jenkins/run_upload_packages.py

   1 #!/usr/bin/env python3
   2
   3 # Copyright (C) The Arvados Authors. All rights reserved.
   4 #
   5 # SPDX-License-Identifier: AGPL-3.0
   6
   7 import argparse
   8 import errno
   9 import functools
  10 import glob
  11 import locale
  12 import logging
  13 import os
  14 import pipes
  15 import re
  16 import shutil
  17 import subprocess
  18 import sys
  19 import time
  20
  21 def run_and_grep(cmd, read_output, *regexps,
  22                  encoding=locale.getpreferredencoding(), **popen_kwargs):
  23     """Run a subprocess and capture output lines matching regexps.
  24
  25     Arguments:
  26     * cmd: The command to run, as a list or string, as for subprocess.Popen.
  27     * read_output: 'stdout' or 'stderr', the name of the output stream to read.
  28     Remaining arguments are regexps to match output, as strings or compiled
  29     regexp objects.  Output lines matching any regexp will be captured.
  30
  31     Keyword arguments:
  32     * encoding: The encoding used to decode the subprocess output.
  33     Remaining keyword arguments are passed directly to subprocess.Popen.
  34
  35     Returns 2-tuple (subprocess returncode, list of matched output lines).
  36     """
  37     regexps = [regexp if hasattr(regexp, 'search') else re.compile(regexp)
  38                for regexp in regexps]
  39     popen_kwargs[read_output] = subprocess.PIPE
  40     proc = subprocess.Popen(cmd, **popen_kwargs)
  41     with open(getattr(proc, read_output).fileno(), encoding=encoding) as output:
  42         matched_lines = []
  43         for line in output:
  44             if any(regexp.search(line) for regexp in regexps):
  45                 matched_lines.append(line)
  46             if read_output == 'stderr':
  47                 print(line, file=sys.stderr, end='')
  48     return proc.wait(), matched_lines
  49
  50
  51 class TimestampFile:
  52     def __init__(self, path):
  53         self.path = path
  54         # Make sure the dirname for `path` exists
  55         p = os.path.dirname(path)
  56         try:
  57             os.makedirs(p)
  58         except OSError as exc:
  59             if exc.errno == errno.EEXIST and os.path.isdir(p):
  60                 pass
  61             else:
  62                 raise
  63         self.start_time = time.time()
  64
  65     def last_upload(self):
  66         try:
  67             return os.path.getmtime(self.path)
  68         except EnvironmentError:
  69             return -1
  70
  71     def update(self):
  72         try:
  73             os.close(os.open(self.path, os.O_CREAT | os.O_APPEND))
  74             os.utime(self.path, (time.time(), self.start_time))
  75         except:
  76             # when the packages directory is created/populated by a build in a
  77             # docker container, as root, the script that runs the upload
  78             # doesn't always have permission to touch a timestamp file there.
  79             # In production, we build/upload from ephemeral machines, which
  80             # means that the timestamp mechanism is not used. We print a
  81             # warning and move on without erroring out.
  82             print("Warning: unable to update timestamp file",self.path,"permission problem?")
  83             pass
  84
  85 class PackageSuite:
  86     NEED_SSH = False
  87
  88     def __init__(self, glob_root, rel_globs):
  89         logger_part = getattr(self, 'LOGGER_PART', os.path.basename(glob_root))
  90         self.logger = logging.getLogger('arvados-dev.upload.' + logger_part)
  91         self.globs = [os.path.join(glob_root, rel_glob)
  92                       for rel_glob in rel_globs]
  93
  94     def files_to_upload(self, since_timestamp):
  95         for abs_glob in self.globs:
  96             for path in glob.glob(abs_glob):
  97                 if os.path.getmtime(path) >= since_timestamp:
  98                     yield path
  99
 100     def upload_file(self, path):
 101         raise NotImplementedError("PackageSuite.upload_file")
 102
 103     def upload_files(self, paths):
 104         for path in paths:
 105             self.logger.info("Uploading %s", path)
 106             self.upload_file(path)
 107
 108     def post_uploads(self, paths):
 109         pass
 110
 111     def update_packages(self, since_timestamp):
 112         upload_paths = list(self.files_to_upload(since_timestamp))
 113         if upload_paths:
 114             self.upload_files(upload_paths)
 115             self.post_uploads(upload_paths)
 116
 117
 118 class PythonPackageSuite(PackageSuite):
 119     LOGGER_PART = 'python'
 120     REUPLOAD_REGEXPS = [
 121         re.compile(
 122             r'^error: Upload failed \(400\): A file named "[^"]+" already exists\b'),
 123         re.compile(
 124             r'^error: Upload failed \(400\): File already exists\b'),
 125         re.compile(
 126             r'^error: Upload failed \(400\): Only one sdist may be uploaded per release\b'),
 127     ]
 128
 129     def __init__(self, glob_root, rel_globs):
 130         super().__init__(glob_root, rel_globs)
 131         self.seen_packages = set()
 132
 133     def upload_file(self, path):
 134         src_dir = os.path.dirname(os.path.dirname(path))
 135         if src_dir in self.seen_packages:
 136             return
 137         self.seen_packages.add(src_dir)
 138         # We also must run `sdist` before `upload`: `upload` uploads any
 139         # distributions previously generated in the command.  It doesn't
 140         # know how to upload distributions already on disk.  We write the
 141         # result to a dedicated directory to avoid interfering with our
 142         # timestamp tracking.
 143         cmd = ['python3', 'setup.py']
 144         if not self.logger.isEnabledFor(logging.INFO):
 145             cmd.append('--quiet')
 146         cmd.extend(['bdist_wheel', '--dist-dir', '.upload_dist'])
 147         cmd.extend(['sdist', '--dist-dir', '.upload_dist'])
 148         cmd.extend(['upload'])
 149         upload_returncode, repushed = run_and_grep(
 150             cmd, 'stderr', *self.REUPLOAD_REGEXPS, cwd=src_dir)
 151         if (upload_returncode != 0) and not repushed:
 152             raise subprocess.CalledProcessError(upload_returncode, cmd)
 153         shutil.rmtree(os.path.join(src_dir, '.upload_dist'))
 154
 155
 156 class GemPackageSuite(PackageSuite):
 157     LOGGER_PART = 'gems'
 158     REUPLOAD_REGEXP = re.compile(r'^Repushing of gem versions is not allowed\.$')
 159
 160     def upload_file(self, path):
 161         cmd = ['gem', 'push', path]
 162         push_returncode, repushed = run_and_grep(cmd, 'stdout', self.REUPLOAD_REGEXP)
 163         if (push_returncode != 0) and not repushed:
 164             raise subprocess.CalledProcessError(push_returncode, cmd)
 165
 166
 167 class DistroPackageSuite(PackageSuite):
 168     NEED_SSH = True
 169     REMOTE_DEST_DIR = 'tmp'
 170
 171     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts):
 172         super().__init__(glob_root, rel_globs)
 173         self.target = target
 174         self.ssh_host = ssh_host
 175         self.ssh_opts = ['-o' + opt for opt in ssh_opts]
 176         if not self.logger.isEnabledFor(logging.INFO):
 177             self.ssh_opts.append('-q')
 178
 179     def _build_cmd(self, base_cmd, *args):
 180         cmd = [base_cmd]
 181         cmd.extend(self.ssh_opts)
 182         cmd.extend(args)
 183         return cmd
 184
 185     def _paths_basenames(self, paths):
 186         return (os.path.basename(path) for path in paths)
 187
 188     def _run_script(self, script, *args):
 189         # SSH will use a shell to run our bash command, so we have to
 190         # quote our arguments.
 191         # self.__class__.__name__ provides $0 for the script, which makes a
 192         # nicer message if there's an error.
 193         subprocess.check_call(self._build_cmd(
 194                 'ssh', self.ssh_host, 'bash', '-ec', pipes.quote(script),
 195                 self.__class__.__name__, *(pipes.quote(s) for s in args)))
 196
 197     def upload_files(self, paths):
 198         dest_dir = os.path.join(self.REMOTE_DEST_DIR, self.target)
 199         mkdir = self._build_cmd('ssh', self.ssh_host, 'install', '-d', dest_dir)
 200         subprocess.check_call(mkdir)
 201         cmd = self._build_cmd('scp', *paths)
 202         cmd.append('{}:{}'.format(self.ssh_host, dest_dir))
 203         subprocess.check_call(cmd)
 204
 205
 206 class DebianPackageSuite(DistroPackageSuite):
 207     APT_SCRIPT = """
 208 cd "$1"; shift
 209 DISTNAME=$1; shift
 210 for package in "$@"; do
 211   set +e
 212   aptly repo search "$DISTNAME" "${package%.deb}" >/dev/null 2>&1
 213   RET=$?
 214   set -e
 215   if [[ $RET -eq 0 ]]; then
 216     echo "Not adding $package, it is already present in repo $DISTNAME"
 217     rm "$package"
 218   else
 219     aptly repo add -remove-files "$DISTNAME" "$package"
 220   fi
 221 done
 222 aptly publish update "$DISTNAME" filesystem:"${DISTNAME%-*}":
 223 """
 224
 225     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts, repo):
 226         super().__init__(glob_root, rel_globs, target, ssh_host, ssh_opts)
 227         self.TARGET_DISTNAMES = {
 228             'debian10': 'buster-'+repo,
 229             'debian11': 'bullseye-'+repo,
 230             'ubuntu1804': 'bionic-'+repo,
 231             'ubuntu2004': 'focal-'+repo,
 232             }
 233
 234     def post_uploads(self, paths):
 235         self._run_script(self.APT_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
 236                          self.TARGET_DISTNAMES[self.target],
 237                          *self._paths_basenames(paths))
 238
 239
 240 class RedHatPackageSuite(DistroPackageSuite):
 241     CREATEREPO_SCRIPT = """
 242 cd "$1"; shift
 243 REPODIR=$1; shift
 244 rpmsign --addsign "$@" </dev/null
 245 mv "$@" "$REPODIR"
 246 createrepo -c ~/.createrepo-cache --update "$REPODIR"
 247 """
 248     REPO_ROOT = '/var/www/rpm.arvados.org/'
 249
 250     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts, repo):
 251         super().__init__(glob_root, rel_globs, target, ssh_host, ssh_opts)
 252         self.TARGET_REPODIRS = {
 253             'centos7': 'CentOS/7/%s/x86_64/' % repo,
 254         }
 255
 256     def post_uploads(self, paths):
 257         repo_dir = os.path.join(self.REPO_ROOT,
 258                                 self.TARGET_REPODIRS[self.target])
 259         self._run_script(self.CREATEREPO_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
 260                          repo_dir, *self._paths_basenames(paths))
 261
 262
 263 def _define_suite(suite_class, *rel_globs, **kwargs):
 264     return functools.partial(suite_class, rel_globs=rel_globs, **kwargs)
 265
 266 PACKAGE_SUITES = {
 267     'python': _define_suite(PythonPackageSuite,
 268                             'sdk/python/dist/*.tar.gz',
 269                             'sdk/cwl/dist/*.tar.gz',
 270                             'services/fuse/dist/*.tar.gz',
 271                             'tools/crunchstat-summary/dist/*.tar.gz',
 272                         ),
 273     'gems': _define_suite(GemPackageSuite,
 274                           'sdk/ruby/*.gem',
 275                           'sdk/cli/*.gem',
 276                           'services/login-sync/*.gem',
 277                       ),
 278     }
 279
 280 def parse_arguments(arguments):
 281     parser = argparse.ArgumentParser(
 282         description="Upload Arvados packages to various repositories")
 283     parser.add_argument(
 284         '--workspace', '-W', default=os.environ.get('WORKSPACE'),
 285         help="Arvados source directory with built packages to upload")
 286     parser.add_argument(
 287         '--ssh-host', '-H',
 288         help="Host specification for distribution repository server")
 289     parser.add_argument('-o', action='append', default=[], dest='ssh_opts',
 290                          metavar='OPTION', help="Pass option to `ssh -o`")
 291     parser.add_argument('--verbose', '-v', action='count', default=0,
 292                         help="Log more information and subcommand output")
 293     parser.add_argument(
 294         '--repo', choices=['dev', 'testing'],
 295         help="Whether to upload to dev (nightly) or testing (release candidate) repository")
 296
 297     parser.add_argument(
 298         'targets', nargs='*', default=['all'], metavar='target',
 299         help="Upload packages to these targets (default all)\nAvailable targets: " +
 300         ', '.join(sorted(PACKAGE_SUITES.keys())))
 301     args = parser.parse_args(arguments)
 302     if 'all' in args.targets:
 303         args.targets = list(PACKAGE_SUITES.keys())
 304
 305     if args.workspace is None:
 306         parser.error("workspace not set from command line or environment")
 307
 308     for target in ['debian10', 'debian11', 'ubuntu1804', 'ubuntu2004']:
 309         PACKAGE_SUITES[target] = _define_suite(
 310             DebianPackageSuite, os.path.join('packages', target, '*.deb'),
 311             target=target, repo=args.repo)
 312     for target in ['centos7']:
 313         PACKAGE_SUITES[target] = _define_suite(
 314             RedHatPackageSuite, os.path.join('packages', target, '*.rpm'),
 315             target=target, repo=args.repo)
 316
 317     for target in args.targets:
 318         try:
 319             suite_class = PACKAGE_SUITES[target].func
 320         except KeyError:
 321             parser.error("unrecognized target {!r}".format(target))
 322         if suite_class.NEED_SSH and (args.ssh_host is None):
 323             parser.error(
 324                 "--ssh-host must be specified to upload distribution packages")
 325     return args
 326
 327 def setup_logger(stream_dest, args):
 328     log_handler = logging.StreamHandler(stream_dest)
 329     log_handler.setFormatter(logging.Formatter(
 330             '%(asctime)s %(name)s[%(process)d] %(levelname)s: %(message)s',
 331             '%Y-%m-%d %H:%M:%S'))
 332     logger = logging.getLogger('arvados-dev.upload')
 333     logger.addHandler(log_handler)
 334     logger.setLevel(max(1, logging.WARNING - (10 * args.verbose)))
 335
 336 def build_suite_and_upload(target, since_timestamp, args):
 337     suite_def = PACKAGE_SUITES[target]
 338     kwargs = {}
 339     if suite_def.func.NEED_SSH:
 340         kwargs.update(ssh_host=args.ssh_host, ssh_opts=args.ssh_opts)
 341     suite = suite_def(args.workspace, **kwargs)
 342     suite.update_packages(since_timestamp)
 343
 344 def main(arguments, stdout=sys.stdout, stderr=sys.stderr):
 345     args = parse_arguments(arguments)
 346     setup_logger(stderr, args)
 347
 348     for target in args.targets:
 349         ts_file = TimestampFile(os.path.join(args.workspace, 'packages',
 350                                              '.last_upload_%s' % target))
 351         last_upload_ts = ts_file.last_upload()
 352         build_suite_and_upload(target, last_upload_ts, args)
 353         ts_file.update()
 354
 355 if __name__ == '__main__':
 356     main(sys.argv[1:])