import filecmp import functools import hashlib import operator import os import os.path import shutil import subprocess import tempfile import types import urllib.parse import urllib.request import xml.dom.minidom from typing import ( Dict, Iterable, List, NewType, Sequence, Tuple, ) Digest16 = NewType('Digest16', str) Digest32 = NewType('Digest32', str) class InfoTableEntry(types.SimpleNamespace): digest: Digest16 file: str size: int url: str class Info(types.SimpleNamespace): channel_html: bytes forwarded_url: str git_revision: str release_name: 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_table(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 release name:') title_name = d.getElementsByTagName( 'title')[0].firstChild.nodeValue.split()[2] h1_name = d.getElementsByTagName('h1')[0].firstChild.nodeValue.split()[2] v.status(title_name) v.result(title_name == h1_name) info.release_name = title_name v.status('Extracting git commit:') git_commit_node = d.getElementsByTagName('tt')[0] info.git_commit = git_commit_node.firstChild.nodeValue v.status(info.git_commit) 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 = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue) info.table[name] = InfoTableEntry(url=url, digest=digest, size=size) v.ok() def digest_file(filename: str) -> Digest16: hasher = hashlib.sha256() with open(filename, 'rb') as f: # pylint: disable=cell-var-from-loop for block in iter(lambda: f.read(4096), b''): hasher.update(block) return Digest16(hasher.hexdigest()) def to_Digest16(v: Verification, digest32: Digest32) -> Digest16: v.status('Converting digest to base16') process = subprocess.run( ['nix', 'to-base16', '--type', 'sha256', digest32], capture_output=True) v.result(process.returncode == 0) return Digest16(process.stdout.decode().strip()) def to_Digest32(v: Verification, digest16: Digest16) -> Digest32: v.status('Converting digest to base32') process = subprocess.run( ['nix', 'to-base32', '--type', 'sha256', digest16], capture_output=True) v.result(process.returncode == 0) return Digest32(process.stdout.decode().strip()) def fetch_with_nix_prefetch_url( v: Verification, url: str, digest: Digest16) -> str: v.status('Fetching %s' % url) process = subprocess.run( ['nix-prefetch-url', '--print-path', url, digest], capture_output=True) v.result(process.returncode == 0) prefetch_digest, path, empty = process.stdout.decode().split('\n') assert empty == '' v.check("Verifying nix-prefetch-url's digest", to_Digest16(v, Digest32(prefetch_digest)) == digest) v.status("Verifying file digest") file_digest = digest_file(path) v.result(file_digest == digest) return path def fetch_resources(v: Verification, info: Info) -> None: for resource in ['git-revision', 'nixexprs.tar.xz']: fields = info.table[resource] url = urllib.parse.urljoin(info.forwarded_url, fields.url) fields.file = fetch_with_nix_prefetch_url(v, url, fields.digest) v.status('Verifying git commit on main page matches git commit in table') v.result( open( info.table['git-revision'].file).read(999) == info.git_commit) def extract_channel(v: Verification, info: Info) -> None: with tempfile.TemporaryDirectory() as d: v.status('Extracting %s' % info.table['nixexprs.tar.xz'].file) 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_table(v, info) fetch_resources(v, info) extract_channel(v, info) print(info) main()