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         fprintf (stderr, "A Xapian exception occurred performing query: %s\n",
 300                  error.get_msg().c_str());
 301         fprintf (stderr, "Query string was: %s\n", query->query_string);
 302         notmuch->exception_reported = TRUE;
 303         talloc_free (messages);
 304         return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
 305     }
 306 }
 307
 308 notmuch_bool_t
 309 _notmuch_mset_messages_valid (notmuch_messages_t *messages)
 310 {
 311     notmuch_mset_messages_t *mset_messages;
 312
 313     mset_messages = (notmuch_mset_messages_t *) messages;
 314
 315     return (mset_messages->iterator != mset_messages->iterator_end);
 316 }
 317
 318 static Xapian::docid
 319 _notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
 320 {
 321     notmuch_mset_messages_t *mset_messages;
 322
 323     mset_messages = (notmuch_mset_messages_t *) messages;
 324
 325     if (! _notmuch_mset_messages_valid (&mset_messages->base))
 326         return 0;
 327
 328     return *mset_messages->iterator;
 329 }
 330
 331 notmuch_message_t *
 332 _notmuch_mset_messages_get (notmuch_messages_t *messages)
 333 {
 334     notmuch_message_t *message;
 335     Xapian::docid doc_id;
 336     notmuch_private_status_t status;
 337     notmuch_mset_messages_t *mset_messages;
 338
 339     mset_messages = (notmuch_mset_messages_t *) messages;
 340
 341     if (! _notmuch_mset_messages_valid (&mset_messages->base))
 342         return NULL;
 343
 344     doc_id = *mset_messages->iterator;
 345
 346     message = _notmuch_message_create (mset_messages,
 347                                        mset_messages->notmuch, doc_id,
 348                                        &status);
 349
 350     if (message == NULL &&
 351        status == NOTMUCH_PRIVATE_STATUS_NO_DOCUMENT_FOUND)
 352     {
 353         INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
 354     }
 355
 356     if (messages->excluded_doc_ids &&
 357         _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
 358         notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
 359
 360     return message;
 361 }
 362
 363 void
 364 _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
 365 {
 366     notmuch_mset_messages_t *mset_messages;
 367
 368     mset_messages = (notmuch_mset_messages_t *) messages;
 369
 370     mset_messages->iterator++;
 371 }
 372
 373 static notmuch_bool_t
 374 _notmuch_doc_id_set_init (void *ctx,
 375                           notmuch_doc_id_set_t *doc_ids,
 376                           GArray *arr)
 377 {
 378     unsigned int max = 0;
 379     unsigned char *bitmap;
 380
 381     for (unsigned int i = 0; i < arr->len; i++)
 382         max = MAX(max, g_array_index (arr, unsigned int, i));
 383     bitmap = talloc_zero_array (ctx, unsigned char, DOCIDSET_WORD(max) + 1);
 384
 385     if (bitmap == NULL)
 386         return FALSE;
 387
 388     doc_ids->bitmap = bitmap;
 389     doc_ids->bound = max + 1;
 390
 391     for (unsigned int i = 0; i < arr->len; i++) {
 392         unsigned int doc_id = g_array_index (arr, unsigned int, i);
 393         bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
 394     }
 395
 396     return TRUE;
 397 }
 398
 399 notmuch_bool_t
 400 _notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
 401                               unsigned int doc_id)
 402 {
 403     if (doc_id >= doc_ids->bound)
 404         return FALSE;
 405     return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
 406 }
 407
 408 void
 409 _notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
 410                             unsigned int doc_id)
 411 {
 412     if (doc_id < doc_ids->bound)
 413         doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
 414 }
 415
 416 /* Glib objects force use to use a talloc destructor as well, (but not
 417  * nearly as ugly as the for messages due to C++ objects). At
 418  * this point, I'd really like to have some talloc-friendly
 419  * equivalents for the few pieces of glib that I'm using. */
 420 static int
 421 _notmuch_threads_destructor (notmuch_threads_t *threads)
 422 {
 423     if (threads->doc_ids)
 424         g_array_unref (threads->doc_ids);
 425
 426     return 0;
 427 }
 428
 429
 430 notmuch_threads_t *
 431 notmuch_query_search_threads (notmuch_query_t *query)
 432 {
 433     notmuch_status_t status;
 434     notmuch_threads_t *threads;
 435     status = notmuch_query_search_threads_st (query, &threads);
 436     if (status)
 437         return NULL;
 438     else
 439         return threads;
 440 }
 441
 442 notmuch_status_t
 443 notmuch_query_search_threads_st (notmuch_query_t *query,
 444                                  notmuch_threads_t **out)
 445 {
 446     notmuch_threads_t *threads;
 447     notmuch_messages_t *messages;
 448     notmuch_status_t status;
 449
 450     threads = talloc (query, notmuch_threads_t);
 451     if (threads == NULL)
 452         return NOTMUCH_STATUS_OUT_OF_MEMORY;
 453     threads->doc_ids = NULL;
 454     talloc_set_destructor (threads, _notmuch_threads_destructor);
 455
 456     threads->query = query;
 457
 458     status = notmuch_query_search_messages_st (query, &messages);
 459     if (status) {
 460         talloc_free (threads);
 461         return status;
 462     }
 463
 464     threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
 465     while (notmuch_messages_valid (messages)) {
 466         unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
 467         g_array_append_val (threads->doc_ids, doc_id);
 468         notmuch_messages_move_to_next (messages);
 469     }
 470     threads->doc_id_pos = 0;
 471
 472     talloc_free (messages);
 473
 474     if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
 475                                     threads->doc_ids)) {
 476         talloc_free (threads);
 477         return NOTMUCH_STATUS_OUT_OF_MEMORY;
 478     }
 479
 480     *out = threads;
 481     return NOTMUCH_STATUS_SUCCESS;
 482 }
 483
 484 void
 485 notmuch_query_destroy (notmuch_query_t *query)
 486 {
 487     talloc_free (query);
 488 }
 489
 490 notmuch_bool_t
 491 notmuch_threads_valid (notmuch_threads_t *threads)
 492 {
 493     unsigned int doc_id;
 494
 495     if (! threads)
 496         return FALSE;
 497
 498     while (threads->doc_id_pos < threads->doc_ids->len) {
 499         doc_id = g_array_index (threads->doc_ids, unsigned int,
 500                                 threads->doc_id_pos);
 501         if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
 502             break;
 503
 504         threads->doc_id_pos++;
 505     }
 506
 507     return threads->doc_id_pos < threads->doc_ids->len;
 508 }
 509
 510 notmuch_thread_t *
 511 notmuch_threads_get (notmuch_threads_t *threads)
 512 {
 513     unsigned int doc_id;
 514
 515     if (! notmuch_threads_valid (threads))
 516         return NULL;
 517
 518     doc_id = g_array_index (threads->doc_ids, unsigned int,
 519                             threads->doc_id_pos);
 520     return _notmuch_thread_create (threads->query,
 521                                    threads->query->notmuch,
 522                                    doc_id,
 523                                    &threads->match_set,
 524                                    threads->query->exclude_terms,
 525                                    threads->query->omit_excluded,
 526                                    threads->query->sort);
 527 }
 528
 529 void
 530 notmuch_threads_move_to_next (notmuch_threads_t *threads)
 531 {
 532     threads->doc_id_pos++;
 533 }
 534
 535 void
 536 notmuch_threads_destroy (notmuch_threads_t *threads)
 537 {
 538     talloc_free (threads);
 539 }
 540
 541 unsigned
 542 notmuch_query_count_messages (notmuch_query_t *query)
 543 {
 544     notmuch_database_t *notmuch = query->notmuch;
 545     const char *query_string = query->query_string;
 546     Xapian::doccount count = 0;
 547
 548     try {
 549         Xapian::Enquire enquire (*notmuch->xapian_db);
 550         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
 551                                                    _find_prefix ("type"),
 552                                                    "mail"));
 553         Xapian::Query string_query, final_query, exclude_query;
 554         Xapian::MSet mset;
 555         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
 556                               Xapian::QueryParser::FLAG_PHRASE |
 557                               Xapian::QueryParser::FLAG_LOVEHATE |
 558                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
 559                               Xapian::QueryParser::FLAG_WILDCARD |
 560                               Xapian::QueryParser::FLAG_PURE_NOT);
 561
 562         if (strcmp (query_string, "") == 0 ||
 563             strcmp (query_string, "*") == 0)
 564         {
 565             final_query = mail_query;
 566         } else {
 567             string_query = notmuch->query_parser->
 568                 parse_query (query_string, flags);
 569             final_query = Xapian::Query (Xapian::Query::OP_AND,
 570                                          mail_query, string_query);
 571         }
 572
 573         exclude_query = _notmuch_exclude_tags (query, final_query);
 574
 575         final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
 576                                          final_query, exclude_query);
 577
 578         enquire.set_weighting_scheme(Xapian::BoolWeight());
 579         enquire.set_docid_order(Xapian::Enquire::ASCENDING);
 580
 581         if (_debug_query ()) {
 582             fprintf (stderr, "Exclude query is:\n%s\n",
 583                      exclude_query.get_description ().c_str ());
 584             fprintf (stderr, "Final query is:\n%s\n",
 585                      final_query.get_description ().c_str ());
 586         }
 587
 588         enquire.set_query (final_query);
 589
 590         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
 591
 592         count = mset.get_matches_estimated();
 593
 594     } catch (const Xapian::Error &error) {
 595         fprintf (stderr, "A Xapian exception occurred: %s\n",
 596                  error.get_msg().c_str());
 597         fprintf (stderr, "Query string was: %s\n", query->query_string);
 598     }
 599
 600     return count;
 601 }
 602
 603 unsigned
 604 notmuch_query_count_threads (notmuch_query_t *query)
 605 {
 606     notmuch_messages_t *messages;
 607     GHashTable *hash;
 608     unsigned int count;
 609     notmuch_sort_t sort;
 610
 611     sort = query->sort;
 612     query->sort = NOTMUCH_SORT_UNSORTED;
 613     messages = notmuch_query_search_messages (query);
 614     query->sort = sort;
 615     if (messages == NULL)
 616         return 0;
 617
 618     hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
 619     if (hash == NULL) {
 620         talloc_free (messages);
 621         return 0;
 622     }
 623
 624     while (notmuch_messages_valid (messages)) {
 625         notmuch_message_t *message = notmuch_messages_get (messages);
 626         const char *thread_id = notmuch_message_get_thread_id (message);
 627         char *thread_id_copy = talloc_strdup (messages, thread_id);
 628         if (unlikely (thread_id_copy == NULL)) {
 629             notmuch_message_destroy (message);
 630             count = 0;
 631             goto DONE;
 632         }
 633         g_hash_table_insert (hash, thread_id_copy, NULL);
 634         notmuch_message_destroy (message);
 635         notmuch_messages_move_to_next (messages);
 636     }
 637
 638     count = g_hash_table_size (hash);
 639
 640   DONE:
 641     g_hash_table_unref (hash);
 642     talloc_free (messages);
 643
 644     return count;
 645 }