Python YAML to JSON to YAML

Your file is losing its formatting because the original dump routine by default writes all leaf nodes in YAML flow-style, whereas your input is block style all the way.

You are also losing the order of the keys, which is first because the JSON parser uses dict, and second because dump sorts the output.

If you look at your intermediate JSON you already see that the key order is gone at that point. To preserve that, use the new API to load your YAML and have a special JSON encoder as a replacement for dump that can handle the subclasses of Mapping in which the YAML is loaded similar to this example from the standard Python doc.

Assuming your YAML is stored in input.yaml:

import sys
import json
from collections.abc import Mapping, Sequence
from collections import OrderedDict
import ruamel.yaml

# if you instantiate a YAML instance as yaml, you have to explicitly import the error
from ruamel.yaml.error import YAMLError


yaml = ruamel.yaml.YAML()  # this uses the new API
# if you have standard indentation, no need to use the following
yaml.indent(sequence=4, offset=2)

input_file = 'input.yaml'
intermediate_file = 'intermediate.json'
output_file = 'output.yaml'


class OrderlyJSONEncoder(json.JSONEncoder):
    def default(self, o):
        if isinstance(o, Mapping):
            return OrderedDict(o)
        elif isinstance(o, Sequence):
            return list(o)
        return json.JSONEncoder.default(self, o)


def yaml_2_json(in_file, out_file):
    with open(in_file, 'r') as stream:
        try:
            datamap = yaml.load(stream)
            with open(out_file, 'w') as output:
                output.write(OrderlyJSONEncoder(indent=2).encode(datamap))
        except YAMLError as exc:
            print(exc)
            return False
    return True


yaml_2_json(input_file, intermediate_file)
with open(intermediate_file) as fp:
    sys.stdout.write(fp.read())

which gives:

{
  "inputs": {
    "webTierCpu": {
      "type": "integer",
      "minimum": 2,
      "default": 2,
      "maximum": 5,
      "title": "Web Server CPU Count",
      "description": "The number of CPUs for the Web nodes"
    }
  }
}

You see that your JSON has the appropriate key order, which we also need to preserve on loading. You can do that without subclassing anything, by specifying the loading of JSON objects into the subclass of Mapping, that the YAML parser is using internally, by providingobject_pairs_hook.

from ruamel.yaml.comments import CommentedMap


def json_2_yaml(in_file, out_file):
    with open(in_file, 'r') as stream:
        try:
            datamap = json.load(stream, object_pairs_hook=CommentedMap)
            # if you need to "restore" literal style scalars, etc.
            # walk_tree(datamap)
            with open(out_file, 'w') as output:
                yaml.dump(datamap, output)
        except yaml.YAMLError as exc:
            print(exc)
            return False
    return True


json_2_yaml(intermediate_file, output_file)
with open(output_file) as fp:
    sys.stdout.write(fp.read())

Which outputs:

inputs:
  webTierCpu:
    type: integer
    minimum: 2
    default: 2
    maximum: 5
    title: Web Server CPU Count
    description: The number of CPUs for the Web nodes

And I hope that that is similar enough to your original input to be acceptable.

Notes:

• When using the new API I tend to use yaml as the name of the instance of ruamel.yaml.YAML(), instead of from ruamel import yaml. That however masks the use of yaml.YAMLError because the error class is not an attribute of YAML()

• If you are developing this kind of stuff, I can recommend removing at least the user input from the actual functionality. It should be trivial to write your parseyaml and parsejson to call yaml_2_json resp. json_2_yaml.

• Any comments in your original YAML file will be lost, although ruamel.yaml can load them. JSON originally did allow comments, but it is not in the specification and no parsers that I know can output comments.

---

Since your real file has literal block scalars you have to use some magic to get those back.

Include the following functions that walk a tree, recursing into dict values and list elements and converting any line with an embedded newline to a type that gets output to YAML as a literal blocks style scalar in place (hence no return value):

from ruamel.yaml.scalarstring import PreservedScalarString, SingleQuotedScalarString
from ruamel.yaml.compat import string_types, MutableMapping, MutableSequence

def preserve_literal(s):
    return PreservedScalarString(s.replace('\r\n', '\n').replace('\r', '\n'))

def walk_tree(base):
    if isinstance(base, MutableMapping):
        for k in base:
            v = base[k]  # type: Text
            if isinstance(v, string_types):
                if '\n' in v:
                    base[k] = preserve_literal(v)
                elif '${' in v or ':' in v:
                    base[k] = SingleQuotedScalarString(v)
            else:
                walk_tree(v)
    elif isinstance(base, MutableSequence):
        for idx, elem in enumerate(base):
            if isinstance(elem, string_types):
                if '\n' in elem:
                    base[idx] = preserve_literal(elem)
                elif '${' in elem or ':' in elem:
                    base[idx] = SingleQuotedScalarString(elem)
            else:
                walk_tree(elem)

And then do

    walk_tree(datamap)

after you load the data from JSON.

With all of the above you should have only one line that differs in your Wordpress.yaml file.

function yaml_validate {
  python -c 'import sys, yaml, json; yaml.safe_load(sys.stdin.read())'
}

function yaml2json {
  python -c 'import sys, yaml, json; print(json.dumps(yaml.safe_load(sys.stdin.read())))'
}

function yaml2json_pretty {
  python -c 'import sys, yaml, json; print(json.dumps(yaml.safe_load(sys.stdin.read()), indent=2, sort_keys=False))'
}

function json_validate {
  python -c 'import sys, yaml, json; json.loads(sys.stdin.read())'
}

function json2yaml {
  python -c 'import sys, yaml, json; print(yaml.dump(json.loads(sys.stdin.read())))'
}

More useful Bash tricks at http://github.com/frgomes/bash-scripts