#!/usr/bin/env python3
from __future__ import annotations

import itertools

import requests
from packaging.metadata import Metadata
from packaging.utils import (
    InvalidWheelFilename,
    canonicalize_name,
    parse_wheel_filename,
)

PACKAGES = [
    "boto3",
    "botocore",
    "urllib3",
    "requests",
    "wheel",
    "certifi",
    "typing-extensions",
    "charset-normalizer",
    "idna",
    "setuptools",
    "pip",
    "python-dateutil",
    "s3transfer",
    "aiobotocore",
    "packaging",
    "six",
    "pyyaml",
    "s3fs",
    "fsspec",
    "numpy",
    "cryptography",
    "google-api-core",
    "grpcio-status",
    "importlib-metadata",
    "cffi",
    "zipp",
    "pycparser",
    "pandas",
    "attrs",
    "protobuf",
    "rsa",
    "pyasn1",
    "jmespath",
    "click",
    "pytz",
    "awscli",
    "pydantic",
    "colorama",
    "markupsafe",
    "jinja2",
    "platformdirs",
    "pyjwt",
    "tomli",
    "googleapis-common-protos",
    "google-auth",
    "cachetools",
    "filelock",
    "wrapt",
    "jsonschema",
    "virtualenv",
    "pluggy",
    "pyparsing",
    "werkzeug",
    "pyarrow",
    "docutils",
    "sqlalchemy",
    "pytest",
    "exceptiongroup",
    "flask",
    "aiohttp",
    "requests-oauthlib",
    "pyasn1-modules",
    "oauthlib",
    "isodate",
    "multidict",
    "scipy",
    "psutil",
    "yarl",
    "async-timeout",
    "soupsieve",
    "iniconfig",
    "frozenlist",
    "beautifulsoup4",
    "aiosignal",
    "grpcio",
    "greenlet",
    "tqdm",
    "pillow",
    "pygments",
    "decorator",
    "importlib-resources",
    "lxml",
    "pyopenssl",
    "requests-toolbelt",
    "openpyxl",
    "et-xmlfile",
    "azure-core",
    "tzdata",
    "asn1crypto",
    "distlib",
    "pydantic-core",
    "coverage",
    "tomlkit",
    "sniffio",
    "more-itertools",
    "pynacl",
    "pexpect",
    "h11",
    "ptyprocess",
    "google-cloud-storage",
]


def check_consistency(package: str) -> None:
    name = canonicalize_name(package)
    metadata_info: dict[str, Metadata | None] = {}
    headers = {"Accept": "application/vnd.pypi.simple.v1+json"}
    rsp = requests.get(f"https://pypi.org/simple/{name}/", headers=headers, timeout=5)

    if not rsp.ok:
        print(f"Failed to get details for {name}: {rsp.text}")
    rsp.raise_for_status()

    info = rsp.json()
    version = info["versions"][-1]

    # Gather up the different metadata hashes referenced by the wheels.
    files = []
    for file in info["files"]:
        filename = file["filename"]
        if not filename.endswith(".whl"):
            continue

        try:
            parsed = parse_wheel_filename(filename)
        except InvalidWheelFilename:
            print(f"Failed to parse wheel filename {filename}")
            continue

        if str(parsed[1]) != version:
            continue

        files.append(file)

        metadata_hash = file["core-metadata"]["sha256"]
        metadata_info[metadata_hash] = None

    # If all metadata hashes are the same, we're done.
    count = len(metadata_info)
    if count < 2:
        print(f"Only {count} metadata hash found for {name} {version}")
        return

    print(f"{count} metadata hashes found for {name} {version}")

    # Download the different metadata files, one for each of the hashes.
    for file in files:
        metadata_hash = file["core-metadata"]["sha256"]
        if metadata_info[metadata_hash] is not None:
            continue

        url = file["url"] + ".metadata"
        rsp = requests.get(url, timeout=5)

        if not rsp.ok:
            print(f"Failed to get metadata from {url}: {rsp.text}")
        rsp.raise_for_status()

        metadata = Metadata.from_email(rsp.content, validate=False)
        metadata_info[metadata_hash] = metadata

    # Metadata class is a bit awkward to work with, make manual checks on the
    # interesting fields.
    inconsistencies = False
    for m1, m2 in itertools.combinations(metadata_info.values(), 2):
        assert m1 is not None
        assert m2 is not None

        if m1.requires_python != m2.requires_python:
            print("Inconsistent requires_python")
            inconsistencies = True

        m1reqs = (
            None
            if m1.requires_dist is None
            else sorted(str(r) for r in m1.requires_dist)
        )
        m2reqs = (
            None
            if m2.requires_dist is None
            else sorted(str(r) for r in m2.requires_dist)
        )
        if m1reqs != m2reqs:
            print("Inconsistent requires_dist")
            inconsistencies = True

    if not inconsistencies:
        print("No interesting inconsistencies found")


if __name__ == "__main__":
    for package in PACKAGES:
        check_consistency(package)