add http search result highlight

src/http_front/http_handler.erl

@@ -117,7 +117,8 @@ search_by_sphinx(Keyword, Page) ->
 		[US, TotalFound, CostTime, DBTime div 1000])),
 	Tip = ?TEXT("<h4>search ~s, ~b results, ~f seconds, db ~f seconds</h4>", 
 		[Keyword, TotalFound, CostTime / 1000, DBTime / 1000 / 1000]),
-	BodyList = format_search_result(Rets),
+	HRets = highlight_search_result(Keyword, Rets),
+	BodyList = format_search_result(HRets),
 	Body = ?TEXT("<ol>~s</ol>", [lists:flatten(BodyList)]),
 	Tip ++ Body ++ append_page_nav(Keyword, Page, TotalFound).
 
@@ -154,10 +155,10 @@ 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) ->
-	SortedFiles = http_common:sort_file_by_size(Files),
+	%SortedFiles = http_common:sort_file_by_size(Files),
 	?TEXT("<li><p class=\"search-title\">
 		<a target='_blank' href=\"/e/http_handler:index?q=~s\">~s</a></p><ul>~s</ul>",
-		[Hash, Name, format_files(SortedFiles, ShowAll)]) ++
+		[Hash, Name, format_files(Files, ShowAll)]) ++
 	?TEXT("<p class=\"search-detail\">Index at: ~s | File count: ~p | Query count: ~p | Total Size: ~s
 		<a href=\"~s\" class=\"download-tip\">  Download</a></p>",
 		[format_time_string(CTime), length(Files), Announce, size_string(http_common:total_size(Files)), format_magnet(Hash)]).
@@ -211,3 +212,15 @@ format_time_string(Secs) ->
 format_date_string(Secs) ->
 	{{Y, M, D}, _} = time_util:seconds_to_local_time(Secs),
 	?TEXT("~b-~2..0b-~2..0b", [Y, M, D]).
+
+% use sphinx excerpt to highlight result
+highlight_search_result(Key, RetList) ->
+	[highlight_one_result(Key, Result) || Result <- RetList].
+
+highlight_one_result(Key, {single, Hash, {Name, Length}, Announce, CTime}) ->
+	HighLName = sphinx_search:highlight_title(Key, Name),
+	{single, Hash, {HighLName, Length}, Announce, CTime};
+highlight_one_result(Key, {multi, Hash, {Name, Files}, Announce, CTime}) ->	
+	HighLName = sphinx_search:highlight_title(Key, Name),
+	{multi, Hash, {HighLName, sphinx_search:highlight_files(Key, Files)}, Announce, CTime}.	
+

src/http_front/sphinx_excerpt.erl

@@ -0,0 +1,74 @@
+%%
+%% sphinx_excerpt.erl
+%% Kevin Lynx
+%% 08.24.2013
+%%
+-module(sphinx_excerpt).
+-export([build_excerpt/5, build_excerpt/3]).
+-compile(export_all).
+
+build_excerpt(Key, Docs, Index) ->
+	build_excerpt(localhost, 9312, Key, Docs, Index).
+
+build_excerpt(IP, Port, Key, Docs, Index) 
+when is_binary(Key), is_binary(Index), is_list(Docs) ->
+	case connect(IP, Port) of
+		{ok, Sock} ->
+			Ret = do_build_excerpt(Sock, Key, Docs, Index),
+          	catch gen_tcp:close(Sock),
+          	Ret;
+		Error ->
+			Error
+	end.
+
+do_build_excerpt(Sock, Key, Docs, Index) ->
+	Flag = 257, % 1 | 256, allow_empty not work
+	BeforeMatch = <<"<span class='highlight'>">>,
+	AfterMatch = <<"</span>">>,
+	ChunkSep = <<"...">>,
+	Limit = 256,
+	Around = 5,
+	LimitPassages = 0,
+	LimitWords = 0,
+	StartPageId = 1,
+	HtmlStripMode = <<"index">>,	
+	PassageBoundary = <<"none">>,
+	Commands = [{32, 0}, {32, Flag},
+		{string, Index}, {string, Key}, {string, BeforeMatch},
+		{string, AfterMatch}, {string, ChunkSep}, {32, Limit},
+		{32, Around}, {32, LimitPassages}, {32, LimitWords},
+		{32, StartPageId}, {string, HtmlStripMode}, {string, PassageBoundary}, {32, length(Docs)}] ++
+		[{string, Doc} || Doc <- Docs],
+  	{Bytes, Size} = giza_protocol:commands_to_bytes(Commands),
+  	giza_protocol:write_number(Sock, 1, 16),
+  	giza_protocol:write_number(Sock, 259, 16),
+  	giza_protocol:write_number(Sock, Size, 32),
+  	gen_tcp:send(Sock, Bytes),
+	parse_excerpt_res(Sock, Docs).
+
+parse_excerpt_res(Sock, Docs) ->
+	{ok, <<_:16, _:16, Len:32>>} = gen_tcp:recv(Sock, 8),
+    true = Len > 0,
+	[read_string_res(Sock, Doc) || Doc <- Docs].
+
+read_string_res(Sock, Doc) ->
+	R = giza_protocol:read_lp_string(Sock),
+	case byte_size(R) == byte_size(Doc) of 
+		true -> <<>>;
+		false -> R
+	end.
+
+connect(Host, Port) ->
+  case gen_tcp:connect(Host, Port,
+                       [binary, {packet, raw},
+                        {active, false}]) of
+    {ok, Sock} ->
+    	{ok, _RawVersion} = gen_tcp:recv(Sock, 4),
+    	giza_protocol:write_number(Sock, 1, 32),
+        {ok, Sock};
+    _ -> error
+  end.
+%
+test() ->
+	build_excerpt(localhost, 9312, <<"avi hi">>, [<<"hello">>, <<"hi, a a a hello avi world">>], <<"xml">>).
+

src/http_front/sphinx_search.erl

@@ -5,7 +5,7 @@
 %%
 -module(sphinx_search).
 -include("vlog.hrl").
--export([search/4, search_hash/3]).
+-export([search/4, search_hash/3, highlight_title/2, highlight_files/2]).
 -define(PORT, 9312).
 -define(INDEX, "xml").
 
@@ -48,5 +48,41 @@ translate_hash({_DocID, Item}) ->
 	40 = length(Hash),
 	Hash.
 
+highlight_title(Key, Name) when is_list(Name) ->
+    R = case catch sphinx_excerpt:build_excerpt(list_to_binary(Key), [list_to_binary(Name)], list_to_binary(?INDEX)) of
+        {'EXIT', Reason} ->
+            ?E(?FMT("highlight_title ~p", [Reason])),
+            <<>>;
+        [Ret] ->
+            Ret
+    end,
+    if byte_size(R) == 0 -> Name; true -> binary_to_list(R) end.
+
+highlight_files(Key, Files) when is_list(Files) ->
+    {Names, Lens} = lists:unzip(Files),
+    BNames = [list_to_binary(Name) || Name <- Names],
+    case catch build_file_excerpts(list_to_binary(Key), BNames, 800) of
+        {'EXIT', Reason} ->
+            ?E(?FMT("highlight_files ~p", [Reason])),
+            Files;
+        Rets ->
+            {L1, L2} = lists:foldl(fun({BName, Name, Len}, Acc) ->
+                {HList, NHList} = Acc,
+                if byte_size(BName) == 0 ->
+                    {HList, [{Name, Len}|NHList]};
+                    true ->
+                    {[{binary_to_list(BName), Len}|HList], NHList}
+                end
+            end, {[], []}, lists:zip3(Rets, Names, Lens)),
+            L1 ++ L2
+    end.
 
+% too many files in one batch will cause error
+build_file_excerpts(BKey, BNames, Batch) ->
+    BIndex = list_to_binary(?INDEX),
+    Cnt = length(BNames) div Batch + 1,
+    SubNamesList = [lists:sublist(BNames, 1 + I * Batch, Batch) ||
+        I <- lists:seq(0, Cnt - 1)],
+    lists:flatten([sphinx_excerpt:build_excerpt(BKey, BName, BIndex) ||
+        BName <- SubNamesList]).
 

www/page.temp

@@ -44,6 +44,10 @@ span.file-size {
 p.page-nav {
     text-align:center;
 }
+.highlight {
+    color:blue;
+    background:yellow;
+}
 </style>
 </head>
 <body>