From a6727897392003115d2173baa152cd23eaf1ced6 Mon Sep 17 00:00:00 2001
From: naruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>
Date: Mon, 23 Jun 2014 20:01:27 +0000
Subject: [PATCH] * lib/uri/mailto.rb: support RFC6068.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@46528 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
---
 ChangeLog               |  4 +++
 lib/uri/mailto.rb       | 65 ++++++++++++++---------------------------
 test/uri/test_mailto.rb |  5 +---
 3 files changed, 27 insertions(+), 47 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index a0feda9eb7..fecacc5df4 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,7 @@
+Tue Jun 24 03:52:35 2014  NARUSE, Yui  <naruse@ruby-lang.org>
+
+	* lib/uri/mailto.rb: support RFC6068.
+
 Mon Jun 23 18:44:45 2014  SHIBATA Hiroshi  <shibata.hiroshi@gmail.com>
 
 	* tool/config_files.rb: rename class ConfigFiles to Downloader.
diff --git a/lib/uri/mailto.rb b/lib/uri/mailto.rb
index 03155fb917..6422e84f37 100644
--- a/lib/uri/mailto.rb
+++ b/lib/uri/mailto.rb
@@ -12,7 +12,7 @@ require 'uri/generic'
 module URI
 
   #
-  # RFC2368, The mailto URL scheme
+  # RFC6068, The mailto URL scheme
   #
   class MailTo < Generic
     include REGEXP
@@ -37,28 +37,12 @@ module URI
     #
     #  Within mailto URLs, the characters "?", "=", "&" are reserved.
 
-    # hname      =  *urlc
-    # hvalue     =  *urlc
-    # header     =  hname "=" hvalue
-    HEADER_PATTERN = "(?:[^?=&]*=[^?=&]*)".freeze
-    HEADER_REGEXP  = Regexp.new(HEADER_PATTERN).freeze
+    HEADER_REGEXP  = /\A(?<hfield>(?:%\h\h|[!$'-.0-;@-Z_a-z~])*=(?:%\h\h|[!$'-.0-;@-Z_a-z~])*)(?:&\g<hfield>)*\z/
     # headers    =  "?" header *( "&" header )
-    # to         =  #mailbox
-    # mailtoURL  =  "mailto:" [ to ] [ headers ]
-    MAILBOX_PATTERN = "(?:#{PATTERN::ESCAPED}|[^(),%?=&])".freeze
-    MAILTO_REGEXP = Regexp.new(" # :nodoc:
-      \\A
-      (#{MAILBOX_PATTERN}*?)                          (?# 1: to)
-      (?:
-        \\?
-        (#{HEADER_PATTERN}(?:\\&#{HEADER_PATTERN})*)  (?# 2: headers)
-      )?
-      (?:
-        \\#
-        (#{PATTERN::FRAGMENT})                        (?# 3: fragment)
-      )?
-      \\z
-    ", Regexp::EXTENDED).freeze
+    # to         =  addr-spec *("," addr-spec )
+    # mailtoURI  =  "mailto:" [ to ] [ hfields ]
+    TO_REGEXP = /\A(?:[!#-'*+\-\/-9=?A-Z^-~]+(?:\.[!#-'*+\-\/-9=?A-Z^-~]+)*|"(?:\\[!-~]|[!#-\[\]-~])*")@(?:[!#-'*+\-\/-9=?A-Z^-~]+(?:\.[!#-'*+\-\/-9=?A-Z^-~]+)*|\[[!-Z^-~]*\])\z/
+    MAILTO_DATA_REGEXP = /\A(?<to>(?<addr-spec>(?<local-part>[!#-'*+\-\/-9=?A-Z^-~]+(?:\.[!#-'*+\-\/-9=?A-Z^-~]+)*|"(?:\\[!-~]|[!#-\[\]-~])*")@(?<domain>[!#-'*+\-\/-9=?A-Z^-~]+(?:\.[!#-'*+\-\/-9=?A-Z^-~]+)*|\[[!-Z^-~]*\]))(?:,\g<addr-spec>)*)?(?<hfields>\?(?<header>(?<hfield>(?:%\h\h|[!$'-.0-;@-Z_a-z~])*=(?:%\h\h|[!$'-.0-;@-Z_a-z~])*)(?:&\g<hfield>)*))?\z/
     # :startdoc:
 
     #
@@ -99,8 +83,8 @@ module URI
 
       if tmp[:headers]
         tmp[:opaque] << '?'
-
-        if tmp[:headers].kind_of?(Array)
+        case tmp[:headers]
+        when Array
           tmp[:opaque] << tmp[:headers].collect { |x|
             if x.kind_of?(Array)
               x[0] + '=' + x[1..-1].join
@@ -108,12 +92,10 @@ module URI
               x.to_s
             end
           }.join('&')
-
-        elsif tmp[:headers].kind_of?(Hash)
+        when Hash
           tmp[:opaque] << tmp[:headers].collect { |h,v|
             h + '=' + v
           }.join('&')
-
         else
           tmp[:opaque] << tmp[:headers].to_s
         end
@@ -137,13 +119,13 @@ module URI
       @to = nil
       @headers = []
 
-      if MAILTO_REGEXP =~ @opaque
+      if m = MAILTO_DATA_REGEXP.match(@opaque)
         if arg[10] # arg_check
-          self.to = $1
-          self.headers = $2
+          self.to = m["to"]
+          self.headers = m["header"]
         else
-          set_to($1)
-          set_headers($2)
+          set_to(m["to"])
+          set_headers(m["header"])
         end
 
       else
@@ -158,14 +140,12 @@ module URI
     # E-mail headers set by the URL, as an Array of Arrays
     attr_reader :headers
 
-    # check the to +v+ component against either
-    # * URI::Parser Regexp for :OPAQUE
-    # * MAILBOX_PATTERN
+    # check the to +v+ component
     def check_to(v)
       return true unless v
       return true if v.size == 0
 
-      if parser.regexp[:OPAQUE] !~ v || /\A#{MAILBOX_PATTERN}*\z/o !~ v
+      if TO_REGEXP !~ v
         raise InvalidComponentError,
           "bad component(expected opaque component): #{v}"
       end
@@ -188,14 +168,13 @@ module URI
     end
 
     # check the headers +v+ component against either
-    # * URI::Parser Regexp for :OPAQUE
-    # * HEADER_PATTERN
+    # * URI::Parser Regexp for :QUERY
+    # * HEADER_REGEXP
     def check_headers(v)
       return true unless v
       return true if v.size == 0
-
-      if parser.regexp[:OPAQUE] !~ v ||
-          /\A(#{HEADER_PATTERN}(?:\&#{HEADER_PATTERN})*)\z/o !~ v
+      if parser.regexp[:QUERY] !~ v ||
+          HEADER_REGEXP !~ v
         raise InvalidComponentError,
           "bad component(expected opaque component): #{v}"
       end
@@ -208,8 +187,8 @@ module URI
     def set_headers(v)
       @headers = []
       if v
-        v.scan(HEADER_REGEXP) do |x|
-          @headers << x.split(/=/o, 2)
+        v.split('&').each do |x|
+          @headers << x.split(/=/, 2)
         end
       end
     end
diff --git a/test/uri/test_mailto.rb b/test/uri/test_mailto.rb
index c2158845e8..3aac447436 100644
--- a/test/uri/test_mailto.rb
+++ b/test/uri/test_mailto.rb
@@ -100,9 +100,6 @@ class TestMailTo < Test::Unit::TestCase
     # mailto:javascript:alert()
     bad << ["javascript:alert()", []]
 
-    # '=' which is in hname or hvalue is wrong.
-    bad << ["foo@example.jp?subject=1+1=2", []]
-
     ok.each do |x|
       assert_equal(x[0],
 		   @u.build(x[1]).to_s)
@@ -111,7 +108,7 @@ class TestMailTo < Test::Unit::TestCase
     end
 
     bad.each do |x|
-      assert_raise(URI::InvalidComponentError) {
+      assert_raise(URI::InvalidComponentError, %[URI::MailTo.build(#{x.inspect})]) {
 	@u.build(x)
       }
     end