]> git.notmuchmail.org Git - notmuch/blobdiff - lib/query.cc
Merge branch 'release'
[notmuch] / lib / query.cc
index 9acf2b03b84da6729f78a8e6ee843a7eb35d0afd..7381a545eda132150461f467d0b6055fb7caa8e8 100644 (file)
 
 #include <glib.h> /* GHashTable, GPtrArray */
 
-#include <xapian.h>
-
 struct _notmuch_query {
     notmuch_database_t *notmuch;
     const char *query_string;
     notmuch_sort_t sort;
+    notmuch_string_list_t *exclude_terms;
+    notmuch_bool_t omit_excluded;
 };
 
 typedef struct _notmuch_mset_messages {
@@ -38,24 +38,47 @@ typedef struct _notmuch_mset_messages {
     Xapian::MSetIterator iterator_end;
 } notmuch_mset_messages_t;
 
-struct _notmuch_threads {
+struct _notmuch_doc_id_set {
+    unsigned char *bitmap;
+    unsigned int bound;
+};
+
+#define DOCIDSET_WORD(bit) ((bit) / CHAR_BIT)
+#define DOCIDSET_BIT(bit) ((bit) % CHAR_BIT)
+
+struct visible _notmuch_threads {
     notmuch_query_t *query;
-    GHashTable *threads;
-    notmuch_messages_t *messages;
 
-    /* This thread ID is our iterator state. */
-    const char *thread_id;
+    /* The ordered list of doc ids matched by the query. */
+    GArray *doc_ids;
+    /* Our iterator's current position in doc_ids. */
+    unsigned int doc_id_pos;
+    /* The set of matched docid's that have not been assigned to a
+     * thread. Initially, this contains every docid in doc_ids. */
+    notmuch_doc_id_set_t match_set;
 };
 
+/* We need this in the message functions so forward declare. */
+static notmuch_bool_t
+_notmuch_doc_id_set_init (void *ctx,
+                         notmuch_doc_id_set_t *doc_ids,
+                         GArray *arr);
+
+static notmuch_bool_t
+_debug_query (void)
+{
+    char *env = getenv ("NOTMUCH_DEBUG_QUERY");
+    return (env && strcmp (env, "") != 0);
+}
+
 notmuch_query_t *
 notmuch_query_create (notmuch_database_t *notmuch,
                      const char *query_string)
 {
     notmuch_query_t *query;
 
-#ifdef DEBUG_QUERY
-    fprintf (stderr, "Query string is:\n%s\n", query_string);
-#endif
+    if (_debug_query ())
+       fprintf (stderr, "Query string is:\n%s\n", query_string);
 
     query = talloc (NULL, notmuch_query_t);
     if (unlikely (query == NULL))
@@ -67,6 +90,10 @@ notmuch_query_create (notmuch_database_t *notmuch,
 
     query->sort = NOTMUCH_SORT_NEWEST_FIRST;
 
+    query->exclude_terms = _notmuch_string_list_create (query);
+
+    query->omit_excluded = TRUE;
+
     return query;
 }
 
@@ -76,6 +103,12 @@ notmuch_query_get_query_string (notmuch_query_t *query)
     return query->query_string;
 }
 
+void
+notmuch_query_set_omit_excluded (notmuch_query_t *query, notmuch_bool_t omit_excluded)
+{
+    query->omit_excluded = omit_excluded;
+}
+
 void
 notmuch_query_set_sort (notmuch_query_t *query, notmuch_sort_t sort)
 {
@@ -88,6 +121,13 @@ notmuch_query_get_sort (notmuch_query_t *query)
     return query->sort;
 }
 
+void
+notmuch_query_add_tag_exclude (notmuch_query_t *query, const char *tag)
+{
+    char *term = talloc_asprintf (query, "%s%s", _find_prefix ("tag"), tag);
+    _notmuch_string_list_append (query->exclude_terms, term);
+}
+
 /* We end up having to call the destructors explicitly because we had
  * to use "placement new" in order to initialize C++ objects within a
  * block that we allocated with talloc. So C++ is making talloc
@@ -103,6 +143,33 @@ _notmuch_messages_destructor (notmuch_mset_messages_t *messages)
     return 0;
 }
 
+/* Return a query that matches messages with the excluded tags
+ * registered with query.  Any tags that explicitly appear in xquery
+ * will not be excluded, and will be removed from the list of exclude
+ * tags.  The caller of this function has to combine the returned
+ * query appropriately.*/
+static Xapian::Query
+_notmuch_exclude_tags (notmuch_query_t *query, Xapian::Query xquery)
+{
+    Xapian::Query exclude_query = Xapian::Query::MatchNothing;
+
+    for (notmuch_string_node_t *term = query->exclude_terms->head; term;
+        term = term->next) {
+       Xapian::TermIterator it = xquery.get_terms_begin ();
+       Xapian::TermIterator end = xquery.get_terms_end ();
+       for (; it != end; it++) {
+           if ((*it).compare (term->string) == 0)
+               break;
+       }
+       if (it == end)
+           exclude_query = Xapian::Query (Xapian::Query::OP_OR,
+                                   exclude_query, Xapian::Query (term->string));
+       else
+           term->string = talloc_strdup (query, "");
+    }
+    return exclude_query;
+}
+
 notmuch_messages_t *
 notmuch_query_search_messages (notmuch_query_t *query)
 {
@@ -128,8 +195,9 @@ notmuch_query_search_messages (notmuch_query_t *query)
        Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
                                                   _find_prefix ("type"),
                                                   "mail"));
-       Xapian::Query string_query, final_query;
+       Xapian::Query string_query, final_query, exclude_query;
        Xapian::MSet mset;
+       Xapian::MSetIterator iterator;
        unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
                              Xapian::QueryParser::FLAG_PHRASE |
                              Xapian::QueryParser::FLAG_LOVEHATE |
@@ -147,6 +215,36 @@ notmuch_query_search_messages (notmuch_query_t *query)
            final_query = Xapian::Query (Xapian::Query::OP_AND,
                                         mail_query, string_query);
        }
+       messages->base.excluded_doc_ids = NULL;
+
+       if (query->exclude_terms) {
+           exclude_query = _notmuch_exclude_tags (query, final_query);
+
+           if (query->omit_excluded)
+               final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
+                                            final_query, exclude_query);
+           else {
+               exclude_query = Xapian::Query (Xapian::Query::OP_AND,
+                                          exclude_query, final_query);
+
+               enquire.set_weighting_scheme (Xapian::BoolWeight());
+               enquire.set_query (exclude_query);
+
+               mset = enquire.get_mset (0, notmuch->xapian_db->get_doccount ());
+
+               GArray *excluded_doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
+
+               for (iterator = mset.begin (); iterator != mset.end (); iterator++) {
+                   unsigned int doc_id = *iterator;
+                   g_array_append_val (excluded_doc_ids, doc_id);
+               }
+               messages->base.excluded_doc_ids = talloc (messages, _notmuch_doc_id_set);
+               _notmuch_doc_id_set_init (query, messages->base.excluded_doc_ids,
+                                         excluded_doc_ids);
+               g_array_unref (excluded_doc_ids);
+           }
+       }
+
 
        enquire.set_weighting_scheme (Xapian::BoolWeight());
 
@@ -164,9 +262,12 @@ notmuch_query_search_messages (notmuch_query_t *query)
            break;
        }
 
-#if DEBUG_QUERY
-       fprintf (stderr, "Final query is:\n%s\n", final_query.get_description().c_str());
-#endif
+       if (_debug_query ()) {
+           fprintf (stderr, "Exclude query is:\n%s\n",
+                    exclude_query.get_description ().c_str ());
+           fprintf (stderr, "Final query is:\n%s\n",
+                    final_query.get_description ().c_str ());
+       }
 
        enquire.set_query (final_query);
 
@@ -197,6 +298,19 @@ _notmuch_mset_messages_valid (notmuch_messages_t *messages)
     return (mset_messages->iterator != mset_messages->iterator_end);
 }
 
+static Xapian::docid
+_notmuch_mset_messages_get_doc_id (notmuch_messages_t *messages)
+{
+    notmuch_mset_messages_t *mset_messages;
+
+    mset_messages = (notmuch_mset_messages_t *) messages;
+
+    if (! _notmuch_mset_messages_valid (&mset_messages->base))
+       return 0;
+
+    return *mset_messages->iterator;
+}
+
 notmuch_message_t *
 _notmuch_mset_messages_get (notmuch_messages_t *messages)
 {
@@ -222,6 +336,10 @@ _notmuch_mset_messages_get (notmuch_messages_t *messages)
        INTERNAL_ERROR ("a messages iterator contains a non-existent document ID.\n");
     }
 
+    if (messages->excluded_doc_ids &&
+       _notmuch_doc_id_set_contains (messages->excluded_doc_ids, doc_id))
+       notmuch_message_set_flag (message, NOTMUCH_MESSAGE_FLAG_EXCLUDED, TRUE);
+
     return message;
 }
 
@@ -235,6 +353,49 @@ _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
     mset_messages->iterator++;
 }
 
+static notmuch_bool_t
+_notmuch_doc_id_set_init (void *ctx,
+                         notmuch_doc_id_set_t *doc_ids,
+                         GArray *arr)
+{
+    unsigned int max = 0;
+    unsigned char *bitmap;
+
+    for (unsigned int i = 0; i < arr->len; i++)
+       max = MAX(max, g_array_index (arr, unsigned int, i));
+    bitmap = talloc_zero_array (ctx, unsigned char, DOCIDSET_WORD(max) + 1);
+
+    if (bitmap == NULL)
+       return FALSE;
+
+    doc_ids->bitmap = bitmap;
+    doc_ids->bound = max + 1;
+
+    for (unsigned int i = 0; i < arr->len; i++) {
+       unsigned int doc_id = g_array_index (arr, unsigned int, i);
+       bitmap[DOCIDSET_WORD(doc_id)] |= 1 << DOCIDSET_BIT(doc_id);
+    }
+
+    return TRUE;
+}
+
+notmuch_bool_t
+_notmuch_doc_id_set_contains (notmuch_doc_id_set_t *doc_ids,
+                             unsigned int doc_id)
+{
+    if (doc_id >= doc_ids->bound)
+       return FALSE;
+    return doc_ids->bitmap[DOCIDSET_WORD(doc_id)] & (1 << DOCIDSET_BIT(doc_id));
+}
+
+void
+_notmuch_doc_id_set_remove (notmuch_doc_id_set_t *doc_ids,
+                            unsigned int doc_id)
+{
+    if (doc_id < doc_ids->bound)
+       doc_ids->bitmap[DOCIDSET_WORD(doc_id)] &= ~(1 << DOCIDSET_BIT(doc_id));
+}
+
 /* Glib objects force use to use a talloc destructor as well, (but not
  * nearly as ugly as the for messages due to C++ objects). At
  * this point, I'd really like to have some talloc-friendly
@@ -242,7 +403,8 @@ _notmuch_mset_messages_move_to_next (notmuch_messages_t *messages)
 static int
 _notmuch_threads_destructor (notmuch_threads_t *threads)
 {
-    g_hash_table_unref (threads->threads);
+    if (threads->doc_ids)
+       g_array_unref (threads->doc_ids);
 
     return 0;
 }
@@ -251,24 +413,37 @@ notmuch_threads_t *
 notmuch_query_search_threads (notmuch_query_t *query)
 {
     notmuch_threads_t *threads;
+    notmuch_messages_t *messages;
 
     threads = talloc (query, notmuch_threads_t);
     if (threads == NULL)
        return NULL;
+    threads->doc_ids = NULL;
+    talloc_set_destructor (threads, _notmuch_threads_destructor);
 
     threads->query = query;
-    threads->threads = g_hash_table_new_full (g_str_hash, g_str_equal,
-                                             free, NULL);
 
-    threads->messages = notmuch_query_search_messages (query);
-    if (threads->messages == NULL) {
+    messages = notmuch_query_search_messages (query);
+    if (messages == NULL) {
            talloc_free (threads);
            return NULL;
     }
 
-    threads->thread_id = NULL;
+    threads->doc_ids = g_array_new (FALSE, FALSE, sizeof (unsigned int));
+    while (notmuch_messages_valid (messages)) {
+       unsigned int doc_id = _notmuch_mset_messages_get_doc_id (messages);
+       g_array_append_val (threads->doc_ids, doc_id);
+       notmuch_messages_move_to_next (messages);
+    }
+    threads->doc_id_pos = 0;
 
-    talloc_set_destructor (threads, _notmuch_threads_destructor);
+    talloc_free (messages);
+
+    if (! _notmuch_doc_id_set_init (threads, &threads->match_set,
+                                   threads->doc_ids)) {
+       talloc_free (threads);
+       return NULL;
+    }
 
     return threads;
 }
@@ -282,51 +457,42 @@ notmuch_query_destroy (notmuch_query_t *query)
 notmuch_bool_t
 notmuch_threads_valid (notmuch_threads_t *threads)
 {
-    notmuch_message_t *message;
-
-    if (threads->thread_id)
-       return TRUE;
+    unsigned int doc_id;
 
-    while (notmuch_messages_valid (threads->messages))
-    {
-       message = notmuch_messages_get (threads->messages);
-
-       threads->thread_id = notmuch_message_get_thread_id (message);
-
-       if (! g_hash_table_lookup_extended (threads->threads,
-                                           threads->thread_id,
-                                           NULL, NULL))
-       {
-           g_hash_table_insert (threads->threads,
-                                xstrdup (threads->thread_id), NULL);
-           notmuch_messages_move_to_next (threads->messages);
-           return TRUE;
-       }
+    while (threads->doc_id_pos < threads->doc_ids->len) {
+       doc_id = g_array_index (threads->doc_ids, unsigned int,
+                               threads->doc_id_pos);
+       if (_notmuch_doc_id_set_contains (&threads->match_set, doc_id))
+           break;
 
-       notmuch_messages_move_to_next (threads->messages);
+       threads->doc_id_pos++;
     }
 
-    threads->thread_id = NULL;
-    return FALSE;
+    return threads->doc_id_pos < threads->doc_ids->len;
 }
 
 notmuch_thread_t *
 notmuch_threads_get (notmuch_threads_t *threads)
 {
+    unsigned int doc_id;
+
     if (! notmuch_threads_valid (threads))
        return NULL;
 
+    doc_id = g_array_index (threads->doc_ids, unsigned int,
+                           threads->doc_id_pos);
     return _notmuch_thread_create (threads->query,
                                   threads->query->notmuch,
-                                  threads->thread_id,
-                                  threads->query->query_string,
+                                  doc_id,
+                                  &threads->match_set,
+                                  threads->query->exclude_terms,
                                   threads->query->sort);
 }
 
 void
 notmuch_threads_move_to_next (notmuch_threads_t *threads)
 {
-    threads->thread_id = NULL;
+    threads->doc_id_pos++;
 }
 
 void
@@ -347,7 +513,7 @@ notmuch_query_count_messages (notmuch_query_t *query)
        Xapian::Query mail_query (talloc_asprintf (query, "%s%s",
                                                   _find_prefix ("type"),
                                                   "mail"));
-       Xapian::Query string_query, final_query;
+       Xapian::Query string_query, final_query, exclude_query;
        Xapian::MSet mset;
        unsigned int flags = (Xapian::QueryParser::FLAG_BOOLEAN |
                              Xapian::QueryParser::FLAG_PHRASE |
@@ -367,12 +533,20 @@ notmuch_query_count_messages (notmuch_query_t *query)
                                         mail_query, string_query);
        }
 
+       exclude_query = _notmuch_exclude_tags (query, final_query);
+
+       final_query = Xapian::Query (Xapian::Query::OP_AND_NOT,
+                                        final_query, exclude_query);
+
        enquire.set_weighting_scheme(Xapian::BoolWeight());
        enquire.set_docid_order(Xapian::Enquire::ASCENDING);
 
-#if DEBUG_QUERY
-       fprintf (stderr, "Final query is:\n%s\n", final_query.get_description().c_str());
-#endif
+       if (_debug_query ()) {
+           fprintf (stderr, "Exclude query is:\n%s\n",
+                    exclude_query.get_description ().c_str ());
+           fprintf (stderr, "Final query is:\n%s\n",
+                    final_query.get_description ().c_str ());
+       }
 
        enquire.set_query (final_query);
 
@@ -388,3 +562,47 @@ notmuch_query_count_messages (notmuch_query_t *query)
 
     return count;
 }
+
+unsigned
+notmuch_query_count_threads (notmuch_query_t *query)
+{
+    notmuch_messages_t *messages;
+    GHashTable *hash;
+    unsigned int count;
+    notmuch_sort_t sort;
+
+    sort = query->sort;
+    query->sort = NOTMUCH_SORT_UNSORTED;
+    messages = notmuch_query_search_messages (query);
+    query->sort = sort;
+    if (messages == NULL)
+       return 0;
+
+    hash = g_hash_table_new_full (g_str_hash, g_str_equal, NULL, NULL);
+    if (hash == NULL) {
+       talloc_free (messages);
+       return 0;
+    }
+
+    while (notmuch_messages_valid (messages)) {
+       notmuch_message_t *message = notmuch_messages_get (messages);
+       const char *thread_id = notmuch_message_get_thread_id (message);
+       char *thread_id_copy = talloc_strdup (messages, thread_id);
+       if (unlikely (thread_id_copy == NULL)) {
+           notmuch_message_destroy (message);
+           count = 0;
+           goto DONE;
+       }
+       g_hash_table_insert (hash, thread_id_copy, NULL);
+       notmuch_message_destroy (message);
+       notmuch_messages_move_to_next (messages);
+    }
+
+    count = g_hash_table_size (hash);
+
+  DONE:
+    g_hash_table_unref (hash);
+    talloc_free (messages);
+
+    return count;
+}