X-Git-Url: http://git.scottworley.com/pinch/blobdiff_plain/41b87c9c5ed19c095f2d51c9180c826378bb2df6..75e95e09d162331bfa20e8eeca5c5695d295d6c2:/pinch.py diff --git a/pinch.py b/pinch.py index 6f3f256..9b845e3 100644 --- a/pinch.py +++ b/pinch.py @@ -1,4 +1,3 @@ -from abc import ABC, abstractmethod import argparse import configparser import filecmp @@ -19,21 +18,25 @@ import urllib.request import xml.dom.minidom from typing import ( + Callable, Dict, Iterable, List, Mapping, NamedTuple, NewType, + Optional, + Set, Tuple, Type, + TypeVar, Union, ) # Use xdg module when it's less painful to have as a dependency -class XDG(types.SimpleNamespace): +class XDG(NamedTuple): XDG_CACHE_HOME: str @@ -109,83 +112,64 @@ class ChannelPin(NamedTuple): Pin = Union[AliasPin, GitPin, ChannelPin] -class SearchPath(types.SimpleNamespace, ABC): +class AliasSearchPath(NamedTuple): + alias_of: str - @abstractmethod - def pin(self, v: Verification) -> Pin: - pass + # pylint: disable=no-self-use + def pin(self, _: Verification, __: Optional[Pin]) -> AliasPin: + return AliasPin() -class AliasSearchPath(SearchPath): - alias_of: str +class GitSearchPath(NamedTuple): + git_ref: str + git_repo: str - def pin(self, v: Verification) -> AliasPin: - assert not hasattr(self, 'git_repo') - return AliasPin() + 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) -# (This lint-disable is for pylint bug https://github.com/PyCQA/pylint/issues/179 -# which is fixed in pylint 2.5.) -class TarrableSearchPath(SearchPath, ABC): # pylint: disable=abstract-method - channel_html: bytes +class ChannelSearchPath(NamedTuple): channel_url: str - forwarded_url: str git_ref: str git_repo: str - old_git_revision: str - table: Dict[str, ChannelTableEntry] - - -class GitSearchPath(TarrableSearchPath): - def pin(self, v: Verification) -> GitPin: - if hasattr(self, 'git_revision'): - self.old_git_revision = self.git_revision - del self.git_revision - - git_fetch(v, self) - return GitPin(release_name=git_revision_name(v, self), - git_revision=self.git_revision) - - def fetch(self, v: Verification, section: str, - conf: configparser.SectionProxy) -> str: - if 'git_repo' not in conf or 'release_name' not in conf: - raise Exception( - 'Cannot update unpinned channel "%s" (Run "pin" before "update")' % - section) - - ensure_git_rev_available(v, self) - return git_get_tarball(v, self) - - -class ChannelSearchPath(TarrableSearchPath): - def pin(self, v: Verification) -> ChannelPin: - if hasattr(self, 'git_revision'): - self.old_git_revision = self.git_revision - del self.git_revision - - fetch(v, self) - parse_channel(v, self) - fetch_resources(v, self) - ensure_git_rev_available(v, self) - check_channel_contents(v, self) + + 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=self.release_name, - tarball_url=self.table['nixexprs.tar.xz'].absolute_url, - tarball_sha256=self.table['nixexprs.tar.xz'].digest, - git_revision=self.git_revision) + 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) - # Lint TODO: Put tarball_url and tarball_sha256 in ChannelSearchPath # pylint: disable=no-self-use - def fetch(self, v: Verification, section: str, - conf: configparser.SectionProxy) -> str: - if 'git_repo' not in conf or 'release_name' not in conf: - raise Exception( - 'Cannot update unpinned channel "%s" (Run "pin" before "update")' % - section) + def fetch(self, v: Verification, pin: Pin) -> str: + assert isinstance(pin, ChannelPin) return fetch_with_nix_prefetch_url( - v, conf['tarball_url'], Digest16( - conf['tarball_sha256'])) + 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]]: @@ -216,18 +200,21 @@ 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: TarrableSearchPath) -> 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() + 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 != channel.forwarded_url) + v.check('Got forwarded', channel.channel_url != forwarded_url) + return channel_html, forwarded_url -def parse_channel(v: Verification, channel: TarrableSearchPath) -> 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:') @@ -236,26 +223,25 @@ def parse_channel(v: Verification, channel: TarrableSearchPath) -> 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: @@ -305,17 +291,18 @@ def fetch_with_nix_prefetch_url( return path # type: ignore # (for old mypy) -def fetch_resources(v: Verification, channel: ChannelSearchPath) -> 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: @@ -325,17 +312,21 @@ def git_cachedir(git_repo: str) -> str: digest_string(git_repo.encode())) -def tarball_cache_file(channel: TarrableSearchPath) -> str: +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()), - channel.git_revision, - channel.release_name)) + pin.git_revision, + pin.release_name)) -def verify_git_ancestry(v: Verification, channel: TarrableSearchPath) -> None: +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', @@ -343,25 +334,29 @@ def verify_git_ancestry(v: Verification, channel: TarrableSearchPath) -> 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: TarrableSearchPath) -> 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 @@ -387,49 +382,53 @@ def git_fetch(v: Verification, channel: TarrableSearchPath) -> 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) + + return new_revision def ensure_git_rev_available( v: Verification, - channel: TarrableSearchPath) -> None: + 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: TarrableSearchPath, + 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) @@ -457,26 +456,24 @@ def compare_tarball_and_git( def extract_tarball( v: Verification, - channel: TarrableSearchPath, + table: Dict[str, ChannelTableEntry], 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.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: 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) @@ -487,8 +484,11 @@ def git_checkout( v.result(git.returncode == 0 and tar.returncode == 0) -def git_get_tarball(v: Verification, channel: TarrableSearchPath) -> str: - cache_file = tarball_cache_file(channel) +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): @@ -496,17 +496,17 @@ def git_get_tarball(v: Verification, channel: TarrableSearchPath) -> str: 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() @@ -526,46 +526,51 @@ def git_get_tarball(v: Verification, channel: TarrableSearchPath) -> str: def check_channel_metadata( v: Verification, - channel: TarrableSearchPath, + 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: TarrableSearchPath) -> None: + 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 git_revision_name(v: Verification, channel: TarrableSearchPath) -> str: +def git_revision_name( + v: Verification, + channel: TarrableSearchPath, + git_revision: str) -> str: v.status('Getting commit date') process = subprocess.run(['git', '-C', @@ -575,20 +580,58 @@ def git_revision_name(v: Verification, channel: TarrableSearchPath) -> str: '--format=%ct-%h', '--abbrev=11', '--no-show-signature', - channel.git_revision], + 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()) -def read_search_path(conf: configparser.SectionProxy) -> SearchPath: - mapping: Mapping[str, Type[SearchPath]] = { - 'alias': AliasSearchPath, - 'channel': ChannelSearchPath, - 'git': GitSearchPath, +K = TypeVar('K') +V = TypeVar('V') + + +def partition_dict(pred: Callable[[K, V], bool], + d: Dict[K, V]) -> Tuple[Dict[K, V], Dict[K, V]]: + selected: Dict[K, V] = {} + remaining: Dict[K, V] = {} + for k, v in d.items(): + if pred(k, v): + selected[k] = v + else: + remaining[k] = v + return selected, remaining + + +def filter_dict(d: Dict[K, V], fields: Set[K] + ) -> Tuple[Dict[K, V], Dict[K, V]]: + return partition_dict(lambda k, v: k in fields, d) + + +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), } - return mapping[conf['type']](**dict(conf.items())) + 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: @@ -616,9 +659,9 @@ def pinCommand(args: argparse.Namespace) -> None: if args.channels and section not in args.channels: continue - sp = read_search_path(config[section]) + sp, old_pin = read_config_section(config[section]) - config[section].update(sp.pin(v)._asdict()) + config[section].update(sp.pin(v, old_pin)._asdict()) with open(args.channels_file, 'w') as configfile: config.write(configfile) @@ -627,20 +670,24 @@ def pinCommand(args: argparse.Namespace) -> None: def updateCommand(args: argparse.Namespace) -> None: v = Verification() exprs: Dict[str, str] = {} - config = read_config_files(args.channels_file) - for section in config: - sp = read_search_path(config[section]) - if isinstance(sp, AliasSearchPath): - assert 'git_repo' not in config[section] - continue - tarball = sp.fetch(v, section, config[section]) + config = { + section: read_pinned_config_section(section, conf) for section, + conf in read_config_files( + args.channels_file).items()} + alias, nonalias = partition_dict( + lambda k, v: isinstance(v[0], AliasSearchPath), config) + + for section, (sp, pin) in nonalias.items(): + assert not isinstance(sp, AliasSearchPath) # mypy can't see through + assert not isinstance(pin, AliasPin) # partition_dict() + tarball = sp.fetch(v, pin) exprs[section] = ( 'f: f { name = "%s"; channelName = "%%s"; src = builtins.storePath "%s"; }' % - (config[section]['release_name'], tarball)) + (pin.release_name, tarball)) - for section in config: - if 'alias_of' in config[section]: - exprs[section] = exprs[str(config[section]['alias_of'])] + for section, (sp, pin) in alias.items(): + assert isinstance(sp, AliasSearchPath) # For mypy + exprs[section] = exprs[sp.alias_of] command = [ 'nix-env',