#!/usr/bin/env python3
"""Which Windows -> IANA mapping tables know Windows' three new Canadian zone IDs?

Windows' fix for the 1 November 2026 changes adds new zones instead of editing
"Mountain/Pacific/Central Standard Time". Software that maps the Windows ID to an
IANA name from an older table finds nothing for a machine switched to them.

Stdlib only; reads public URLs; writes nothing. Pinned URLs reproduce the frozen
result; "moving" rows show today's branch heads. Prints one row per source with
the count of each ID (ASCII or UTF-16LE, since ICU .res and .NET literals are UTF-16)
and the sha256 of the bytes read.

    python3 -I windows-ids-check.py
"""
import gzip
import hashlib
import io
import sys
import urllib.request
import zipfile

IDS = ("Alberta Standard Time", "British Columbia Standard Time", "Manitoba Standard Time")
RAW = "https://raw.githubusercontent.com"

SOURCES = [
    # (label, kind, url)
    ("CLDR main (commit c4bb2c199c, 2026-10-09)", "pinned",
     f"{RAW}/unicode-org/cldr/c4bb2c199c/common/supplemental/windowsZones.xml"),
    ("CLDR 49 beta4 (release-49-beta4)", "pinned",
     f"{RAW}/unicode-org/cldr/release-49-beta4/common/supplemental/windowsZones.xml"),
    ("ICU 78 (maint/maint-78)", "moving",
     f"{RAW}/unicode-org/icu/maint/maint-78/icu4c/source/data/misc/windowsZones.txt"),
    ("ICU 79 (maint/maint-79)", "moving",
     f"{RAW}/unicode-org/icu/maint/maint-79/icu4c/source/data/misc/windowsZones.txt"),
    ("ICU 79.1rc (release-79.1rc)", "pinned",
     f"{RAW}/unicode-org/icu/release-79.1rc/icu4c/source/data/misc/windowsZones.txt"),
    ("ICU tz update files 2026e (windowsZones.res)", "pinned",
     f"{RAW}/unicode-org/icu-data/bae5ca647fa41703393dbf00240ec1b807a9c353/tzdata/icunew/2026e/44/le/windowsZones.res"),
    ("tzlocal 5.4.4 (tag 42dfd05)", "pinned",
     f"{RAW}/regebro/tzlocal/42dfd05/tzlocal/windows_tz.py"),
    ("tzlocal master", "moving",
     f"{RAW}/regebro/tzlocal/master/tzlocal/windows_tz.py"),
    ("OpenJDK master tzmappings.override", "moving",
     f"{RAW}/openjdk/jdk/master/src/java.base/share/data/tzdata/tzmappings.override"),
    ("PostgreSQL master initdb win32 map", "moving",
     f"{RAW}/postgres/postgres/master/src/bin/initdb/findtimezone.c"),
    ("Go master abbreviation table", "moving",
     f"{RAW}/golang/go/master/src/time/zoneinfo_abbrs_windows.go"),
    ("TimeZoneConverter 8.0.0 (NuGet, net8.0 dll)", "pinned",
     "https://api.nuget.org/v3-flatcontainer/timezoneconverter/8.0.0/timezoneconverter.8.0.0.nupkg"),
    ("TimeZoneConverter 7.2.0 (NuGet, net8.0 dll)", "pinned",
     "https://api.nuget.org/v3-flatcontainer/timezoneconverter/7.2.0/timezoneconverter.7.2.0.nupkg"),
]


def fetch(url):
    req = urllib.request.Request(url, headers={"User-Agent": "windows-ids-check"})
    with urllib.request.urlopen(req, timeout=60) as r:
        return r.read()


def payload(url, data):
    if url.endswith(".nupkg"):
        z = zipfile.ZipFile(io.BytesIO(data))
        return z.read("lib/net8.0/TimeZoneConverter.dll")
    if data[:2] == b"\x1f\x8b":
        return gzip.decompress(data)
    return data


def count(data, s):
    return data.count(s.encode("ascii")) + data.count(s.encode("utf-16-le"))


def main():
    print("source | kind | " + " | ".join(IDS) + " | sha256(bytes read)")
    failed = 0
    for label, kind, url in SOURCES:
        try:
            raw = fetch(url)
        except Exception as e:  # report and keep going; a missing row stays UNKNOWN
            print(f"{label} | {kind} | UNKNOWN ({type(e).__name__}: {e}) | {url}")
            failed += 1
            continue
        data = payload(url, raw)
        counts = [count(data, s) for s in IDS]
        print(f"{label} | {kind} | " + " | ".join(str(c) for c in counts)
              + f" | {hashlib.sha256(raw).hexdigest()[:16]}")
    return 1 if failed else 0


if __name__ == "__main__":
    sys.exit(main())