url_utils.py

"""
Contains common utilities for URL escaping.
"""
import re


def quote_slashes(text):
    """
    Quote '/' characters so that they aren't visible to
    django's url quoting, unquoting, or url regex matching.

    Escapes '/'' to the sequence ';_', and ';' to the sequence
    ';;'. By making the escape sequence fixed length, and escaping
    identifier character ';', we are able to reverse the escaping.
    """
    return re.sub(ur'[;/]', _quote_slashes, text)


def unquote_slashes(text):
    """
    Unquote slashes quoted by `quote_slashes`
    """
    return re.sub(r'(;;|;_)', _unquote_slashes, text)


def _quote_slashes(match):
    """
    Helper function for `quote_slashes`
    """
    matched = match.group(0)
    # We have to escape ';', because that is our
    # escape sequence identifier (otherwise, the escaping)
    # couldn't distinguish between us adding ';_' to the string
    # and ';_' appearing naturally in the string
    if matched == ';':
        return ';;'
    elif matched == '/':
        return ';_'
    else:
        return matched


def _unquote_slashes(match):
    """
    Helper function for `unquote_slashes`
    """
    matched = match.group(0)
    if matched == ';;':
        return ';'
    elif matched == ';_':
        return '/'
    else:
        return matched