X-Git-Url: http://git.scottworley.com/pinch/blobdiff_plain/89e79125a08a7dbc1882ce356270682836ac0ebb..f9cd7bdc50fad6c24ac226a6532ecfb81d78cbe2:/pinch.py diff --git a/pinch.py b/pinch.py index cdc0ba8..c77d891 100644 --- a/pinch.py +++ b/pinch.py @@ -5,6 +5,7 @@ import operator import os import os.path import shutil +import subprocess import tempfile import types import urllib.parse @@ -15,24 +16,30 @@ from typing import ( Dict, Iterable, List, - Sequence, + NewType, Tuple, ) +Digest16 = NewType('Digest16', str) +Digest32 = NewType('Digest32', str) -class InfoTableEntry(types.SimpleNamespace): - content: bytes - digest: str + +class ChannelTableEntry(types.SimpleNamespace): + digest: Digest16 file: str size: int url: str -class Info(types.SimpleNamespace): +class Channel(types.SimpleNamespace): channel_html: bytes forwarded_url: str + git_cachedir: str + git_ref: str + git_repo: str git_revision: str - table: Dict[str, InfoTableEntry] + release_name: str + table: Dict[str, ChannelTableEntry] url: str @@ -71,10 +78,7 @@ class Verification: self.result(True) -def compare(a: str, - b: str) -> Tuple[Sequence[str], - Sequence[str], - Sequence[str]]: +def compare(a: str, b: str) -> Tuple[List[str], List[str], List[str]]: def throw(error: OSError) -> None: raise error @@ -102,87 +106,268 @@ def compare(a: str, return filecmp.cmpfiles(a, b, files, shallow=False) -def fetch(v: Verification, channel_url: str) -> Info: - info = Info() - info.url = channel_url +def fetch(v: Verification, channel: Channel) -> None: 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() + request = urllib.request.urlopen(channel.url, timeout=10) + channel.channel_html = request.read() + channel.forwarded_url = request.geturl() v.result(request.status == 200) - v.check('Got forwarded', info.url != info.forwarded_url) - return info + v.check('Got forwarded', channel.url != channel.forwarded_url) -def parse(v: Verification, info: Info) -> None: +def parse_channel(v: Verification, channel: Channel) -> None: v.status('Parsing channel description as XML') - d = xml.dom.minidom.parseString(info.channel_html) + d = xml.dom.minidom.parseString(channel.channel_html) v.ok() - v.status('Extracting git commit') + 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) + channel.release_name = title_name + + v.status('Extracting git commit:') git_commit_node = d.getElementsByTagName('tt')[0] - info.git_commit = git_commit_node.firstChild.nodeValue + channel.git_commit = git_commit_node.firstChild.nodeValue + v.status(channel.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 = {} + channel.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) + digest = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue) + channel.table[name] = ChannelTableEntry( + url=url, digest=digest, size=size) v.ok() -def fetch_resources(v: Verification, info: Info) -> None: +def digest_string(s: bytes) -> Digest16: + return Digest16(hashlib.sha256(s).hexdigest()) + + +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, channel: Channel) -> 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') + fields = channel.table[resource] + url = urllib.parse.urljoin(channel.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( + channel.table['git-revision'].file).read(999) == channel.git_commit) + + +def git_fetch(v: Verification, channel: Channel) -> None: + # It would be nice if we could share the nix git cache, but as of the time + # of writing it is transitioning from gitv2 (deprecated) to gitv3 (not ready + # yet), and trying to straddle them both is too far into nix implementation + # details for my comfort. So we re-implement here half of nix.fetchGit. + # :( + + # TODO: Consider using pyxdg to find this path. + channel.git_cachedir = os.path.expanduser( + '~/.cache/nix-pin-channel/git/%s' % + digest_string( + channel.url.encode())) + if not os.path.exists(channel.git_cachedir): + v.status("Initializing git repo") + process = subprocess.run( + ['git', 'init', '--bare', channel.git_cachedir]) + v.result(process.returncode == 0) + + v.status('Checking if we already have this rev:') + process = subprocess.run( + ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_commit]) + if process.returncode == 0: + v.status('yes') + if process.returncode == 1: + v.status('no') + v.result(process.returncode == 0 or process.returncode == 1) + if process.returncode == 1: + v.status('Fetching ref "%s"' % channel.git_ref) + # We don't use --force here because we want to abort and freak out if forced + # updates are happening. + process = subprocess.run(['git', + '-C', + channel.git_cachedir, + 'fetch', + channel.git_repo, + '%s:%s' % (channel.git_ref, + channel.git_ref)]) + v.result(process.returncode == 0) + v.status('Verifying that fetch retrieved this rev') + process = subprocess.run( + ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_commit]) + v.result(process.returncode == 0) + + v.status('Verifying rev is an ancestor of ref') + process = subprocess.run(['git', + '-C', + channel.git_cachedir, + 'merge-base', + '--is-ancestor', + channel.git_commit, + channel.git_ref]) + v.result(process.returncode == 0) + + +def compare_tarball_and_git( + v: Verification, + channel: Channel, + channel_contents: str, + git_contents: str) -> None: + v.status('Comparing channel tarball with git checkout') + match, mismatch, errors = compare(os.path.join( + channel_contents, channel.release_name), git_contents) + v.ok() + v.check('%d files match' % len(match), len(match) > 0) + v.check('%d files differ' % len(mismatch), len(mismatch) == 0) + expected_errors = [ + '.git-revision', + '.version-suffix', + 'nixpkgs', + 'programs.sqlite', + 'svn-revision'] + benign_errors = [] + for ee in expected_errors: + if ee in errors: + errors.remove(ee) + benign_errors.append(ee) + v.check( + '%d unexpected incomparable files' % + len(errors), + len(errors) == 0) + v.check( + '(%d of %d expected incomparable files)' % + (len(benign_errors), + len(expected_errors)), + len(benign_errors) == len(expected_errors)) + + +def extract_tarball(v: Verification, channel: Channel, dest: str) -> None: + v.status('Extracting tarball %s' % + channel.table['nixexprs.tar.xz'].file) + shutil.unpack_archive( + channel.table['nixexprs.tar.xz'].file, + dest) + v.ok() + + +def git_checkout(v: Verification, channel: Channel, dest: str) -> None: + v.status('Checking out corresponding git revision') + git = subprocess.Popen(['git', + '-C', + channel.git_cachedir, + 'archive', + channel.git_commit], + stdout=subprocess.PIPE) + tar = subprocess.Popen( + ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout) + git.stdout.close() + tar.wait() + git.wait() + v.result(git.returncode == 0 and tar.returncode == 0) + + +def check_channel_metadata( + v: Verification, + channel: Channel, + channel_contents: str) -> None: + v.status('Verifying git commit in channel tarball') + v.result( + open( + os.path.join( + channel_contents, + channel.release_name, + '.git-revision')).read(999) == channel.git_commit) + + v.status( + 'Verifying version-suffix is a suffix of release name %s:' % + channel.release_name) + version_suffix = open( + os.path.join( + channel_contents, + channel.release_name, + '.version-suffix')).read(999) + v.status(version_suffix) + v.result(channel.release_name.endswith(version_suffix)) + + +def check_channel_contents(v: Verification, channel: Channel) -> None: + with tempfile.TemporaryDirectory() as channel_contents, \ + tempfile.TemporaryDirectory() as git_contents: + + extract_tarball(v, channel, channel_contents) + check_channel_metadata(v, channel, channel_contents) + + git_checkout(v, channel, git_contents) + + compare_tarball_and_git(v, channel, channel_contents, git_contents) + + v.status('Removing temporary directories') 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) + channel = Channel(url='https://channels.nixos.org/nixos-20.03', + git_repo='https://github.com/NixOS/nixpkgs.git', + git_ref='nixos-20.03') + fetch(v, channel) + parse_channel(v, channel) + fetch_resources(v, channel) + git_fetch(v, channel) + check_channel_contents(v, channel) + print(channel) main()