From 5fe4c673bb8ee10d0fccadb4da14d7a500c2b8ec Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?P=C3=A9ter=20Dimitrov?= Date: Wed, 18 Oct 2017 15:48:04 +0200 Subject: stdlib: Implement compose_query and dissect_query --- lib/stdlib/src/uri_string.erl | 226 +++++++++++++++++++++++++++++++---- lib/stdlib/test/uri_string_SUITE.erl | 38 +++++- 2 files changed, 240 insertions(+), 24 deletions(-) (limited to 'lib/stdlib') diff --git a/lib/stdlib/src/uri_string.erl b/lib/stdlib/src/uri_string.erl index 7d180f73b8..1b8f8b828f 100644 --- a/lib/stdlib/src/uri_string.erl +++ b/lib/stdlib/src/uri_string.erl @@ -226,8 +226,9 @@ %%------------------------------------------------------------------------- %% External API %%------------------------------------------------------------------------- --export([compose_query/1, create_uri_reference/2, dissect_query/1, normalize/1, - parse/1, recompose/1, resolve_uri_reference/2, transcode/2]). +-export([compose_query/1, compose_query/2, create_uri_reference/2, + dissect_query/1, normalize/1, parse/1, + recompose/1, resolve_uri_reference/2, transcode/2]). -export_type([uri_map/0, uri_string/0]). @@ -377,28 +378,66 @@ transcode(URIString, Options) when is_list(URIString) -> %%------------------------------------------------------------------------- -%% Working with query strings -%% HTML 2.0 - application/x-www-form-urlencoded -%% RFC 1866 [8.2.1] +%% Functions for working with the query part of a URI as a list +%% of key/value pairs. +%% HTML 2.0 (RFC 1866) defines a media type application/x-www-form-urlencoded +%% in section [8.2.1] "The form-urlencoded Media Type". %%------------------------------------------------------------------------- %%------------------------------------------------------------------------- %% Compose urlencoded query string from a list of unescaped key/value pairs. %%------------------------------------------------------------------------- -spec compose_query(QueryList) -> QueryString when - QueryList :: [{unicode:chardata(), unicode:chardata()}], - QueryString :: uri_string(). -compose_query(_) -> - "". + QueryList :: [{uri_string(), uri_string()}], + QueryString :: string(). +compose_query(List) -> + compose_query(List, []). + + +-spec compose_query(QueryList, Options) -> QueryString when + QueryList :: [{uri_string(), uri_string()}], + Options :: [{separator, atom()}], + QueryString :: string(). +compose_query([],_Options) -> + []; +compose_query(List, Options) -> + try compose_query(List, Options, []) of + Result -> Result + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end. +%% +compose_query([{Key,Value}|Rest], Options, Acc) -> + Separator = get_separator(Options, Acc), + K = form_urlencode(Key), + V = form_urlencode(Value), + compose_query(Rest, Options, Acc ++ Separator ++ K ++ "=" ++ V); +compose_query([], _Options, Acc) -> + Acc. + %%------------------------------------------------------------------------- %% Dissect a query string into a list of unescaped key/value pairs. %%------------------------------------------------------------------------- -spec dissect_query(QueryString) -> QueryList when QueryString :: uri_string(), - QueryList :: [{unicode:chardata(), unicode:chardata()}]. -dissect_query(_) -> - "". + QueryList :: [{string(), string()}]. +dissect_query([]) -> + []; +dissect_query(QueryString) when is_binary(QueryString) -> + L = convert_list(QueryString, utf8), + try dissect_query_key(L, [], [], []) of + Result -> Result + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end; +dissect_query(QueryString) -> + L = flatten_list(QueryString, utf8), + try dissect_query_key(L, [], [], []) of + Result -> Result + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end. %%%======================================================================== @@ -1705,10 +1744,10 @@ transcode_pct([], Acc, B, InEncoding, OutEncoding) -> %% Convert to binary convert_binary(Binary, InEncoding, OutEncoding) -> case unicode:characters_to_binary(Binary, InEncoding, OutEncoding) of - {error, List, RestData} -> - throw({error, List, RestData}); - {incomplete, List, RestData} -> - throw({error, List, RestData}); + {error, _List, RestData} -> + throw({error, unicode, RestData}); + {incomplete, _List, RestData} -> + throw({error, unicode, RestData}); Result -> Result end. @@ -1717,10 +1756,10 @@ convert_binary(Binary, InEncoding, OutEncoding) -> %% Convert to list convert_list(Binary, InEncoding) -> case unicode:characters_to_list(Binary, InEncoding) of - {error, List, RestData} -> - throw({error, List, RestData}); - {incomplete, List, RestData} -> - throw({error, List, RestData}); + {error, _List, RestData} -> + throw({error, unicode, RestData}); + {incomplete, _List, RestData} -> + throw({error, unicode, RestData}); Result -> Result end. @@ -1740,8 +1779,153 @@ flatten_list([H|T], InEnc, Acc) when is_list(H) -> flatten_list([H|T], InEnc, Acc) -> flatten_list(T, InEnc, [H|Acc]); flatten_list([], _InEnc, Acc) -> - lists:reverse(Acc). + lists:reverse(Acc); +flatten_list(Arg, _, _) -> + throw({error, badarg, Arg}). + percent_encode_segment(Segment) -> percent_encode_binary(Segment, <<>>). + + +%%------------------------------------------------------------------------- +%% Helper functions for compose_query +%%------------------------------------------------------------------------- + +%% Returns separator to be used between key-value pairs +get_separator(_, Acc) when length(Acc) =:= 0 -> + []; +get_separator([], _Acc) -> + "&"; +get_separator([{separator, amp}], _Acc) -> + "&"; +get_separator([{separator, semicolon}], _Acc) -> + ";". + + +%% Form-urlencode input based on RFC 1866 [8.2.1] +form_urlencode(Cs) when is_binary(Cs) -> + L = convert_list(Cs, utf8), + form_urlencode(L, []); +form_urlencode(Cs) -> + L = flatten_list(Cs, utf8), + form_urlencode(L, []). +%% +form_urlencode([], Acc) -> + lists:reverse(Acc); +form_urlencode([$ |T], Acc) -> + form_urlencode(T, [$+|Acc]); +form_urlencode([H|T], Acc) -> + case is_url_char(H) of + true -> + form_urlencode(T, [H|Acc]); + false -> + E = urlencode_char(H), + form_urlencode(T, lists:reverse(E) ++ Acc) + end. + + +urlencode_char(C) -> + B = percent_encode_binary(C), + unicode:characters_to_list(B). + + +%% Return true if input char can appear in URL according to +%% RFC 1738 "Uniform Resource Locators". +is_url_char(C) + when 0 =< C, C =< 31; + 128 =< C, C =< 255 -> false; +is_url_char(127) -> false; +is_url_char(C) -> + not (is_reserved(C) orelse is_unsafe(C)). + + +%% Reserved characters (RFC 1738) +is_reserved($;) -> true; +is_reserved($/) -> true; +is_reserved($?) -> true; +is_reserved($:) -> true; +is_reserved($@) -> true; +is_reserved($=) -> true; +is_reserved($&) -> true; +is_reserved(_) -> false. + + +%% Unsafe characters (RFC 1738) +is_unsafe(${) -> true; +is_unsafe($}) -> true; +is_unsafe($|) -> true; +is_unsafe($\\) -> true; +is_unsafe($^) -> true; +is_unsafe($~) -> true; +is_unsafe($[) -> true; +is_unsafe($]) -> true; +is_unsafe($`) -> true; +is_unsafe(_) -> false. + + +%%------------------------------------------------------------------------- +%% Helper functions for dissect_query +%%------------------------------------------------------------------------- +dissect_query_key([$=|T], Acc, Key, Value) -> + dissect_query_value(T, Acc, Key, Value); +dissect_query_key([H|T], Acc, Key, Value) -> + dissect_query_key(T, Acc, [H|Key], Value); +dissect_query_key(L, _, _, _) -> + throw({error, missing_value, L}). + + +dissect_query_value([$&|_] = L, Acc, Key, Value) -> + K = form_urldecode(lists:reverse(Key)), + V = form_urldecode(lists:reverse(Value)), + dissect_query_separator_amp(L, [{K,V}|Acc], [], []); +dissect_query_value([$;|_] = L, Acc, Key, Value) -> + K = form_urldecode(lists:reverse(Key)), + V = form_urldecode(lists:reverse(Value)), + dissect_query_separator_semicolon(L, [{K,V}|Acc], [], []); +dissect_query_value([H|T], Acc, Key, Value) -> + dissect_query_value(T, Acc, Key, [H|Value]); +dissect_query_value([], Acc, Key, Value) -> + K = form_urldecode(lists:reverse(Key)), + V = form_urldecode(lists:reverse(Value)), + lists:reverse([{K,V}|Acc]). + + +dissect_query_separator_amp("&" ++ T, Acc, Key, Value) -> + dissect_query_key(T, Acc, Key, Value); +dissect_query_separator_amp(L, _, _, _) -> + throw({error, invalid_separator, L}). + + +dissect_query_separator_semicolon([$;|T], Acc, Key, Value) -> + dissect_query_key(T, Acc, Key, Value). + + +%% Form-urldecode input based on RFC 1866 [8.2.1] +form_urldecode(Cs) -> + B = convert_binary(Cs, utf8, utf8), + Result = form_urldecode(B, <<>>), + convert_list(Result, utf8). +%% +form_urldecode(<<>>, Acc) -> + convert_list(Acc, utf8); +form_urldecode(<<$+,T/binary>>, Acc) -> + form_urlencode(T, [$ |Acc]); +form_urldecode(<<$%,C0,C1,T/binary>>, Acc) -> + case is_hex_digit(C0) andalso is_hex_digit(C1) of + true -> + V = ?HEX2DEC(C0)*16+?HEX2DEC(C1), + form_urldecode(T, <>); + false -> + L = convert_list(<<$%,C0,C1,T/binary>>, utf8), + throw({error, urldecode, L}) + end; +form_urldecode(<>, Acc) -> + case is_url_char(H) of + true -> + form_urldecode(T, <>); + false -> + L = convert_list(<>, utf8), + throw({error, urldecode, L}) + end. diff --git a/lib/stdlib/test/uri_string_SUITE.erl b/lib/stdlib/test/uri_string_SUITE.erl index 901d38a4da..beb534e023 100644 --- a/lib/stdlib/test/uri_string_SUITE.erl +++ b/lib/stdlib/test/uri_string_SUITE.erl @@ -37,7 +37,9 @@ recompose_query/1, recompose_parse_query/1, recompose_path/1, recompose_parse_path/1, recompose_autogen/1, parse_recompose_autogen/1, - transcode_basic/1, transcode_options/1, transcode_mixed/1, transcode_negative/1 + transcode_basic/1, transcode_options/1, transcode_mixed/1, transcode_negative/1, + compose_query/1, compose_query_negative/1, + dissect_query/1, dissect_query_negative/1 ]). @@ -104,7 +106,11 @@ all() -> transcode_basic, transcode_options, transcode_mixed, - transcode_negative + transcode_negative, + compose_query, + compose_query_negative, + dissect_query, + dissect_query_negative ]. groups() -> @@ -803,5 +809,31 @@ transcode_mixed(_Config) -> transcode_negative(_Config) -> {invalid_input,"foo","BX"} = uri_string:transcode(<<"foo%C3%BXbar"/utf8>>, [{in_encoding, utf8},{out_encoding, utf32}]), - {invalid_input,<<>>,<<"ö">>} = + {invalid_input,unicode,<<"ö">>} = uri_string:transcode("foo%F6bar", [{in_encoding, utf8},{out_encoding, utf8}]). + +compose_query(_Config) -> + [] = uri_string:compose_query([]), + "foo=1&bar=2" = uri_string:compose_query([{<<"foo">>,"1"}, {"bar", "2"}]), + "foo=1&bar=2" = uri_string:compose_query([{"foo","1"}, {"bar", "2"}],[{separator,amp}]), + "foo=1;bar=2" = uri_string:compose_query([{"foo","1"}, {"bar", "2"}],[{separator,semicolon}]), + "foo+bar=1&%C3%B6=2" = uri_string:compose_query([{"foo bar","1"}, {"ö", "2"}]). + +compose_query_negative(_Config) -> + {error,badarg,4} = uri_string:compose_query([{"",4}]). + +dissect_query(_Config) -> + [] = uri_string:dissect_query(""), + [{"foo","1"}, {"bar", "2"}] = uri_string:dissect_query("foo=1&bar=2"), + [{"foo","1"}, {"bar", "2"}] = uri_string:dissect_query("foo=1;bar=2"), + [{"foo","1"}, {"bar", "222"}] = uri_string:dissect_query([<<"foo=1;bar=2">>,"22"]), + [{"foo","ö"}, {"bar", "2"}] = uri_string:dissect_query("foo=%C3%B6&bar=2"). + +dissect_query_negative(_Config) -> + {error,invalid_separator,"≈bar=2"} = + uri_string:dissect_query("foo=1≈bar=2"), + {error,urldecode,"&bar"} = + uri_string:dissect_query("foo1&bar=2"), + {error,urldecode,"%XX%B6"} = uri_string:dissect_query("foo=%XX%B6&bar=2"), + {error,unicode,<<153,182>>} = + uri_string:dissect_query("foo=%99%B6&bar=2"). -- cgit v1.2.3