Invalid timezones in latest data about opentraveldata

[da115115]

Thanks @alexprengere!

Yes, that happens when the POR are unknown from both OpenTravelData (opentraveldata/optd_por_best_known_so_far.csv CSV file) and Geonames. A time-zone code is then assigned based only on the country (as seen in the opentraveldata/optd_tz_light.csv CSV file) by the tools/extract_non_geonames_por.awk AWK script.

As some countries span multiple time-zones, that process may be refined thanks to the opentraveldata/optd_por_tz.csv CSV file, as seen in the tools/extract_non_geonames_por.awk AWK script again.

But, at that stage, rather than adding an entry to the opentraveldata/optd_por_tz.csv CSV file, it is much better to add the POR in Geonames and in OpenTravelData (opentraveldata/optd_por_best_known_so_far.csv CSV file)... which I have done for both the POR you mentioned above :)
So, the issue should have been fixed.

Could you please give the script which you have run in order to find out that issue?

Also, please do not hesitate to contribute to the Service Delivery Quality (SDQ) project, which is a collection of QA scripts in Python for monitoring the quality of the OpenTravelData data.

[alexprengere]

I just checked the latest file, the issue is fixed indeed.

The code I used is part of the NeoBase test suite. To create a Neobase instance from a specific optd file, you can either set the environment variable OPTD_POR_FILE, or use NeoBase(open(filename)).

Here is an extraction of those tests in an independant script that checks both timezones and geocodes.

import pytz

def test_timezones(base):
    for por in base:
        tz = base.get(por, 'timezone')
        try:
            pytz.timezone(tz)
        except pytz.exceptions.UnknownTimeZoneError:
            print('Unknown timezone for {0} ({1})'.format(por, tz))


def test_geocodes(base):
    for por in base:
        if base.get_location(por) is None:
            print('Missing geocode for {0}'.format(por))


if __name__ == '__main__':
    from neobase import NeoBase
    test_timezones(NeoBase())
    test_geocodes(NeoBase())

You need to install the latest pytz, otherwise you might get false positives. Then run the tests with:

OPTD_POR_FILE=./optd_por_public.csv python test.py