ec60e2e45c01226a9ed326b8af22f1add8b77c8b
[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 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_database_t *notmuch = query->notmuch;
178     const char *query_string = query->query_string;
179     notmuch_mset_messages_t *messages;
180
181     messages = talloc (query, notmuch_mset_messages_t);
182     if (unlikely (messages == NULL))
183         return NULL;
184
185     try {
186
187         messages->base.is_of_list_type = FALSE;
188         messages->base.iterator = NULL;
189         messages->notmuch = notmuch;
190         new (&messages->iterator) Xapian::MSetIterator ();
191         new (&messages->iterator_end) Xapian::MSetIterator ();
192
193         talloc_set_destructor (messages, _notmuch_messages_destructor);
194
195         Xapian::Enquire enquire (*notmuch->xapian_db);
196         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
197                                                    _find_prefix ("type"),
198                                                    "mail"));
199         Xapian::Query string_query, final_query, exclude_query;
200         Xapian::MSet mset;
201         Xapian::MSetIterator iterator;
202         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
203                               Xapian::QueryParser::FLAG_PHRASE |
204                               Xapian::QueryParser::FLAG_LOVEHATE |
205                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
206                               Xapian::QueryParser::FLAG_WILDCARD |
207                               Xapian::QueryParser::FLAG_PURE_NOT);
208
209         if (strcmp (query_string, "") == 0 ||
210             strcmp (query_string, "*") == 0)
211         {
212             final_query = mail_query;
213         } else {
214             string_query = notmuch->query_parser->
215                 parse_query (query_string, flags);
216             final_query = Xapian::Query (Xapian::Query::OP_AND,
217                                          mail_query, string_query);
218         }
219         messages->base.excluded_doc_ids = NULL;
220
221         if ((query->omit_excluded != NOTMUCH_EXCLUDE_FALSE) && (query->exclude_terms)) {
222             exclude_query = _notmuch_exclude_tags (query, final_query);
223
224             if (query->omit_excluded == NOTMUCH_EXCLUDE_TRUE ||
225                 query->omit_excluded == NOTMUCH_EXCLUDE_ALL)
226             {
227                 final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
228                                              final_query, exclude_query);
229             } else { /* NOTMUCH_EXCLUDE_FLAG */
230                 exclude_query = Xapian::Query (Xapian::Query::OP_AND,
231                                            exclude_query, final_query);
232
233                 enquire.set_weighting_scheme (Xapian::BoolWeight());
234                 enquire.set_query (exclude_query);
235
236                 mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
237
238                 GArray *excluded_doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
239
240                 for (iterator = mset.begin (); iterator != mset.end (); iterator++) {
241                     unsigned int doc_id = *iterator;
242                     g_array_append_val (excluded_doc_ids, doc_id);
243                 }
244                 messages->base.excluded_doc_ids = talloc (messages, _notmuch_doc_id_set);
245                 _notmuch_doc_id_set_init (query, messages->base.excluded_doc_ids,
246                                           excluded_doc_ids);
247                 g_array_unref (excluded_doc_ids);
248             }
249         }
250
251
252         enquire.set_weighting_scheme (Xapian::BoolWeight());
253
254         switch (query->sort) {
255         case NOTMUCH_SORT_OLDEST_FIRST:
256             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, FALSE);
257             break;
258         case NOTMUCH_SORT_NEWEST_FIRST:
259             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, TRUE);
260             break;
261         case NOTMUCH_SORT_MESSAGE_ID:
262             enquire.set_sort_by_value (NOTMUCH_VALUE_MESSAGE_ID, FALSE);
263             break;
264         case NOTMUCH_SORT_UNSORTED:
265             break;
266         }
267
268         if (_debug_query ()) {
269             fprintf (stderr, "Exclude query is:\n%s\n",
270                      exclude_query.get_description ().c_str ());
271             fprintf (stderr, "Final query is:\n%s\n",
272                      final_query.get_description ().c_str ());
273         }
274
275         enquire.set_query (final_query);
276
277         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
278
279         messages->iterator = mset.begin ();
280         messages->iterator_end = mset.end ();
281
282         return &messages->base;
283
284     } catch (const Xapian::Error &error) {
285         fprintf (stderr, "A Xapian exception occurred performing query: %s\n",
286                  error.get_msg().c_str());
287         fprintf (stderr, "Query string was: %s\n", query->query_string);
288         notmuch->exception_reported = TRUE;
289         talloc_free (messages);
290         return NULL;
291     }
292 }
293
294 notmuch_bool_t
295 _notmuch_mset_messages_valid (notmuch_messages_t *messages)
296 {
297     notmuch_mset_messages_t *mset_messages;
298
299     mset_messages = (notmuch_mset_messages_t *) messages;
300
301     return (mset_messages->iterator != mset_messages->iterator_end);
302 }
303
304 static Xapian::docid
305 _notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
306 {
307     notmuch_mset_messages_t *mset_messages;
308
309     mset_messages = (notmuch_mset_messages_t *) messages;
310
311     if (! _notmuch_mset_messages_valid (&mset_messages->base))
312         return 0;
313
314     return *mset_messages->iterator;
315 }
316
317 notmuch_message_t *
318 _notmuch_mset_messages_get (notmuch_messages_t *messages)
319 {
320     notmuch_message_t *message;
321     Xapian::docid doc_id;
322     notmuch_private_status_t status;
323     notmuch_mset_messages_t *mset_messages;
324
325     mset_messages = (notmuch_mset_messages_t *) messages;
326
327     if (! _notmuch_mset_messages_valid (&mset_messages->base))
328         return NULL;
329
330     doc_id = *mset_messages->iterator;
331
332     message = _notmuch_message_create (mset_messages,
333                                        mset_messages->notmuch, doc_id,
334                                        &status);
335
336     if (message == NULL &&
337        status == NOTMUCH_PRIVATE_STATUS_NO_DOCUMENT_FOUND)
338     {
339         INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
340     }
341
342     if (messages->excluded_doc_ids &&
343         _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
344         notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
345
346     return message;
347 }
348
349 void
350 _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
351 {
352     notmuch_mset_messages_t *mset_messages;
353
354     mset_messages = (notmuch_mset_messages_t *) messages;
355
356     mset_messages->iterator++;
357 }
358
359 static notmuch_bool_t
360 _notmuch_doc_id_set_init (void *ctx,
361                           notmuch_doc_id_set_t *doc_ids,
362                           GArray *arr)
363 {
364     unsigned int max = 0;
365     unsigned char *bitmap;
366
367     for (unsigned int i = 0; i < arr->len; i++)
368         max = MAX(max, g_array_index (arr, unsigned int, i));
369     bitmap = talloc_zero_array (ctx, unsigned char, DOCIDSET_WORD(max) + 1);
370
371     if (bitmap == NULL)
372         return FALSE;
373
374     doc_ids->bitmap = bitmap;
375     doc_ids->bound = max + 1;
376
377     for (unsigned int i = 0; i < arr->len; i++) {
378         unsigned int doc_id = g_array_index (arr, unsigned int, i);
379         bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
380     }
381
382     return TRUE;
383 }
384
385 notmuch_bool_t
386 _notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
387                               unsigned int doc_id)
388 {
389     if (doc_id >= doc_ids->bound)
390         return FALSE;
391     return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
392 }
393
394 void
395 _notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
396                             unsigned int doc_id)
397 {
398     if (doc_id < doc_ids->bound)
399         doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
400 }
401
402 /* Glib objects force use to use a talloc destructor as well, (but not
403  * nearly as ugly as the for messages due to C++ objects). At
404  * this point, I'd really like to have some talloc-friendly
405  * equivalents for the few pieces of glib that I'm using. */
406 static int
407 _notmuch_threads_destructor (notmuch_threads_t *threads)
408 {
409     if (threads->doc_ids)
410         g_array_unref (threads->doc_ids);
411
412     return 0;
413 }
414
415 notmuch_threads_t *
416 notmuch_query_search_threads (notmuch_query_t *query)
417 {
418     notmuch_threads_t *threads;
419     notmuch_messages_t *messages;
420
421     threads = talloc (query, notmuch_threads_t);
422     if (threads == NULL)
423         return NULL;
424     threads->doc_ids = NULL;
425     talloc_set_destructor (threads, _notmuch_threads_destructor);
426
427     threads->query = query;
428
429     messages = notmuch_query_search_messages (query);
430     if (messages == NULL) {
431             talloc_free (threads);
432             return NULL;
433     }
434
435     threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
436     while (notmuch_messages_valid (messages)) {
437         unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
438         g_array_append_val (threads->doc_ids, doc_id);
439         notmuch_messages_move_to_next (messages);
440     }
441     threads->doc_id_pos = 0;
442
443     talloc_free (messages);
444
445     if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
446                                     threads->doc_ids)) {
447         talloc_free (threads);
448         return NULL;
449     }
450
451     return threads;
452 }
453
454 void
455 notmuch_query_destroy (notmuch_query_t *query)
456 {
457     talloc_free (query);
458 }
459
460 notmuch_bool_t
461 notmuch_threads_valid (notmuch_threads_t *threads)
462 {
463     unsigned int doc_id;
464
465     while (threads->doc_id_pos < threads->doc_ids->len) {
466         doc_id = g_array_index (threads->doc_ids, unsigned int,
467                                 threads->doc_id_pos);
468         if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
469             break;
470
471         threads->doc_id_pos++;
472     }
473
474     return threads->doc_id_pos < threads->doc_ids->len;
475 }
476
477 notmuch_thread_t *
478 notmuch_threads_get (notmuch_threads_t *threads)
479 {
480     unsigned int doc_id;
481
482     if (! notmuch_threads_valid (threads))
483         return NULL;
484
485     doc_id = g_array_index (threads->doc_ids, unsigned int,
486                             threads->doc_id_pos);
487     return _notmuch_thread_create (threads->query,
488                                    threads->query->notmuch,
489                                    doc_id,
490                                    &threads->match_set,
491                                    threads->query->exclude_terms,
492                                    threads->query->omit_excluded,
493                                    threads->query->sort);
494 }
495
496 void
497 notmuch_threads_move_to_next (notmuch_threads_t *threads)
498 {
499     threads->doc_id_pos++;
500 }
501
502 void
503 notmuch_threads_destroy (notmuch_threads_t *threads)
504 {
505     talloc_free (threads);
506 }
507
508 unsigned
509 notmuch_query_count_messages (notmuch_query_t *query)
510 {
511     notmuch_database_t *notmuch = query->notmuch;
512     const char *query_string = query->query_string;
513     Xapian::doccount count = 0;
514
515     try {
516         Xapian::Enquire enquire (*notmuch->xapian_db);
517         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
518                                                    _find_prefix ("type"),
519                                                    "mail"));
520         Xapian::Query string_query, final_query, exclude_query;
521         Xapian::MSet mset;
522         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
523                               Xapian::QueryParser::FLAG_PHRASE |
524                               Xapian::QueryParser::FLAG_LOVEHATE |
525                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
526                               Xapian::QueryParser::FLAG_WILDCARD |
527                               Xapian::QueryParser::FLAG_PURE_NOT);
528
529         if (strcmp (query_string, "") == 0 ||
530             strcmp (query_string, "*") == 0)
531         {
532             final_query = mail_query;
533         } else {
534             string_query = notmuch->query_parser->
535                 parse_query (query_string, flags);
536             final_query = Xapian::Query (Xapian::Query::OP_AND,
537                                          mail_query, string_query);
538         }
539
540         exclude_query = _notmuch_exclude_tags (query, final_query);
541
542         final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
543                                          final_query, exclude_query);
544
545         enquire.set_weighting_scheme(Xapian::BoolWeight());
546         enquire.set_docid_order(Xapian::Enquire::ASCENDING);
547
548         if (_debug_query ()) {
549             fprintf (stderr, "Exclude query is:\n%s\n",
550                      exclude_query.get_description ().c_str ());
551             fprintf (stderr, "Final query is:\n%s\n",
552                      final_query.get_description ().c_str ());
553         }
554
555         enquire.set_query (final_query);
556
557         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
558
559         count = mset.get_matches_estimated();
560
561     } catch (const Xapian::Error &error) {
562         fprintf (stderr, "A Xapian exception occurred: %s\n",
563                  error.get_msg().c_str());
564         fprintf (stderr, "Query string was: %s\n", query->query_string);
565     }
566
567     return count;
568 }
569
570 unsigned
571 notmuch_query_count_threads (notmuch_query_t *query)
572 {
573     notmuch_messages_t *messages;
574     GHashTable *hash;
575     unsigned int count;
576     notmuch_sort_t sort;
577
578     sort = query->sort;
579     query->sort = NOTMUCH_SORT_UNSORTED;
580     messages = notmuch_query_search_messages (query);
581     query->sort = sort;
582     if (messages == NULL)
583         return 0;
584
585     hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
586     if (hash == NULL) {
587         talloc_free (messages);
588         return 0;
589     }
590
591     while (notmuch_messages_valid (messages)) {
592         notmuch_message_t *message = notmuch_messages_get (messages);
593         const char *thread_id = notmuch_message_get_thread_id (message);
594         char *thread_id_copy = talloc_strdup (messages, thread_id);
595         if (unlikely (thread_id_copy == NULL)) {
596             notmuch_message_destroy (message);
597             count = 0;
598             goto DONE;
599         }
600         g_hash_table_insert (hash, thread_id_copy, NULL);
601         notmuch_message_destroy (message);
602         notmuch_messages_move_to_next (messages);
603     }
604
605     count = g_hash_table_size (hash);
606
607   DONE:
608     g_hash_table_unref (hash);
609     talloc_free (messages);
610
611     return count;
612 }