#!/usr/bin/env python3
"""Regenerate ui/frame_timezones.py, the time zone -> country table usage statistics use.

From the IANA tz database's zone.tab (each zone's country) and tzdata.zi (the old
names that link to them), as installed on macOS and most Linux systems. It prints the
old names left out because their zone is shared by several countries: check any new
one, and add it to OLD_SPELLINGS if it's just an old spelling of a zone.tab zone.

Run: python3 scripts/timezones.py [/usr/share/zoneinfo]
"""
import sys
from pathlib import Path

src = Path(sys.argv[1] if len(sys.argv) > 1 else '/usr/share/zoneinfo')
country = {}
for line in (src / 'zone.tab').read_text().splitlines():
    if line and not line.startswith('#'):
        cc, _, zone = line.split('\t')[:3]
        country[zone] = cc
links = {}
for line in (src / 'tzdata.zi').read_text().splitlines():
    if line.startswith('L '):
        _, target, alias = line.split()
        links[alias] = target
def resolve(z):
    for _ in range(5):
        z = links.get(z, z)
    return z


# An old name links to a zone with the same clocks, which can be in another country
# (Africa/Asmera, Eritrea, links to Africa/Nairobi), so it only gets a country when every
# zone.tab zone sharing that target is in one country.
countries_of = {}
for zone, cc in country.items():
    countries_of.setdefault(resolve(zone), set()).add(cc)
# Clock rules rather than places (EET links to Athens, WET to Lisbon): no country.
GENERIC = {'CET', 'EET', 'MET', 'WET', 'EST', 'MST', 'HST', 'CST6CDT', 'EST5EDT', 'MST7MDT', 'PST8PDT'}
# Places that have left zone.tab, so their target's group looks like another country's.
PLACES = {'Pacific/Johnston': 'UM'}
dropped = []
for alias in links:
    if alias in country or alias in GENERIC or alias.startswith('Etc/'):
        continue
    if alias in PLACES:
        country[alias] = PLACES[alias]
        continue
    ccs = countries_of.get(resolve(alias), set())
    if len(ccs) == 1:
        country[alias] = next(iter(ccs))
    else:
        dropped.append(alias)
# Old names for a place in a zone.tab zone whose target is shared, checked by hand.
OLD_SPELLINGS = {
    'Africa/Asmera': 'Africa/Asmara', 'Africa/Timbuktu': 'Africa/Bamako', 'America/Coral_Harbour': 'America/Atikokan',
    'America/Montreal': 'America/Toronto', 'America/Nipigon': 'America/Toronto', 'America/Thunder_Bay': 'America/Toronto',
    'Canada/Eastern': 'America/Toronto', 'America/Virgin': 'America/St_Thomas', 'Antarctica/South_Pole': 'Antarctica/McMurdo',
    'Asia/Rangoon': 'Asia/Yangon', 'Atlantic/Jan_Mayen': 'Arctic/Longyearbyen', 'Europe/Belfast': 'Europe/London',
    'GB': 'Europe/London', 'GB-Eire': 'Europe/London', 'Iceland': 'Atlantic/Reykjavik', 'NZ': 'Pacific/Auckland',
    'Pacific/Samoa': 'Pacific/Pago_Pago', 'US/Samoa': 'Pacific/Pago_Pago', 'Singapore': 'Asia/Singapore',
    'US/Arizona': 'America/Phoenix', 'Pacific/Truk': 'Pacific/Chuuk', 'Pacific/Ponape': 'Pacific/Pohnpei',
    'Pacific/Yap': 'Pacific/Chuuk'}
for alias, zone in OLD_SPELLINGS.items():
    if alias in links and zone in country:
        country[alias] = country[zone]
        dropped.remove(alias)
version = (src / '+VERSION').read_text().strip() if (src / '+VERSION').exists() else 'unknown'
rows = ''.join(f'    {z!r}: {c!r},\n' for z, c in sorted(country.items()))
out = Path(__file__).resolve().parent.parent / 'ui' / 'frame_timezones.py'
out.write_text(f'''"""Time zone name -> two-letter country, from the IANA tz database {version} (zone.tab, and
the old names in tzdata.zi that link to its zones). Generated by scripts/timezones.py.
Zones that aren't in one country, such as UTC, aren't here."""
COUNTRY = {{
{rows}}}
''')
print(f'{len(country)} zones, tz {version} -> {out}')
print('no country (shared by several):', ' '.join(sorted(dropped)) or 'none')