git.scottworley.com Git - pinch/blob - pinch.py

   1 import argparse
   2 import configparser
   3 import filecmp
   4 import functools
   5 import hashlib
   6 import operator
   7 import os
   8 import os.path
   9 import shutil
  10 import subprocess
  11 import tempfile
  12 import types
  13 import urllib.parse
  14 import urllib.request
  15 import xml.dom.minidom
  16
  17 from typing import (
  18     Dict,
  19     Iterable,
  20     List,
  21     NewType,
  22     Tuple,
  23 )
  24
  25 Digest16 = NewType('Digest16', str)
  26 Digest32 = NewType('Digest32', str)
  27
  28
  29 class ChannelTableEntry(types.SimpleNamespace):
  30     absolute_url: str
  31     digest: Digest16
  32     file: str
  33     size: int
  34     url: str
  35
  36
  37 class Channel(types.SimpleNamespace):
  38     channel_html: bytes
  39     channel_url: str
  40     forwarded_url: str
  41     git_cachedir: str
  42     git_ref: str
  43     git_repo: str
  44     git_revision: str
  45     old_git_revision: str
  46     release_name: str
  47     table: Dict[str, ChannelTableEntry]
  48
  49
  50 class VerificationError(Exception):
  51     pass
  52
  53
  54 class Verification:
  55
  56     def __init__(self) -> None:
  57         self.line_length = 0
  58
  59     def status(self, s: str) -> None:
  60         print(s, end=' ', flush=True)
  61         self.line_length += 1 + len(s)  # Unicode??
  62
  63     @staticmethod
  64     def _color(s: str, c: int) -> str:
  65         return '\033[%2dm%s\033[00m' % (c, s)
  66
  67     def result(self, r: bool) -> None:
  68         message, color = {True: ('OK ', 92), False: ('FAIL', 91)}[r]
  69         length = len(message)
  70         cols = shutil.get_terminal_size().columns
  71         pad = (cols - (self.line_length + length)) % cols
  72         print(' ' * pad + self._color(message, color))
  73         self.line_length = 0
  74         if not r:
  75             raise VerificationError()
  76
  77     def check(self, s: str, r: bool) -> None:
  78         self.status(s)
  79         self.result(r)
  80
  81     def ok(self) -> None:
  82         self.result(True)
  83
  84
  85 def compare(a: str, b: str) -> Tuple[List[str], List[str], List[str]]:
  86
  87     def throw(error: OSError) -> None:
  88         raise error
  89
  90     def join(x: str, y: str) -> str:
  91         return y if x == '.' else os.path.join(x, y)
  92
  93     def recursive_files(d: str) -> Iterable[str]:
  94         all_files: List[str] = []
  95         for path, dirs, files in os.walk(d, onerror=throw):
  96             rel = os.path.relpath(path, start=d)
  97             all_files.extend(join(rel, f) for f in files)
  98             for dir_or_link in dirs:
  99                 if os.path.islink(join(path, dir_or_link)):
 100                     all_files.append(join(rel, dir_or_link))
 101         return all_files
 102
 103     def exclude_dot_git(files: Iterable[str]) -> Iterable[str]:
 104         return (f for f in files if not f.startswith('.git/'))
 105
 106     files = functools.reduce(
 107         operator.or_, (set(
 108             exclude_dot_git(
 109                 recursive_files(x))) for x in [a, b]))
 110     return filecmp.cmpfiles(a, b, files, shallow=False)
 111
 112
 113 def fetch(v: Verification, channel: Channel) -> None:
 114     v.status('Fetching channel')
 115     request = urllib.request.urlopen(channel.channel_url, timeout=10)
 116     channel.channel_html = request.read()
 117     channel.forwarded_url = request.geturl()
 118     v.result(request.status == 200)
 119     v.check('Got forwarded', channel.channel_url != channel.forwarded_url)
 120
 121
 122 def parse_channel(v: Verification, channel: Channel) -> None:
 123     v.status('Parsing channel description as XML')
 124     d = xml.dom.minidom.parseString(channel.channel_html)
 125     v.ok()
 126
 127     v.status('Extracting release name:')
 128     title_name = d.getElementsByTagName(
 129         'title')[0].firstChild.nodeValue.split()[2]
 130     h1_name = d.getElementsByTagName('h1')[0].firstChild.nodeValue.split()[2]
 131     v.status(title_name)
 132     v.result(title_name == h1_name)
 133     channel.release_name = title_name
 134
 135     v.status('Extracting git commit:')
 136     git_commit_node = d.getElementsByTagName('tt')[0]
 137     channel.git_revision = git_commit_node.firstChild.nodeValue
 138     v.status(channel.git_revision)
 139     v.ok()
 140     v.status('Verifying git commit label')
 141     v.result(git_commit_node.previousSibling.nodeValue == 'Git commit ')
 142
 143     v.status('Parsing table')
 144     channel.table = {}
 145     for row in d.getElementsByTagName('tr')[1:]:
 146         name = row.childNodes[0].firstChild.firstChild.nodeValue
 147         url = row.childNodes[0].firstChild.getAttribute('href')
 148         size = int(row.childNodes[1].firstChild.nodeValue)
 149         digest = Digest16(row.childNodes[2].firstChild.firstChild.nodeValue)
 150         channel.table[name] = ChannelTableEntry(
 151             url=url, digest=digest, size=size)
 152     v.ok()
 153
 154
 155 def digest_string(s: bytes) -> Digest16:
 156     return Digest16(hashlib.sha256(s).hexdigest())
 157
 158
 159 def digest_file(filename: str) -> Digest16:
 160     hasher = hashlib.sha256()
 161     with open(filename, 'rb') as f:
 162         # pylint: disable=cell-var-from-loop
 163         for block in iter(lambda: f.read(4096), b''):
 164             hasher.update(block)
 165     return Digest16(hasher.hexdigest())
 166
 167
 168 def to_Digest16(v: Verification, digest32: Digest32) -> Digest16:
 169     v.status('Converting digest to base16')
 170     process = subprocess.run(
 171         ['nix', 'to-base16', '--type', 'sha256', digest32], capture_output=True)
 172     v.result(process.returncode == 0)
 173     return Digest16(process.stdout.decode().strip())
 174
 175
 176 def to_Digest32(v: Verification, digest16: Digest16) -> Digest32:
 177     v.status('Converting digest to base32')
 178     process = subprocess.run(
 179         ['nix', 'to-base32', '--type', 'sha256', digest16], capture_output=True)
 180     v.result(process.returncode == 0)
 181     return Digest32(process.stdout.decode().strip())
 182
 183
 184 def fetch_with_nix_prefetch_url(
 185         v: Verification,
 186         url: str,
 187         digest: Digest16) -> str:
 188     v.status('Fetching %s' % url)
 189     process = subprocess.run(
 190         ['nix-prefetch-url', '--print-path', url, digest], capture_output=True)
 191     v.result(process.returncode == 0)
 192     prefetch_digest, path, empty = process.stdout.decode().split('\n')
 193     assert empty == ''
 194     v.check("Verifying nix-prefetch-url's digest",
 195             to_Digest16(v, Digest32(prefetch_digest)) == digest)
 196     v.status("Verifying file digest")
 197     file_digest = digest_file(path)
 198     v.result(file_digest == digest)
 199     return path
 200
 201
 202 def fetch_resources(v: Verification, channel: Channel) -> None:
 203     for resource in ['git-revision', 'nixexprs.tar.xz']:
 204         fields = channel.table[resource]
 205         fields.absolute_url = urllib.parse.urljoin(
 206             channel.forwarded_url, fields.url)
 207         fields.file = fetch_with_nix_prefetch_url(
 208             v, fields.absolute_url, fields.digest)
 209     v.status('Verifying git commit on main page matches git commit in table')
 210     v.result(
 211         open(
 212             channel.table['git-revision'].file).read(999) == channel.git_revision)
 213
 214
 215 def git_fetch(v: Verification, channel: Channel) -> None:
 216     # It would be nice if we could share the nix git cache, but as of the time
 217     # of writing it is transitioning from gitv2 (deprecated) to gitv3 (not ready
 218     # yet), and trying to straddle them both is too far into nix implementation
 219     # details for my comfort.  So we re-implement here half of nix.fetchGit.
 220     # :(
 221
 222     # TODO: Consider using pyxdg to find this path.
 223     channel.git_cachedir = os.path.expanduser(
 224         '~/.cache/nix-pin-channel/git/%s' %
 225         digest_string(
 226             channel.git_repo.encode()))
 227     if not os.path.exists(channel.git_cachedir):
 228         v.status("Initializing git repo")
 229         process = subprocess.run(
 230             ['git', 'init', '--bare', channel.git_cachedir])
 231         v.result(process.returncode == 0)
 232
 233     have_rev = False
 234     if hasattr(channel, 'git_revision'):
 235         v.status('Checking if we already have this rev:')
 236         process = subprocess.run(
 237             ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_revision])
 238         if process.returncode == 0:
 239             v.status('yes')
 240         if process.returncode == 1:
 241             v.status('no')
 242         v.result(process.returncode == 0 or process.returncode == 1)
 243         have_rev = process.returncode == 0
 244
 245     if not have_rev:
 246         v.status(
 247             'Fetching ref "%s" from %s' %
 248             (channel.git_ref, channel.git_repo))
 249         # We don't use --force here because we want to abort and freak out if forced
 250         # updates are happening.
 251         process = subprocess.run(['git',
 252                                   '-C',
 253                                   channel.git_cachedir,
 254                                   'fetch',
 255                                   channel.git_repo,
 256                                   '%s:%s' % (channel.git_ref,
 257                                              channel.git_ref)])
 258         v.result(process.returncode == 0)
 259         if hasattr(channel, 'git_revision'):
 260             v.status('Verifying that fetch retrieved this rev')
 261             process = subprocess.run(
 262                 ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_revision])
 263             v.result(process.returncode == 0)
 264
 265     if not hasattr(channel, 'git_revision'):
 266         channel.git_revision = open(
 267             os.path.join(
 268                 channel.git_cachedir,
 269                 'refs',
 270                 'heads',
 271                 channel.git_ref)).read(999).strip()
 272
 273     v.status('Verifying rev is an ancestor of ref')
 274     process = subprocess.run(['git',
 275                               '-C',
 276                               channel.git_cachedir,
 277                               'merge-base',
 278                               '--is-ancestor',
 279                               channel.git_revision,
 280                               channel.git_ref])
 281     v.result(process.returncode == 0)
 282
 283     if hasattr(channel, 'old_git_revision'):
 284         v.status(
 285             'Verifying rev is an ancestor of previous rev %s' %
 286             channel.old_git_revision)
 287         process = subprocess.run(['git',
 288                                   '-C',
 289                                   channel.git_cachedir,
 290                                   'merge-base',
 291                                   '--is-ancestor',
 292                                   channel.old_git_revision,
 293                                   channel.git_revision])
 294         v.result(process.returncode == 0)
 295
 296
 297 def compare_tarball_and_git(
 298         v: Verification,
 299         channel: Channel,
 300         channel_contents: str,
 301         git_contents: str) -> None:
 302     v.status('Comparing channel tarball with git checkout')
 303     match, mismatch, errors = compare(os.path.join(
 304         channel_contents, channel.release_name), git_contents)
 305     v.ok()
 306     v.check('%d files match' % len(match), len(match) > 0)
 307     v.check('%d files differ' % len(mismatch), len(mismatch) == 0)
 308     expected_errors = [
 309         '.git-revision',
 310         '.version-suffix',
 311         'nixpkgs',
 312         'programs.sqlite',
 313         'svn-revision']
 314     benign_errors = []
 315     for ee in expected_errors:
 316         if ee in errors:
 317             errors.remove(ee)
 318             benign_errors.append(ee)
 319     v.check(
 320         '%d unexpected incomparable files' %
 321         len(errors),
 322         len(errors) == 0)
 323     v.check(
 324         '(%d of %d expected incomparable files)' %
 325         (len(benign_errors),
 326          len(expected_errors)),
 327         len(benign_errors) == len(expected_errors))
 328
 329
 330 def extract_tarball(v: Verification, channel: Channel, dest: str) -> None:
 331     v.status('Extracting tarball %s' %
 332              channel.table['nixexprs.tar.xz'].file)
 333     shutil.unpack_archive(
 334         channel.table['nixexprs.tar.xz'].file,
 335         dest)
 336     v.ok()
 337
 338
 339 def git_checkout(v: Verification, channel: Channel, dest: str) -> None:
 340     v.status('Checking out corresponding git revision')
 341     git = subprocess.Popen(['git',
 342                             '-C',
 343                             channel.git_cachedir,
 344                             'archive',
 345                             channel.git_revision],
 346                            stdout=subprocess.PIPE)
 347     tar = subprocess.Popen(
 348         ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout)
 349     git.stdout.close()
 350     tar.wait()
 351     git.wait()
 352     v.result(git.returncode == 0 and tar.returncode == 0)
 353
 354
 355 def check_channel_metadata(
 356         v: Verification,
 357         channel: Channel,
 358         channel_contents: str) -> None:
 359     v.status('Verifying git commit in channel tarball')
 360     v.result(
 361         open(
 362             os.path.join(
 363                 channel_contents,
 364                 channel.release_name,
 365                 '.git-revision')).read(999) == channel.git_revision)
 366
 367     v.status(
 368         'Verifying version-suffix is a suffix of release name %s:' %
 369         channel.release_name)
 370     version_suffix = open(
 371         os.path.join(
 372             channel_contents,
 373             channel.release_name,
 374             '.version-suffix')).read(999)
 375     v.status(version_suffix)
 376     v.result(channel.release_name.endswith(version_suffix))
 377
 378
 379 def check_channel_contents(v: Verification, channel: Channel) -> None:
 380     with tempfile.TemporaryDirectory() as channel_contents, \
 381             tempfile.TemporaryDirectory() as git_contents:
 382
 383         extract_tarball(v, channel, channel_contents)
 384         check_channel_metadata(v, channel, channel_contents)
 385
 386         git_checkout(v, channel, git_contents)
 387
 388         compare_tarball_and_git(v, channel, channel_contents, git_contents)
 389
 390         v.status('Removing temporary directories')
 391     v.ok()
 392
 393
 394 def pin_channel(v: Verification, channel: Channel) -> None:
 395     fetch(v, channel)
 396     parse_channel(v, channel)
 397     fetch_resources(v, channel)
 398     git_fetch(v, channel)
 399     check_channel_contents(v, channel)
 400
 401
 402 def git_revision_name(v: Verification, channel: Channel) -> str:
 403     v.status('Getting commit date')
 404     process = subprocess.run(['git',
 405                               '-C',
 406                               channel.git_cachedir,
 407                               'lo',
 408                               '-n1',
 409                               '--format=%ct-%h',
 410                               '--abbrev=11',
 411                               channel.git_revision],
 412                              capture_output=True)
 413     v.result(process.returncode == 0 and process.stdout != '')
 414     return '%s-%s' % (os.path.basename(channel.git_repo),
 415                       process.stdout.decode().strip())
 416
 417
 418 def make_channel(conf: configparser.SectionProxy) -> Channel:
 419     channel = Channel(**dict(conf.items()))
 420     if hasattr(channel, 'git_revision'):
 421         channel.old_git_revision = channel.git_revision
 422         del channel.git_revision
 423     return channel
 424
 425
 426 def pin(args: argparse.Namespace) -> None:
 427     v = Verification()
 428     config = configparser.ConfigParser()
 429     config.read_file(open(args.channels_file), args.channels_file)
 430     for section in config.sections():
 431         channel = make_channel(config[section])
 432         if 'channel_url' in config[section]:
 433             pin_channel(v, channel)
 434             config[section]['name'] = channel.release_name
 435             config[section]['tarball_url'] = channel.table['nixexprs.tar.xz'].absolute_url
 436             config[section]['tarball_sha256'] = channel.table['nixexprs.tar.xz'].digest
 437         else:
 438             git_fetch(v, channel)
 439             config[section]['name'] = git_revision_name(v, channel)
 440         config[section]['git_revision'] = channel.git_revision
 441
 442     with open(args.channels_file, 'w') as configfile:
 443         config.write(configfile)
 444
 445
 446 def main() -> None:
 447     parser = argparse.ArgumentParser(prog='pinch')
 448     subparsers = parser.add_subparsers(dest='mode', required=True)
 449     parser_pin = subparsers.add_parser('pin')
 450     parser_pin.add_argument('channels_file', type=str)
 451     parser_pin.set_defaults(func=pin)
 452     args = parser.parse_args()
 453     args.func(args)
 454
 455
 456 main()