)) {
 235             exclude_query = _notmuch_exclude_tags (query, final_query);
 236
 237             if (query->omit_excluded == NOTMUCH_EXCLUDE_TRUE ||
 238                 query->omit_excluded == NOTMUCH_EXCLUDE_ALL)
 239             {
 240                 final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
 241                                              final_query, exclude_query);
 242             } else { /* NOTMUCH_EXCLUDE_FLAG */
 243                 exclude_query = Xapian::Query (Xapian::Query::OP_AND,
 244                                            exclude_query, final_query);
 245
 246                 enquire.set_weighting_scheme (Xapian::BoolWeight());
 247                 enquire.set_query (exclude_query);
 248
 249                 mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
 250
 251                 GArray *excluded_doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
 252
 253                 for (iterator = mset.begin (); iterator != mset.end (); iterator++) {
 254                     unsigned int doc_id = *iterator;
 255                     g_array_append_val (excluded_doc_ids, doc_id);
 256                 }
 257                 messages->base.excluded_doc_ids = talloc (messages, _notmuch_doc_id_set);
 258                 _notmuch_doc_id_set_init (query, messages->base.excluded_doc_ids,
 259                                           excluded_doc_ids);
 260                 g_array_unref (excluded_doc_ids);
 261             }
 262         }
 263
 264
 265         enquire.set_weighting_scheme (Xapian::BoolWeight());
 266
 267         switch (query->sort) {
 268         case NOTMUCH_SORT_OLDEST_FIRST:
 269             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, FALSE);
 270             break;
 271         case NOTMUCH_SORT_NEWEST_FIRST:
 272             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, TRUE);
 273             break;
 274         case NOTMUCH_SORT_MESSAGE_ID:
 275             enquire.set_sort_by_value (NOTMUCH_VALUE_MESSAGE_ID, FALSE);
 276             break;
 277         case NOTMUCH_SORT_UNSORTED:
 278             break;
 279         }
 280
 281         if (_debug_query ()) {
 282             fprintf (stderr, "Exclude query is:\n%s\n",
 283                      exclude_query.get_description ().c_str ());
 284             fprintf (stderr, "Final query is:\n%s\n",
 285                      final_query.get_description ().c_str ());
 286         }
 287
 288         enquire.set_query (final_query);
 289
 290         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
 291
 292         messages->iterator = mset.begin ();
 293         messages->iterator_end = mset.end ();
 294
 295         *out = &messages->base;
 296         return NOTMUCH_STATUS_SUCCESS;
 297
 298     } catch (const Xapian::Error &error) {
 299         _notmuch_database_log (notmuch,
 300                                "A Xapian exception occurred performing query: %s\n"
 301                                "Query string was: %s\n",
 302                                error.get_msg().c_str(),
 303                                query->query_string);
 304
 305         notmuch->exception_reported = TRUE;
 306         talloc_free (messages);
 307         return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
 308     }
 309 }
 310
 311 notmuch_bool_t
 312 _notmuch_mset_messages_valid (notmuch_messages_t *messages)
 313 {
 314     notmuch_mset_messages_t *mset_messages;
 315
 316     mset_messages = (notmuch_mset_messages_t *) messages;
 317
 318     return (mset_messages->iterator != mset_messages->iterator_end);
 319 }
 320
 321 static Xapian::docid
 322 _notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
 323 {
 324     notmuch_mset_messages_t *mset_messages;
 325
 326     mset_messages = (notmuch_mset_messages_t *) messages;
 327
 328     if (! _notmuch_mset_messages_valid (&mset_messages->base))
 329         return 0;
 330
 331     return *mset_messages->iterator;
 332 }
 333
 334 notmuch_message_t *
 335 _notmuch_mset_messages_get (notmuch_messages_t *messages)
 336 {
 337     notmuch_message_t *message;
 338     Xapian::docid doc_id;
 339     notmuch_private_status_t status;
 340     notmuch_mset_messages_t *mset_messages;
 341
 342     mset_messages = (notmuch_mset_messages_t *) messages;
 343
 344     if (! _notmuch_mset_messages_valid (&mset_messages->base))
 345         return NULL;
 346
 347     doc_id = *mset_messages->iterator;
 348
 349     message = _notmuch_message_create (mset_messages,
 350                                        mset_messages->notmuch, doc_id,
 351                                        &status);
 352
 353     if (message == NULL &&
 354        status == NOTMUCH_PRIVATE_STATUS_NO_DOCUMENT_FOUND)
 355     {
 356         INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
 357     }
 358
 359     if (messages->excluded_doc_ids &&
 360         _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
 361         notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
 362
 363     return message;
 364 }
 365
 366 void
 367 _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
 368 {
 369     notmuch_mset_messages_t *mset_messages;
 370
 371     mset_messages = (notmuch_mset_messages_t *) messages;
 372
 373     mset_messages->iterator++;
 374 }
 375
 376 static notmuch_bool_t
 377 _notmuch_doc_id_set_init (void *ctx,
 378                           notmuch_doc_id_set_t *doc_ids,
 379                           GArray *arr)
 380 {
 381     unsigned int max = 0;
 382     unsigned char *bitmap;
 383
 384     for (unsigned int i = 0; i < arr->len; i++)
 385         max = MAX(max, g_array_index (arr, unsigned int, i));
 386     bitmap = talloc_zero_array (ctx, unsigned char, DOCIDSET_WORD(max) + 1);
 387
 388     if (bitmap == NULL)
 389         return FALSE;
 390
 391     doc_ids->bitmap = bitmap;
 392     doc_ids->bound = max + 1;
 393
 394     for (unsigned int i = 0; i < arr->len; i++) {
 395         unsigned int doc_id = g_array_index (arr, unsigned int, i);
 396         bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
 397     }
 398
 399     return TRUE;
 400 }
 401
 402 notmuch_bool_t
 403 _notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
 404                               unsigned int doc_id)
 405 {
 406     if (doc_id >= doc_ids->bound)
 407         return FALSE;
 408     return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
 409 }
 410
 411 void
 412 _notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
 413                             unsigned int doc_id)
 414 {
 415     if (doc_id < doc_ids->bound)
 416         doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
 417 }
 418
 419 /* Glib objects force use to use a talloc destructor as well, (but not
 420  * nearly as ugly as the for messages due to C++ objects). At
 421  * this point, I'd really like to have some talloc-friendly
 422  * equivalents for the few pieces of glib that I'm using. */
 423 static int
 424 _notmuch_threads_destructor (notmuch_threads_t *threads)
 425 {
 426     if (threads->doc_ids)
 427         g_array_unref (threads->doc_ids);
 428
 429     return 0;
 430 }
 431
 432
 433 notmuch_threads_t *
 434 notmuch_query_search_threads (notmuch_query_t *query)
 435 {
 436     notmuch_status_t status;
 437     notmuch_threads_t *threads;
 438     status = notmuch_query_search_threads_st (query, &threads);
 439     if (status)
 440         return NULL;
 441     else
 442         return threads;
 443 }
 444
 445 notmuch_status_t
 446 notmuch_query_search_threads_st (notmuch_query_t *query,
 447                                  notmuch_threads_t **out)
 448 {
 449     notmuch_threads_t *threads;
 450     notmuch_messages_t *messages;
 451     notmuch_status_t status;
 452
 453     threads = talloc (query, notmuch_threads_t);
 454     if (threads == NULL)
 455         return NOTMUCH_STATUS_OUT_OF_MEMORY;
 456     threads->doc_ids = NULL;
 457     talloc_set_destructor (threads, _notmuch_threads_destructor);
 458
 459     threads->query = query;
 460
 461     status = notmuch_query_search_messages_st (query, &messages);
 462     if (status) {
 463         talloc_free (threads);
 464         return status;
 465     }
 466
 467     threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
 468     while (notmuch_messages_valid (messages)) {
 469         unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
 470         g_array_append_val (threads->doc_ids, doc_id);
 471         notmuch_messages_move_to_next (messages);
 472     }
 473     threads->doc_id_pos = 0;
 474
 475     talloc_free (messages);
 476
 477     if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
 478                                     threads->doc_ids)) {
 479         talloc_free (threads);
 480         return NOTMUCH_STATUS_OUT_OF_MEMORY;
 481     }
 482
 483     *out = threads;
 484     return NOTMUCH_STATUS_SUCCESS;
 485 }
 486
 487 void
 488 notmuch_query_destroy (notmuch_query_t *query)
 489 {
 490     talloc_free (query);
 491 }
 492
 493 notmuch_bool_t
 494 notmuch_threads_valid (notmuch_threads_t *threads)
 495 {
 496     unsigned int doc_id;
 497
 498     if (! threads)
 499         return FALSE;
 500
 501     while (threads->doc_id_pos < threads->doc_ids->len) {
 502         doc_id = g_array_index (threads->doc_ids, unsigned int,
 503                                 threads->doc_id_pos);
 504         if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
 505             break;
 506
 507         threads->doc_id_pos++;
 508     }
 509
 510     return threads->doc_id_pos < threads->doc_ids->len;
 511 }
 512
 513 notmuch_thread_t *
 514 notmuch_threads_get (notmuch_threads_t *threads)
 515 {
 516     unsigned int doc_id;
 517
 518     if (! notmuch_threads_valid (threads))
 519         return NULL;
 520
 521     doc_id = g_array_index (threads->doc_ids, unsigned int,
 522                             threads->doc_id_pos);
 523     return _notmuch_thread_create (threads->query,
 524                                    threads->query->notmuch,
 525                                    doc_id,
 526                                    &threads->match_set,
 527                                    threads->query->exclude_terms,
 528                                    threads->query->omit_excluded,
 529                                    threads->query->sort);
 530 }
 531
 532 void
 533 notmuch_threads_move_to_next (notmuch_threads_t *threads)
 534 {
 535     threads->doc_id_pos++;
 536 }
 537
 538 void
 539 notmuch_threads_destroy (notmuch_threads_t *threads)
 540 {
 541     talloc_free (threads);
 542 }
 543
 544 unsigned
 545 notmuch_query_count_messages (notmuch_query_t *query)
 546 {
 547     notmuch_database_t *notmuch = query->notmuch;
 548     const char *query_string = query->query_string;
 549     Xapian::doccount count = 0;
 550
 551     try {
 552         Xapian::Enquire enquire (*notmuch->xapian_db);
 553         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
 554                                                    _find_prefix ("type"),
 555                                                    "mail"));
 556         Xapian::Query string_query, final_query, exclude_query;
 557         Xapian::MSet mset;
 558         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
 559                               Xapian::QueryParser::FLAG_PHRASE |
 560                               Xapian::QueryParser::FLAG_LOVEHATE |
 561                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
 562                               Xapian::QueryParser::FLAG_WILDCARD |
 563                               Xapian::QueryParser::FLAG_PURE_NOT);
 564
 565         if (strcmp (query_string, "") == 0 ||
 566             strcmp (query_string, "*") == 0)
 567         {
 568             final_query = mail_query;
 569         } else {
 570             string_query = notmuch->query_parser->
 571                 parse_query (query_string, flags);
 572             final_query = Xapian::Query (Xapian::Query::OP_AND,
 573                                          mail_query, string_query);
 574         }
 575
 576         exclude_query = _notmuch_exclude_tags (query, final_query);
 577
 578         final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
 579                                          final_query, exclude_query);
 580
 581         enquire.set_weighting_scheme(Xapian::BoolWeight());
 582         enquire.set_docid_order(Xapian::Enquire::ASCENDING);
 583
 584         if (_debug_query ()) {
 585             fprintf (stderr, "Exclude query is:\n%s\n",
 586                      exclude_query.get_description ().c_str ());
 587             fprintf (stderr, "Final query is:\n%s\n",
 588                      final_query.get_description ().c_str ());
 589         }
 590
 591         enquire.set_query (final_query);
 592
 593         /*
 594          * Set the checkatleast parameter to the number of documents
 595          * in the database to make get_matches_estimated() exact.
 596          */
 597         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount (),
 598                                  notmuch->xapian_db->get_doccount ());
 599
 600         count = mset.get_matches_estimated();
 601
 602     } catch (const Xapian::Error &error) {
 603         _notmuch_database_log (notmuch,
 604                                "A Xapian exception occurred performing query: %s\n"
 605                                "Query string was: %s\n",
 606                                error.get_msg().c_str(),
 607                                query->query_string);
 608
 609     }
 610
 611     return count;
 612 }
 613
 614 unsigned
 615 notmuch_query_count_threads (notmuch_query_t *query)
 616 {
 617     notmuch_messages_t *messages;
 618     GHashTable *hash;
 619     unsigned int count;
 620     notmuch_sort_t sort;
 621
 622     sort = query->sort;
 623     query->sort = NOTMUCH_SORT_UNSORTED;
 624     messages = notmuch_query_search_messages (query);
 625     query->sort = sort;
 626     if (messages == NULL)
 627         return 0;
 628
 629     hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
 630     if (hash == NULL) {
 631         talloc_free (messages);
 632         return 0;
 633     }
 634
 635     while (notmuch_messages_valid (messages)) {
 636         notmuch_message_t *message = notmuch_messages_get (messages);
 637         const char *thread_id = notmuch_message_get_thread_id (message);
 638         char *thread_id_copy = talloc_strdup (messages, thread_id);
 639         if (unlikely (thread_id_copy == NULL)) {
 640             notmuch_message_destroy (message);
 641             count = 0;
 642             goto DONE;
 643         }
 644         g_hash_table_insert (hash, thread_id_copy, NULL);
 645         notmuch_message_destroy (message);
 646         notmuch_messages_move_to_next (messages);
 647     }
 648
 649     count = g_hash_table_size (hash);
 650
 651   DONE:
 652     g_hash_table_unref (hash);
 653     talloc_free (messages);
 654
 655     return count;
 656 }
 657
 658 notmuch_database_t *
 659 notmuch_query_get_database (notmuch_query_t *query)
 660 {
 661     return query->notmuch;
 662 }