stdlib: Allow undefined port in uri_map()

uri_map() updated to allow 'undefined' ports in order to align
the implementation with RFC 3986:

   port        = *DIGIT

An 'undefined' port is mapped to a ":" during recompose operation.

3 files changed, 43 insertions, 12 deletions

diff --git a/lib/stdlib/src/uri_string.erl b/lib/stdlib/src/uri_string.erl
index ca212284d2..16650d5005 100644
--- a/lib/stdlib/src/uri_string.erl
+++ b/lib/stdlib/src/uri_string.erl
@@ -283,7 +283,7 @@
   #{fragment => unicode:chardata(),
     host => unicode:chardata(),
     path => unicode:chardata(),
-    port => non_neg_integer(),
+    port => non_neg_integer() | undefined,
     query => unicode:chardata(),
     scheme => unicode:chardata(),
     userinfo => unicode:chardata()} | #{}.
@@ -807,7 +807,7 @@ is_userinfo(Char) -> is_unreserved(Char) orelse is_sub_delim(Char).
 parse_host(?STRING_REST($:, Rest), URI) ->
     {T, URI1} = parse_port(Rest, URI),
     H = calculate_parsed_host_port(Rest, T),
-    Port = binary_to_integer(H),
+    Port = get_port(H),
     {Rest, URI1#{port => Port}};
 parse_host(?STRING_REST($/, Rest), URI) ->
     {T, URI1} = parse_segment(Rest, URI),  % path-abempty
@@ -836,7 +836,7 @@ parse_host(?STRING_EMPTY, URI) ->
 parse_reg_name(?STRING_REST($:, Rest), URI) ->
     {T, URI1} = parse_port(Rest, URI),
     H = calculate_parsed_host_port(Rest, T),
-    Port = binary_to_integer(H),
+    Port = get_port(H),
     {Rest, URI1#{port => Port}};
 parse_reg_name(?STRING_REST($/, Rest), URI) ->
     {T, URI1} = parse_segment(Rest, URI),  % path-abempty
@@ -869,7 +869,7 @@ parse_ipv4_bin(?STRING_REST($:, Rest), Acc, URI) ->
     _ = validate_ipv4_address(lists:reverse(Acc)),
     {T, URI1} = parse_port(Rest, URI),
     H = calculate_parsed_host_port(Rest, T),
-    Port = binary_to_integer(H),
+    Port = get_port(H),
     {Rest, URI1#{port => Port}};
 parse_ipv4_bin(?STRING_REST($/, Rest), Acc, URI) ->
     _ = validate_ipv4_address(lists:reverse(Acc)),
@@ -932,7 +932,7 @@ is_ipv6(Char) -> is_hex_digit(Char).
 parse_ipv6_bin_end(?STRING_REST($:, Rest), URI) ->
     {T, URI1} = parse_port(Rest, URI),
     H = calculate_parsed_host_port(Rest, T),
-    Port = binary_to_integer(H),
+    Port = get_port(H),
     {Rest, URI1#{port => Port}};
 parse_ipv6_bin_end(?STRING_REST($/, Rest), URI) ->
     {T, URI1} = parse_segment(Rest, URI),  % path-abempty
@@ -1148,7 +1148,7 @@ calculate_parsed_userinfo(Input, Unparsed) ->
 
 -spec calculate_parsed_host_port(binary(), binary()) -> binary().
 calculate_parsed_host_port(Input, <<>>) ->
-    strip_last_char(Input, [$?,$#,$/]);
+    strip_last_char(Input, [$:,$?,$#,$/]);
 calculate_parsed_host_port(Input, Unparsed) ->
     get_parsed_binary(Input, Unparsed).
 
@@ -1159,6 +1159,18 @@ calculate_parsed_query_fragment(Input, Unparsed) ->
     get_parsed_binary(Input, Unparsed).
 
 
+get_port(<<>>) ->
+    undefined;
+get_port(B) ->
+    try binary_to_integer(B) of
+        Port ->
+            Port
+    catch
+        error:badarg ->
+            throw({error, invalid_uri, B})
+    end.
+
+
 %% Strip last char if it is in list
 strip_last_char(<<>>, _) -> <<>>;
 strip_last_char(Input, [C0]) ->
@@ -1187,6 +1199,19 @@ strip_last_char(Input, [C0,C1,C2]) ->
             init_binary(Input);
         _Else ->
             Input
+    end;
+strip_last_char(Input, [C0,C1,C2,C3]) ->
+    case binary:last(Input) of
+        C0 ->
+            init_binary(Input);
+        C1 ->
+            init_binary(Input);
+        C2 ->
+            init_binary(Input);
+        C3 ->
+            init_binary(Input);
+        _Else ->
+            Input
     end.
 
 
@@ -1530,6 +1555,8 @@ update_host(#{}, URI) ->
 
 
 %% URI cannot be empty for ports. E.g. ":8080" is not a valid URI
+update_port(#{port := undefined}, URI) ->
+    concat(URI, <<":">>);
 update_port(#{port := Port}, URI) ->
     concat(URI,add_colon(encode_port(Port)));
 update_port(#{}, URI) ->
diff --git a/lib/stdlib/test/property_test/uri_string_recompose.erl b/lib/stdlib/test/property_test/uri_string_recompose.erl
index 97f9d727a0..e51a671172 100644
--- a/lib/stdlib/test/property_test/uri_string_recompose.erl
+++ b/lib/stdlib/test/property_test/uri_string_recompose.erl
@@ -267,8 +267,9 @@ host_uri() ->
 %% Port, Query, Fragment
 %%-------------------------------------------------------------------------
 port() ->
-    range(1,65535).
-
+    frequency([{10, undefined},
+               {10, range(1,65535)}
+              ]).
 
 query_map() ->
     unicode().
diff --git a/lib/stdlib/test/uri_string_SUITE.erl b/lib/stdlib/test/uri_string_SUITE.erl
index 95a49f5eb3..9ee321c509 100644
--- a/lib/stdlib/test/uri_string_SUITE.erl
+++ b/lib/stdlib/test/uri_string_SUITE.erl
@@ -692,7 +692,9 @@ parse_special2(_Config) ->
     #{host := [],path := [],userinfo := []} = uri_string:parse("//@"),
     #{host := [],path := [],scheme := "foo",userinfo := []} = uri_string:parse("foo://@"),
     #{host := [],path := "/",userinfo := []} = uri_string:parse("//@/"),
-    #{host := [],path := "/",scheme := "foo",userinfo := []} = uri_string:parse("foo://@/").
+    #{host := [],path := "/",scheme := "foo",userinfo := []} = uri_string:parse("foo://@/"),
+    #{host := "localhost",path := "/",port := undefined} = uri_string:parse("//localhost:/"),
+    #{host := [],path := [],port := undefined} = uri_string:parse("//:").
 
 parse_negative(_Config) ->
     {error,invalid_uri,"å"} = uri_string:parse("å"),
@@ -702,7 +704,8 @@ parse_negative(_Config) ->
     {error,invalid_uri,"ö"} = uri_string:parse("//host/path#foö"),
     {error,invalid_uri,"127.256.0.1"} = uri_string:parse("//127.256.0.1"),
     {error,invalid_uri,":::127.0.0.1"} = uri_string:parse("//[:::127.0.0.1]"),
-    {error,non_utf8,<<0,0,0,246>>} = uri_string:parse("//%00%00%00%F6").
+    {error,invalid_utf8,<<0,0,0,246>>} = uri_string:parse("//%00%00%00%F6"),
+    {error,invalid_uri,"A"} = uri_string:parse("//localhost:A8").
 
 
 %%-------------------------------------------------------------------------
@@ -836,8 +839,8 @@ compose_query(_Config) ->
         uri_string:compose_query([{<<"foo bar">>,<<"1">>}, {<<"ö"/utf8>>, <<"2">>}]).
 
 compose_query_negative(_Config) ->
-    {error,badarg,4} = uri_string:compose_query([{"",4}]),
-    {error,badarg,5} = uri_string:compose_query([{5,""}]),
+    {error,invalid_input,4} = uri_string:compose_query([{"",4}]),
+    {error,invalid_input,5} = uri_string:compose_query([{5,""}]),
     {error,invalid_utf8,<<"ö">>} = uri_string:compose_query([{"foo bar","1"}, {<<"ö">>, "2"}]).
 
 dissect_query(_Config) ->