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