From 8ad01e4aa804925165023483b6ff1d20b450dcb8 Mon Sep 17 00:00:00 2001
From: djcb <djcb@djcbsoftware.nl>
Date: Wed, 17 Oct 2012 17:59:33 +0300
Subject: [PATCH] * lib: threading: pre-sort the threaded results, add some
 more comments

---
 lib/mu-container.c |  8 ++++----
 lib/mu-msg-iter.cc |  3 +++
 lib/mu-query.cc    | 24 +++++++++++++++---------
 lib/mu-threader.c  | 12 +++++-------
 4 files changed, 27 insertions(+), 20 deletions(-)

diff --git a/lib/mu-container.c b/lib/mu-container.c
index ed4ce056..73587c92 100644
--- a/lib/mu-container.c
+++ b/lib/mu-container.c
@@ -548,10 +548,10 @@ thread_info_destroy (MuMsgIterThreadInfo *ti)
 
 
 struct _ThreadInfo {
-	GHashTable		*hash;
-	const char*		 format;
+	GHashTable *hash;
+	const char *format;
 };
-typedef struct _ThreadInfo	 ThreadInfo;
+typedef struct _ThreadInfo ThreadInfo;
 
 
 static void
@@ -589,7 +589,7 @@ thread_segment_format_string (size_t matchnum)
 	/* get the number of digits needed in a hex-representation of
 	 * matchnum */
 	digitnum = (unsigned) (ceil (log(matchnum)/log(16)));
-	snprintf (frmt, sizeof(frmt),"%%0%ux", digitnum);
+	snprintf (frmt, sizeof(frmt), "%%0%ux", digitnum);
 
 	return frmt;
 }
diff --git a/lib/mu-msg-iter.cc b/lib/mu-msg-iter.cc
index 3f6ecbc7..ee0f9bd7 100644
--- a/lib/mu-msg-iter.cc
+++ b/lib/mu-msg-iter.cc
@@ -57,6 +57,9 @@ public:
 
 		_matches = _enq.get_mset (0, maxnum);
 
+		/* when threading, we calculate the threads for the
+		 * set of matches, then requery/sort based on the
+		 * threading */
 		if (threads && !_matches.empty()) {
 
 			_matches.fetch();
diff --git a/lib/mu-query.cc b/lib/mu-query.cc
index 90366ba5..71846da7 100644
--- a/lib/mu-query.cc
+++ b/lib/mu-query.cc
@@ -335,12 +335,11 @@ get_enquire (MuQuery *self, const char *searchexpr, gboolean threads,
 {
 	Xapian::Enquire enq (self->db());
 
-	/* note, when our result will be *threaded*, we sort
-	 * in our threading code (mu-threader etc.), and don't
-	 * let Xapian do any sorting */
-	if (!threads && sortfieldid != MU_MSG_FIELD_ID_NONE)
-			enq.set_sort_by_value ((Xapian::valueno)sortfieldid,
-					       revert ? true : false);
+	if (sortfieldid != MU_MSG_FIELD_ID_NONE)
+		enq.set_sort_by_value ((Xapian::valueno)sortfieldid,
+				       revert ? true : false);
+
+	/* empty or "" means "matchall" */
 	if (!mu_str_is_empty(searchexpr) &&
 	    g_strcmp0 (searchexpr, "\"\"") != 0) /* NULL or "" or """" */
 			enq.set_query(get_query (self, searchexpr, err));
@@ -367,15 +366,22 @@ mu_query_run (MuQuery *self, const char* searchexpr, gboolean threads,
 		MuMsgIter *iter;
 		Xapian::Enquire enq (get_enquire(self, searchexpr, threads,
 						 sortfieldid, revert, err));
+
+		/* get the 'real' maxnum if it was specified as < 0 */
+		maxnum <= 0 ? self->db().get_doccount() : maxnum;
+
 		iter = mu_msg_iter_new (
 			reinterpret_cast<XapianEnquire*>(&enq),
-			maxnum <= 0 ? self->db().get_doccount() : maxnum,
-			threads, threads ? sortfieldid : MU_MSG_FIELD_ID_NONE,
+			maxnum,	threads,
+			/* in we were *not* using threads, no further sorting
+			 * is needed since Xapian already sorted */
+			threads ? sortfieldid : MU_MSG_FIELD_ID_NONE,
 			revert,	err);
 
 		if (err && *err && (*err)->code == MU_ERROR_XAPIAN_MODIFIED) {
 			g_clear_error (err);
-			return try_requery (self, searchexpr, threads, sortfieldid,
+			return try_requery (self, searchexpr, threads,
+					    sortfieldid,
 					    revert, maxnum, err);
 		} else
 			return iter;
diff --git a/lib/mu-threader.c b/lib/mu-threader.c
index 587574a7..326818cf 100644
--- a/lib/mu-threader.c
+++ b/lib/mu-threader.c
@@ -48,12 +48,12 @@
  *
  */
 
-
 /* step 1 */ static GHashTable* create_containers (MuMsgIter *iter);
 /* step 2 */ static MuContainer *find_root_set (GHashTable *ids);
 static MuContainer* prune_empty_containers (MuContainer *root);
 /* static void group_root_set_by_subject (GSList *root_set); */
-GHashTable* create_doc_id_thread_path_hash (MuContainer *root, size_t match_num);
+GHashTable* create_doc_id_thread_path_hash (MuContainer *root,
+					    size_t match_num);
 
 /* msg threading algorithm, based on JWZ's algorithm,
  * http://www.jwz.org/doc/threading.html */
@@ -125,9 +125,8 @@ assert_no_duplicates (GHashTable *ids)
 }
 
 
-
-
-/* a referred message is a message that is refered by some other message */
+/* a referred message is a message that is refered by some other
+ * message */
 static MuContainer*
 find_or_create_referred (GHashTable *id_table, const char *msgid,
 			 gboolean *created)
@@ -297,8 +296,7 @@ static GHashTable*
 create_containers (MuMsgIter *iter)
 {
 	GHashTable *id_table;
-	id_table = g_hash_table_new_full (g_str_hash,
-					  g_str_equal,
+	id_table = g_hash_table_new_full (g_str_hash, g_str_equal,
 					  NULL,
 					  (GDestroyNotify)mu_container_destroy);