X-Git-Url: http://git.scottworley.com/pinch/blobdiff_plain/ca2c3eddd11811b0c494763e3d99614e9f6c62e7..038f00114125f80eb24b500ed4001881b785460a:/pinch.py diff --git a/pinch.py b/pinch.py index a8b6103..dd70718 100644 --- a/pinch.py +++ b/pinch.py @@ -1,3 +1,4 @@ +import configparser import filecmp import functools import hashlib @@ -6,6 +7,7 @@ import os import os.path import shutil import subprocess +import sys import tempfile import types import urllib.parse @@ -17,7 +19,6 @@ from typing import ( Iterable, List, NewType, - Sequence, Tuple, ) @@ -26,6 +27,7 @@ Digest32 = NewType('Digest32', str) class ChannelTableEntry(types.SimpleNamespace): + absolute_url: str digest: Digest16 file: str size: int @@ -35,6 +37,9 @@ class ChannelTableEntry(types.SimpleNamespace): class Channel(types.SimpleNamespace): channel_html: bytes forwarded_url: str + git_cachedir: str + git_ref: str + git_repo: str git_revision: str release_name: str table: Dict[str, ChannelTableEntry] @@ -76,10 +81,7 @@ 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 @@ -131,8 +133,8 @@ def parse_channel(v: Verification, channel: Channel) -> None: v.status('Extracting git commit:') git_commit_node = d.getElementsByTagName('tt')[0] - channel.git_commit = git_commit_node.firstChild.nodeValue - v.status(channel.git_commit) + channel.git_revision = git_commit_node.firstChild.nodeValue + v.status(channel.git_revision) v.ok() v.status('Verifying git commit label') v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ') @@ -144,10 +146,15 @@ def parse_channel(v: Verification, channel: Channel) -> None: url = row.childNodes[0].firstChild.getAttribute('href') size = int(row.childNodes[1].firstChild.nodeValue) digest = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue) - channel.table[name] = ChannelTableEntry(url=url, digest=digest, size=size) + channel.table[name] = ChannelTableEntry( + url=url, digest=digest, size=size) v.ok() +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: @@ -194,31 +201,183 @@ def fetch_with_nix_prefetch_url( def fetch_resources(v: Verification, channel: Channel) -> None: for resource in ['git-revision', 'nixexprs.tar.xz']: fields = channel.table[resource] - url = urllib.parse.urljoin(channel.forwarded_url, fields.url) - fields.file = fetch_with_nix_prefetch_url(v, url, fields.digest) + fields.absolute_url = urllib.parse.urljoin( + channel.forwarded_url, fields.url) + fields.file = fetch_with_nix_prefetch_url( + v, fields.absolute_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) + channel.table['git-revision'].file).read(999) == channel.git_revision) + + +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_revision]) + 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_revision]) + 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_revision, + channel.git_ref]) + v.result(process.returncode == 0) -def check_channel_contents(v: Verification, channel: Channel) -> None: - with tempfile.TemporaryDirectory() as d: - v.status('Extracting %s' % channel.table['nixexprs.tar.xz'].file) - shutil.unpack_archive(channel.table['nixexprs.tar.xz'].file, d) - v.ok() - v.status('Removing temporary directory') +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 main() -> None: - v = Verification() - channel = Channel(url='https://channels.nixos.org/nixos-20.03') - fetch(v, channel) - parse_channel(v, channel) - fetch_resources(v, channel) - check_channel_contents(v, channel) - print(channel) +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_revision], + 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_revision) + + 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) -main() + compare_tarball_and_git(v, channel, channel_contents, git_contents) + + v.status('Removing temporary directories') + v.ok() + + +def main(argv: List[str]) -> None: + v = Verification() + config = configparser.ConfigParser() + config.read_file(open(argv[1]), argv[1]) + for section in config.sections(): + channel = Channel(**dict(config[section].items())) + fetch(v, channel) + parse_channel(v, channel) + fetch_resources(v, channel) + git_fetch(v, channel) + check_channel_contents(v, channel) + config[section]['git_rev'] = channel.git_revision + config[section]['tarball_url'] = channel.table['nixexprs.tar.xz'].absolute_url + config[section]['tarball_sha256'] = channel.table['nixexprs.tar.xz'].digest + with open(argv[1], 'w') as configfile: + config.write(configfile) + + +main(sys.argv)