Helper function for cleaning up user input geographic coordinates. Call get_coords(crappy_x, crappy_y) to get cleaned coordinates. No guarantees, check your results, etc.

coord_cleaner.py

def clean_point(point):
    # junk
    point = point.strip()
    junk = [u'\ufeff', ' ', u'\xa0', u'\u2013']  # [,  ,  , –]
    for symbol in junk:
        point = point.replace(symbol, '')

    point = point.lstrip('+')
    point = point.lstrip("'")  # leading quote chars to fool excel
    point = point.lstrip(',')
    point = point.rstrip(',')
    point = point.rstrip('.')
    point = point.rstrip('N')
    point = point.lstrip('N')
    point = point.rstrip('E')
    point = point.lstrip('E')

    # e.g 1.23°S -> -1.23°
    sw_patterns = [re.compile(u'[SW](.*)'), re.compile(u'(.*)[SW]')]
    for p in sw_patterns:
        match = p.match(point)
        if match:
            coord = match.groups()[0]
            if not coord.startswith('-'):
                point = u'-%s' % coord

    # degrees
    degrees = [u'\xb0', u'\xba']  # [°, º]
    for symbol in degrees:
        # e.g. 50.12345°N
        p = re.compile(u'(-?\d+(\.\d+)?)%s$' % symbol)
        match = p.match(point)
        if match:
            # If all we have is a number with a trailing degree symbol then we
            # can just strip it because we're already decimal
            point = match.groups()[0]

    for symbol in degrees:
        point = point.replace(symbol, '.')

    # minutes
    minutes = [u'\u2019', u'\u2032', u'\xb4', "'", ':']  # [’, ′, ´, ', :]
    for symbol in minutes:
        # if we got degrees and decimal minutes (e.g "65° 42.00714'", which by
        # now would be "65.42.00714'") we don't want need to replace the minute
        # symbol with a point because it's already been done.
        p = re.compile(u'(-?\d+\.\d+(\.\d+)?)%s$' % symbol)
        match = p.match(point)
        if match:
            point = match.groups()[0]
            break

    for symbol in minutes:
        point = point.replace(symbol, '.')

    # seconds
    point = point.rstrip(u'\xa8')    # ¨
    point = point.rstrip(u'\u2033')  # ″
    point = point.rstrip(u'\u201d')  # ”
    point = point.rstrip('"')
    return point

    
def dms_to_dec(point):
    # split on points and multiply by dms (then sum)
    point_array = point.split('.', 2)
    total = 0
    multiplications = [1.0, 60.0, 3600.0]

    for i, value in enumerate(point_array):
        total += float(value) / multiplications[i]

    return total


def get_coords(lat, lon):
    def _get_coord(point):
        point = point.strip()
        if point in ['TBC', 'N/A', '??', '']:
            return None

        point = clean_point(point)
        try:
            point = float(point)
        except:
            point = dms_to_dec(point)

        return point

    return _get_coord(lat), _get_coord(lon)