Merge pull request #1322 from tsloughter/binary_uri_encode

Add unicode binary  support to http_uri functions 
OTP-14404

2 files changed, 70 insertions, 11 deletions

diff --git a/lib/inets/src/http_lib/http_uri.erl b/lib/inets/src/http_lib/http_uri.erl
index cb3e107ccf..4568d165e7 100644
--- a/lib/inets/src/http_lib/http_uri.erl
+++ b/lib/inets/src/http_lib/http_uri.erl
@@ -102,16 +102,23 @@ parse(AbsURI, Opts) ->
 
 reserved() ->
     sets:from_list([$;, $:, $@, $&, $=, $+, $,, $/, $?,
-		    $#, $[, $], $<, $>, $\", ${, $}, $|,
+            $#, $[, $], $<, $>, $\", ${, $}, $|, %"
 			       $\\, $', $^, $%, $ ]).
 
-encode(URI) ->
+encode(URI) when is_list(URI) ->
     Reserved = reserved(), 
-    lists:append([uri_encode(Char, Reserved) || Char <- URI]).
+    lists:append([uri_encode(Char, Reserved) || Char <- URI]);
+encode(URI) when is_binary(URI) ->
+    Reserved = reserved(),
+    << <<(uri_encode_binary(Char, Reserved))/binary>> || <<Char>> <= URI >>.
 
-decode(String) ->
-    do_decode(String).
+decode(String) when is_list(String) ->
+    do_decode(String);
+decode(String) when is_binary(String) ->
+    do_decode_binary(String).
 
+do_decode([$+|Rest]) ->
+    [$ |do_decode(Rest)];
 do_decode([$%,Hex1,Hex2|Rest]) ->
     [hex2dec(Hex1)*16+hex2dec(Hex2)|do_decode(Rest)];
 do_decode([First|Rest]) ->
@@ -119,6 +126,14 @@ do_decode([First|Rest]) ->
 do_decode([]) ->
     [].
 
+do_decode_binary(<<$+, Rest/bits>>) ->
+    <<$ , (do_decode_binary(Rest))/binary>>;
+do_decode_binary(<<$%, Hex:2/binary, Rest/bits>>) ->
+    <<(binary_to_integer(Hex, 16)), (do_decode_binary(Rest))/binary>>;
+do_decode_binary(<<First:1/binary, Rest/bits>>) ->
+    <<First/binary, (do_decode_binary(Rest))/binary>>;
+do_decode_binary(<<>>) ->
+    <<>>.
 
 %%%========================================================================
 %%% Internal functions
@@ -162,9 +177,30 @@ extract_scheme(Str, Opts) ->
 		    {error, Error}
 	    end;
 	_ ->
-	    {ok, list_to_atom(http_util:to_lower(Str))}
+        {ok, to_atom(http_util:to_lower(Str))}
     end.
 
+to_atom(S) when is_list(S) ->
+    list_to_atom(S);
+to_atom(S) when is_binary(S) ->
+    binary_to_atom(S, unicode).
+
+parse_uri_rest(Scheme, DefaultPort, <<"//", URIPart/binary>>, Opts) ->
+    {Authority, PathQueryFragment} =
+        split_uri(URIPart, "[/?#]", {URIPart, <<"">>}, 1, 0),
+    {RawPath, QueryFragment} =
+        split_uri(PathQueryFragment, "[?#]", {PathQueryFragment, <<"">>}, 1, 0),
+    {Query, Fragment} =
+        split_uri(QueryFragment, "#", {QueryFragment, <<"">>}, 1, 0),
+    {UserInfo, HostPort} = split_uri(Authority, "@", {<<"">>, Authority}, 1, 1),
+    {Host, Port}         = parse_host_port(Scheme, DefaultPort, HostPort, Opts),
+    Path                 = path(RawPath),
+    case lists:keyfind(fragment, 1, Opts) of
+        {fragment, true} ->
+            {ok, {Scheme, UserInfo, Host, Port, Path, Query, Fragment}};
+        _ ->
+            {ok, {Scheme, UserInfo, Host, Port, Path, Query}}
+    end;
 parse_uri_rest(Scheme, DefaultPort, "//" ++ URIPart, Opts) ->
     {Authority, PathQueryFragment} =
         split_uri(URIPart, "[/?#]", {URIPart, ""}, 1, 0),
@@ -185,6 +221,11 @@ parse_uri_rest(Scheme, DefaultPort, "//" ++ URIPart, Opts) ->
 
 %% In this version of the function, we no longer need 
 %% the Scheme argument, but just in case...
+parse_host_port(_Scheme, DefaultPort, <<"[", HostPort/binary>>, Opts) -> %ipv6
+    {Host, ColonPort} = split_uri(HostPort, "\\]", {HostPort, <<"">>}, 1, 1),
+    Host2 = maybe_ipv6_host_with_brackets(Host, Opts),
+    {_, Port} = split_uri(ColonPort, ":", {<<"">>, DefaultPort}, 0, 1),
+    {Host2, int_port(Port)};
 parse_host_port(_Scheme, DefaultPort, "[" ++ HostPort, Opts) -> %ipv6
     {Host, ColonPort} = split_uri(HostPort, "\\]", {HostPort, ""}, 1, 1),
     Host2 = maybe_ipv6_host_with_brackets(Host, Opts),
@@ -198,12 +239,19 @@ parse_host_port(_Scheme, DefaultPort, HostPort, _Opts) ->
 split_uri(UriPart, SplitChar, NoMatchResult, SkipLeft, SkipRight) ->
     case re:run(UriPart, SplitChar, [{capture, first}]) of
 	{match, [{Match, _}]} ->
-	    {string:substr(UriPart, 1, Match + 1 - SkipLeft),
-	     string:substr(UriPart, Match + 1 + SkipRight, length(UriPart))};
+        {string:slice(UriPart, 0, Match + 1 - SkipLeft),
+         string:slice(UriPart, Match + SkipRight, string:length(UriPart))};
 	nomatch ->
 	    NoMatchResult
     end.
 
+maybe_ipv6_host_with_brackets(Host, Opts) when is_binary(Host) ->
+    case lists:keysearch(ipv6_host_with_brackets, 1, Opts) of
+    {value, {ipv6_host_with_brackets, true}} ->
+        <<"[", Host/binary, "]">>;
+    _ ->
+        Host
+    end;
 maybe_ipv6_host_with_brackets(Host, Opts) ->
     case lists:keysearch(ipv6_host_with_brackets, 1, Opts) of
 	{value, {ipv6_host_with_brackets, true}} ->
@@ -212,15 +260,18 @@ maybe_ipv6_host_with_brackets(Host, Opts) ->
 	    Host
     end.
 
-
 int_port(Port) when is_integer(Port) ->
     Port;
+int_port(Port) when is_binary(Port) ->
+    binary_to_integer(Port);
 int_port(Port) when is_list(Port) ->
     list_to_integer(Port);
 %% This is the case where no port was found and there was no default port
 int_port(no_default_port) ->
     throw({error, no_default_port}).
 
+path(<<"">>) ->
+    <<"/">>;
 path("") ->
     "/";
 path(Path) ->
@@ -234,6 +285,14 @@ uri_encode(Char, Reserved) ->
 	    [Char]
     end.
 
+uri_encode_binary(Char, Reserved) ->
+    case sets:is_element(Char, Reserved) of
+        true ->
+            << $%, (integer_to_binary(Char, 16))/binary >>;
+        false ->
+            <<Char>>
+    end.
+
 hex2dec(X) when (X>=$0) andalso (X=<$9) -> X-$0;
 hex2dec(X) when (X>=$A) andalso (X=<$F) -> X-$A+10;
 hex2dec(X) when (X>=$a) andalso (X=<$f) -> X-$a+10.
diff --git a/lib/inets/src/http_lib/http_util.erl b/lib/inets/src/http_lib/http_util.erl
index 20c342dd66..78b0aa2885 100644
--- a/lib/inets/src/http_lib/http_util.erl
+++ b/lib/inets/src/http_lib/http_util.erl
@@ -35,10 +35,10 @@
 %%%  Internal application API
 %%%=========================================================================
 to_upper(Str) ->
-    string:to_upper(Str).
+    string:uppercase(Str).
 
 to_lower(Str) ->
-    string:to_lower(Str).
+    string:lowercase(Str).
 
 %% Example: Mon, 09-Dec-2002 13:46:00 GMT
 convert_netscapecookie_date([_D,_A,_Y, $,, $ ,