Python split string without splitting escaped character

The edited version of Henry's answer with Python3 compatibility, tests and fix some issues:

def split_unescape(s, delim, escape='\\', unescape=True):
    """
    >>> split_unescape('foo,bar', ',')
    ['foo', 'bar']
    >>> split_unescape('foo$,bar', ',', '$')
    ['foo,bar']
    >>> split_unescape('foo$$,bar', ',', '$', unescape=True)
    ['foo$', 'bar']
    >>> split_unescape('foo$$,bar', ',', '$', unescape=False)
    ['foo$$', 'bar']
    >>> split_unescape('foo$', ',', '$', unescape=True)
    ['foo$']
    """
    ret = []
    current = []
    itr = iter(s)
    for ch in itr:
        if ch == escape:
            try:
                # skip the next character; it has been escaped!
                if not unescape:
                    current.append(escape)
                current.append(next(itr))
            except StopIteration:
                if unescape:
                    current.append(escape)
        elif ch == delim:
            # split! (add current to the list and reset it)
            ret.append(''.join(current))
            current = []
        else:
            current.append(ch)
    ret.append(''.join(current))
    return ret

As Ignacio says, yes, but not trivially in one go. The issue is that you need lookback to determine if you're at an escaped delimiter or not, and the basic string.split doesn't provide that functionality.

If this isn't inside a tight loop so performance isn't a significant issue, you can do it by first splitting on the escaped delimiters, then performing the split, and then merging. Ugly demo code follows:

# Bear in mind this is not rigorously tested!
def escaped_split(s, delim):
    # split by escaped, then by not-escaped
    escaped_delim = '\\'+delim
    sections = [p.split(delim) for p in s.split(escaped_delim)] 
    ret = []
    prev = None
    for parts in sections: # for each list of "real" splits
        if prev is None:
            if len(parts) > 1:
                # Add first item, unless it's also the last in its section
                ret.append(parts[0])
        else:
            # Add the previous last item joined to the first item
            ret.append(escaped_delim.join([prev, parts[0]]))
        for part in parts[1:-1]:
            # Add all the items in the middle
            ret.append(part)
        prev = parts[-1]
    return ret

s = r'http\://www.example.url:ftp\://www.example.url'
print (escaped_split(s, ':')) 
# >>> ['http\\://www.example.url', 'ftp\\://www.example.url']

Alternately, it might be easier to follow the logic if you just split the string by hand.

def escaped_split(s, delim):
    ret = []
    current = []
    itr = iter(s)
    for ch in itr:
        if ch == '\\':
            try:
                # skip the next character; it has been escaped!
                current.append('\\')
                current.append(next(itr))
            except StopIteration:
                pass
        elif ch == delim:
            # split! (add current to the list and reset it)
            ret.append(''.join(current))
            current = []
        else:
            current.append(ch)
    ret.append(''.join(current))
    return ret

Note that this second version behaves slightly differently when it encounters double-escapes followed by a delimiter: this function allows escaped escape characters, so that escaped_split(r'a\\:b', ':') returns ['a\\\\', 'b'], because the first \ escapes the second one, leaving the : to be interpreted as a real delimiter. So that's something to watch out for.

There is a much easier way using a regex with a negative lookbehind assertion:

re.split(r'(?<!\\):', str)