#!/usr/bin/env python
# -*- coding: ascii -*-
"""
Split a stream of JSON-objects/arrays into separate strings.

:Version:   2008-09-20-alpha
:Status:    experimental

:Example:
    TODO

:SeeAlso:   JSON specification
:Warning:
    .. Warning::
        This is **experimental** code!
:Bug:

:Author:    Roland Koebler (r.koebler@yahoo.de)
:Copyright: 2008 by Roland Koebler (r.koebler@yahoo.de)
:License:   see __license__
:Changelog:
        - 2008-09-19:     created

TODO:
    - change to "yield"
    - test
    - enhance
"""

__version__ = "2008-09-20-alpha"
__author__   = "Roland Koebler <r.koebler@yahoo.de>"
__license__  = """Copyright (c) 2008 by Roland Koebler (r.koebler(at)yahoo.de)"""

#=========================================

def json_split(s):
    """Split a stream of JSON-objects/arrays into separate strings.

    Every resulting string then contains exactly 1 JSON-array/object.

    This is especially useful if you send JSON-objects/arrays over
    a streaming connection, and want to separate the JSON-objects/arrays
    before deserialization.

    :Parameters: s: (unicode-) JSON-string
    :Returns:    (LIST_OF_JSON_STRINGS, REMAINING_STRING)
    :Note:       no error-handling is included
    :Example:

        verbose example::

            >>> import simplejson
            >>> test = {u'a': u'b', u'1': 2, u'c': {u'1': [1, 2], u'3': [{u'd': [u'}']}], u'2': {u'3': 4}}, u'xy': u'x ] } " [ { y'}
            >>> json = simplejson.dumps(test)
            >>> stream = json*5
            >>> print stream[0:120]   #show first 120 characters of stream
            {"a": "b", "1": 2, "c": {"1": [1, 2], "3": [{"d": ["}"]}], "2": {"3": 4}}, "xy": "x ] } \\" [ { y"}{"a": "b", "1": 2, "c"
            >>> list = json_split(stream)
            >>> result = simplejson.loads(list[0][0])
            >>> result == test
            True

        the same in much shorter notation::

            >>> import simplejson
            >>> test = {u'a': u'b', u'1': 2, u'c': {u'1': [1, 2], u'3': [{u'd': [u'}']}], u'2': {u'3': 4}}, u'xy': u'x ] } " [ { y'}
            >>> result = simplejson.loads(json_split(simplejson.dumps(test)*5)[0][0])
            >>> result == test
            True

    :Author:     Roland Koebler (r.koebler@yahoo.de)
    :Version:    2008-09-20-pre
    """
    json_strings = []

    state = 0   #0=search start, 1=array,  11=string inside array 
                #                2=object, 12=string inside object
    b = 0       #begin of string-part
    i = 0       #current position in string
    cnt = 0
    while True:
        if i >= len(s):
            break

        if 0 == state:          #find start of json-array or object
            #TODO: what to do with s[b:i]
            #TODO: skip chars after \\ ?
            if s[i] == '[':
                state = 1
                b = i
                cnt = 1
            elif s[i] == '{':
                state = 2
                b = i
                cnt = 1
        elif state > 10:        #inside string
            if s[i] == '\\':    # skip char after \
                i += 1
            elif s[i] == '"':   # end of string
                state -= 10
        elif 1 == state:        #inside array
            if s[i] == '"':
                state += 10
            elif s[i] == '[':
                cnt += 1
            elif s[i] == ']':
                cnt -= 1
                if cnt == 0:
                    json_strings.append(s[b:i+1])
                    state = 0
                    b = i+1
        elif 2 == state:        #inside object
            if s[i] == '"':
                state += 10
            elif s[i] == '{':
                cnt += 1
            elif s[i] == '}':
                cnt -= 1
                if cnt == 0:
                    json_strings.append(s[b:i+1])
                    state = 0
                    b = i+1
        i += 1

    return (json_strings, s[b:])

#=========================================
# doctest

if __name__ == "__main__":
    import doctest
    doctest.testmod()

#=========================================