Factor out pin check

[pinch] / pinch.py

import argparse
import configparser
import filecmp
import functools
import getpass
import hashlib
import operator
import os
import os.path
import shlex
import shutil
import subprocess
import sys
import tempfile
import types
import urllib.parse
import urllib.request
import xml.dom.minidom

from typing import (
    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(NamedTuple):
    XDG_CACHE_HOME: str


xdg = XDG(
    XDG_CACHE_HOME=os.getenv(
        'XDG_CACHE_HOME',
        os.path.expanduser('~/.cache')))


class VerificationError(Exception):
    pass


class Verification:

    def __init__(self) -> None:
        self.line_length = 0

    def status(self, s: str) -> None:
        print(s, end=' ', file=sys.stderr, 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 or 80
        pad = (cols - (self.line_length + length)) % cols
        print(' ' * pad + self._color(message, color), file=sys.stderr)
        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)


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:
        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_channel(
        v: Verification, channel: ChannelSearchPath) -> Tuple[str, str]:
    v.status('Fetching channel')
    request = urllib.request.urlopen(channel.channel_url, timeout=10)
    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_html: str) \
        -> Tuple[Dict[str, ChannelTableEntry], GitPin]:
    v.status('Parsing channel description as XML')
    d = xml.dom.minidom.parseString(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)

    v.status('Extracting git commit:')
    git_commit_node = d.getElementsByTagName('tt')[0]
    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')
    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)
        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:
    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], stdout=subprocess.PIPE)
    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], stdout=subprocess.PIPE)
    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], stdout=subprocess.PIPE)
    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  # type: ignore  # (for old mypy)


def fetch_resources(
        v: Verification,
        pin: GitPin,
        forwarded_url: str,
        table: Dict[str, ChannelTableEntry]) -> None:
    for resource in ['git-revision', 'nixexprs.tar.xz']:
        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(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()))


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',
                              '-C',
                              cachedir,
                              'merge-base',
                              '--is-ancestor',
                              new_revision,
                              channel.git_ref])
    v.result(process.returncode == 0)

    if old_revision is not None:
        v.status(
            'Verifying rev is an ancestor of previous rev %s' %
            old_revision)
        process = subprocess.run(['git',
                                  '-C',
                                  cachedir,
                                  'merge-base',
                                  '--is-ancestor',
                                  old_revision,
                                  new_revision])
        v.result(process.returncode == 0)


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
    # details for my comfort.  So we re-implement here half of nix.fetchGit.
    # :(

    cachedir = git_cachedir(channel.git_repo)
    if not os.path.exists(cachedir):
        v.status("Initializing git repo")
        process = subprocess.run(
            ['git', 'init', '--bare', cachedir])
        v.result(process.returncode == 0)

    v.status('Fetching ref "%s" from %s' % (channel.git_ref, channel.git_repo))
    # We don't use --force here because we want to abort and freak out if forced
    # updates are happening.
    process = subprocess.run(['git',
                              '-C',
                              cachedir,
                              'fetch',
                              channel.git_repo,
                              '%s:%s' % (channel.git_ref,
                                         channel.git_ref)])
    v.result(process.returncode == 0)

    if desired_revision is not None:
        v.status('Verifying that fetch retrieved this rev')
        process = subprocess.run(
            ['git', '-C', cachedir, 'cat-file', '-e', desired_revision])
        v.result(process.returncode == 0)

    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,
        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', 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, pin.git_revision, old_revision)
            return
    git_fetch(v, channel, pin.git_revision, old_revision)


def compare_tarball_and_git(
        v: Verification,
        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, 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)
    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,
        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: TarrableSearchPath,
        pin: GitPin,
        dest: str) -> None:
    v.status('Checking out corresponding git revision')
    git = subprocess.Popen(['git',
                            '-C',
                            git_cachedir(channel.git_repo),
                            'archive',
                            pin.git_revision],
                           stdout=subprocess.PIPE)
    tar = subprocess.Popen(
        ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout)
    if git.stdout:
        git.stdout.close()
    tar.wait()
    git.wait()
    v.result(git.returncode == 0 and tar.returncode == 0)


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, pin.release_name + '.tar.xz')
        with open(output_filename, 'w') as output_file:
            v.status(
                'Generating tarball for git revision %s' %
                pin.git_revision)
            git = subprocess.Popen(['git',
                                    '-C',
                                    git_cachedir(channel.git_repo),
                                    'archive',
                                    '--prefix=%s/' % pin.release_name,
                                    pin.git_revision],
                                   stdout=subprocess.PIPE)
            xz = subprocess.Popen(['xz'], stdin=git.stdout, stdout=output_file)
            xz.wait()
            git.wait()
            v.result(git.returncode == 0 and xz.returncode == 0)

        v.status('Putting tarball in Nix store')
        process = subprocess.run(
            ['nix-store', '--add', output_filename], stdout=subprocess.PIPE)
        v.result(process.returncode == 0)
        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,
        pin: GitPin,
        channel_contents: str) -> None:
    v.status('Verifying git commit in channel tarball')
    v.result(
        open(
            os.path.join(
                channel_contents,
                pin.release_name,
                '.git-revision')).read(999) == pin.git_revision)

    v.status(
        'Verifying version-suffix is a suffix of release name %s:' %
        pin.release_name)
    version_suffix = open(
        os.path.join(
            channel_contents,
            pin.release_name,
            '.version-suffix')).read(999)
    v.status(version_suffix)
    v.result(pin.release_name.endswith(version_suffix))


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, table, channel_contents)
        check_channel_metadata(v, pin, channel_contents)

        git_checkout(v, channel, pin, 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,
        git_revision: str) -> str:
    v.status('Getting commit date')
    process = subprocess.run(['git',
                              '-C',
                              git_cachedir(channel.git_repo),
                              'log',
                              '-n1',
                              '--format=%ct-%h',
                              '--abbrev=11',
                              '--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 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

        sp, old_pin = read_config_section(config[section])

        config[section].update(sp.pin(v, old_pin)._asdict())

    with open(args.channels_file, 'w') as configfile:
        config.write(configfile)


def updateCommand(args: argparse.Namespace) -> None:
    v = Verification()
    exprs: Dict[str, str] = {}
    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 section in config:
        if 'alias_of' in config[section]:
            exprs[section] = exprs[str(config[section]['alias_of'])]

    command = [
        'nix-env',
        '--profile',
        '/nix/var/nix/profiles/per-user/%s/channels' %
        getpass.getuser(),
        '--show-trace',
        '--file',
        '<nix/unpack-channel.nix>',
        '--install',
        '--from-expression'] + [exprs[name] % name for name in sorted(exprs.keys())]
    if args.dry_run:
        print(' '.join(map(shlex.quote, command)))
    else:
        v.status('Installing channels with nix-env')
        process = subprocess.run(command)
        v.result(process.returncode == 0)


def main() -> None:
    parser = argparse.ArgumentParser(prog='pinch')
    subparsers = parser.add_subparsers(dest='mode', required=True)
    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=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=updateCommand)
    args = parser.parse_args()
    args.func(args)


if __name__ == '__main__':
    main()