From 2f1c5789c1882413df0346fec1f29eed24f37698 Mon Sep 17 00:00:00 2001
From: Godfrey Chan <godfreykfc@gmail.com>
Date: Wed, 27 Nov 2013 02:36:10 -0800
Subject: Fixed a long-standing bug in `json_escape` that strips quotation
 marks

---
 actionview/CHANGELOG.md                            |  4 ++
 .../core_ext/string/output_safety.rb               | 61 +++++++++++++++++-----
 2 files changed, 53 insertions(+), 12 deletions(-)

diff --git a/actionview/CHANGELOG.md b/actionview/CHANGELOG.md
index d53b321f97..65d045d1f7 100644
--- a/actionview/CHANGELOG.md
+++ b/actionview/CHANGELOG.md
@@ -1,3 +1,7 @@
+*   Fixed a long-standing bug in `json_escape` that causes quotation marks to be stripped.
+
+    *Godfrey Chan*
+
 *   `ActionView::MissingTemplate` includes underscore when raised for a partial.
 
     Fixes #13002.
diff --git a/activesupport/lib/active_support/core_ext/string/output_safety.rb b/activesupport/lib/active_support/core_ext/string/output_safety.rb
index c21650bb83..d54265f1b1 100644
--- a/activesupport/lib/active_support/core_ext/string/output_safety.rb
+++ b/activesupport/lib/active_support/core_ext/string/output_safety.rb
@@ -6,7 +6,7 @@ class ERB
     HTML_ESCAPE = { '&' => '&amp;',  '>' => '&gt;',   '<' => '&lt;', '"' => '&quot;', "'" => '&#39;' }
     JSON_ESCAPE = { '&' => '\u0026', '>' => '\u003E', '<' => '\u003C' }
     HTML_ESCAPE_ONCE_REGEXP = /["><']|&(?!([a-zA-Z]+|(#\d+));)/
-    JSON_ESCAPE_REGEXP = /[&"><]/
+    JSON_ESCAPE_REGEXP = /[&><]/
 
     # A utility method for escaping HTML tag characters.
     # This method is also aliased as <tt>h</tt>.
@@ -48,17 +48,54 @@ class ERB
 
     module_function :html_escape_once
 
-    # A utility method for escaping HTML entities in JSON strings
-    # using \uXXXX JavaScript escape sequences for string literals:
-    #
-    #   json_escape('is a > 0 & a < 10?')
-    #   # => is a \u003E 0 \u0026 a \u003C 10?
-    #
-    # Note that after this operation is performed the output is not
-    # valid JSON. In particular double quotes are removed:
-    #
-    #   json_escape('{"name":"john","created_at":"2010-04-28T01:39:31Z","id":1}')
-    #   # => {name:john,created_at:2010-04-28T01:39:31Z,id:1}
+    # A utility method for escaping HTML entities in JSON strings. Specifically, the
+    # &, > and < characters are replaced with their equivilant unicode escaped form -
+    # \u0026, \u003e, and \u003c. These sequences has identical meaning as the original
+    # characters inside the context of a JSON string, so assuming the input is a valid
+    # and well-formed JSON value, the output will have equivilant meaning when parsed:
+    # 
+    #   json = JSON.generate({ name: "</script><script>alert('PWNED!!!')</script>"})
+    #   # => "{\"name\":\"</script><script>alert('PWNED!!!')</script>\"}"
+    # 
+    #   json_escape(json)
+    #   # => "{\"name\":\"\\u003C/script\\u003E\\u003Cscript\\u003Ealert('PWNED!!!')\\u003C/script\\u003E\"}"
+    # 
+    #   JSON.parse(json) == JSON.parse(json_escape(json))
+    #   # => true
+    # 
+    # The intended use case for this method is to escape JSON strings before including
+    # them inside a script tag to avoid XSS vulnerability:
+    # 
+    #   <script type="application/javascript">
+    #     var currentUser = <%= json_escape current_user.to_json %>;
+    #   </script>
+    # 
+    # WARNING: this helper only works with valid JSON. Using this on non-JSON values
+    # will open up serious XSS vulnerabilities. For example, if you replace the
+    # +current_user.to_json+ in the example above with user input instead, the browser
+    # will happily eval() that string as JavaScript.
+    # 
+    # The escaping performed in this method is identical to those performed in the
+    # ActiveSupport JSON encoder when +ActiveSupport.escape_html_entities_in_json+ is
+    # set to true. Because this transformation is idempotent, this helper can be
+    # applied even if +ActiveSupport.escape_html_entities_in_json+ is already true.
+    # 
+    # Therefore, when you are unsure if +ActiveSupport.escape_html_entities_in_json+
+    # is enabled, or if you are unsure where your JSON string originated from, it
+    # is recommended that you always apply this helper (other libraries, such as the
+    # JSON gem, does not provide this kind of protection by default; also some gems
+    # might override +#to_json+ to bypass ActiveSupport's encoder).
+    # 
+    # The output of this helper method is marked as HTML safe so that you can directly
+    # include it inside a +<script>+ tag as shown above.
+    # 
+    # However, it is NOT safe to use the output of this inside an HTML attribute,
+    # because quotation marks are not escaped. Doing so might break your page's layout.
+    # If you intend to use this inside an HTML attribute, you should use the 
+    # +html_escape+ helper (or its +h+ alias) instead:
+    # 
+    #   <div data-user-info="<%= h current_user.to_json %>">...</div>
+    # 
     def json_escape(s)
       result = s.to_s.gsub(JSON_ESCAPE_REGEXP, JSON_ESCAPE)
       s.html_safe? ? result.html_safe : result
-- 
cgit v1.2.3