json_or_py_encoding.py

#!/usr/bin/env python
# -*- coding: utf-8
from __future__ import (print_function, )

import six
from six import ensure_binary, text_type, ensure_text
import json


__all__ = ['json_str_or_bytes_repr_as_safe_text']


def json_str_or_bytes_repr_as_safe_text(v):
    # type: (unicode) -> unicode

    """Stackstorm action pack - python 2&3 notes.

    Either installed as python2 or python3 pack, if action result is a ``dict``, then all key value
    should of type str(raw bytes string as of python2, unicode string as of python3, but both case
    typed as str, in simple words, python3: str type, python2: str type).

    But awkward situation is string with CJK characters of action result dict values, when action
    was installed in python3 pack for some reason st2 engine didn't treat them as utf-8 but as ascii
    therefore did'nt get encoded to python3 str(unicode) , as a consequence, the action result dict
    just got transformed to string or such action is taken failed. This situation makes it really
    annoying implementing python 2&3 compatible action with CJK charsets.

    This function accept a unicode object representing repr(str object) or json string. and return
    it as a python2 str object or python3 object.

    _debug(u'你好'.encode('utf-8'))                                           # <class 'str'>你好
    _debug(json_str_or_bytes_repr_as_safe_text('你好'))                       # <class 'str'>你好
    _debug(json_str_or_bytes_repr_as_safe_text(u'你好'.encode('utf-8')))      # <class 'str'>你好
    _debug(json_str_or_bytes_repr_as_safe_text(json.dumps(u'你好')))          # <class 'str'>你好
    _debug(json_str_or_bytes_repr_as_safe_text(json.dumps(u'你好')[1:-1]))    # <class 'str'>你好
    """

    try:
        open('a', 'wb')
        bv = ensure_binary(v, 'utf-8')
        if not v:
            return ''
        if bv[:2] == b'b\\':
            eval('tv = {}'.format(v))
            return ensure_text(tv, 'utf-8')
        elif b'\\u' in bv:
            # is json

            # ensure wrapped in double quotes
            if (chr(bv[0]) if six.PY3 else bv[0]) != chr(34):
                bv = b'"' + bv + b'"'
            loaded_json = json.loads(bv)
            if type(loaded_json) != text_type:
                return v
            else:
                return loaded_json
        else:
            return ensure_text(v)
    except:
        return v


def _debug(x):
    print(type(x), x)


if __name__ == '__main__':
    _debug(u'你好'.encode('utf-8'))
    _debug(json_str_or_bytes_repr_as_safe_text('你好'))
    _debug(json_str_or_bytes_repr_as_safe_text(u'你好'.encode('utf-8')))
    _debug(json_str_or_bytes_repr_as_safe_text(json.dumps(u'你好')))
    _debug(json_str_or_bytes_repr_as_safe_text(json.dumps(u'你好')[1:-1]))