phparray.py

"""
Quick POC for parsing a PHP array from python.
"""

from pyparsing import *

"""
array
    ( members )
    ()

members
    key => value
    members , key => value

key
    string | int

value
    string | number | array | null | true | false
"""

NULL = Keyword('NULL').setParseAction(replaceWith(None))
TRUE = Keyword('TRUE').setParseAction(replaceWith(True))
FALSE = Keyword('FALSE').setParseAction(replaceWith(False))

LPAREN = Suppress("array (")
RPAREN = Suppress(")")
ARROW = Suppress("=>")
COMMA = Suppress(",")

array = Forward()
value = Forward()
key = Forward()

string = sglQuotedString().setParseAction(removeQuotes)
number = pyparsing_common.number()

key << (number | string)
value << (string | number | Group(array) | NULL | FALSE | TRUE)

members = delimitedList(Group(key + ARROW + value))

array << Dict(LPAREN + Optional(members) + Optional(COMMA).suppress() + RPAREN)


def from_php(chars):
    """
    Parse a PHP array from a string created
    with `var_export`.
    """
    result = array.parseString(chars)
    # For some reason `asDict` coerces int keys to strings.
    return result.asDict(), result.asList()


if __name__ == "__main__":
    import pprint
    test_data = """
        array (
          'astring' => '74525',
          'ind_code' =>
           array (
                  0 => '74525',
                  'fpp' => 'IND_420'),
          'sector_code' => 423.423,
          1 => FALSE,
        )
        """
    pprint.pprint(from_php(test_data))