X-Git-Url: http://git.scottworley.com/pinch/blobdiff_plain/26125a281f6733d08ffb488668b052ee9c20061a..e8bd4979a02d7bb9e8fb7d978fd0e986d7ee60d1:/pinch.py?ds=inline diff --git a/pinch.py b/pinch.py index 1827281..1959f45 100644 --- a/pinch.py +++ b/pinch.py @@ -21,36 +21,28 @@ from typing import ( Dict, Iterable, List, + Mapping, + NamedTuple, NewType, + Optional, + Set, Tuple, + Type, + TypeVar, + Union, ) -import xdg +# Use xdg module when it's less painful to have as a dependency -Digest16 = NewType('Digest16', str) -Digest32 = NewType('Digest32', str) +class XDG(NamedTuple): + XDG_CACHE_HOME: str -class ChannelTableEntry(types.SimpleNamespace): - absolute_url: str - digest: Digest16 - file: str - size: int - url: str - - -class Channel(types.SimpleNamespace): - alias_of: str - channel_html: bytes - channel_url: str - forwarded_url: str - git_ref: str - git_repo: str - git_revision: str - old_git_revision: str - release_name: str - table: Dict[str, ChannelTableEntry] +xdg = XDG( + XDG_CACHE_HOME=os.getenv( + 'XDG_CACHE_HOME', + os.path.expanduser('~/.cache'))) class VerificationError(Exception): @@ -73,7 +65,7 @@ class Verification: def result(self, r: bool) -> None: message, color = {True: ('OK ', 92), False: ('FAIL', 91)}[r] length = len(message) - cols = shutil.get_terminal_size().columns + cols = shutil.get_terminal_size().columns or 80 pad = (cols - (self.line_length + length)) % cols print(' ' * pad + self._color(message, color), file=sys.stderr) self.line_length = 0 @@ -88,6 +80,97 @@ class Verification: self.result(True) +Digest16 = NewType('Digest16', str) +Digest32 = NewType('Digest32', str) + + +class ChannelTableEntry(types.SimpleNamespace): + absolute_url: str + digest: Digest16 + file: str + size: int + url: str + + +class AliasPin(NamedTuple): + pass + + +class GitPin(NamedTuple): + git_revision: str + release_name: str + + +class ChannelPin(NamedTuple): + git_revision: str + release_name: str + tarball_url: str + tarball_sha256: str + + +Pin = Union[AliasPin, GitPin, ChannelPin] + + +class AliasSearchPath(NamedTuple): + alias_of: str + + # pylint: disable=no-self-use + def pin(self, _: Verification, __: Optional[Pin]) -> AliasPin: + return AliasPin() + + +class GitSearchPath(NamedTuple): + git_ref: str + git_repo: str + + def pin(self, v: Verification, old_pin: Optional[Pin]) -> GitPin: + if old_pin is not None: + assert isinstance(old_pin, GitPin) + old_revision = old_pin.git_revision if old_pin is not None else None + + new_revision = git_fetch(v, self, None, old_revision) + return GitPin(release_name=git_revision_name(v, self, new_revision), + git_revision=new_revision) + + def fetch(self, v: Verification, pin: Pin) -> str: + assert isinstance(pin, GitPin) + ensure_git_rev_available(v, self, pin, None) + return git_get_tarball(v, self, pin) + + +class ChannelSearchPath(NamedTuple): + channel_url: str + git_ref: str + git_repo: str + + def pin(self, v: Verification, old_pin: Optional[Pin]) -> ChannelPin: + if old_pin is not None: + assert isinstance(old_pin, ChannelPin) + old_revision = old_pin.git_revision if old_pin is not None else None + + channel_html, forwarded_url = fetch_channel(v, self) + table, new_gitpin = parse_channel(v, channel_html) + fetch_resources(v, new_gitpin, forwarded_url, table) + ensure_git_rev_available(v, self, new_gitpin, old_revision) + check_channel_contents(v, self, table, new_gitpin) + return ChannelPin( + release_name=new_gitpin.release_name, + tarball_url=table['nixexprs.tar.xz'].absolute_url, + tarball_sha256=table['nixexprs.tar.xz'].digest, + git_revision=new_gitpin.git_revision) + + # pylint: disable=no-self-use + def fetch(self, v: Verification, pin: Pin) -> str: + assert isinstance(pin, ChannelPin) + + return fetch_with_nix_prefetch_url( + v, pin.tarball_url, Digest16(pin.tarball_sha256)) + + +SearchPath = Union[AliasSearchPath, GitSearchPath, ChannelSearchPath] +TarrableSearchPath = Union[GitSearchPath, ChannelSearchPath] + + def compare(a: str, b: str) -> Tuple[List[str], List[str], List[str]]: def throw(error: OSError) -> None: @@ -116,18 +199,21 @@ def compare(a: str, b: str) -> Tuple[List[str], List[str], List[str]]: return filecmp.cmpfiles(a, b, files, shallow=False) -def fetch(v: Verification, channel: Channel) -> None: +def fetch_channel( + v: Verification, channel: ChannelSearchPath) -> Tuple[str, str]: v.status('Fetching channel') request = urllib.request.urlopen(channel.channel_url, timeout=10) - channel.channel_html = request.read() - channel.forwarded_url = request.geturl() - v.result(request.status == 200) - v.check('Got forwarded', channel.channel_url != channel.forwarded_url) + channel_html = request.read() + forwarded_url = request.geturl() + v.result(request.status == 200) # type: ignore # (for old mypy) + v.check('Got forwarded', channel.channel_url != forwarded_url) + return channel_html, forwarded_url -def parse_channel(v: Verification, channel: Channel) -> None: +def parse_channel(v: Verification, channel_html: str) \ + -> Tuple[Dict[str, ChannelTableEntry], GitPin]: v.status('Parsing channel description as XML') - d = xml.dom.minidom.parseString(channel.channel_html) + d = xml.dom.minidom.parseString(channel_html) v.ok() v.status('Extracting release name:') @@ -136,26 +222,25 @@ def parse_channel(v: Verification, channel: Channel) -> None: 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] - channel.git_revision = git_commit_node.firstChild.nodeValue - v.status(channel.git_revision) + git_revision = git_commit_node.firstChild.nodeValue + v.status(git_revision) v.ok() v.status('Verifying git commit label') v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ') v.status('Parsing table') - channel.table = {} + table: Dict[str, ChannelTableEntry] = {} 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) - channel.table[name] = ChannelTableEntry( - url=url, digest=digest, size=size) + table[name] = ChannelTableEntry(url=url, digest=digest, size=size) v.ok() + return table, GitPin(release_name=title_name, git_revision=git_revision) def digest_string(s: bytes) -> Digest16: @@ -174,7 +259,7 @@ def digest_file(filename: str) -> Digest16: 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) + ['nix', 'to-base16', '--type', 'sha256', digest32], stdout=subprocess.PIPE) v.result(process.returncode == 0) return Digest16(process.stdout.decode().strip()) @@ -182,7 +267,7 @@ def to_Digest16(v: Verification, digest32: Digest32) -> Digest16: 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) + ['nix', 'to-base32', '--type', 'sha256', digest16], stdout=subprocess.PIPE) v.result(process.returncode == 0) return Digest32(process.stdout.decode().strip()) @@ -193,7 +278,7 @@ def fetch_with_nix_prefetch_url( digest: Digest16) -> str: v.status('Fetching %s' % url) process = subprocess.run( - ['nix-prefetch-url', '--print-path', url, digest], capture_output=True) + ['nix-prefetch-url', '--print-path', url, digest], stdout=subprocess.PIPE) v.result(process.returncode == 0) prefetch_digest, path, empty = process.stdout.decode().split('\n') assert empty == '' @@ -202,31 +287,45 @@ def fetch_with_nix_prefetch_url( v.status("Verifying file digest") file_digest = digest_file(path) v.result(file_digest == digest) - return path + return path # type: ignore # (for old mypy) -def fetch_resources(v: Verification, channel: Channel) -> None: +def fetch_resources( + v: Verification, + pin: GitPin, + forwarded_url: str, + table: Dict[str, ChannelTableEntry]) -> None: for resource in ['git-revision', 'nixexprs.tar.xz']: - fields = channel.table[resource] - fields.absolute_url = urllib.parse.urljoin( - channel.forwarded_url, fields.url) + fields = table[resource] + fields.absolute_url = urllib.parse.urljoin(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_revision) + v.result(open(table['git-revision'].file).read(999) == pin.git_revision) def git_cachedir(git_repo: str) -> str: return os.path.join( xdg.XDG_CACHE_HOME, 'pinch/git', - digest_string( - git_repo.encode())) + digest_string(git_repo.encode())) -def verify_git_ancestry(v: Verification, channel: Channel) -> None: +def tarball_cache_file(channel: TarrableSearchPath, pin: GitPin) -> str: + return os.path.join( + xdg.XDG_CACHE_HOME, + 'pinch/git-tarball', + '%s-%s-%s' % + (digest_string(channel.git_repo.encode()), + pin.git_revision, + pin.release_name)) + + +def verify_git_ancestry( + v: Verification, + channel: TarrableSearchPath, + new_revision: str, + old_revision: Optional[str]) -> None: cachedir = git_cachedir(channel.git_repo) v.status('Verifying rev is an ancestor of ref') process = subprocess.run(['git', @@ -234,25 +333,29 @@ def verify_git_ancestry(v: Verification, channel: Channel) -> None: cachedir, 'merge-base', '--is-ancestor', - channel.git_revision, + new_revision, channel.git_ref]) v.result(process.returncode == 0) - if hasattr(channel, 'old_git_revision'): + if old_revision is not None: v.status( 'Verifying rev is an ancestor of previous rev %s' % - channel.old_git_revision) + old_revision) process = subprocess.run(['git', '-C', cachedir, 'merge-base', '--is-ancestor', - channel.old_git_revision, - channel.git_revision]) + old_revision, + new_revision]) v.result(process.returncode == 0) -def git_fetch(v: Verification, channel: Channel) -> None: +def git_fetch( + v: Verification, + channel: TarrableSearchPath, + desired_revision: Optional[str], + old_revision: Optional[str]) -> str: # 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 @@ -278,47 +381,53 @@ def git_fetch(v: Verification, channel: Channel) -> None: channel.git_ref)]) v.result(process.returncode == 0) - if hasattr(channel, 'git_revision'): + if desired_revision is not None: v.status('Verifying that fetch retrieved this rev') process = subprocess.run( - ['git', '-C', cachedir, 'cat-file', '-e', channel.git_revision]) + ['git', '-C', cachedir, 'cat-file', '-e', desired_revision]) v.result(process.returncode == 0) - else: - channel.git_revision = open( - os.path.join( - cachedir, - 'refs', - 'heads', - channel.git_ref)).read(999).strip() - verify_git_ancestry(v, channel) + new_revision = open( + os.path.join( + cachedir, + 'refs', + 'heads', + channel.git_ref)).read(999).strip() + verify_git_ancestry(v, channel, new_revision, old_revision) -def ensure_git_rev_available(v: Verification, channel: Channel) -> None: + return new_revision + + +def ensure_git_rev_available( + v: Verification, + channel: TarrableSearchPath, + pin: GitPin, + old_revision: Optional[str]) -> None: cachedir = git_cachedir(channel.git_repo) if os.path.exists(cachedir): v.status('Checking if we already have this rev:') process = subprocess.run( - ['git', '-C', cachedir, 'cat-file', '-e', channel.git_revision]) + ['git', '-C', cachedir, 'cat-file', '-e', pin.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 == 0: - verify_git_ancestry(v, channel) + verify_git_ancestry(v, channel, pin.git_revision, old_revision) return - git_fetch(v, channel) + git_fetch(v, channel, pin.git_revision, old_revision) def compare_tarball_and_git( v: Verification, - channel: Channel, + pin: GitPin, 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) + channel_contents, pin.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) @@ -344,22 +453,26 @@ def compare_tarball_and_git( 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) +def extract_tarball( + v: Verification, + table: Dict[str, ChannelTableEntry], + dest: str) -> None: + v.status('Extracting tarball %s' % table['nixexprs.tar.xz'].file) + shutil.unpack_archive(table['nixexprs.tar.xz'].file, dest) v.ok() -def git_checkout(v: Verification, channel: Channel, dest: str) -> None: +def git_checkout( + v: Verification, + channel: TarrableSearchPath, + pin: GitPin, + dest: str) -> None: v.status('Checking out corresponding git revision') git = subprocess.Popen(['git', '-C', git_cachedir(channel.git_repo), 'archive', - channel.git_revision], + pin.git_revision], stdout=subprocess.PIPE) tar = subprocess.Popen( ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout) @@ -370,20 +483,29 @@ def git_checkout(v: Verification, channel: Channel, dest: str) -> None: v.result(git.returncode == 0 and tar.returncode == 0) -def git_get_tarball(v: Verification, channel: Channel) -> str: +def git_get_tarball( + v: Verification, + channel: TarrableSearchPath, + pin: GitPin) -> str: + cache_file = tarball_cache_file(channel, pin) + if os.path.exists(cache_file): + cached_tarball = open(cache_file).read(9999) + if os.path.exists(cached_tarball): + return cached_tarball + with tempfile.TemporaryDirectory() as output_dir: output_filename = os.path.join( - output_dir, channel.release_name + '.tar.xz') + output_dir, pin.release_name + '.tar.xz') with open(output_filename, 'w') as output_file: v.status( 'Generating tarball for git revision %s' % - channel.git_revision) + pin.git_revision) git = subprocess.Popen(['git', '-C', git_cachedir(channel.git_repo), 'archive', - '--prefix=%s/' % channel.release_name, - channel.git_revision], + '--prefix=%s/' % pin.release_name, + pin.git_revision], stdout=subprocess.PIPE) xz = subprocess.Popen(['xz'], stdin=git.stdout, stdout=output_file) xz.wait() @@ -392,144 +514,169 @@ def git_get_tarball(v: Verification, channel: Channel) -> str: v.status('Putting tarball in Nix store') process = subprocess.run( - ['nix-store', '--add', output_filename], capture_output=True) + ['nix-store', '--add', output_filename], stdout=subprocess.PIPE) v.result(process.returncode == 0) - return process.stdout.decode().strip() + store_tarball = process.stdout.decode().strip() + + os.makedirs(os.path.dirname(cache_file), exist_ok=True) + open(cache_file, 'w').write(store_tarball) + return store_tarball # type: ignore # (for old mypy) def check_channel_metadata( v: Verification, - channel: Channel, + pin: GitPin, 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) + pin.release_name, + '.git-revision')).read(999) == pin.git_revision) v.status( 'Verifying version-suffix is a suffix of release name %s:' % - channel.release_name) + pin.release_name) version_suffix = open( os.path.join( channel_contents, - channel.release_name, + pin.release_name, '.version-suffix')).read(999) v.status(version_suffix) - v.result(channel.release_name.endswith(version_suffix)) + v.result(pin.release_name.endswith(version_suffix)) -def check_channel_contents(v: Verification, channel: Channel) -> None: +def check_channel_contents( + v: Verification, + channel: TarrableSearchPath, + table: Dict[str, ChannelTableEntry], + pin: GitPin) -> 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) + extract_tarball(v, table, channel_contents) + check_channel_metadata(v, pin, channel_contents) - git_checkout(v, channel, git_contents) + git_checkout(v, channel, pin, git_contents) - compare_tarball_and_git(v, channel, channel_contents, git_contents) + compare_tarball_and_git(v, pin, channel_contents, git_contents) v.status('Removing temporary directories') v.ok() -def pin_channel(v: Verification, channel: Channel) -> None: - fetch(v, channel) - parse_channel(v, channel) - fetch_resources(v, channel) - ensure_git_rev_available(v, channel) - check_channel_contents(v, channel) - - -def git_revision_name(v: Verification, channel: Channel) -> str: +def git_revision_name( + v: Verification, + channel: TarrableSearchPath, + git_revision: str) -> str: v.status('Getting commit date') process = subprocess.run(['git', '-C', git_cachedir(channel.git_repo), - 'lo', + 'log', '-n1', '--format=%ct-%h', '--abbrev=11', - channel.git_revision], - capture_output=True) + '--no-show-signature', + git_revision], + stdout=subprocess.PIPE) v.result(process.returncode == 0 and process.stdout != b'') return '%s-%s' % (os.path.basename(channel.git_repo), process.stdout.decode().strip()) +K = TypeVar('K') +V = TypeVar('V') + + +def filter_dict(d: Dict[K, V], fields: Set[K] + ) -> Tuple[Dict[K, V], Dict[K, V]]: + selected: Dict[K, V] = {} + remaining: Dict[K, V] = {} + for k, v in d.items(): + if k in fields: + selected[k] = v + else: + remaining[k] = v + return selected, remaining + + +def read_config_section( + conf: configparser.SectionProxy) -> Tuple[SearchPath, Optional[Pin]]: + mapping: Mapping[str, Tuple[Type[SearchPath], Type[Pin]]] = { + 'alias': (AliasSearchPath, AliasPin), + 'channel': (ChannelSearchPath, ChannelPin), + 'git': (GitSearchPath, GitPin), + } + SP, P = mapping[conf['type']] + _, all_fields = filter_dict(dict(conf.items()), set(['type'])) + pin_fields, remaining_fields = filter_dict(all_fields, set(P._fields)) + # Error suppression works around https://github.com/python/mypy/issues/9007 + pin_present = pin_fields != {} or P._fields == () + pin = P(**pin_fields) if pin_present else None # type:ignore[call-arg] + return SP(**remaining_fields), pin + + +def read_pinned_config_section( + section: str, conf: configparser.SectionProxy) -> Tuple[SearchPath, Pin]: + sp, pin = read_config_section(conf) + if pin is None: + raise Exception( + 'Cannot update unpinned channel "%s" (Run "pin" before "update")' % + section) + return sp, pin + + def read_config(filename: str) -> configparser.ConfigParser: config = configparser.ConfigParser() config.read_file(open(filename), filename) return config -def pin(args: argparse.Namespace) -> None: +def read_config_files( + filenames: Iterable[str]) -> Dict[str, configparser.SectionProxy]: + merged_config: Dict[str, configparser.SectionProxy] = {} + for file in filenames: + config = read_config(file) + for section in config.sections(): + if section in merged_config: + raise Exception('Duplicate channel "%s"' % section) + merged_config[section] = config[section] + return merged_config + + +def pinCommand(args: argparse.Namespace) -> None: v = Verification() config = read_config(args.channels_file) for section in config.sections(): if args.channels and section not in args.channels: continue - channel = Channel(**dict(config[section].items())) - - if hasattr(channel, 'alias_of'): - assert not hasattr(channel, 'git_repo') - continue + sp, old_pin = read_config_section(config[section]) - if hasattr(channel, 'git_revision'): - channel.old_git_revision = channel.git_revision - del channel.git_revision - - if 'channel_url' in config[section]: - pin_channel(v, channel) - config[section]['release_name'] = channel.release_name - config[section]['tarball_url'] = channel.table['nixexprs.tar.xz'].absolute_url - config[section]['tarball_sha256'] = channel.table['nixexprs.tar.xz'].digest - else: - git_fetch(v, channel) - config[section]['release_name'] = git_revision_name(v, channel) - config[section]['git_revision'] = channel.git_revision + config[section].update(sp.pin(v, old_pin)._asdict()) with open(args.channels_file, 'w') as configfile: config.write(configfile) -def update(args: argparse.Namespace) -> None: +def updateCommand(args: argparse.Namespace) -> None: v = Verification() - config = configparser.ConfigParser() exprs: Dict[str, str] = {} - configs = [read_config(filename) for filename in args.channels_file] - for config in configs: - for section in config.sections(): - - if 'alias_of' in config[section]: - assert 'git_repo' not in config[section] - continue - - if 'channel_url' in config[section]: - tarball = fetch_with_nix_prefetch_url( - v, config[section]['tarball_url'], Digest16( - config[section]['tarball_sha256'])) - else: - channel = Channel(**dict(config[section].items())) - ensure_git_rev_available(v, channel) - tarball = git_get_tarball(v, channel) - - if section in exprs: - raise Exception('Duplicate channel "%s"' % section) - exprs[section] = ( - 'f: f { name = "%s"; channelName = "%%s"; src = builtins.storePath "%s"; }' % - (config[section]['release_name'], tarball)) + config = read_config_files(args.channels_file) + for section in config: + sp, pin = read_pinned_config_section(section, config[section]) + if isinstance(sp, AliasSearchPath): + continue + tarball = sp.fetch(v, pin) + exprs[section] = ( + 'f: f { name = "%s"; channelName = "%%s"; src = builtins.storePath "%s"; }' % + (config[section]['release_name'], tarball)) - for config in configs: - for section in config.sections(): - if 'alias_of' in config[section]: - if section in exprs: - raise Exception('Duplicate channel "%s"' % section) - exprs[section] = exprs[str(config[section]['alias_of'])] + for section in config: + if 'alias_of' in config[section]: + exprs[section] = exprs[str(config[section]['alias_of'])] command = [ 'nix-env', @@ -555,13 +702,14 @@ def main() -> None: parser_pin = subparsers.add_parser('pin') parser_pin.add_argument('channels_file', type=str) parser_pin.add_argument('channels', type=str, nargs='*') - parser_pin.set_defaults(func=pin) + parser_pin.set_defaults(func=pinCommand) parser_update = subparsers.add_parser('update') parser_update.add_argument('--dry-run', action='store_true') parser_update.add_argument('channels_file', type=str, nargs='+') - parser_update.set_defaults(func=update) + parser_update.set_defaults(func=updateCommand) args = parser.parse_args() args.func(args) -main() +if __name__ == '__main__': + main()