From 2658040f3a2b23d90d7fe37c059300894be17e0c Mon Sep 17 00:00:00 2001
From: Kevin Lynx <kevinlynx@gmail.com>
Date: Sat, 20 Jul 2013 10:56:41 +0800
Subject: [PATCH] add simple `get' json api, fix http search space decode

---
 src/http_front/api.erl          | 77 +++++++++++++++++++++++++++++++++
 src/http_front/crawler_http.erl | 12 ++++-
 src/http_front/http_handler.erl | 10 +----
 3 files changed, 89 insertions(+), 10 deletions(-)
 create mode 100644 src/http_front/api.erl

diff --git a/src/http_front/api.erl b/src/http_front/api.erl
new file mode 100644
index 0000000..85f52d7
--- /dev/null
+++ b/src/http_front/api.erl
@@ -0,0 +1,77 @@
+%%
+%% api.erl
+%% Kevin Lynx
+%% 07.19.2013
+%% HTTP API
+%%
+-module(api).
+-export([search/3]).
+-compile(export_all).
+-define(TEXT(Fmt, Args), lists:flatten(io_lib:format(Fmt, Args))).
+-define(MAX_FILE, 10).
+
+search(SessionID, _Env, Input) ->
+	Res = case crawler_http:get_search_keyword(Input) of
+		[] -> 
+			"{\"error\":\"null input\", \"suggest\":\"api/search?q=keyword\"}";
+		Keyword ->
+			do_search(Keyword)
+	end,	
+	mod_esi:deliver(SessionID, ["Content-Type: application/json\r\n\r\n", Res]).
+
+do_search(Keyword) ->
+	{Rets, Stats} = http_cache:search(Keyword),
+	{_Found, Cost, Scanned} = Stats,
+	Tip = ?TEXT("{\"keyword\":\"~s\", \"found\":~p, \"cost\":~p,", [Keyword, Scanned, Cost div 1000]),
+	BodyList = format_search_result(Rets),
+	Body = ?TEXT("\"results\":[~s]}", [BodyList]),
+	Tip ++ Body.
+	
+format_search_result([]) ->
+	"";
+format_search_result([First|Rest]) ->
+	S = [format_one_result(First, false)] ++ ["," ++ format_one_result(Ret, false) || Ret <- Rest],
+	lists:flatten(S).
+
+format_one_result({single, Hash, {Name, Length}, Announce, CTime}, ShowAll) ->
+	format_one_result(Hash, Name, [{Name, Length}], Announce, CTime, ShowAll);
+
+format_one_result({multi, Hash, {Name, Files}, Announce, CTime}, ShowAll) ->	
+	format_one_result(Hash, Name, Files, Announce, CTime, ShowAll).
+
+format_one_result(Hash, Name, Files, Announce, CTime, ShowAll) ->
+	?TEXT("{\"hash\":\"~s\", \"name\":\"~s\", \"created_at\":~p, \"req\":~p,",
+		[Hash, Name, CTime, Announce]) ++
+	"\"files\":[" ++ format_files(Files, ShowAll) ++ "]}".
+
+format_files([], _) ->
+	"";
+
+format_files(Files, false) ->
+	Max = ?MAX_FILE,
+	Sub = case length(Files) >= Max of
+		true ->
+			lists:sublist(Files, Max) ++ [{more, length(Files) - Max}];
+		false ->
+			Files
+	end,
+	do_format_files(Sub);
+
+format_files(Files, true) ->
+	do_format_files(Files).
+
+do_format_files([First|Rest]) ->
+	S = [format_file(First)] ++ ["," ++ format_file(File) || File <- Rest],
+	lists:flatten(S).
+
+format_file({more, Len}) ->	
+	?TEXT("{\"name\":\"__more__\",\"size\":~b}", [Len]);
+
+format_file({Name, Length}) ->
+	?TEXT("{\"name\":\"~s\",\"size\":~b}", [Name, Length]).
+
+%%
+test_search(Keyword) ->
+	Filename = ?TEXT("search_~s.html", [Keyword]),
+	Body = do_search(Keyword),
+	file:write_file(Filename, Body).
diff --git a/src/http_front/crawler_http.erl b/src/http_front/crawler_http.erl
index 8a11cce..8d15fa5 100644
--- a/src/http_front/crawler_http.erl
+++ b/src/http_front/crawler_http.erl
@@ -15,6 +15,7 @@
 		 start/4,
 		 start/1,
 		 page_temp/0,
+		 get_search_keyword/1,
 	     stop/0]).
 -record(state, {html_temp, httpid}).
 
@@ -58,7 +59,7 @@ init([DBHost, DBPort, Port, PoolSize]) ->
   	{document_root, "www"},
   	{server_root, "."},
     {directory_index, ["index.html"]},
-  	{erl_script_alias, {"/e", [http_handler]}},
+  	{erl_script_alias, {"/e", [http_handler, api]}},
   	{mime_types, [{"html","text/html"}, 
   				  {"css","text/css"}, {"js","application/x-javascript"}]}]),
 	{ok, B} = file:read_file("www/page.temp"),
@@ -90,3 +91,12 @@ code_change(_, _, State) ->
 handle_info(_, State) ->
     {noreply, State}.
 
+get_search_keyword(Input) ->
+	D = urldecode:decode(Input),
+	ReqList = httpd:parse_query(D),
+	case proplists:get_value("q", ReqList) of
+		undefined -> 
+			"";
+		Keyword ->
+			Keyword
+	end.
diff --git a/src/http_front/http_handler.erl b/src/http_front/http_handler.erl
index 7285feb..e5e0f43 100644
--- a/src/http_front/http_handler.erl
+++ b/src/http_front/http_handler.erl
@@ -17,7 +17,7 @@
 -define(CONTENT_TYPE, "Content-Type: text/html\r\n\r\n").
 
 search(SessionID, _Env, Input) ->
-	{K, Body} = case get_search_keyword(Input) of
+	{K, Body} = case crawler_http:get_search_keyword(Input) of
 		[] -> 
 			{"", "invalid input"};
 		Key ->
@@ -73,14 +73,6 @@ index(SessionID, _Env, Input) ->
 	Response = simple_html("", Body),
 	mod_esi:deliver(SessionID, [?CONTENT_TYPE, Response]).
 
-get_search_keyword(Input) ->
-	case string:equal(string:substr(Input, 1, 2), "q=") of
-		true ->
-			urldecode:decode(string:substr(Input, 3));
-		false ->
-			[]
-	end.
-
 get_index_hash(Input) ->
 	case string:equal(string:substr(Input, 1, 2), "q=") of
 		true ->