From b5692682e2580e92bb7b1ae4f905d1ca06e12ebd Mon Sep 17 00:00:00 2001
From: Kevin Lynx <kevinlynx@gmail.com>
Date: Mon, 15 Jul 2013 21:40:22 +0800
Subject: [PATCH] integrate cache_index for hash_reader, default is disabled

---
 src/cache_indexer/db_hash_index.erl   |  4 ++--
 src/hash_reader/hash_reader.erl       | 30 +++++++++++++++++++++++++--
 src/hash_reader/hash_reader_stats.erl | 12 ++++++++++-
 src/hash_reader/hash_reader_sup.erl   |  1 +
 4 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/src/cache_indexer/db_hash_index.erl b/src/cache_indexer/db_hash_index.erl
index 987b12f..4adad08 100644
--- a/src/cache_indexer/db_hash_index.erl
+++ b/src/cache_indexer/db_hash_index.erl
@@ -22,8 +22,8 @@ do_insert(Conn, Hash) ->
 
 exist(Conn, Hash) when is_list(Hash) ->
 	Sel = {'_id', list_to_binary(Hash)},
-	{Doc} = mongo:do(safe, master, Conn, ?DBNAME, fun() ->
+	Doc = mongo:do(safe, master, Conn, ?DBNAME, fun() ->
 		mongo:find_one(?COLLNAME, Sel)
 	end),
-	Doc == {}.
+	Doc /= {}.
 
diff --git a/src/hash_reader/hash_reader.erl b/src/hash_reader/hash_reader.erl
index 6a813fc..5b164bc 100644
--- a/src/hash_reader/hash_reader.erl
+++ b/src/hash_reader/hash_reader.erl
@@ -41,6 +41,11 @@ terminate(_, State) ->
 code_change(_, _, State) ->
     {ok, State}.
 
+handle_info(filter_torrent, State) ->
+	Conn = db_conn(State),
+	try_next_download(Conn),
+	{noreply, State};
+
 handle_info({got_torrent, failed, _Hash}, State) ->
 	#state{downloading = D} = State,
 	Conn = db_conn(State),
@@ -87,7 +92,7 @@ handle_info(process_download_hash, State) ->
 			try_next_download(Conn),
 			% until the max downloader count reaches
 			timer:send_after(?DOWNLOAD_INTERVAL, process_download_hash),
-			D + 1
+			D % + 1, bug here ?
 	end,
 	{noreply, State#state{downloading = NewD}};
 
@@ -206,7 +211,28 @@ insert_to_download_wait(Conn, Doc) ->
 
 try_next_download(Conn) ->
 	Doc = load_delete_next(Conn, ?HASH_DOWNLOAD_COLL),
-	schedule_next(Doc, true).
+	check_in_index_cache(Conn, Doc).
+
+check_in_index_cache(_, {}) ->
+	ok;
+check_in_index_cache(Conn, {Doc}) ->
+	{Hash} = bson:lookup(hash, Doc),
+	ListHash = binary_to_list(Hash),
+	Try = should_try_download(config:get(check_cache, false), Conn, ListHash),
+	case Try of
+		true ->
+			schedule_next({Doc}, true);
+		false -> 
+			% not in the local cache index, which means it may not exist on the server
+			% so give it up
+			hash_reader_stats:handle_cache_filtered(),
+			self() ! filter_torrent
+	end.
+
+should_try_download(true, Conn, Hash) ->
+	db_hash_index:exist(Conn, Hash);
+should_try_download(false, _, _) ->
+	true.
 
 % if there's no hash, try `wait_download' 
 try_next(Conn) ->
diff --git a/src/hash_reader/hash_reader_stats.erl b/src/hash_reader/hash_reader_stats.erl
index cfb655a..e6ac1c4 100644
--- a/src/hash_reader/hash_reader_stats.erl
+++ b/src/hash_reader/hash_reader_stats.erl
@@ -19,9 +19,10 @@
 		 handle_used_cache/0,
 		 handle_download_ok/0,
 		 handle_download_failed/0,
+		 handle_cache_filtered/0,
 		 dump/0]).
 -record(state, {tref, count, start, name, cache_used = 0, 
-	updated = 0, inserted = 0,
+	updated = 0, inserted = 0, cache_filtered = 0,
 	download_ok = 0, download_failed = 0}).
 -define(STATS_INTERVAL, 10*60*1000).
 -define(TEXT(Fmt, Arg), lists:flatten(io_lib:format(Fmt, Arg))).
@@ -53,6 +54,9 @@ handle_download_ok() ->
 handle_download_failed() ->
 	gen_server:cast(srv_name(), inc_download_failed).
 
+handle_cache_filtered() ->
+	gen_server:cast(srv_name(), inc_cache_filtered).
+
 srv_name() ->
 	?MODULE.
 
@@ -99,6 +103,10 @@ handle_cast(inc_download_ok, State) ->
 handle_cast(inc_download_failed, State) ->
 	#state{download_failed = D} = State,
 	{noreply, State#state{download_failed = D + 1}};
+
+handle_cast(inc_cache_filtered, State) ->
+	#state{cache_filtered = D} = State,
+	{noreply, State#state{cache_filtered = D + 1}};
 	
 handle_cast(stop, State) ->
 	{stop, normal, State}.
@@ -131,6 +139,7 @@ date_string() ->
 
 format_stats(State) ->
 	#state{count = C, start = Start, cache_used = Cache, 
+		cache_filtered = CacheFiltered, 
 		download_ok = DO, download_failed = DF,
 		updated = U, inserted = I} = State,
 	{Day, {H, M, S}} = stats_time(Start),
@@ -146,6 +155,7 @@ format_stats(State) ->
 	?TEXT("  Download torrents speed ~p tor/min~n", [I div TotalMins]) ++
 	?TEXT("  Download success ~p~n", [DO]) ++
 	?TEXT("  Download failed ~p~n", [DF]) ++
+	?TEXT("  Cache Index Filtered ~p~n", [CacheFiltered]) ++
 	?TEXT("  Updated ~p~n", [U]) ++
 	?TEXT("  Inserted ~p~n", [I]) ++
 	?TEXT("  Inserted percentage ~.2f%~n", [InsertPercent]) ++
diff --git a/src/hash_reader/hash_reader_sup.erl b/src/hash_reader/hash_reader_sup.erl
index 990489a..d4a6774 100644
--- a/src/hash_reader/hash_reader_sup.erl
+++ b/src/hash_reader/hash_reader_sup.erl
@@ -74,4 +74,5 @@ config_default() ->
 	 {save_to_file, true},
 	 {load_from_db, false},
 	 {text_seg, simple},
+	 {check_cache, false},
 	 {torrent_path, "torrents/"}].