lib: Expose query debug output via an environment variable
[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_bool_t omit_excluded_messages;
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 int *bitmap;
43     unsigned int bound;
44 };
45
46 #define DOCIDSET_WORD(bit) ((bit) / sizeof (unsigned int))
47 #define DOCIDSET_BIT(bit) ((bit) % sizeof (unsigned int))
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 (NULL, 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_messages = FALSE;
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_messages (notmuch_query_t *query, notmuch_bool_t omit)
108 {
109     query->omit_excluded_messages = omit;
110 }
111
112 void
113 notmuch_query_set_sort (notmuch_query_t *query, notmuch_sort_t sort)
114 {
115     query->sort = sort;
116 }
117
118 notmuch_sort_t
119 notmuch_query_get_sort (notmuch_query_t *query)
120 {
121     return query->sort;
122 }
123
124 void
125 notmuch_query_add_tag_exclude (notmuch_query_t *query, const char *tag)
126 {
127     char *term = talloc_asprintf (query, "%s%s", _find_prefix ("tag"), tag);
128     _notmuch_string_list_append (query->exclude_terms, term);
129 }
130
131 /* We end up having to call the destructors explicitly because we had
132  * to use "placement new" in order to initialize C++ objects within a
133  * block that we allocated with talloc. So C++ is making talloc
134  * slightly less simple to use, (we wouldn't need
135  * talloc_set_destructor at all otherwise).
136  */
137 static int
138 _notmuch_messages_destructor (notmuch_mset_messages_t *messages)
139 {
140     messages->iterator.~MSetIterator ();
141     messages->iterator_end.~MSetIterator ();
142
143     return 0;
144 }
145
146 /* Return a query that matches messages with the excluded tags
147  * registered with query.  Any tags that explicitly appear in xquery
148  * will not be excluded, and will be removed from the list of exclude
149  * tags.  The caller of this function has to combine the returned
150  * query appropriately.*/
151 static Xapian::Query
152 _notmuch_exclude_tags (notmuch_query_t *query, Xapian::Query xquery)
153 {
154     Xapian::Query exclude_query = Xapian::Query::MatchNothing;
155
156     for (notmuch_string_node_t *term = query->exclude_terms->head; term;
157          term = term->next) {
158         Xapian::TermIterator it = xquery.get_terms_begin ();
159         Xapian::TermIterator end = xquery.get_terms_end ();
160         for (; it != end; it++) {
161             if ((*it).compare (term->string) == 0)
162                 break;
163         }
164         if (it == end)
165             exclude_query = Xapian::Query (Xapian::Query::OP_OR,
166                                     exclude_query, Xapian::Query (term->string));
167         else
168             term->string = talloc_strdup (query, "");
169     }
170     return exclude_query;
171 }
172
173 notmuch_messages_t *
174 notmuch_query_search_messages (notmuch_query_t *query)
175 {
176     notmuch_database_t *notmuch = query->notmuch;
177     const char *query_string = query->query_string;
178     notmuch_mset_messages_t *messages;
179
180     messages = talloc (query, notmuch_mset_messages_t);
181     if (unlikely (messages == NULL))
182         return NULL;
183
184     try {
185
186         messages->base.is_of_list_type = FALSE;
187         messages->base.iterator = NULL;
188         messages->notmuch = notmuch;
189         new (&messages->iterator) Xapian::MSetIterator ();
190         new (&messages->iterator_end) Xapian::MSetIterator ();
191
192         talloc_set_destructor (messages, _notmuch_messages_destructor);
193
194         Xapian::Enquire enquire (*notmuch->xapian_db);
195         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
196                                                    _find_prefix ("type"),
197                                                    "mail"));
198         Xapian::Query string_query, final_query, exclude_query;
199         Xapian::MSet mset;
200         Xapian::MSetIterator iterator;
201         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
202                               Xapian::QueryParser::FLAG_PHRASE |
203                               Xapian::QueryParser::FLAG_LOVEHATE |
204                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
205                               Xapian::QueryParser::FLAG_WILDCARD |
206                               Xapian::QueryParser::FLAG_PURE_NOT);
207
208         if (strcmp (query_string, "") == 0 ||
209             strcmp (query_string, "*") == 0)
210         {
211             final_query = mail_query;
212         } else {
213             string_query = notmuch->query_parser->
214                 parse_query (query_string, flags);
215             final_query = Xapian::Query (Xapian::Query::OP_AND,
216                                          mail_query, string_query);
217         }
218         messages->base.excluded_doc_ids = NULL;
219
220         if (query->exclude_terms) {
221             exclude_query = _notmuch_exclude_tags (query, final_query);
222             exclude_query = Xapian::Query (Xapian::Query::OP_AND,
223                                            exclude_query, final_query);
224
225             if (query->omit_excluded_messages)
226                 final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
227                                              final_query, exclude_query);
228             else {
229                 enquire.set_weighting_scheme (Xapian::BoolWeight());
230                 enquire.set_query (exclude_query);
231
232                 mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
233
234                 GArray *excluded_doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
235
236                 for (iterator = mset.begin (); iterator != mset.end (); iterator++) {
237                     unsigned int doc_id = *iterator;
238                     g_array_append_val (excluded_doc_ids, doc_id);
239                 }
240                 messages->base.excluded_doc_ids = talloc (messages, _notmuch_doc_id_set);
241                 _notmuch_doc_id_set_init (query, messages->base.excluded_doc_ids,
242                                           excluded_doc_ids);
243                 g_array_unref (excluded_doc_ids);
244             }
245         }
246
247
248         enquire.set_weighting_scheme (Xapian::BoolWeight());
249
250         switch (query->sort) {
251         case NOTMUCH_SORT_OLDEST_FIRST:
252             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, FALSE);
253             break;
254         case NOTMUCH_SORT_NEWEST_FIRST:
255             enquire.set_sort_by_value (NOTMUCH_VALUE_TIMESTAMP, TRUE);
256             break;
257         case NOTMUCH_SORT_MESSAGE_ID:
258             enquire.set_sort_by_value (NOTMUCH_VALUE_MESSAGE_ID, FALSE);
259             break;
260         case NOTMUCH_SORT_UNSORTED:
261             break;
262         }
263
264         if (_debug_query ())
265             fprintf (stderr, "Final query is:\n%s\n",
266                      final_query.get_description ().c_str ());
267
268         enquire.set_query (final_query);
269
270         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
271
272         messages->iterator = mset.begin ();
273         messages->iterator_end = mset.end ();
274
275         return &messages->base;
276
277     } catch (const Xapian::Error &error) {
278         fprintf (stderr, "A Xapian exception occurred performing query: %s\n",
279                  error.get_msg().c_str());
280         fprintf (stderr, "Query string was: %s\n", query->query_string);
281         notmuch->exception_reported = TRUE;
282         talloc_free (messages);
283         return NULL;
284     }
285 }
286
287 notmuch_bool_t
288 _notmuch_mset_messages_valid (notmuch_messages_t *messages)
289 {
290     notmuch_mset_messages_t *mset_messages;
291
292     mset_messages = (notmuch_mset_messages_t *) messages;
293
294     return (mset_messages->iterator != mset_messages->iterator_end);
295 }
296
297 static Xapian::docid
298 _notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
299 {
300     notmuch_mset_messages_t *mset_messages;
301
302     mset_messages = (notmuch_mset_messages_t *) messages;
303
304     if (! _notmuch_mset_messages_valid (&mset_messages->base))
305         return 0;
306
307     return *mset_messages->iterator;
308 }
309
310 notmuch_message_t *
311 _notmuch_mset_messages_get (notmuch_messages_t *messages)
312 {
313     notmuch_message_t *message;
314     Xapian::docid doc_id;
315     notmuch_private_status_t status;
316     notmuch_mset_messages_t *mset_messages;
317
318     mset_messages = (notmuch_mset_messages_t *) messages;
319
320     if (! _notmuch_mset_messages_valid (&mset_messages->base))
321         return NULL;
322
323     doc_id = *mset_messages->iterator;
324
325     message = _notmuch_message_create (mset_messages,
326                                        mset_messages->notmuch, doc_id,
327                                        &status);
328
329     if (message == NULL &&
330        status == NOTMUCH_PRIVATE_STATUS_NO_DOCUMENT_FOUND)
331     {
332         INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
333     }
334
335     if (messages->excluded_doc_ids &&
336         _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
337         notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
338
339     return message;
340 }
341
342 void
343 _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
344 {
345     notmuch_mset_messages_t *mset_messages;
346
347     mset_messages = (notmuch_mset_messages_t *) messages;
348
349     mset_messages->iterator++;
350 }
351
352 static notmuch_bool_t
353 _notmuch_doc_id_set_init (void *ctx,
354                           notmuch_doc_id_set_t *doc_ids,
355                           GArray *arr)
356 {
357     unsigned int max = 0;
358     unsigned int *bitmap;
359
360     for (unsigned int i = 0; i < arr->len; i++)
361         max = MAX(max, g_array_index (arr, unsigned int, i));
362     bitmap = talloc_zero_array (ctx, unsigned int, 1 + max / sizeof (*bitmap));
363
364     if (bitmap == NULL)
365         return FALSE;
366
367     doc_ids->bitmap = bitmap;
368     doc_ids->bound = max + 1;
369
370     for (unsigned int i = 0; i < arr->len; i++) {
371         unsigned int doc_id = g_array_index (arr, unsigned int, i);
372         bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
373     }
374
375     return TRUE;
376 }
377
378 notmuch_bool_t
379 _notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
380                               unsigned int doc_id)
381 {
382     if (doc_id >= doc_ids->bound)
383         return FALSE;
384     return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
385 }
386
387 void
388 _notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
389                             unsigned int doc_id)
390 {
391     if (doc_id < doc_ids->bound)
392         doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
393 }
394
395 /* Glib objects force use to use a talloc destructor as well, (but not
396  * nearly as ugly as the for messages due to C++ objects). At
397  * this point, I'd really like to have some talloc-friendly
398  * equivalents for the few pieces of glib that I'm using. */
399 static int
400 _notmuch_threads_destructor (notmuch_threads_t *threads)
401 {
402     if (threads->doc_ids)
403         g_array_unref (threads->doc_ids);
404
405     return 0;
406 }
407
408 notmuch_threads_t *
409 notmuch_query_search_threads (notmuch_query_t *query)
410 {
411     notmuch_threads_t *threads;
412     notmuch_messages_t *messages;
413
414     threads = talloc (query, notmuch_threads_t);
415     if (threads == NULL)
416         return NULL;
417     threads->doc_ids = NULL;
418     talloc_set_destructor (threads, _notmuch_threads_destructor);
419
420     threads->query = query;
421
422     messages = notmuch_query_search_messages (query);
423     if (messages == NULL) {
424             talloc_free (threads);
425             return NULL;
426     }
427
428     threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
429     while (notmuch_messages_valid (messages)) {
430         unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
431         g_array_append_val (threads->doc_ids, doc_id);
432         notmuch_messages_move_to_next (messages);
433     }
434     threads->doc_id_pos = 0;
435
436     talloc_free (messages);
437
438     if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
439                                     threads->doc_ids)) {
440         talloc_free (threads);
441         return NULL;
442     }
443
444     return threads;
445 }
446
447 void
448 notmuch_query_destroy (notmuch_query_t *query)
449 {
450     talloc_free (query);
451 }
452
453 notmuch_bool_t
454 notmuch_threads_valid (notmuch_threads_t *threads)
455 {
456     unsigned int doc_id;
457
458     while (threads->doc_id_pos < threads->doc_ids->len) {
459         doc_id = g_array_index (threads->doc_ids, unsigned int,
460                                 threads->doc_id_pos);
461         if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
462             break;
463
464         threads->doc_id_pos++;
465     }
466
467     return threads->doc_id_pos < threads->doc_ids->len;
468 }
469
470 notmuch_thread_t *
471 notmuch_threads_get (notmuch_threads_t *threads)
472 {
473     unsigned int doc_id;
474
475     if (! notmuch_threads_valid (threads))
476         return NULL;
477
478     doc_id = g_array_index (threads->doc_ids, unsigned int,
479                             threads->doc_id_pos);
480     return _notmuch_thread_create (threads->query,
481                                    threads->query->notmuch,
482                                    doc_id,
483                                    &threads->match_set,
484                                    threads->query->exclude_terms,
485                                    threads->query->sort);
486 }
487
488 void
489 notmuch_threads_move_to_next (notmuch_threads_t *threads)
490 {
491     threads->doc_id_pos++;
492 }
493
494 void
495 notmuch_threads_destroy (notmuch_threads_t *threads)
496 {
497     talloc_free (threads);
498 }
499
500 unsigned
501 notmuch_query_count_messages (notmuch_query_t *query)
502 {
503     notmuch_database_t *notmuch = query->notmuch;
504     const char *query_string = query->query_string;
505     Xapian::doccount count = 0;
506
507     try {
508         Xapian::Enquire enquire (*notmuch->xapian_db);
509         Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
510                                                    _find_prefix ("type"),
511                                                    "mail"));
512         Xapian::Query string_query, final_query, exclude_query;
513         Xapian::MSet mset;
514         unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
515                               Xapian::QueryParser::FLAG_PHRASE |
516                               Xapian::QueryParser::FLAG_LOVEHATE |
517                               Xapian::QueryParser::FLAG_BOOLEAN_ANY_CASE |
518                               Xapian::QueryParser::FLAG_WILDCARD |
519                               Xapian::QueryParser::FLAG_PURE_NOT);
520
521         if (strcmp (query_string, "") == 0 ||
522             strcmp (query_string, "*") == 0)
523         {
524             final_query = mail_query;
525         } else {
526             string_query = notmuch->query_parser->
527                 parse_query (query_string, flags);
528             final_query = Xapian::Query (Xapian::Query::OP_AND,
529                                          mail_query, string_query);
530         }
531
532         exclude_query = _notmuch_exclude_tags (query, final_query);
533
534         final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
535                                          final_query, exclude_query);
536
537         enquire.set_weighting_scheme(Xapian::BoolWeight());
538         enquire.set_docid_order(Xapian::Enquire::ASCENDING);
539
540         if (_debug_query ())
541             fprintf (stderr, "Final query is:\n%s\n",
542                      final_query.get_description ().c_str ());
543
544         enquire.set_query (final_query);
545
546         mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
547
548         count = mset.get_matches_estimated();
549
550     } catch (const Xapian::Error &error) {
551         fprintf (stderr, "A Xapian exception occurred: %s\n",
552                  error.get_msg().c_str());
553         fprintf (stderr, "Query string was: %s\n", query->query_string);
554     }
555
556     return count;
557 }
558
559 unsigned
560 notmuch_query_count_threads (notmuch_query_t *query)
561 {
562     notmuch_messages_t *messages;
563     GHashTable *hash;
564     unsigned int count;
565     notmuch_sort_t sort;
566
567     sort = query->sort;
568     query->sort = NOTMUCH_SORT_UNSORTED;
569     messages = notmuch_query_search_messages (query);
570     query->sort = sort;
571     if (messages == NULL)
572         return 0;
573
574     hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
575     if (hash == NULL) {
576         talloc_free (messages);
577         return 0;
578     }
579
580     while (notmuch_messages_valid (messages)) {
581         notmuch_message_t *message = notmuch_messages_get (messages);
582         const char *thread_id = notmuch_message_get_thread_id (message);
583         char *thread_id_copy = talloc_strdup (messages, thread_id);
584         if (unlikely (thread_id_copy == NULL)) {
585             notmuch_message_destroy (message);
586             count = 0;
587             goto DONE;
588         }
589         g_hash_table_insert (hash, thread_id_copy, NULL);
590         notmuch_message_destroy (message);
591         notmuch_messages_move_to_next (messages);
592     }
593
594     count = g_hash_table_size (hash);
595
596   DONE:
597     g_hash_table_unref (hash);
598     talloc_free (messages);
599
600     return count;
601 }