From da11b15aef87f392a807b4756bf285160e15a194 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?P=C3=A9ter=20Dimitrov?= <peterdmv@erlang.org>
Date: Mon, 23 Oct 2017 12:02:16 +0200
Subject: stdlib: Update supported separators (query string)

Update list of supported separators:
 - escaped_amp (default): "&amp;"
 - amp: "&"
 - semicolon: ";"
---
 lib/stdlib/doc/src/uri_string.xml    | 10 +++++-----
 lib/stdlib/src/uri_string.erl        |  4 ++++
 lib/stdlib/test/uri_string_SUITE.erl |  6 ++++--
 3 files changed, 13 insertions(+), 7 deletions(-)

(limited to 'lib')

diff --git a/lib/stdlib/doc/src/uri_string.xml b/lib/stdlib/doc/src/uri_string.xml
index 496573ae2f..97b38ea93e 100644
--- a/lib/stdlib/doc/src/uri_string.xml
+++ b/lib/stdlib/doc/src/uri_string.xml
@@ -28,9 +28,10 @@
     <rev>A</rev>
   </header>
   <module>uri_string</module>
-  <modulesummary>RFC 3986 compliant URI processing functions.</modulesummary>
+  <modulesummary>URI processing functions.</modulesummary>
   <description>
-    <p>This module contains functions for parsing and handling RFC 3986 compliant URIs.</p>
+    <p>This module contains functions for parsing and handling URIs (RFC 3986) and
+    form-urlencoded query strings (RFC 1866).</p>
     <p>A URI is an identifier consisting of a sequence of characters   matching the syntax
     rule named <em>URI</em> in <em>RFC 3986</em>.</p>
     <p>  The generic URI syntax consists of a hierarchical sequence of components referred
@@ -109,7 +110,7 @@
 	<c><anno>QueryList</anno></c>, a list of unescaped key-value pairs.
 	Media type <c>application/x-www-form-urlencoded</c> is defined in section
 	8.2.1 of <c>RFC 1866</c> (HTML 2.0). Reserved and unsafe characters, as
-	defined by RFC 1738 (Uniform Resource Locators), are procent-encoded.
+	defined by RFC 1738 (Uniform Resource Locators), are percent-encoded.
 	</p>
         <p><em>Example:</em></p>
         <pre>
@@ -125,8 +126,7 @@
       <desc>
         <p>Same as <c>compose_query/1</c> but with an additional
 	<c><anno>Options</anno></c> parameter, that controls the type of separator used
-	between key-value pairs. There are two supported separator types: <c>amp</c> (<![CDATA[&amp;]]>)
-	and <c>semicolon</c> (;).</p>
+	between key-value pairs. There are three supported separator types: <c>amp</c> (<![CDATA[&]]>), <c>escaped_amp</c> (<![CDATA[&amp;]]>) and <c>semicolon</c> (;). If the parameter <c><anno>Options</anno></c> is empty, separator takes the default value (<c>escaped_amp</c>).</p>
         <p><em>Example:</em></p>
         <pre>
 1> <input>uri_string:compose_query([{"foo bar","1"},{"city","örebro"}],</input>
diff --git a/lib/stdlib/src/uri_string.erl b/lib/stdlib/src/uri_string.erl
index 8723d3f183..a4fd9c66f4 100644
--- a/lib/stdlib/src/uri_string.erl
+++ b/lib/stdlib/src/uri_string.erl
@@ -1806,6 +1806,8 @@ get_separator(_, Acc) when length(Acc) =:= 0 ->
 get_separator([], _Acc) ->
     "&amp;";
 get_separator([{separator, amp}], _Acc) ->
+    "&";
+get_separator([{separator, escaped_amp}], _Acc) ->
     "&amp;";
 get_separator([{separator, semicolon}], _Acc) ->
     ";".
@@ -1901,6 +1903,8 @@ dissect_query_value([], Acc, Key, Value) ->
 
 dissect_query_separator_amp("&amp;" ++ T, Acc, Key, Value) ->
     dissect_query_key(T, Acc, Key, Value);
+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}).
 
diff --git a/lib/stdlib/test/uri_string_SUITE.erl b/lib/stdlib/test/uri_string_SUITE.erl
index b70cb842de..fe832ac82c 100644
--- a/lib/stdlib/test/uri_string_SUITE.erl
+++ b/lib/stdlib/test/uri_string_SUITE.erl
@@ -827,7 +827,8 @@ transcode_negative(_Config) ->
 compose_query(_Config) ->
     [] = uri_string:compose_query([]),
     "foo=1&amp;bar=2" = uri_string:compose_query([{<<"foo">>,"1"}, {"bar", "2"}]),
-    "foo=1&amp;bar=2" = uri_string:compose_query([{"foo","1"}, {"bar", "2"}],[{separator,amp}]),
+    "foo=1&amp;bar=2" = uri_string:compose_query([{"foo","1"}, {"bar", "2"}],[{separator,escaped_amp}]),
+    "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&amp;%C3%B6=2" = uri_string:compose_query([{"foo bar","1"}, {"ö", "2"}]).
 
@@ -837,12 +838,13 @@ compose_query_negative(_Config) ->
 dissect_query(_Config) ->
     [] = uri_string:dissect_query(""),
     [{"foo","1"}, {"bar", "2"}] = uri_string:dissect_query("foo=1&amp;bar=2"),
+    [{"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&amp;bar=2").
 
 dissect_query_negative(_Config) ->
-    {error,invalid_separator,"&ap;bar=2"} =
+    {error,urldecode,";bar"} =
         uri_string:dissect_query("foo=1&ap;bar=2"),
     {error,urldecode,"&amp;bar"} =
         uri_string:dissect_query("foo1&amp;bar=2"),
-- 
cgit v1.2.3