YAML for Ruby: How to turn off the boolean interpretations?

I have YAML documents generated in Perl and Python that I process with
Ruby.

Problem: Ruby wants to auto-interpret boolean flagged words, example:

···

---
words:
  - yes
  - put
  - it
  - on
  - the
  - off
  - setting

gets loaded in Ruby as

[true, "put", "it", true, "the", false, "setting"]

Basically the latest Perl and Python YAML modules I could find won't
quote some of the words that Ruby's YAML would. (Things like on, off,
yes, no, +) Perhaps they conform to a different version of the YAML
spec. I find the incompatibility a bit frustrating. I can and have
easily hacked the other modules to do the compatible thing but there
are political admin maintenance issues surrounding my patching all the
company computers with this.

I'll not comment right now on whether I think it was a good idea to
flag these items as special to begin with. I'd just like to know if
there's a simple module over-ride hack or module parameter setting that
I can set or clip into my scripts to deal with this problem.

Thanks.

mikshir wrote:

Problem: Ruby wants to auto-interpret boolean flagged words, [...]

Hi Mikshir,

Hadn't heard about those :-? Interesting.

I'll not comment right now on whether I think it was a good idea to
flag these items as special to begin with. I'd just like to know if
there's a simple module over-ride hack or module parameter setting that
I can set or clip into my scripts to deal with this problem.

Thanks.

This isn't anything official, so caution is recommended.

Successive changes to YAML.rb mean that I've had to stitch
two fixes. The relevant one should auto-select and it
seems to work for the following Rubys:

       1.8.0 (2003-08-30)
       1.8.2 (2004-12-25)
       1.8.3 (2005-07-22)
       1.8.4 (2005-12-01)
       1.9.0 (2005-10-31)

It re-categorises all bool types as normal strings and
doesn't touch any other types.

Note that, as well as inserting the module addition in your
scripts, you must use:
           YAML.parse(foo).transform
           ^^^^^^^^^^^^^^^^^^^^^^^^^
in place of YAML.load
http://yaml4r.sourceforge.net/doc/class/yaml_parser_parse_method.htm

<fix/test below> - email me if any problem(s).

daz

···

--

require 'yaml'

#----------------------------------------------------------------------
module YAML::Syck
  BOOL_RE = /bool#\w+\z/
  if defined?(Loader)
    class Loader
      alias_method :o_transfer, :transfer
      def transfer(type_id, val)
        (String === type_id) and type_id.gsub!(BOOL_RE, 'str')
        o_transfer(type_id, val)
      end
    end
  else
    class Node
      alias_method :o_transform, :transform
      def transform
        kind == :scalar and self.type_id = type_id.gsub(BOOL_RE, 'str')
        o_transform
      end
    end
  end
end
#----------------------------------------------------------------------

hobj = YAML.parse(<<EoS).transform
---
words:
  - yes
  - put
  - it
  - on
  - the
  - off
  - setting
EoS

if hobj['words'] == %w{ yes put it on the off setting }
  puts ' ***---> S U C C E S S <---***'
else
  p hobj
end

Well, I was looking at the YAML site, and apparently yes/no on/off are part of a draft specification, it looks like ruby is just ahead of the curve, so to speak.

http://yaml.org/type/bool.html

This of course, fails to help you.

The only solution I could come up with, in a word, sucks.

require 'yaml'

test = YAML.parse(FIle.read("test.yml")) # where test.yml contains that ---\n words: -yes -no etc.
test.children[0].value.each do |x|
       if x.type_id =~ /bool#(?:yes|no)\z/
             x.type_id = "tag:yaml.org,2002:str"
       end
end

results = test.transform #=> {"words"=>["yes", "put", "it", "on", "the", "off", "setting"]}

Of course this iteration will only work for that file, since I made some massive assumptions about the structure. Your other option is to do a search and replace on the yess, ons,offs, etc and quote them before parsing.

···

On Dec 9, 2005, at 3:57 PM, mikshir wrote:

I have YAML documents generated in Perl and Python that I process with
Ruby.

Problem: Ruby wants to auto-interpret boolean flagged words, example:

---
words:
  - yes
  - put
  - it
  - on
  - the
  - off
  - setting

gets loaded in Ruby as

[true, "put", "it", true, "the", false, "setting"]

Basically the latest Perl and Python YAML modules I could find won't
quote some of the words that Ruby's YAML would. (Things like on, off,
yes, no, +) Perhaps they conform to a different version of the YAML
spec. I find the incompatibility a bit frustrating. I can and have
easily hacked the other modules to do the compatible thing but there
are political admin maintenance issues surrounding my patching all the
company computers with this.

I'll not comment right now on whether I think it was a good idea to
flag these items as special to begin with. I'd just like to know if
there's a simple module over-ride hack or module parameter setting that
I can set or clip into my scripts to deal with this problem.

Thanks.

why not simply conform to the spec? :

     harp:~ > cat a.rb
     doc = <<-yaml

···

On Sat, 10 Dec 2005, mikshir wrote:

I have YAML documents generated in Perl and Python that I process with
Ruby.

Problem: Ruby wants to auto-interpret boolean flagged words, example:

---
words:
- yes
- put
- it
- on
- the
- off
- setting

gets loaded in Ruby as

[true, "put", "it", true, "the", false, "setting"]

Basically the latest Perl and Python YAML modules I could find won't
quote some of the words that Ruby's YAML would. (Things like on, off,
yes, no, +) Perhaps they conform to a different version of the YAML
spec. I find the incompatibility a bit frustrating. I can and have
easily hacked the other modules to do the compatible thing but there
are political admin maintenance issues surrounding my patching all the
company computers with this.

I'll not comment right now on whether I think it was a good idea to
flag these items as special to begin with. I'd just like to know if
there's a simple module over-ride hack or module parameter setting that
I can set or clip into my scripts to deal with this problem.

Thanks.

     ---
     "words":
      - "yes"
      - "put"
      - "it"
      - "on"
      - "the"
      - "off"
      - "setting"
     yaml

     doc = STDIN.read unless STDIN.tty?

     require "yaml" and y(YAML::load(doc))

     harp:~ > ruby a.rb
     words:
     - "yes"
     - put
     - it
     - "on"
     - the
     - "off"
     - setting

     harp:~ > ruby a.rb|ruby a.rb
     words:
     - "yes"
     - put
     - it
     - "on"
     - the
     - "off"
     - setting

as you can see these words need to be quoted to be handled correctly -
basically the error is both in the person(s) writing the docs and in the other
decoders. i'll bet a quoted "yes" or "no" works in perl and python decoders
too though.

regards.

-a
--

ara [dot] t [dot] howard [at] noaa [dot] gov
all happiness comes from the desire for others to be happy. all misery
comes from the desire for oneself to be happy.
-- bodhicaryavatara

===============================================================================

I wrote:

http://yaml4r.sourceforge.net/doc/class/yaml_parser_parse_method.htm

Huh? It changed in transit :confused: Should be:
  http://yaml4r.sourceforge.net/doc/class/yaml_parse_method.htm

I believe that the documents are being generated by Python and/or Perl and he is (for whatever reason) not allowed to mess with that part.

···

On Dec 10, 2005, at 2:16 AM, ara.t.howard@noaa.gov wrote:

why not simply conform to the spec? :

Thanks for the followups, gang.

Yes, I have my own modified version of PyYaml to make it conform to the
latest Ruby implemention. (I haven't done this for the Perl YAML
module yet.) The immediate issue is that for certain reasons I can't
update the modules system-wide and so I was looking for a quick and
easy user-script solution; like throwing in some re-definitions,
overrides, or extensions to force in some backwards compatability and
something I can throw in a library or with a switch.

Quoting the flagged words does the right thing (forces them to be
strings) and pre/post processing all Python and Perl generated Yaml
docs has been a stop-gap measure but is clumsy or not convenient.

Anyhow, thanks very much Daz. I will test out your method. /salute