4ccd81046128052f3461622e8a58c5c63af782f0
[notmuch] / 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 https://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 (const 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 (const 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     return _notmuch_query_search_documents (query, "mail", out);
191 }
192
193 notmuch_status_t
194 _notmuch_query_search_documents (notmuch_query_t *query,
195                                  const char *type,
196                                  notmuch_messages_t **out)
197 {
198     notmuch_database_t *notmuch = query->notmuch;
199     const char *query_string = query->query_string;
200     notmuch_mset_messages_t *messages;
201
202     messages = talloc (query, notmuch_mset_messages_t);
203     if (unlikely (messages == NULL))
204         return NOTMUCH_STATUS_OUT_OF_MEMORY;
205
206     try {
207
208         messages->base.is_of_list_type = FALSE;
209         messages->base.iterator = NULL;
210         messages->notmuch = notmuch;
211         new (&messages->iterator) Xapian::MSetIterator ();
212         new (&messages->iterator_end) Xapian::MSetIterator ();
213
214         talloc_set_destructor (messages, _notmuch_messages_destructor);
215
216         Xapian::Enquire enquire (*notmuch->xapian_db);
217         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
218                                                    _find_prefix ("type"),
219                                                    type));
220         Xapian::Query string_query, final_query, exclude_query;
221         Xapian::MSet mset;
222         Xapian::MSetIterator iterator;
223
224         if (strcmp (query_string, "") == 0 ||
225             strcmp (query_string, "*") == 0)
226         {
227             final_query = mail_query;
228         } else {
229             string_query = notmuch->query_parser->
230                 parse_query (query_string, NOTMUCH_QUERY_PARSER_FLAGS);
231             final_query = Xapian::Query (Xapian::Query::OP_AND,
232                                          mail_query, string_query);
233         }
234         messages->base.excluded_doc_ids = NULL;
235
236         if ((query->omit_excluded != NOTMUCH_EXCLUDE_FALSE) && (query->exclude_terms)) {
237             exclude_query = _notmuch_exclude_tags (query, final_query);
238
239             if (query->omit_excluded == NOTMUCH_EXCLUDE_TRUE ||
240                 query->omit_excluded == NOTMUCH_EXCLUDE_ALL)
241             {
242                 final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
243                                              final_query, exclude_query);
244             } else { /* NOTMUCH_EXCLUDE_FLAG */
245                 exclude_query = Xapian::Query (Xapian::Query::OP_AND,
246                                            exclude_query, final_query);
247
248                 enquire.set_weighting_scheme (Xapian::BoolWeight());
249                 enquire.set_query (exclude_query);
250
251                 mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
252
253                 GArray *excluded_doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
254
255                 for (iterator = mset.begin (); iterator != mset.end (); iterator++) {
256                     unsigned int doc_id = *iterator;
257                     g_array_append_val (excluded_doc_ids, doc_id);
258                 }
259                 messages->base.excluded_doc_ids = talloc (messages, _notmuch_doc_id_set);
260                 _notmuch_doc_id_set_init (query, messages->base.excluded_doc_ids,
261                                           excluded_doc_ids);
262                 g_array_unref (excluded_doc_ids);
263             }
264         }
265
266
267         enquire.set_weighting_scheme (Xapian::BoolWeight());
268
269         switch (query->sort) {
270         case NOTMUCH_SORT_OLDEST_FIRST:
271             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, FALSE);
272             break;
273         case NOTMUCH_SORT_NEWEST_FIRST:
274             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, TRUE);
275             break;
276         case NOTMUCH_SORT_MESSAGE_ID:
277             enquire.set_sort_by_value (NOTMUCH_VALUE_MESSAGE_ID, FALSE);
278             break;
279         case NOTMUCH_SORT_UNSORTED:
280             break;
281         }
282
283         if (_debug_query ()) {
284             fprintf (stderr, "Exclude query is:\n%s\n",
285                      exclude_query.get_description ().c_str ());
286             fprintf (stderr, "Final query is:\n%s\n",
287                      final_query.get_description ().c_str ());
288         }
289
290         enquire.set_query (final_query);
291
292         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
293
294         messages->iterator = mset.begin ();
295         messages->iterator_end = mset.end ();
296
297         *out = &messages->base;
298         return NOTMUCH_STATUS_SUCCESS;
299
300     } catch (const Xapian::Error &error) {
301         _notmuch_database_log (notmuch,
302                                "A Xapian exception occurred performing query: %s\n",
303                                error.get_msg().c_str());
304         _notmuch_database_log_append (notmuch,
305                                "Query string was: %s\n",
306                                query->query_string);
307
308         notmuch->exception_reported = TRUE;
309         talloc_free (messages);
310         return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
311     }
312 }
313
314 notmuch_bool_t
315 _notmuch_mset_messages_valid (notmuch_messages_t *messages)
316 {
317     notmuch_mset_messages_t *mset_messages;
318
319     mset_messages = (notmuch_mset_messages_t *) messages;
320
321     return (mset_messages->iterator != mset_messages->iterator_end);
322 }
323
324 static Xapian::docid
325 _notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
326 {
327     notmuch_mset_messages_t *mset_messages;
328
329     mset_messages = (notmuch_mset_messages_t *) messages;
330
331     if (! _notmuch_mset_messages_valid (&mset_messages->base))
332         return 0;
333
334     return *mset_messages->iterator;
335 }
336
337 notmuch_message_t *
338 _notmuch_mset_messages_get (notmuch_messages_t *messages)
339 {
340     notmuch_message_t *message;
341     Xapian::docid doc_id;
342     notmuch_private_status_t status;
343     notmuch_mset_messages_t *mset_messages;
344
345     mset_messages = (notmuch_mset_messages_t *) messages;
346
347     if (! _notmuch_mset_messages_valid (&mset_messages->base))
348         return NULL;
349
350     doc_id = *mset_messages->iterator;
351
352     message = _notmuch_message_create (mset_messages,
353                                        mset_messages->notmuch, doc_id,
354                                        &status);
355
356     if (message == NULL &&
357        status == NOTMUCH_PRIVATE_STATUS_NO_DOCUMENT_FOUND)
358     {
359         INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
360     }
361
362     if (messages->excluded_doc_ids &&
363         _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
364         notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
365
366     return message;
367 }
368
369 void
370 _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
371 {
372     notmuch_mset_messages_t *mset_messages;
373
374     mset_messages = (notmuch_mset_messages_t *) messages;
375
376     mset_messages->iterator++;
377 }
378
379 static notmuch_bool_t
380 _notmuch_doc_id_set_init (void *ctx,
381                           notmuch_doc_id_set_t *doc_ids,
382                           GArray *arr)
383 {
384     unsigned int max = 0;
385     unsigned char *bitmap;
386
387     for (unsigned int i = 0; i < arr->len; i++)
388         max = MAX(max, g_array_index (arr, unsigned int, i));
389     bitmap = talloc_zero_array (ctx, unsigned char, DOCIDSET_WORD(max) + 1);
390
391     if (bitmap == NULL)
392         return FALSE;
393
394     doc_ids->bitmap = bitmap;
395     doc_ids->bound = max + 1;
396
397     for (unsigned int i = 0; i < arr->len; i++) {
398         unsigned int doc_id = g_array_index (arr, unsigned int, i);
399         bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
400     }
401
402     return TRUE;
403 }
404
405 notmuch_bool_t
406 _notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
407                               unsigned int doc_id)
408 {
409     if (doc_id >= doc_ids->bound)
410         return FALSE;
411     return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
412 }
413
414 void
415 _notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
416                             unsigned int doc_id)
417 {
418     if (doc_id < doc_ids->bound)
419         doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
420 }
421
422 /* Glib objects force use to use a talloc destructor as well, (but not
423  * nearly as ugly as the for messages due to C++ objects). At
424  * this point, I'd really like to have some talloc-friendly
425  * equivalents for the few pieces of glib that I'm using. */
426 static int
427 _notmuch_threads_destructor (notmuch_threads_t *threads)
428 {
429     if (threads->doc_ids)
430         g_array_unref (threads->doc_ids);
431
432     return 0;
433 }
434
435
436 notmuch_threads_t *
437 notmuch_query_search_threads (notmuch_query_t *query)
438 {
439     notmuch_status_t status;
440     notmuch_threads_t *threads;
441     status = notmuch_query_search_threads_st (query, &threads);
442     if (status)
443         return NULL;
444     else
445         return threads;
446 }
447
448 notmuch_status_t
449 notmuch_query_search_threads_st (notmuch_query_t *query,
450                                  notmuch_threads_t **out)
451 {
452     notmuch_threads_t *threads;
453     notmuch_messages_t *messages;
454     notmuch_status_t status;
455
456     threads = talloc (query, notmuch_threads_t);
457     if (threads == NULL)
458         return NOTMUCH_STATUS_OUT_OF_MEMORY;
459     threads->doc_ids = NULL;
460     talloc_set_destructor (threads, _notmuch_threads_destructor);
461
462     threads->query = query;
463
464     status = notmuch_query_search_messages_st (query, &messages);
465     if (status) {
466         talloc_free (threads);
467         return status;
468     }
469
470     threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
471     while (notmuch_messages_valid (messages)) {
472         unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
473         g_array_append_val (threads->doc_ids, doc_id);
474         notmuch_messages_move_to_next (messages);
475     }
476     threads->doc_id_pos = 0;
477
478     talloc_free (messages);
479
480     if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
481                                     threads->doc_ids)) {
482         talloc_free (threads);
483         return NOTMUCH_STATUS_OUT_OF_MEMORY;
484     }
485
486     *out = threads;
487     return NOTMUCH_STATUS_SUCCESS;
488 }
489
490 void
491 notmuch_query_destroy (notmuch_query_t *query)
492 {
493     talloc_free (query);
494 }
495
496 notmuch_bool_t
497 notmuch_threads_valid (notmuch_threads_t *threads)
498 {
499     unsigned int doc_id;
500
501     if (! threads)
502         return FALSE;
503
504     while (threads->doc_id_pos < threads->doc_ids->len) {
505         doc_id = g_array_index (threads->doc_ids, unsigned int,
506                                 threads->doc_id_pos);
507         if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
508             break;
509
510         threads->doc_id_pos++;
511     }
512
513     return threads->doc_id_pos < threads->doc_ids->len;
514 }
515
516 notmuch_thread_t *
517 notmuch_threads_get (notmuch_threads_t *threads)
518 {
519     unsigned int doc_id;
520
521     if (! notmuch_threads_valid (threads))
522         return NULL;
523
524     doc_id = g_array_index (threads->doc_ids, unsigned int,
525                             threads->doc_id_pos);
526     return _notmuch_thread_create (threads->query,
527                                    threads->query->notmuch,
528                                    doc_id,
529                                    &threads->match_set,
530                                    threads->query->exclude_terms,
531                                    threads->query->omit_excluded,
532                                    threads->query->sort);
533 }
534
535 void
536 notmuch_threads_move_to_next (notmuch_threads_t *threads)
537 {
538     threads->doc_id_pos++;
539 }
540
541 void
542 notmuch_threads_destroy (notmuch_threads_t *threads)
543 {
544     talloc_free (threads);
545 }
546
547 unsigned int
548 notmuch_query_count_messages (notmuch_query_t *query)
549 {
550     notmuch_status_t status;
551     unsigned int count;
552
553     status = notmuch_query_count_messages_st (query, &count);
554     return status ? 0 : count;
555 }
556
557 notmuch_status_t
558 notmuch_query_count_messages_st (notmuch_query_t *query, unsigned *count_out)
559 {
560     return _notmuch_query_count_documents (query, "mail", count_out);
561 }
562
563 notmuch_status_t
564 _notmuch_query_count_documents (notmuch_query_t *query, const char *type, unsigned *count_out)
565 {
566     notmuch_database_t *notmuch = query->notmuch;
567     const char *query_string = query->query_string;
568     Xapian::doccount count = 0;
569
570     try {
571         Xapian::Enquire enquire (*notmuch->xapian_db);
572         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
573                                                    _find_prefix ("type"),
574                                                    type));
575         Xapian::Query string_query, final_query, exclude_query;
576         Xapian::MSet mset;
577
578         if (strcmp (query_string, "") == 0 ||
579             strcmp (query_string, "*") == 0)
580         {
581             final_query = mail_query;
582         } else {
583             string_query = notmuch->query_parser->
584                 parse_query (query_string, NOTMUCH_QUERY_PARSER_FLAGS);
585             final_query = Xapian::Query (Xapian::Query::OP_AND,
586                                          mail_query, string_query);
587         }
588
589         exclude_query = _notmuch_exclude_tags (query, final_query);
590
591         final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
592                                          final_query, exclude_query);
593
594         enquire.set_weighting_scheme(Xapian::BoolWeight());
595         enquire.set_docid_order(Xapian::Enquire::ASCENDING);
596
597         if (_debug_query ()) {
598             fprintf (stderr, "Exclude query is:\n%s\n",
599                      exclude_query.get_description ().c_str ());
600             fprintf (stderr, "Final query is:\n%s\n",
601                      final_query.get_description ().c_str ());
602         }
603
604         enquire.set_query (final_query);
605
606         /*
607          * Set the checkatleast parameter to the number of documents
608          * in the database to make get_matches_estimated() exact.
609          * Set the max parameter to 0 to avoid fetching documents we will discard.
610          */
611         mset = enquire.get_mset (0, 0,
612                                  notmuch->xapian_db->get_doccount ());
613
614         count = mset.get_matches_estimated();
615
616     } catch (const Xapian::Error &error) {
617         _notmuch_database_log (notmuch,
618                                "A Xapian exception occurred performing query: %s\n",
619                                error.get_msg().c_str());
620         _notmuch_database_log_append (notmuch,
621                                       "Query string was: %s\n",
622                                       query->query_string);
623         return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
624     }
625
626     *count_out = count;
627     return NOTMUCH_STATUS_SUCCESS;
628 }
629
630 unsigned
631 notmuch_query_count_threads (notmuch_query_t *query)
632 {
633     notmuch_status_t status;
634     unsigned int count;
635
636     status = notmuch_query_count_threads_st (query, &count);
637     return status ? 0 : count;
638 }
639
640 notmuch_status_t
641 notmuch_query_count_threads_st (notmuch_query_t *query, unsigned *count)
642 {
643     notmuch_messages_t *messages;
644     GHashTable *hash;
645     notmuch_sort_t sort;
646     notmuch_status_t ret = NOTMUCH_STATUS_SUCCESS;
647
648     sort = query->sort;
649     query->sort = NOTMUCH_SORT_UNSORTED;
650     ret = notmuch_query_search_messages_st (query, &messages);
651     if (ret)
652         return ret;
653     query->sort = sort;
654     if (messages == NULL)
655         return NOTMUCH_STATUS_XAPIAN_EXCEPTION;
656
657     hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
658     if (hash == NULL) {
659         talloc_free (messages);
660         return NOTMUCH_STATUS_OUT_OF_MEMORY;
661     }
662
663     while (notmuch_messages_valid (messages)) {
664         notmuch_message_t *message = notmuch_messages_get (messages);
665         const char *thread_id = notmuch_message_get_thread_id (message);
666         char *thread_id_copy = talloc_strdup (messages, thread_id);
667         if (unlikely (thread_id_copy == NULL)) {
668             notmuch_message_destroy (message);
669             ret = NOTMUCH_STATUS_OUT_OF_MEMORY;
670             goto DONE;
671         }
672         g_hash_table_insert (hash, thread_id_copy, NULL);
673         notmuch_message_destroy (message);
674         notmuch_messages_move_to_next (messages);
675     }
676
677     *count = g_hash_table_size (hash);
678
679   DONE:
680     g_hash_table_unref (hash);
681     talloc_free (messages);
682
683     return ret;
684 }
685
686 notmuch_database_t *
687 notmuch_query_get_database (const notmuch_query_t *query)
688 {
689     return query->notmuch;
690 }