Using redbaron to find all open()s in text mode so we can review them for python 3

find_text_opens

from redbaron import RedBaron
import os
import sys
import io


def is_bad_open(node):
    # must be function invocation (not method invocation)
    if node.parent.type != 'atomtrailers':
        return False

    if node.index_on_parent != 0:
        return False

    # must be called()
    if node.next is None or node.next.type != 'call':
        return False

    call = node.next
    arglist = call.value

    # if it only has one argument, its text
    if len(arglist) <= 1:
        return True

    def is_binary_mode(node):
        return node.value.type == 'string' and "b" in str(node.value)

    # second argument is not a kvarg, must be mode
    if arglist[1].target is None:        
        # we can ignore it if its a string literal containing 'b'
        return not is_binary_mode(arglist[1])
    
    # second argument is a kvarg, so try to find mode in
    # any remaining args
    for arg in arglist[1:]:
        if arg.target.value == "mode":
            return not is_binary_mode(arg)

    return True


def process(src):
    """
    >>> process('open(x)')
    '1:open(x)\\n'
    >>> process('open(x, "wb")')
    ''
    >>> process('open(x, "w")')
    '1:open(x, "w")\\n'
    >>> process('open(x, something=blah)')
    '1:open(x, something=blah)\\n'
    >>> process('open(x, something=blah, mode="r")')
    '1:open(x, something=blah, mode="r")\\n'
    """
    output = io.BytesIO()
    for sort in RedBaron(src).find_all('name', value='open'):
        if is_bad_open(sort):
            output.write("%d:%s\n" %
                (sort.absolute_bounding_box.top_left.line, sort.parent))

    return output.getvalue()


if __name__ == "__main__":
    if len(sys.argv) > 1:
        for f in sys.argv[1:]:
            s = process(open(f).read())
            if s:
                print f
                print s
    else:
        import doctest
        doctest.testmod()