Use SimpleNamespace for typed records

[pinch] / pinch.py

import filecmp
import functools
import hashlib
import operator
import os
import os.path
import shutil
import tempfile
import types
import urllib.parse
import urllib.request
import xml.dom.minidom

from typing import (
    Dict,
    Iterable,
    List,
    Sequence,
    Tuple,
)


class InfoTableEntry(types.SimpleNamespace):
    content: bytes
    digest: str
    file: str
    size: int
    url: str


class Info(types.SimpleNamespace):
    channel_html: bytes
    forwarded_url: str
    git_revision: str
    table: Dict[str, InfoTableEntry]
    url: str


class VerificationError(Exception):
    pass


class Verification:

    def __init__(self) -> None:
        self.line_length = 0

    def status(self, s: str) -> None:
        print(s, end=' ', flush=True)
        self.line_length += 1 + len(s)  # Unicode??

    @staticmethod
    def _color(s: str, c: int) -> str:
        return '\033[%2dm%s\033[00m' % (c, s)

    def result(self, r: bool) -> None:
        message, color = {True: ('OK ', 92), False: ('FAIL', 91)}[r]
        length = len(message)
        cols = shutil.get_terminal_size().columns
        pad = (cols - (self.line_length + length)) % cols
        print(' ' * pad + self._color(message, color))
        self.line_length = 0
        if not r:
            raise VerificationError()

    def check(self, s: str, r: bool) -> None:
        self.status(s)
        self.result(r)

    def ok(self) -> None:
        self.result(True)


def compare(a: str,
            b: str) -> Tuple[Sequence[str],
                             Sequence[str],
                             Sequence[str]]:

    def throw(error: OSError) -> None:
        raise error

    def join(x: str, y: str) -> str:
        return y if x == '.' else os.path.join(x, y)

    def recursive_files(d: str) -> Iterable[str]:
        all_files: List[str] = []
        for path, dirs, files in os.walk(d, onerror=throw):
            rel = os.path.relpath(path, start=d)
            all_files.extend(join(rel, f) for f in files)
            for dir_or_link in dirs:
                if os.path.islink(join(path, dir_or_link)):
                    all_files.append(join(rel, dir_or_link))
        return all_files

    def exclude_dot_git(files: Iterable[str]) -> Iterable[str]:
        return (f for f in files if not f.startswith('.git/'))

    files = functools.reduce(
        operator.or_, (set(
            exclude_dot_git(
                recursive_files(x))) for x in [a, b]))
    return filecmp.cmpfiles(a, b, files, shallow=False)


def fetch(v: Verification, channel_url: str) -> Info:
    info = Info()
    info.url = channel_url
    v.status('Fetching channel')
    request = urllib.request.urlopen(
        'https://channels.nixos.org/nixos-20.03', timeout=10)
    info.channel_html = request.read()
    info.forwarded_url = request.geturl()
    v.result(request.status == 200)
    v.check('Got forwarded', info.url != info.forwarded_url)
    return info


def parse(v: Verification, info: Info) -> None:
    v.status('Parsing channel description as XML')
    d = xml.dom.minidom.parseString(info.channel_html)
    v.ok()

    v.status('Extracting git commit')
    git_commit_node = d.getElementsByTagName('tt')[0]
    info.git_commit = git_commit_node.firstChild.nodeValue
    v.ok()
    v.status('Verifying git commit label')
    v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ')

    v.status('Parsing table')
    info.table = {}
    for row in d.getElementsByTagName('tr')[1:]:
        name = row.childNodes[0].firstChild.firstChild.nodeValue
        url = row.childNodes[0].firstChild.getAttribute('href')
        size = int(row.childNodes[1].firstChild.nodeValue)
        digest = row.childNodes[2].firstChild.firstChild.nodeValue
        info.table[name] = InfoTableEntry(url=url, digest=digest, size=size)
    v.ok()


def fetch_resources(v: Verification, info: Info) -> None:

    for resource in ['git-revision', 'nixexprs.tar.xz']:
        fields = info.table[resource]
        v.status('Fetching resource "%s"' % resource)
        url = urllib.parse.urljoin(info.forwarded_url, fields.url)
        request = urllib.request.urlopen(url, timeout=10)
        if fields.size < 4096:
            fields.content = request.read()
        else:
            with tempfile.NamedTemporaryFile(suffix='.nixexprs.tar.xz', delete=False) as tmp_file:
                shutil.copyfileobj(request, tmp_file)
                fields.file = tmp_file.name
        v.result(request.status == 200)
        v.status('Verifying digest for "%s"' % resource)
        if fields.size < 4096:
            actual_hash = hashlib.sha256(fields.content).hexdigest()
        else:
            hasher = hashlib.sha256()
            with open(fields.file, 'rb') as f:
                # pylint: disable=cell-var-from-loop
                for block in iter(lambda: f.read(4096), b''):
                    hasher.update(block)
            actual_hash = hasher.hexdigest()
        v.result(actual_hash == fields.digest)
    v.check('Verifying git commit on main page matches git commit in table',
            info.table['git-revision'].content.decode() == info.git_commit)


def extract_channel(v: Verification, info: Info) -> None:
    with tempfile.TemporaryDirectory() as d:
        v.status('Extracting nixexprs.tar.xz')
        shutil.unpack_archive(info.table['nixexprs.tar.xz'].file, d)
        v.ok()
        v.status('Removing temporary directory')
    v.ok()


def main() -> None:
    v = Verification()
    info = fetch(v, 'https://channels.nixos.org/nixos-20.03')
    parse(v, info)
    fetch_resources(v, info)
    extract_channel(v, info)
    print(info)


main()