Forum: Ruby Ruby 1.9.1 built-in JSON troubles

Posted by Aaron D. Gifford (Guest)
on 2010-03-06 21:55
(Received via mailing list)
I'm puzzled.  On a box running Ruby 1.9.1 I try this:

user@host:/path$ irb
irb(main):001:0> require 'json'
=> true
irb(main):002:0> JSON.parse(JSON.generate('\\'))
JSON::ParserError: 598: unexpected token at '"\\"'
        from /usr/local/lib/ruby/1.9/json/common.rb:122:in `parse'
        from /usr/local/lib/ruby/1.9/json/common.rb:122:in `parse'
        from (irb):2
        from /usr/local/bin/irb:12:in `<main>'
irb(main):003:0> RUBY_VERSION
=> "1.9.1"
irb(main):004:0> ^D

This is the JSON module that came with 1.9--I don't have a JSON gem 
installed.

Any ideas where to go looking?  A little web searching showed me
various similar problems, but most were under 1.8.

Looking for a clue,
Aaron out.
Posted by Brian Candler (candlerb)
on 2010-03-08 13:39
Aaron D. Gifford wrote:
> irb(main):002:0> JSON.parse(JSON.generate('\\'))
> JSON::ParserError: 598: unexpected token at '"\\"'

This is simply because JSON.parse only parses JSON *objects* and 
*arrays*, not strings or numbers.

>> JSON.parse('{"foo":"bar"}')
=> {"foo"=>"bar"}
>> JSON.parse('["foo","bar"]')
=> ["foo", "bar"]
>> JSON.parse('"bar"')
JSON::ParserError: 574: unexpected token at '"bar"'

I use the following workaround:

>> json = JSON.generate('\\')
=> "\"\\\\\""
>> JSON.parse("[" + json + "]").first
=> "\\"

HTH,

Brian.
Posted by Aaron D. Gifford (Guest)
on 2010-03-08 18:46
(Received via mailing list)
On Mon, Mar 8, 2010 at 5:39 AM, Brian Candler <b.candler@pobox.com> 
wrote:
> => ["foo", "bar"]
> HTH,
>
> Brian.

Thanks for the note, Brian.

The JSON documentation does NOT make that clear.

And that limitation is a terrible limitation.  Ugly!

I ended up rolling my own pure Ruby JSON parser that doesn't have that
work-around requirement.  It looks like I could have avoided that with
your work-around (even though it's sad that the JSON library requires
such an ugly kludge).

Aaron out.
Posted by Aaron D. Gifford (Guest)
on 2010-03-08 19:02
(Received via mailing list)
As per Brian's suggestion, I used his workaround this way:
(IMHO, JSON ought to handle this automagically.)

require 'json'
def json_parse_kludge(data)
  return JSON.parse('[' + data + ']')[0] if data.is_a?(String) &&
data[0,1] == '"'
  JSON.parse(data)
end

Fixes my issue:  json_parse_kludge(JSON.generate('\\'))

irb(main):001:0> require 'json'
=> true
irb(main):002:0> def json_parse_kludge(data)
irb(main):003:1>   return JSON.parse('[' + data + ']')[0] if
data.is_a?(String) && data[0,1] == '"'
irb(main):004:1>   JSON.parse(data)
irb(main):005:1> end
=> nil
irb(main):006:0> json_parse_kludge(JSON.generate('//'))
=> "//"
irb(main):007:0>

Thanks again, Brian!

Aaron out.
Posted by Brian Candler (candlerb)
on 2010-03-08 20:19
Aaron D. Gifford wrote:
> Thanks for the note, Brian.
> 
> The JSON documentation does NOT make that clear.

I could understand if it only allowed an object at the top level (which 
is what CouchDB requires), but I agree it doesn't make sense to allow 
two types of values but not the other types.

Your kludge is a bit messy, it won't parse ' "foo"' for example (with a 
leading space). If you are afraid of building an extra string, then how 
about:

def jparse(str)
  return JSON.parse(str) if str =~ /\A\s*[{\[]/
  JSON.parse("[#{str}]")[0]
end
Posted by Aaron D. Gifford (Guest)
on 2010-03-09 01:41
(Received via mailing list)
On Mon, Mar 8, 2010 at 12:19 PM, Brian Candler <b.candler@pobox.com> 
wrote:
>... If you are afraid of building an extra string, then how
> about:
>
> def jparse(str)
>  return JSON.parse(str) if str =~ /\A\s*[{\[]/
>  JSON.parse("[#{str}]")[0]
> end

Definitely more robust, since that will handle numbers too.  Thanks

Aaron out.
Please log in before posting. Registration is free and takes only a minute.
Existing account (Switch to SSL-encrypted connection)
NEW: Do you have a Google/GoogleMail or Yahoo account? No registration required!
Log in with Google account | Log in with Yahoo account
No account? Register here.