OpeningHoursBot.py

#!/usr/bin/env python
# encoding: utf-8
# Helper script to fix common mistakes in opening_hours like values in OSM
# automatically.
# For discussion see:
# http://permalink.gmane.org/gmane.comp.gis.openstreetmap.tagging/16810

# modules {{{
import logging
import sys
import re
import xml.etree.ElementTree as ET
sys.path.append('pyopening_hours/')
from pyopening_hours import OpeningHours, ParseException
# }}}


class OpeningHoursBot:
    def fix_xml_object(self, osm_xml_root):
        """Fix the given OSM XML object using the defended rule set."""
        osm_xml = osm_xml_root.getroot()

        logging.info(
            u"Parsing OSM file version '{}' generated by '{}' …".format(
                osm_xml.attrib['version'],
                osm_xml.attrib['generator']
            )
        )
        for node in osm_xml_root.findall('node'):
            self._for_object(node)
        for way in osm_xml_root.findall('way'):
            self._for_object(way)

    def _fix_ruleset_1_time_error(self, wrong_val):
        """ '0930-0630' -> '09:30-06:30'

        Use with caution. Does conflict with goal b) (as defined here:
        http://permalink.gmane.org/gmane.comp.gis.openstreetmap.tagging/16810)

        Should not be enabled because the time could be specified in 12 hours
        format or in 24 hours format (which is the recommended format:
        http://wiki.openstreetmap.org/wiki/Key:opening_hours:specification).
        """
        regex = re.compile(r'\A(?P<day_list>(?:(?:Mo|Tu|We|Th|Fr|Sa|Su)-?){2}\s*|)(?P<start_hour>[0-1][0-9]|2[0-4])(?P<start_min>[:1-5][0-9]|0[0-9])\s*(?P<sep>-)\s*(?P<end_hour>[0-1][0-9]|2[0-4])(?P<end_min>[:1-5][0-9]|0[0-9])\Z')

        re_object = re.search(regex, wrong_val)
        if not re_object:
            return wrong_val
        return '{}{}:{}{}{}:{}'.format(
            re_object.group('day_list'),
            re_object.group('start_hour'), re_object.group('start_min'),
            re_object.group('sep'),
            re_object.group('end_hour'), re_object.group('end_min')
        )

    def _fix_ruleset_2(self, wrong_val):
        """
        Changes: 'nach vereinbarung' -> '"nach Vereinbarung"'
        (most often used value for this).
        """
        regex = re.compile(r'\A(?P<start_word>(?:Termin|nur) |)nach vereinbarung\Z', flags=re.IGNORECASE)

        re_object = re.search(regex, wrong_val)
        if not re_object:
            return wrong_val
        return '"{}nach Vereinbarung"'.format(
            re_object.group('start_word') if re_object.group('start_word') else ''
        )

    fixing_functions = [
        #  _fix_ruleset_1_time_error,
        _fix_ruleset_2,
    ]

    # helper functions {{{
    def _is_opening_hours_ok(self, value):
        """Returns True if there are no warnings and False if OpeningHours raises
        an exception. It will return None in all other cases."""
        error = False
        try:
            oh = OpeningHours(value)
        except ParseException:
            error = True

        if not error and not oh.getWarnings():
            if value != oh.prettifyValue():
                logging.error(
                    "OpeningHoursBot did generate an opening_hours value ('{}') which does not match the prettified value.".format(value)
                )
            return True
        elif error:
            return False
        else:
            return None
    # }}}

    # do the work {{{
    def _fix_opening_hours(self, tag):
        """Fix value of the tag with functions in fixing_functions list.

        :param tag: XML tag node.
        :returns: Boolean which is True if fixing was successful.
        """
        value = tag.attrib['v']
        if self._is_opening_hours_ok(value) is False:
            correct_value = value
            for fix_oh_val_func in self.fixing_functions:
                correct_value = fix_oh_val_func(self, value)
                if value != correct_value:
                    break
            if value != correct_value:
                if self._is_opening_hours_ok(correct_value):
                    logging.debug(
                        u"Fixed value '{}' -> '{}'".format(
                            value,
                            correct_value
                        )
                    )
                    tag.set('v', correct_value)
                    return True
                else:
                    logging.critical(
                        "OpeningHoursBot did generate a wrong opening_hours value ('{}' -> '{}')".format(
                            value,
                            correct_value
                        )
                    )
        return False

    def _for_object(self, node):
        """Called for each object (node, way)."""
        fixing_successful = None
        for tag in node.findall('tag'):
            if tag.attrib['k'] == 'opening_hours':
                fixing_successful = self._fix_opening_hours(tag)
        if fixing_successful:
            node.set('action', 'modify')
    # }}}

def main(osm_file):

    logging.basicConfig(
        format='%(levelname)s: %(message)s',
        level=logging.DEBUG,
        # level=logging.INFO,
    )

    if osm_file[-4:] == '.osm':
        osm_file_basename = osm_file[:-4]

    osm_xml_root = ET.parse(osm_file)

    oh_bot = OpeningHoursBot()

    oh_bot.fix_xml_object(osm_xml_root)

    osm_xml_root.write('{}-fixed.osm'.format(osm_file_basename))

if __name__ == '__main__':
    if len(sys.argv) > 1:
        osm_file = sys.argv[1]
    else:
        logging.error(
            'Not enough parameters.'
            + ' 1. File path OSM file.'
        )
        sys.exit(1)
    main(osm_file)