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

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