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     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     forwarded_url: str
  40     git_cachedir: str
  41     git_ref: str
  42     git_repo: str
  43     git_revision: str
  44     release_name: str
  45     table: Dict[str, ChannelTableEntry]
  46     url: str
  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.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.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
 214 def git_fetch(v: Verification, channel: Channel) -> None:
 215     # It would be nice if we could share the nix git cache, but as of the time
 216     # of writing it is transitioning from gitv2 (deprecated) to gitv3 (not ready
 217     # yet), and trying to straddle them both is too far into nix implementation
 218     # details for my comfort.  So we re-implement here half of nix.fetchGit.
 219     # :(
 220
 221     # TODO: Consider using pyxdg to find this path.
 222     channel.git_cachedir = os.path.expanduser(
 223         '~/.cache/nix-pin-channel/git/%s' %
 224         digest_string(
 225             channel.url.encode()))
 226     if not os.path.exists(channel.git_cachedir):
 227         v.status("Initializing git repo")
 228         process = subprocess.run(
 229             ['git', 'init', '--bare', channel.git_cachedir])
 230         v.result(process.returncode == 0)
 231
 232     v.status('Checking if we already have this rev:')
 233     process = subprocess.run(
 234         ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_revision])
 235     if process.returncode == 0:
 236         v.status('yes')
 237     if process.returncode == 1:
 238         v.status('no')
 239     v.result(process.returncode == 0 or process.returncode == 1)
 240     if process.returncode == 1:
 241         v.status('Fetching ref "%s"' % channel.git_ref)
 242         # We don't use --force here because we want to abort and freak out if forced
 243         # updates are happening.
 244         process = subprocess.run(['git',
 245                                   '-C',
 246                                   channel.git_cachedir,
 247                                   'fetch',
 248                                   channel.git_repo,
 249                                   '%s:%s' % (channel.git_ref,
 250                                              channel.git_ref)])
 251         v.result(process.returncode == 0)
 252         v.status('Verifying that fetch retrieved this rev')
 253         process = subprocess.run(
 254             ['git', '-C', channel.git_cachedir, 'cat-file', '-e', channel.git_revision])
 255         v.result(process.returncode == 0)
 256
 257     v.status('Verifying rev is an ancestor of ref')
 258     process = subprocess.run(['git',
 259                               '-C',
 260                               channel.git_cachedir,
 261                               'merge-base',
 262                               '--is-ancestor',
 263                               channel.git_revision,
 264                               channel.git_ref])
 265     v.result(process.returncode == 0)
 266
 267
 268 def compare_tarball_and_git(
 269         v: Verification,
 270         channel: Channel,
 271         channel_contents: str,
 272         git_contents: str) -> None:
 273     v.status('Comparing channel tarball with git checkout')
 274     match, mismatch, errors = compare(os.path.join(
 275         channel_contents, channel.release_name), git_contents)
 276     v.ok()
 277     v.check('%d files match' % len(match), len(match) > 0)
 278     v.check('%d files differ' % len(mismatch), len(mismatch) == 0)
 279     expected_errors = [
 280         '.git-revision',
 281         '.version-suffix',
 282         'nixpkgs',
 283         'programs.sqlite',
 284         'svn-revision']
 285     benign_errors = []
 286     for ee in expected_errors:
 287         if ee in errors:
 288             errors.remove(ee)
 289             benign_errors.append(ee)
 290     v.check(
 291         '%d unexpected incomparable files' %
 292         len(errors),
 293         len(errors) == 0)
 294     v.check(
 295         '(%d of %d expected incomparable files)' %
 296         (len(benign_errors),
 297          len(expected_errors)),
 298         len(benign_errors) == len(expected_errors))
 299
 300
 301 def extract_tarball(v: Verification, channel: Channel, dest: str) -> None:
 302     v.status('Extracting tarball %s' %
 303              channel.table['nixexprs.tar.xz'].file)
 304     shutil.unpack_archive(
 305         channel.table['nixexprs.tar.xz'].file,
 306         dest)
 307     v.ok()
 308
 309
 310 def git_checkout(v: Verification, channel: Channel, dest: str) -> None:
 311     v.status('Checking out corresponding git revision')
 312     git = subprocess.Popen(['git',
 313                             '-C',
 314                             channel.git_cachedir,
 315                             'archive',
 316                             channel.git_revision],
 317                            stdout=subprocess.PIPE)
 318     tar = subprocess.Popen(
 319         ['tar', 'x', '-C', dest, '-f', '-'], stdin=git.stdout)
 320     git.stdout.close()
 321     tar.wait()
 322     git.wait()
 323     v.result(git.returncode == 0 and tar.returncode == 0)
 324
 325
 326 def check_channel_metadata(
 327         v: Verification,
 328         channel: Channel,
 329         channel_contents: str) -> None:
 330     v.status('Verifying git commit in channel tarball')
 331     v.result(
 332         open(
 333             os.path.join(
 334                 channel_contents,
 335                 channel.release_name,
 336                 '.git-revision')).read(999) == channel.git_revision)
 337
 338     v.status(
 339         'Verifying version-suffix is a suffix of release name %s:' %
 340         channel.release_name)
 341     version_suffix = open(
 342         os.path.join(
 343             channel_contents,
 344             channel.release_name,
 345             '.version-suffix')).read(999)
 346     v.status(version_suffix)
 347     v.result(channel.release_name.endswith(version_suffix))
 348
 349
 350 def check_channel_contents(v: Verification, channel: Channel) -> None:
 351     with tempfile.TemporaryDirectory() as channel_contents, \
 352             tempfile.TemporaryDirectory() as git_contents:
 353
 354         extract_tarball(v, channel, channel_contents)
 355         check_channel_metadata(v, channel, channel_contents)
 356
 357         git_checkout(v, channel, git_contents)
 358
 359         compare_tarball_and_git(v, channel, channel_contents, git_contents)
 360
 361         v.status('Removing temporary directories')
 362     v.ok()
 363
 364
 365 def main(argv: List[str]) -> None:
 366     v = Verification()
 367     config = configparser.ConfigParser()
 368     config.read_file(open(argv[1]), argv[1])
 369     for section in config.sections():
 370         channel = Channel(**dict(config[section].items()))
 371         fetch(v, channel)
 372         parse_channel(v, channel)
 373         fetch_resources(v, channel)
 374         git_fetch(v, channel)
 375         check_channel_contents(v, channel)
 376         config[section]['git_rev'] = channel.git_revision
 377         config[section]['tarball_url'] = channel.table['nixexprs.tar.xz'].absolute_url
 378         config[section]['tarball_sha256'] = channel.table['nixexprs.tar.xz'].digest
 379     with open(argv[1], 'w') as configfile:
 380         config.write(configfile)
 381
 382
 383 main(sys.argv)