From 1ff534f6e410c4904b6e65dbfc9135d34445685d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?P=C3=A9ter=20Dimitrov?= Date: Fri, 8 Jun 2018 13:24:02 +0200 Subject: stdlib: Fix normalization function in uri_string - Fix parsing of hostnames that start with a number. - Update uri_string:parse/1 to be only responsible for parsing input URIs into URI components. Implicit percent-encoding normalization has been removed. - Implement percent-encoding normalization. - Update uri_string:normalize/{1,2} to include percent-encoding normalization. - Update test suites according to the new semantics. - Add new property test: normalize Change-Id: I6f37dcae2b3fcb4b29d286dbb0dfc563e8f211ae --- lib/stdlib/src/uri_string.erl | 206 ++++++++++------ .../test/property_test/uri_string_recompose.erl | 260 ++++++++++++++++++-- lib/stdlib/test/uri_string_SUITE.erl | 261 ++++++++++++++++++--- lib/stdlib/test/uri_string_property_test_SUITE.erl | 7 +- 4 files changed, 593 insertions(+), 141 deletions(-) (limited to 'lib/stdlib') diff --git a/lib/stdlib/src/uri_string.erl b/lib/stdlib/src/uri_string.erl index 28d36ea229..48cce90d68 100644 --- a/lib/stdlib/src/uri_string.erl +++ b/lib/stdlib/src/uri_string.erl @@ -297,7 +297,10 @@ NormalizedURI :: uri_string() | error(). normalize(URIMap) -> - normalize(URIMap, []). + try normalize(URIMap, []) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end. -spec normalize(URI, Options) -> NormalizedURI when @@ -523,34 +526,34 @@ parse_relative_part(?STRING_REST("//", Rest), URI) -> {T, URI1} -> Userinfo = calculate_parsed_userinfo(Rest, T), URI2 = maybe_add_path(URI1), - URI2#{userinfo => decode_userinfo(Userinfo)} + URI2#{userinfo => Userinfo} catch throw:{_,_,_} -> {T, URI1} = parse_host(Rest, URI), Host = calculate_parsed_host_port(Rest, T), URI2 = maybe_add_path(URI1), - URI2#{host => decode_host(remove_brackets(Host))} + URI2#{host => remove_brackets(Host)} end; parse_relative_part(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-absolute Path = calculate_parsed_part(Rest, T), - URI1#{path => decode_path(?STRING_REST($/, Path))}; + URI1#{path => ?STRING_REST($/, Path)}; parse_relative_part(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), URI2 = maybe_add_path(URI1), - URI2#{query => decode_query(Query)}; + URI2#{query => Query}; parse_relative_part(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), URI2 = maybe_add_path(URI1), - URI2#{fragment => decode_fragment(Fragment)}; + URI2#{fragment => Fragment}; parse_relative_part(?STRING_REST(Char, Rest), URI) -> case is_segment_nz_nc(Char) of true -> {T, URI1} = parse_segment_nz_nc(Rest, URI), % path-noscheme Path = calculate_parsed_part(Rest, T), - URI1#{path => decode_path(?STRING_REST(Char, Path))}; + URI1#{path => ?STRING_REST(Char, Path)}; false -> throw({error,invalid_uri,[Char]}) end. @@ -593,11 +596,11 @@ parse_segment(?STRING_REST($/, Rest), URI) -> parse_segment(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_segment(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_segment(?STRING_REST(Char, Rest), URI) -> case is_pchar(Char) of true -> parse_segment(Rest, URI); @@ -616,11 +619,11 @@ parse_segment_nz_nc(?STRING_REST($/, Rest), URI) -> parse_segment_nz_nc(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_segment_nz_nc(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_segment_nz_nc(?STRING_REST(Char, Rest), URI) -> case is_segment_nz_nc(Char) of true -> parse_segment_nz_nc(Rest, URI); @@ -709,31 +712,31 @@ parse_hier(?STRING_REST("//", Rest), URI) -> try parse_userinfo(Rest, URI) of {T, URI1} -> Userinfo = calculate_parsed_userinfo(Rest, T), - {Rest, URI1#{userinfo => decode_userinfo(Userinfo)}} + {Rest, URI1#{userinfo => Userinfo}} catch throw:{_,_,_} -> {T, URI1} = parse_host(Rest, URI), Host = calculate_parsed_host_port(Rest, T), - {Rest, URI1#{host => decode_host(remove_brackets(Host))}} + {Rest, URI1#{host => remove_brackets(Host)}} end; parse_hier(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-absolute Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_hier(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_hier(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_hier(?STRING_REST(Char, Rest), URI) -> % path-rootless case is_pchar(Char) of true -> % segment_nz {T, URI1} = parse_segment(Rest, URI), Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST(Char, Path))}}; + {Rest, URI1#{path => ?STRING_REST(Char, Path)}}; false -> throw({error,invalid_uri,[Char]}) end; parse_hier(?STRING_EMPTY, URI) -> @@ -770,7 +773,7 @@ parse_userinfo(?CHAR($@), URI) -> parse_userinfo(?STRING_REST($@, Rest), URI) -> {T, URI1} = parse_host(Rest, URI), Host = calculate_parsed_host_port(Rest, T), - {Rest, URI1#{host => decode_host(remove_brackets(Host))}}; + {Rest, URI1#{host => remove_brackets(Host)}}; parse_userinfo(?STRING_REST(Char, Rest), URI) -> case is_userinfo(Char) of true -> parse_userinfo(Rest, URI); @@ -836,20 +839,25 @@ parse_host(?STRING_REST($:, Rest), URI) -> parse_host(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-abempty Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_host(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_host(?STRING_REST($[, Rest), URI) -> parse_ipv6_bin(Rest, [], URI); parse_host(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_host(?STRING_REST(Char, Rest), URI) -> case is_digit(Char) of - true -> parse_ipv4_bin(Rest, [Char], URI); + true -> + try parse_ipv4_bin(Rest, [Char], URI) + catch + throw:{_,_,_} -> + parse_reg_name(?STRING_REST(Char, Rest), URI) + end; false -> parse_reg_name(?STRING_REST(Char, Rest), URI) end; parse_host(?STRING_EMPTY, URI) -> @@ -865,15 +873,15 @@ parse_reg_name(?STRING_REST($:, Rest), URI) -> parse_reg_name(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-abempty Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_reg_name(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_reg_name(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_reg_name(?STRING_REST(Char, Rest), URI) -> case is_reg_name(Char) of true -> parse_reg_name(Rest, URI); @@ -899,17 +907,17 @@ parse_ipv4_bin(?STRING_REST($/, Rest), Acc, URI) -> _ = validate_ipv4_address(lists:reverse(Acc)), {T, URI1} = parse_segment(Rest, URI), % path-abempty Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_ipv4_bin(?STRING_REST($?, Rest), Acc, URI) -> _ = validate_ipv4_address(lists:reverse(Acc)), {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_ipv4_bin(?STRING_REST($#, Rest), Acc, URI) -> _ = validate_ipv4_address(lists:reverse(Acc)), {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_ipv4_bin(?STRING_REST(Char, Rest), Acc, URI) -> case is_ipv4(Char) of true -> parse_ipv4_bin(Rest, [Char|Acc], URI); @@ -961,15 +969,15 @@ parse_ipv6_bin_end(?STRING_REST($:, Rest), URI) -> parse_ipv6_bin_end(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-abempty Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_ipv6_bin_end(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_ipv6_bin_end(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_ipv6_bin_end(?STRING_REST(Char, Rest), URI) -> case is_ipv6(Char) of true -> parse_ipv6_bin_end(Rest, URI); @@ -999,15 +1007,15 @@ validate_ipv6_address(Addr) -> parse_port(?STRING_REST($/, Rest), URI) -> {T, URI1} = parse_segment(Rest, URI), % path-abempty Path = calculate_parsed_part(Rest, T), - {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; + {Rest, URI1#{path => ?STRING_REST($/, Path)}}; parse_port(?STRING_REST($?, Rest), URI) -> {T, URI1} = parse_query(Rest, URI), % path-empty ?query Query = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{query => decode_query(Query)}}; + {Rest, URI1#{query => Query}}; parse_port(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), % path-empty Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_port(?STRING_REST(Char, Rest), URI) -> case is_digit(Char) of true -> parse_port(Rest, URI); @@ -1033,7 +1041,7 @@ parse_port(?STRING_EMPTY, URI) -> parse_query(?STRING_REST($#, Rest), URI) -> {T, URI1} = parse_fragment(Rest, URI), Fragment = calculate_parsed_query_fragment(Rest, T), - {Rest, URI1#{fragment => decode_fragment(Fragment)}}; + {Rest, URI1#{fragment => Fragment}}; parse_query(?STRING_REST(Char, Rest), URI) -> case is_query(Char) of true -> parse_query(Rest, URI); @@ -1088,6 +1096,31 @@ is_fragment(Char) -> is_pchar(Char). %% %%------------------------------------------------------------------------- +%% Return true if input char is reserved. +-spec is_reserved(char()) -> boolean(). +is_reserved($:) -> true; +is_reserved($/) -> true; +is_reserved($?) -> true; +is_reserved($#) -> true; +is_reserved($[) -> true; +is_reserved($]) -> true; +is_reserved($@) -> true; + +is_reserved($!) -> true; +is_reserved($$) -> true; +is_reserved($&) -> true; +is_reserved($') -> true; +is_reserved($() -> true; +is_reserved($)) -> true; + +is_reserved($*) -> true; +is_reserved($+) -> true; +is_reserved($,) -> true; +is_reserved($;) -> true; +is_reserved($=) -> true; +is_reserved(_) -> false. + + %% Check if char is sub-delim. -spec is_sub_delim(char()) -> boolean(). is_sub_delim($!) -> true; @@ -1276,36 +1309,6 @@ byte_size_exl_head(Binary) -> byte_size(Binary) + 1. %% %% pct-encoded = "%" HEXDIG HEXDIG %%------------------------------------------------------------------------- --spec decode_userinfo(binary()) -> binary(). -decode_userinfo(Cs) -> - check_utf8(decode(Cs, fun is_userinfo/1, <<>>)). - --spec decode_host(binary()) -> binary(). -decode_host(Cs) -> - check_utf8(decode(Cs, fun is_host/1, <<>>)). - --spec decode_path(binary()) -> binary(). -decode_path(Cs) -> - check_utf8(decode(Cs, fun is_path/1, <<>>)). - --spec decode_query(binary()) -> binary(). -decode_query(Cs) -> - check_utf8(decode(Cs, fun is_query/1, <<>>)). - --spec decode_fragment(binary()) -> binary(). -decode_fragment(Cs) -> - check_utf8(decode(Cs, fun is_fragment/1, <<>>)). - - -%% Returns Cs if it is utf8 encoded. -check_utf8(Cs) -> - case unicode:characters_to_list(Cs) of - {incomplete,_,_} -> - throw({error,invalid_utf8,Cs}); - {error,_,_} -> - throw({error,invalid_utf8,Cs}); - _ -> Cs - end. %%------------------------------------------------------------------------- %% Percent-encode @@ -1351,20 +1354,56 @@ encode_fragment(Cs) -> %%------------------------------------------------------------------------- %% Helper funtions for percent-decode %%------------------------------------------------------------------------- -decode(<<$%,C0,C1,Cs/binary>>, Fun, Acc) -> + +-spec decode(list()|binary()) -> list() | binary(). +decode(Cs) -> + decode(Cs, <<>>). +%% +decode(L, Acc) when is_list(L) -> + B0 = unicode:characters_to_binary(L), + B1 = decode(B0, Acc), + unicode:characters_to_list(B1); +decode(<<$%,C0,C1,Cs/binary>>, Acc) -> case is_hex_digit(C0) andalso is_hex_digit(C1) of true -> B = ?HEX2DEC(C0)*16+?HEX2DEC(C1), - decode(Cs, Fun, <>); + case is_reserved(B) of + true -> + %% [2.2] Characters in the reserved set are protected from + %% normalization. + %% [2.1] For consistency, URI producers and normalizers should + %% use uppercase hexadecimal digits for all percent- + %% encodings. + H0 = hex_to_upper(C0), + H1 = hex_to_upper(C1), + decode(Cs, <>); + false -> + decode(Cs, <>) + end; false -> throw({error,invalid_percent_encoding,<<$%,C0,C1>>}) end; -decode(<>, Fun, Acc) -> - case Fun(C) of - true -> decode(Cs, Fun, <>); - false -> throw({error,invalid_percent_encoding,<>}) - end; -decode(<<>>, _Fun, Acc) -> - Acc. +decode(<>, Acc) -> + decode(Cs, <>); +decode(<<>>, Acc) -> + check_utf8(Acc). + +%% Returns Cs if it is utf8 encoded. +check_utf8(Cs) -> + case unicode:characters_to_list(Cs) of + {incomplete,_,_} -> + throw({error,invalid_utf8,Cs}); + {error,_,_} -> + throw({error,invalid_utf8,Cs}); + _ -> Cs + end. + +%% Convert hex digit to uppercase form +hex_to_upper(H) when $a =< H, H =< $f -> + H - 32; +hex_to_upper(H) when $0 =< H, H =< $9;$A =< H, H =< $F-> + H; +hex_to_upper(H) -> + throw({error,invalid_input, H}). %% Check if char is allowed in host -spec is_host(char()) -> boolean(). @@ -1925,9 +1964,10 @@ base10_decode_unicode(<>, _, _) -> %%------------------------------------------------------------------------- normalize_map(URIMap) -> - normalize_path_segment( - normalize_scheme_based( - normalize_case(URIMap))). + normalize_path_segment( + normalize_scheme_based( + normalize_percent_encoding( + normalize_case(URIMap)))). %% 6.2.2.1. Case Normalization @@ -1942,6 +1982,18 @@ normalize_case(#{} = Map) -> Map. +%% 6.2.2.2. Percent-Encoding Normalization +normalize_percent_encoding(Map) -> + Fun = fun (K,V) when K =:= userinfo; K =:= host; K =:= path; + K =:= query; K =:= fragment -> + decode(V); + %% Handle port and scheme + (_,V) -> + V + end, + maps:map(Fun, Map). + + to_lower(Cs) when is_list(Cs) -> B = convert_to_binary(Cs, utf8, utf8), convert_to_list(to_lower(B), utf8); diff --git a/lib/stdlib/test/property_test/uri_string_recompose.erl b/lib/stdlib/test/property_test/uri_string_recompose.erl index e51a671172..35b3a50b9c 100644 --- a/lib/stdlib/test/property_test/uri_string_recompose.erl +++ b/lib/stdlib/test/property_test/uri_string_recompose.erl @@ -65,15 +65,29 @@ -define(QUERY, {query, query_map()}). -define(FRAGMENT, {fragment, fragment_map()}). +%% Non-unicode +-define(USER_NU, {userinfo, non_unicode()}). +-define(HOST_NU, {host, host_map_nu()}). +-define(PATH_ABE_NU, {path, path_abempty_map_nu()}). +-define(PATH_ABS_NU, {path, path_absolute_map_nu()}). +-define(PATH_NOS_NU, {path, path_noscheme_map_nu()}). +-define(PATH_ROO_NU, {path, path_rootless_map_nu()}). +-define(QUERY_NU, {query, query_map_nu()}). +-define(FRAGMENT_NU, {fragment, fragment_map_nu()}). %%%======================================================================== %%% Properties %%%======================================================================== prop_recompose() -> + ?FORALL(Map, map_no_unicode(), + Map =:= uri_string:parse(uri_string:recompose(Map))). + +prop_normalize() -> ?FORALL(Map, map(), - Map =:= uri_string:parse(uri_string:recompose(Map)) - ). + uri_string:normalize(Map, [return_map]) =:= + uri_string:normalize(uri_string:parse(uri_string:recompose(Map)), + [return_map])). %% Stats prop_map_key_length_collect() -> @@ -96,6 +110,9 @@ prop_scheme_collect() -> map() -> ?LET(Gen, comp_proplist(), proplist_to_map(Gen)). +map_no_unicode() -> + ?LET(Gen, comp_proplist_nu(), proplist_to_map(Gen)). + comp_proplist() -> frequency([ {2, [?SCHEME,?PATH_ABS]}, @@ -166,6 +183,76 @@ comp_proplist() -> {2, [?USER,?HOST,?PORT,?PATH_ABE,?QUERY,?FRAGMENT]} ]). +comp_proplist_nu() -> + frequency([ + {2, [?SCHEME,?PATH_ABS_NU]}, + {2, [?SCHEME,?PATH_ROO_NU]}, + {2, [?SCHEME,?PATH_EMP]}, + {2, [?SCHEME,?HOST_NU,?PATH_ABE_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PATH_ABE_NU]}, + {2, [?SCHEME,?HOST_NU,?PORT,?PATH_ABE_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU]}, + + {2, [?PATH_ABS_NU]}, + {2, [?PATH_NOS_NU]}, + {2, [?PATH_EMP]}, + {2, [?HOST_NU,?PATH_ABE_NU]}, + {2, [?USER_NU,?HOST_NU,?PATH_ABE_NU]}, + {2, [?HOST_NU,?PORT,?PATH_ABE_NU]}, + {2, [?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU]}, + + + {2, [?SCHEME,?PATH_ABS_NU,?QUERY_NU]}, + {2, [?SCHEME,?PATH_ROO_NU,?QUERY_NU]}, + {2, [?SCHEME,?PATH_EMP,?QUERY_NU]}, + {2, [?SCHEME,?HOST_NU,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?SCHEME,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU]}, + + {2, [?PATH_ABS_NU,?QUERY_NU]}, + {2, [?PATH_NOS_NU,?QUERY_NU]}, + {2, [?PATH_EMP,?QUERY_NU]}, + {2, [?HOST_NU,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?USER_NU,?HOST_NU,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU]}, + {2, [?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU]}, + + + {2, [?SCHEME,?PATH_ABS_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?PATH_ROO_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?PATH_EMP,?FRAGMENT_NU]}, + {2, [?SCHEME,?HOST_NU,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?HOST_NU,?PORT,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?FRAGMENT_NU]}, + + {2, [?PATH_ABS_NU,?FRAGMENT_NU]}, + {2, [?PATH_NOS_NU,?FRAGMENT_NU]}, + {2, [?PATH_EMP,?FRAGMENT_NU]}, + {2, [?HOST_NU,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?USER_NU,?HOST_NU,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?HOST_NU,?PORT,?PATH_ABE_NU,?FRAGMENT_NU]}, + {2, [?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?FRAGMENT_NU]}, + + + {2, [?SCHEME,?PATH_ABS_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?PATH_ROO_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?PATH_EMP,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?HOST_NU,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?SCHEME,?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + + {2, [?PATH_ABS_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?PATH_NOS_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?PATH_EMP,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?HOST_NU,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?USER_NU,?HOST_NU,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]}, + {2, [?USER_NU,?HOST_NU,?PORT,?PATH_ABE_NU,?QUERY_NU,?FRAGMENT_NU]} + ]). + %%------------------------------------------------------------------------- %% Path @@ -174,6 +261,11 @@ path_abempty_map() -> frequency([{90, path_abe_map()}, {10, path_empty_map()}]). +path_abempty_map_nu() -> + frequency([{90, path_abe_map_nu()}, + {10, path_empty_map()}]). + + path_abe_map() -> ?SIZED(Length, path_abe_map(Length, [])). %% @@ -182,6 +274,14 @@ path_abe_map(0, Segments) -> path_abe_map(N, Segments) -> path_abe_map(N-1, [slash(),segment()|Segments]). +path_abe_map_nu() -> + ?SIZED(Length, path_abe_map_nu(Length, [])). +%% +path_abe_map_nu(0, Segments) -> + ?LET(Gen, Segments, lists:append(Gen)); +path_abe_map_nu(N, Segments) -> + path_abe_map_nu(N-1, [slash(),segment_nu()|Segments]). + path_absolute_map() -> ?SIZED(Length, path_absolute_map(Length, [])). @@ -191,6 +291,14 @@ path_absolute_map(0, Segments) -> path_absolute_map(N, Segments) -> path_absolute_map(N-1, [slash(),segment()|Segments]). +path_absolute_map_nu() -> + ?SIZED(Length, path_absolute_map_nu(Length, [])). +%% +path_absolute_map_nu(0, Segments) -> + ?LET(Gen, [slash(),segment_nz_nu()|Segments], lists:append(Gen)); +path_absolute_map_nu(N, Segments) -> + path_absolute_map_nu(N-1, [slash(),segment_nu()|Segments]). + path_noscheme_map() -> ?SIZED(Length, path_noscheme_map(Length, [])). @@ -200,6 +308,15 @@ path_noscheme_map(0, Segments) -> path_noscheme_map(N, Segments) -> path_noscheme_map(N-1, [slash(),segment()|Segments]). +path_noscheme_map_nu() -> + ?SIZED(Length, path_noscheme_map_nu(Length, [])). +%% +path_noscheme_map_nu(0, Segments) -> + ?LET(Gen, [segment_nz_nc_nu()|Segments], lists:append(Gen)); +path_noscheme_map_nu(N, Segments) -> + path_noscheme_map_nu(N-1, [slash(),segment_nu()|Segments]). + + path_rootless_map() -> ?SIZED(Length, path_rootless_map(Length, [])). %% @@ -208,24 +325,59 @@ path_rootless_map(0, Segments) -> path_rootless_map(N, Segments) -> path_rootless_map(N-1, [slash(),segment()|Segments]). +path_rootless_map_nu() -> + ?SIZED(Length, path_rootless_map_nu(Length, [])). +%% +path_rootless_map_nu(0, Segments) -> + ?LET(Gen, [segment_nz_nu()|Segments], lists:append(Gen)); +path_rootless_map_nu(N, Segments) -> + path_rootless_map_nu(N-1, [slash(),segment_nu()|Segments]). + segment_nz() -> non_empty(segment()). -segment_nz_nc() -> - non_empty(list(frequency([{30, unreserved()}, - {10, sub_delims()}, - {10, unicode_char()}, - {5, oneof([$@])} - ]))). +segment_nz_nu() -> + non_empty(segment_nu()). +segment_nz_nc() -> + ?LET(Gen, + non_empty(list(frequency([{30, unreserved()}, + {10, ptc_encoded_reserved()}, + {10, sub_delims()}, + {10, unicode_char()}, + {5, oneof([$@])} + ]))), + lists:flatten(Gen)). + +segment_nz_nc_nu() -> + ?LET(Gen, + non_empty(list(frequency([{30, unreserved()}, + {10, ptc_encoded_reserved()}, + {10, sub_delims()}, + {5, oneof([$@])} + ]))), + lists:flatten(Gen)). + segment() -> - list(frequency([{30, unreserved()}, - {10, sub_delims()}, - {10, unicode_char()}, - {5, oneof([$:, $@])} - ])). + ?LET(Gen, + list(frequency([{30, unreserved()}, + {10, ptc_encoded_reserved()}, + {10, sub_delims()}, + {10, unicode_char()}, + {5, oneof([$:, $@])} + ])), + lists:flatten(Gen)). + +segment_nu() -> + ?LET(Gen, + list(frequency([{30, unreserved()}, + {10, ptc_encoded_reserved()}, + {10, sub_delims()}, + {5, oneof([$:, $@])} + ])), + lists:flatten(Gen)). slash() -> "/". @@ -235,19 +387,35 @@ path_empty_map() -> %%------------------------------------------------------------------------- -%% Path +%% Host %%------------------------------------------------------------------------- host_map() -> frequency([{30, reg_name()}, {30, ip_address()} ]). +host_map_nu() -> + frequency([{30, reg_name_nu()}, + {30, ip_address()} + ]). reg_name() -> - list(frequency([{30, alpha()}, - {10, sub_delims()}, - {10, unicode_char()} - ])). + ?LET(Gen, + list(frequency([{30, alpha()}, + {10, sub_delims()}, + {10, ptc_encoded_reserved()}, + {10, unicode_char()} + ])), + lists:flatten(Gen)). + +reg_name_nu() -> + ?LET(Gen, + list(frequency([{30, alpha()}, + {10, sub_delims()}, + {10, ptc_encoded_reserved()} + ])), + lists:flatten(Gen)). + ip_address() -> oneof(["127.0.0.1", "::127.0.0.1", @@ -258,10 +426,13 @@ ip_address() -> %% Generating only reg-names host_uri() -> - non_empty(list(frequency([{30, unreserved()}, - {10, sub_delims()}, - {10, pct_encoded()} - ]))). + ?LET(Gen, + non_empty(list(frequency([{30, unreserved()}, + {10, sub_delims()}, + {10, ptc_encoded_reserved()}, + {10, pct_encoded()} + ]))), + lists:flatten(Gen)). %%------------------------------------------------------------------------- %% Port, Query, Fragment @@ -274,6 +445,9 @@ port() -> query_map() -> unicode(). +query_map_nu() -> + non_unicode(). + query_uri() -> [$?| non_empty(list(frequency([{20, pchar()}, @@ -283,6 +457,10 @@ query_uri() -> fragment_map() -> unicode(). +fragment_map_nu() -> + non_unicode(). + + fragment_uri() -> [$?| non_empty(list(frequency([{20, pchar()}, {5, oneof([$/, $?])} % punctuation @@ -311,9 +489,14 @@ scheme(N, L) -> %%------------------------------------------------------------------------- unicode() -> list(frequency([{20, alpha()}, % alpha - {10, digit()}, % digit - {10, unicode_char()} % unicode - ])). + {10, digit()}, % digit + {10, unicode_char()} % unicode + ])). + +non_unicode() -> + list(frequency([{20, alpha()}, % alpha + {10, digit()} % digit + ])). scheme_char() -> frequency([{20, alpha()}, % alpha @@ -327,6 +510,7 @@ sub_delims() -> pchar() -> frequency([{20, unreserved()}, + {5, ptc_encoded_reserved()}, {5, pct_encoded()}, {5, sub_delims()}, {1, oneof([$:, $@])} % punctuation @@ -351,6 +535,22 @@ digit() -> pct_encoded() -> oneof(["%C3%A4", "%C3%A5", "%C3%B6"]). +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 2.2. Reserved Characters] +%% +%% reserved = gen-delims / sub-delims +%% +%% gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@" +%% 3A 2F 3F 23 5B 5D 40 +%% sub-delims = "!" / "$" / "&" / "'" / "(" / ")" +%% 21 24 26 27 28 29 +%% / "*" / "+" / "," / ";" / "=" +%% 2A 2B 2C 3B 3D +%%------------------------------------------------------------------------- +ptc_encoded_reserved() -> + oneof(["%3A","%2F","%3F","%23","%5B","%5D","%40", + "%21","%24","%26","%27","%28","%29", + "%2A","%2B","%2C","%3B","3D"]). %%%======================================================================== %%% Helpers @@ -359,3 +559,13 @@ proplist_to_map(L) -> lists:foldl(fun({K,V},M) -> M#{K => V}; (_,M) -> M end, #{}, L). + +map_scheme_host_to_lower(Map) -> + Fun = fun (scheme,V) -> + string:to_lower(V); + (host,V) -> + string:to_lower(V); + (_,V) -> + V + end, + maps:map(Fun, Map). diff --git a/lib/stdlib/test/uri_string_SUITE.erl b/lib/stdlib/test/uri_string_SUITE.erl index 92f8bb3292..2aa399525d 100644 --- a/lib/stdlib/test/uri_string_SUITE.erl +++ b/lib/stdlib/test/uri_string_SUITE.erl @@ -23,6 +23,12 @@ -export([all/0, suite/0,groups/0, normalize/1, normalize_map/1, normalize_return_map/1, normalize_negative/1, + normalize_binary_pct_encoded_userinfo/1, + normalize_binary_pct_encoded_query/1, + normalize_binary_pct_encoded_fragment/1, + normalize_pct_encoded_userinfo/1, + normalize_pct_encoded_query/1, + normalize_pct_encoded_fragment/1, parse_binary_fragment/1, parse_binary_host/1, parse_binary_host_ipv4/1, parse_binary_host_ipv6/1, parse_binary_path/1, parse_binary_pct_encoded_fragment/1, parse_binary_pct_encoded_query/1, @@ -41,7 +47,8 @@ transcode_basic/1, transcode_options/1, transcode_mixed/1, transcode_negative/1, compose_query/1, compose_query_latin1/1, compose_query_negative/1, dissect_query/1, dissect_query_negative/1, - interop_query_latin1/1, interop_query_utf8/1 + interop_query_latin1/1, interop_query_utf8/1, + regression_parse/1, regression_recompose/1, regression_normalize/1 ]). @@ -71,6 +78,12 @@ all() -> normalize_map, normalize_return_map, normalize_negative, + normalize_binary_pct_encoded_userinfo, + normalize_binary_pct_encoded_query, + normalize_binary_pct_encoded_fragment, + normalize_pct_encoded_userinfo, + normalize_pct_encoded_query, + normalize_pct_encoded_fragment, parse_binary_scheme, parse_binary_userinfo, parse_binary_pct_encoded_userinfo, @@ -120,7 +133,10 @@ all() -> dissect_query, dissect_query_negative, interop_query_latin1, - interop_query_utf8 + interop_query_utf8, + regression_parse, + regression_recompose, + regression_normalize ]. groups() -> @@ -338,20 +354,23 @@ parse_binary_userinfo(_Config) -> uri_string:parse(<<"foo://user:password@localhost">>). parse_binary_pct_encoded_userinfo(_Config) -> - #{scheme := <<"user">>, path := <<"合@気道"/utf8>>} = + #{scheme := <<"user">>, path := <<"%E5%90%88@%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"user:%E5%90%88@%E6%B0%97%E9%81%93">>), - #{path := <<"合気道@"/utf8>>} = uri_string:parse(<<"%E5%90%88%E6%B0%97%E9%81%93@">>), - #{path := <<"/合気道@"/utf8>>} = uri_string:parse(<<"/%E5%90%88%E6%B0%97%E9%81%93@">>), - #{path := <<"合@気道"/utf8>>} = uri_string:parse(<<"%E5%90%88@%E6%B0%97%E9%81%93">>), - #{userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + #{path := <<"%E5%90%88%E6%B0%97%E9%81%93@">>} = + uri_string:parse(<<"%E5%90%88%E6%B0%97%E9%81%93@">>), + #{path := <<"/%E5%90%88%E6%B0%97%E9%81%93@">>} = + uri_string:parse(<<"/%E5%90%88%E6%B0%97%E9%81%93@">>), + #{path := <<"%E5%90%88@%E6%B0%97%E9%81%93">>} = + uri_string:parse(<<"%E5%90%88@%E6%B0%97%E9%81%93">>), + #{userinfo := <<"%E5%90%88">>, host := <<"%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"//%E5%90%88@%E6%B0%97%E9%81%93">>), - #{userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + #{userinfo := <<"%E5%90%88:%E6%B0%97">>, host := <<"%E9%81%93">>} = uri_string:parse(<<"//%E5%90%88:%E6%B0%97@%E9%81%93">>), - #{scheme := <<"foo">>, path := <<"/合気道@"/utf8>>} = + #{scheme := <<"foo">>, path := <<"/%E5%90%88%E6%B0%97%E9%81%93@">>} = uri_string:parse(<<"foo:/%E5%90%88%E6%B0%97%E9%81%93@">>), - #{scheme := <<"foo">>, userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + #{scheme := <<"foo">>, userinfo := <<"%E5%90%88">>, host := <<"%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93">>), - #{scheme := <<"foo">>, userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + #{scheme := <<"foo">>, userinfo := <<"%E5%90%88:%E6%B0%97">>, host := <<"%E9%81%93">>} = uri_string:parse(<<"foo://%E5%90%88:%E6%B0%97@%E9%81%93">>), {error,invalid_uri,"@"} = uri_string:parse(<<"//%E5%90%88@%E6%B0%97%E9%81%93@">>), {error,invalid_uri,":"} = uri_string:parse(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93@">>). @@ -369,8 +388,8 @@ parse_binary_host_ipv4(_Config) -> #{host := <<"127.0.0.1">>, query := <<"name=ferret">>} = uri_string:parse(<<"//127.0.0.1?name=ferret">>), #{host := <<"127.0.0.1">>, fragment := <<"nose">>} = uri_string:parse(<<"//127.0.0.1#nose">>), - {error,invalid_uri,"x"} = uri_string:parse(<<"//127.0.0.x">>), - {error,invalid_uri,"1227.0.0.1"} = uri_string:parse(<<"//1227.0.0.1">>). + #{host := <<"127.0.0.x">>,path := <<>>} = uri_string:parse(<<"//127.0.0.x">>), + #{host := <<"1227.0.0.1">>,path := <<>>} = uri_string:parse(<<"//1227.0.0.1">>). parse_binary_host_ipv6(_Config) -> #{host := <<"::127.0.0.1">>} = uri_string:parse(<<"//[::127.0.0.1]">>), @@ -439,9 +458,9 @@ parse_binary_query(_Config) -> parse_binary_pct_encoded_query(_Config) -> #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/">>, - query := <<"name=合気道"/utf8>>} = + query := <<"name=%E5%90%88%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>), - #{host := <<"example.com">>, path := <<"/">>, query := <<"name=合気道"/utf8>>} = + #{host := <<"example.com">>, path := <<"/">>, query := <<"name=%E5%90%88%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>). parse_binary_fragment(_Config) -> @@ -472,9 +491,11 @@ parse_binary_fragment(_Config) -> uri_string:parse(<<"//example.com/#nose">>). parse_binary_pct_encoded_fragment(_Config) -> - #{scheme := <<"foo">>, host := <<"example.com">>, fragment := <<"合気道"/utf8>>} = + #{scheme := <<"foo">>, host := <<"example.com">>, + fragment := <<"%E5%90%88%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"foo://example.com#%E5%90%88%E6%B0%97%E9%81%93">>), - #{host := <<"example.com">>, path := <<"/">>, fragment := <<"合気道"/utf8>>} = + #{host := <<"example.com">>, path := <<"/">>, + fragment := <<"%E5%90%88%E6%B0%97%E9%81%93">>} = uri_string:parse(<<"//example.com/#%E5%90%88%E6%B0%97%E9%81%93">>). parse_scheme(_Config) -> @@ -506,25 +527,27 @@ parse_userinfo(_Config) -> uri_string:parse("foo://user:password@localhost"). parse_pct_encoded_userinfo(_Config) -> - #{scheme := "user", path := "合@気道"} = + #{scheme := "user", path := "%E5%90%88@%E6%B0%97%E9%81%93"} = uri_string:parse("user:%E5%90%88@%E6%B0%97%E9%81%93"), - #{path := "合気道@"} = uri_string:parse("%E5%90%88%E6%B0%97%E9%81%93@"), - #{path := "/合気道@"} = uri_string:parse("/%E5%90%88%E6%B0%97%E9%81%93@"), - #{path := "合@気道"} = uri_string:parse("%E5%90%88@%E6%B0%97%E9%81%93"), - #{userinfo := "合", host := "気道"} = + #{path := "%E5%90%88%E6%B0%97%E9%81%93@"} = + uri_string:parse("%E5%90%88%E6%B0%97%E9%81%93@"), + #{path := "/%E5%90%88%E6%B0%97%E9%81%93@"} = + uri_string:parse("/%E5%90%88%E6%B0%97%E9%81%93@"), + #{path := "%E5%90%88@%E6%B0%97%E9%81%93"} = + uri_string:parse("%E5%90%88@%E6%B0%97%E9%81%93"), + #{userinfo := "%E5%90%88", host := "%E6%B0%97%E9%81%93"} = uri_string:parse("//%E5%90%88@%E6%B0%97%E9%81%93"), - #{userinfo := "合:気", host := "道"} = + #{userinfo := "%E5%90%88:%E6%B0%97", host := "%E9%81%93"} = uri_string:parse("//%E5%90%88:%E6%B0%97@%E9%81%93"), - #{scheme := "foo", path := "/合気道@"} = + #{scheme := "foo", path := "/%E5%90%88%E6%B0%97%E9%81%93@"} = uri_string:parse("foo:/%E5%90%88%E6%B0%97%E9%81%93@"), - #{scheme := "foo", userinfo := "合", host := "気道"} = + #{scheme := "foo", userinfo := "%E5%90%88", host := "%E6%B0%97%E9%81%93"} = uri_string:parse("foo://%E5%90%88@%E6%B0%97%E9%81%93"), - #{scheme := "foo", userinfo := "合:気", host := "道"} = + #{scheme := "foo", userinfo := "%E5%90%88:%E6%B0%97", host := "%E9%81%93"} = uri_string:parse("foo://%E5%90%88:%E6%B0%97@%E9%81%93"), {error,invalid_uri,"@"} = uri_string:parse("//%E5%90%88@%E6%B0%97%E9%81%93@"), {error,invalid_uri,":"} = uri_string:parse("foo://%E5%90%88@%E6%B0%97%E9%81%93@"). - parse_host(_Config) -> #{host := "hostname"} = uri_string:parse("//hostname"), #{host := "hostname",scheme := "foo"} = uri_string:parse("foo://hostname"), @@ -538,8 +561,8 @@ parse_host_ipv4(_Config) -> #{host := "127.0.0.1", path := "/over/there"} = uri_string:parse("//127.0.0.1/over/there"), #{host := "127.0.0.1", query := "name=ferret"} = uri_string:parse("//127.0.0.1?name=ferret"), #{host := "127.0.0.1", fragment := "nose"} = uri_string:parse("//127.0.0.1#nose"), - {error,invalid_uri,"x"} = uri_string:parse("//127.0.0.x"), - {error,invalid_uri,"1227.0.0.1"} = uri_string:parse("//1227.0.0.1"). + #{host := "127.0.0.x",path := []} = uri_string:parse("//127.0.0.x"), + #{host := "1227.0.0.1",path := []} = uri_string:parse("//1227.0.0.1"). parse_host_ipv6(_Config) -> #{host := "::127.0.0.1"} = uri_string:parse("//[::127.0.0.1]"), @@ -602,9 +625,9 @@ parse_query(_Config) -> parse_pct_encoded_query(_Config) -> #{scheme := "foo", host := "example.com", path := "/", - query := "name=合気道"} = + query := "name=%E5%90%88%E6%B0%97%E9%81%93"} = uri_string:parse("foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93"), - #{host := "example.com", path := "/", query := "name=合気道"} = + #{host := "example.com", path := "/", query := "name=%E5%90%88%E6%B0%97%E9%81%93"} = uri_string:parse("//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93"). parse_fragment(_Config) -> @@ -635,9 +658,11 @@ parse_fragment(_Config) -> uri_string:parse("//example.com/#nose"). parse_pct_encoded_fragment(_Config) -> - #{scheme := "foo", host := "example.com", fragment := "合気道"} = + #{scheme := "foo", host := "example.com", + fragment := "%E5%90%88%E6%B0%97%E9%81%93"} = uri_string:parse("foo://example.com#%E5%90%88%E6%B0%97%E9%81%93"), - #{host := "example.com", path := "/", fragment := "合気道"} = + #{host := "example.com", path := "/", + fragment := "%E5%90%88%E6%B0%97%E9%81%93"} = uri_string:parse("//example.com/#%E5%90%88%E6%B0%97%E9%81%93"). parse_list(_Config) -> @@ -711,9 +736,7 @@ parse_negative(_Config) -> {error,invalid_uri,":"} = uri_string:parse("foo://usär@host"), {error,invalid_uri,"ö"} = uri_string:parse("//host/path?foö=bar"), {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,invalid_utf8,<<0,0,0,246>>} = uri_string:parse("//%00%00%00%F6"), {error,invalid_uri,"A"} = uri_string:parse("//localhost:A8"). @@ -913,7 +936,9 @@ normalize(_Config) -> <<"sftp://localhost">> = uri_string:normalize(<<"sftp://localhost:22">>), <<"tftp://localhost">> = - uri_string:normalize(<<"tftp://localhost:69">>). + uri_string:normalize(<<"tftp://localhost:69">>), + <<"/foo/%2F/bar">> = + uri_string:normalize(<<"/foo/%2f/%62ar">>). normalize_map(_Config) -> "/a/g" = uri_string:normalize(#{path => "/a/b/c/./../../g"}), @@ -942,7 +967,9 @@ normalize_map(_Config) -> host => <<"localhost">>}), <<"tftp://localhost">> = uri_string:normalize(#{scheme => <<"tftp">>,port => 69,path => <<>>, - host => <<"localhost">>}). + host => <<"localhost">>}), + "/foo/%2F/bar" = + uri_string:normalize(#{path => "/foo/%2f/%62ar"}). normalize_return_map(_Config) -> #{scheme := "http",path := "/a/g",host := "localhost-örebro"} = @@ -963,7 +990,82 @@ normalize_negative(_Config) -> {error,invalid_uri,":"} = uri_string:normalize("http://[192.168.0.1]", [return_map]), {error,invalid_uri,":"} = - uri_string:normalize(<<"http://[192.168.0.1]">>, [return_map]). + uri_string:normalize(<<"http://[192.168.0.1]">>, [return_map]), + {error,invalid_utf8,<<0,0,0,246>>} = uri_string:normalize("//%00%00%00%F6"). + +normalize_binary_pct_encoded_userinfo(_Config) -> + #{scheme := <<"user">>, path := <<"合@気道"/utf8>>} = + uri_string:normalize(<<"user:%E5%90%88@%E6%B0%97%E9%81%93">>, [return_map]), + #{path := <<"合気道@"/utf8>>} = + uri_string:normalize(<<"%E5%90%88%E6%B0%97%E9%81%93@">>, [return_map]), + #{path := <<"/合気道@"/utf8>>} = + uri_string:normalize(<<"/%E5%90%88%E6%B0%97%E9%81%93@">>, [return_map]), + #{path := <<"合@気道"/utf8>>} = + uri_string:normalize(<<"%E5%90%88@%E6%B0%97%E9%81%93">>, [return_map]), + #{userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + uri_string:normalize(<<"//%E5%90%88@%E6%B0%97%E9%81%93">>, [return_map]), + #{userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + uri_string:normalize(<<"//%E5%90%88:%E6%B0%97@%E9%81%93">>, [return_map]), + #{scheme := <<"foo">>, path := <<"/合気道@"/utf8>>} = + uri_string:normalize(<<"foo:/%E5%90%88%E6%B0%97%E9%81%93@">>, [return_map]), + #{scheme := <<"foo">>, userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + uri_string:normalize(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93">>, [return_map]), + #{scheme := <<"foo">>, userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + uri_string:normalize(<<"foo://%E5%90%88:%E6%B0%97@%E9%81%93">>, [return_map]), + {error,invalid_uri,"@"} = + uri_string:normalize(<<"//%E5%90%88@%E6%B0%97%E9%81%93@">>, [return_map]), + {error,invalid_uri,":"} = + uri_string:normalize(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93@">>, [return_map]). + +normalize_binary_pct_encoded_query(_Config) -> + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/">>, + query := <<"name=合気道"/utf8>>} = + uri_string:normalize(<<"foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>, [return_map]), + #{host := <<"example.com">>, path := <<"/">>, query := <<"name=合気道"/utf8>>} = + uri_string:normalize(<<"//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>, [return_map]). + +normalize_binary_pct_encoded_fragment(_Config) -> + #{scheme := <<"foo">>, host := <<"example.com">>, fragment := <<"合気道"/utf8>>} = + uri_string:normalize(<<"foo://example.com#%E5%90%88%E6%B0%97%E9%81%93">>, [return_map]), + #{host := <<"example.com">>, path := <<"/">>, fragment := <<"合気道"/utf8>>} = + uri_string:normalize(<<"//example.com/#%E5%90%88%E6%B0%97%E9%81%93">>, [return_map]). + +normalize_pct_encoded_userinfo(_Config) -> + #{scheme := "user", path := "合@気道"} = + uri_string:normalize("user:%E5%90%88@%E6%B0%97%E9%81%93", [return_map]), + #{path := "合気道@"} = + uri_string:normalize("%E5%90%88%E6%B0%97%E9%81%93@", [return_map]), + #{path := "/合気道@"} = + uri_string:normalize("/%E5%90%88%E6%B0%97%E9%81%93@", [return_map]), + #{path := "合@気道"} = + uri_string:normalize("%E5%90%88@%E6%B0%97%E9%81%93", [return_map]), + #{userinfo := "合", host := "気道"} = + uri_string:normalize("//%E5%90%88@%E6%B0%97%E9%81%93", [return_map]), + #{userinfo := "合:気", host := "道"} = + uri_string:normalize("//%E5%90%88:%E6%B0%97@%E9%81%93", [return_map]), + #{scheme := "foo", path := "/合気道@"} = + uri_string:normalize("foo:/%E5%90%88%E6%B0%97%E9%81%93@", [return_map]), + #{scheme := "foo", userinfo := "合", host := "気道"} = + uri_string:normalize("foo://%E5%90%88@%E6%B0%97%E9%81%93", [return_map]), + #{scheme := "foo", userinfo := "合:気", host := "道"} = + uri_string:normalize("foo://%E5%90%88:%E6%B0%97@%E9%81%93", [return_map]), + {error,invalid_uri,"@"} = + uri_string:normalize("//%E5%90%88@%E6%B0%97%E9%81%93@", [return_map]), + {error,invalid_uri,":"} = + uri_string:normalize("foo://%E5%90%88@%E6%B0%97%E9%81%93@", [return_map]). + +normalize_pct_encoded_query(_Config) -> + #{scheme := "foo", host := "example.com", path := "/", + query := "name=合気道"} = + uri_string:normalize("foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93", [return_map]), + #{host := "example.com", path := "/", query := "name=合気道"} = + uri_string:normalize("//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93", [return_map]). + +normalize_pct_encoded_fragment(_Config) -> + #{scheme := "foo", host := "example.com", fragment := "合気道"} = + uri_string:normalize("foo://example.com#%E5%90%88%E6%B0%97%E9%81%93", [return_map]), + #{host := "example.com", path := "/", fragment := "合気道"} = + uri_string:normalize("//example.com/#%E5%90%88%E6%B0%97%E9%81%93", [return_map]). interop_query_utf8(_Config) -> Q = uri_string:compose_query([{"foo bar","1"}, {"合", "2"}]), @@ -977,3 +1079,86 @@ interop_query_latin1(_Config) -> Uri1 = uri_string:transcode(Uri, [{in_encoding, latin1}]), #{query := Q1} = uri_string:parse(Uri1), [{"foo bar","1"}, {"合", "2"}] = uri_string:dissect_query(Q1). + +regression_parse(_Config) -> + #{host := "Bar",path := [],scheme := "FOo"} = + uri_string:parse("FOo://Bar"), + #{host := "bar",path := [],scheme := "foo"} = + uri_string:parse("foo://bar"), + #{host := "A%2f",path := "/%62ar",scheme := "foo"} = + uri_string:parse("foo://A%2f/%62ar"), + #{host := "a%2F",path := "/bar",scheme := "foo"} = + uri_string:parse("foo://a%2F/bar"), + #{host := "%C3%B6",path := [],scheme := "FOo"} = + uri_string:parse("FOo://%C3%B6"). + +regression_recompose(_Config) -> + "FOo://Bar" = + uri_string:recompose(#{host => "Bar",path => [],scheme => "FOo"}), + "foo://bar" = + uri_string:recompose(#{host => "bar",path => [],scheme => "foo"}), + "foo://A%2f/%62ar" = + uri_string:recompose(#{host => "A%2f",path => "/%62ar",scheme => "foo"}), + "foo://a%2F/bar" = + uri_string:recompose(#{host => "a%2F",path => "/bar",scheme => "foo"}), + "FOo://%C3%B6" = + uri_string:recompose(#{host => "%C3%B6",path => [],scheme => "FOo"}), + "FOo://%C3%B6" = + uri_string:recompose(#{host => "ö",path => [],scheme => "FOo"}). + +regression_normalize(_Config) -> + "foo://bar" = + uri_string:normalize("FOo://Bar"), + #{host := "bar",path := [],scheme := "foo"} = + uri_string:normalize("FOo://Bar", [return_map]), + + "foo://bar" = + uri_string:normalize("foo://bar"), + #{host := "bar",path := [],scheme := "foo"} = + uri_string:normalize("foo://bar", [return_map]), + + "foo://a%2F/bar" = + uri_string:normalize("foo://A%2f/%62ar"), + #{host := "a%2F",path := "/bar",scheme := "foo"} = + uri_string:normalize("foo://A%2f/%62ar", [return_map]), + + "foo://a%2F/bar" = + uri_string:normalize("foo://a%2F/bar"), + #{host := "a%2F",path := "/bar",scheme := "foo"} = + uri_string:normalize("foo://a%2F/bar", [return_map]), + + "foo://%C3%B6" = + uri_string:normalize("FOo://%C3%B6"), + #{host := "ö",path := [],scheme := "foo"} = + uri_string:normalize("FOo://%C3%B6", [return_map]), + + + "foo://bar" = + uri_string:normalize(#{host => "Bar",path => [],scheme => "FOo"}), + #{host := "bar",path := [],scheme := "foo"} = + uri_string:normalize(#{host => "Bar",path => [],scheme => "FOo"}, [return_map]), + + "foo://bar" = + uri_string:normalize(#{host => "bar",path => [],scheme => "foo"}), + #{host := "bar",path := [],scheme := "foo"} = + uri_string:normalize(#{host => "bar",path => [],scheme => "foo"}, [return_map]), + + "foo://a%2F/bar" = + uri_string:normalize(#{host => "A%2f",path => "/%62ar",scheme => "foo"}), + #{host := "a%2F",path := "/bar",scheme := "foo"} = + uri_string:normalize(#{host => "A%2f",path => "/%62ar",scheme => "foo"}, [return_map]), + + "foo://a%2F/bar" = + uri_string:normalize(#{host => "a%2F",path => "/bar",scheme => "foo"}), + #{host := "a%2F",path := "/bar",scheme := "foo"} = + uri_string:normalize(#{host => "a%2F",path => "/bar",scheme => "foo"}, [return_map]), + + "foo://%C3%B6" = + uri_string:normalize(#{host => "%C3%B6",path => [],scheme => "FOo"}), + #{host := "ö",path := [],scheme := "foo"} = + uri_string:normalize(#{host => "%C3%B6",path => [],scheme => "FOo"}, [return_map]), + + "foo://%C3%B6" = + uri_string:normalize(#{host => "ö",path => [],scheme => "FOo"}), + #{host := "ö",path := [],scheme := "foo"} = + uri_string:normalize(#{host => "ö",path => [],scheme => "FOo"}, [return_map]). diff --git a/lib/stdlib/test/uri_string_property_test_SUITE.erl b/lib/stdlib/test/uri_string_property_test_SUITE.erl index ae2c61c7aa..b01dd9bf65 100644 --- a/lib/stdlib/test/uri_string_property_test_SUITE.erl +++ b/lib/stdlib/test/uri_string_property_test_SUITE.erl @@ -22,7 +22,7 @@ -include_lib("common_test/include/ct.hrl"). -compile(export_all). -all() -> [recompose]. +all() -> [recompose, normalize]. init_per_suite(Config) -> ct_property_test:init_per_suite(Config). @@ -37,3 +37,8 @@ recompose(Config) -> ct_property_test:quickcheck( uri_string_recompose:prop_recompose(), Config). + +normalize(Config) -> + ct_property_test:quickcheck( + uri_string_recompose:prop_normalize(), + Config). -- cgit v1.2.3 From 24231f76f5e3fed3ec64abdfca22928928f46933 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?P=C3=A9ter=20Dimitrov?= Date: Mon, 11 Jun 2018 14:56:04 +0200 Subject: stdlib: Fix release_tests target in Makefile 'make release_tests' copies the property_test directory. Change-Id: Ife75711aec1ef6e85ef091b1ba28a221be3d1110 --- lib/stdlib/test/Makefile | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'lib/stdlib') diff --git a/lib/stdlib/test/Makefile b/lib/stdlib/test/Makefile index ae2e3d0e2b..bbe3cefa42 100644 --- a/lib/stdlib/test/Makefile +++ b/lib/stdlib/test/Makefile @@ -88,6 +88,7 @@ MODULES= \ unicode_SUITE \ unicode_util_SUITE \ uri_string_SUITE \ + uri_string_property_test_SUITE \ win32reg_SUITE \ y2k_SUITE \ select_SUITE \ @@ -152,6 +153,6 @@ release_tests_spec: make_emakefile $(INSTALL_DATA) stdlib.spec stdlib_bench.spec $(EMAKEFILE) \ $(ERL_FILES) $(COVERFILE) "$(RELSYSDIR)" chmod -R u+w "$(RELSYSDIR)" - @tar cf - *_SUITE_data | (cd "$(RELSYSDIR)"; tar xf -) + @tar cf - *_SUITE_data property_test | (cd "$(RELSYSDIR)"; tar xf -) release_docs_spec: -- cgit v1.2.3