Bug #6566: JSON.dump can generate invalid UTF-8 sequence

• Status: Closed
• Assignee: naruse (Yui NARUSE)
• Target version: -
• ruby -v: ruby 2.0.0dev (2012-06-09) [x86_64-linux]
• Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN

[shyouhei (Shyouhei Urabe)]

=begin
Look, in the following code JSON.dump outputs a sequence invalid as UTF-8.

-- encoding: utf-8 --

require 'json'
IO.popen('hexdump -C', 'w') do |fp|
JSON.dump(["\xea"], fp)
end

RFC4627 says that to encode JSON as a Unicode is a "SHALL". So this is an RFC violation.

=end

[nobu (Nobuyoshi Nakada)]

• File bug-6566.diff bug-6566.diff added

=begin
A bit simpler, it seems wrong that
JSON.generate(["\xea"]).valid_encoding?
returns (({false})).

I think this would be a bug in json generator, but what should happen
in this case? Seems (({convert_UTF8_to_JSON_ASCII()})) wants to reject
invalid sequence.
=end

[naruse (Yui NARUSE)]

json is not only for 1.9, so nobu's patch is not acceptable.
I made https://github.com/flori/json/pull/139 .

[naruse (Yui NARUSE)]

• Target version changed from 2.0.0 to 2.6

[naruse (Yui NARUSE)]

• Target version deleted (2.6)

[jeremyevans0 (Jeremy Evans)]

• Status changed from Assigned to Closed

I think this was fixed upstream last year: https://github.com/flori/json/commit/91a24ecac3d722bc3e545fdd0657ed225e91c2df