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 functools
   9 import glob
  10 import locale
  11 import logging
  12 import os
  13 import pipes
  14 import re
  15 import shutil
  16 import subprocess
  17 import sys
  18 import time
  19
  20 def run_and_grep(cmd, read_output, *regexps,
  21                  encoding=locale.getpreferredencoding(), **popen_kwargs):
  22     """Run a subprocess and capture output lines matching regexps.
  23
  24     Arguments:
  25     * cmd: The command to run, as a list or string, as for subprocess.Popen.
  26     * read_output: 'stdout' or 'stderr', the name of the output stream to read.
  27     Remaining arguments are regexps to match output, as strings or compiled
  28     regexp objects.  Output lines matching any regexp will be captured.
  29
  30     Keyword arguments:
  31     * encoding: The encoding used to decode the subprocess output.
  32     Remaining keyword arguments are passed directly to subprocess.Popen.
  33
  34     Returns 2-tuple (subprocess returncode, list of matched output lines).
  35     """
  36     regexps = [regexp if hasattr(regexp, 'search') else re.compile(regexp)
  37                for regexp in regexps]
  38     popen_kwargs[read_output] = subprocess.PIPE
  39     proc = subprocess.Popen(cmd, **popen_kwargs)
  40     with open(getattr(proc, read_output).fileno(), encoding=encoding) as output:
  41         matched_lines = []
  42         for line in output:
  43             if any(regexp.search(line) for regexp in regexps):
  44                 matched_lines.append(line)
  45             if read_output == 'stderr':
  46                 print(line, file=sys.stderr, end='')
  47     return proc.wait(), matched_lines
  48
  49
  50 class TimestampFile:
  51     def __init__(self, path):
  52         self.path = path
  53         self.start_time = time.time()
  54
  55     def last_upload(self):
  56         try:
  57             return os.path.getmtime(self.path)
  58         except EnvironmentError:
  59             return -1
  60
  61     def update(self):
  62         os.close(os.open(self.path, os.O_CREAT | os.O_APPEND))
  63         os.utime(self.path, (time.time(), self.start_time))
  64
  65
  66 class PackageSuite:
  67     NEED_SSH = False
  68
  69     def __init__(self, glob_root, rel_globs):
  70         logger_part = getattr(self, 'LOGGER_PART', os.path.basename(glob_root))
  71         self.logger = logging.getLogger('arvados-dev.upload.' + logger_part)
  72         self.globs = [os.path.join(glob_root, rel_glob)
  73                       for rel_glob in rel_globs]
  74
  75     def files_to_upload(self, since_timestamp):
  76         for abs_glob in self.globs:
  77             for path in glob.glob(abs_glob):
  78                 if os.path.getmtime(path) >= since_timestamp:
  79                     yield path
  80
  81     def upload_file(self, path):
  82         raise NotImplementedError("PackageSuite.upload_file")
  83
  84     def upload_files(self, paths):
  85         for path in paths:
  86             self.logger.info("Uploading %s", path)
  87             self.upload_file(path)
  88
  89     def post_uploads(self, paths):
  90         pass
  91
  92     def update_packages(self, since_timestamp):
  93         upload_paths = list(self.files_to_upload(since_timestamp))
  94         if upload_paths:
  95             self.upload_files(upload_paths)
  96             self.post_uploads(upload_paths)
  97
  98
  99 class PythonPackageSuite(PackageSuite):
 100     LOGGER_PART = 'python'
 101     REUPLOAD_REGEXPS = [
 102         re.compile(
 103             r'^error: Upload failed \(400\): A file named "[^"]+" already exists\b'),
 104         re.compile(
 105             r'^error: Upload failed \(400\): File already exists\b'),
 106         re.compile(
 107             r'^error: Upload failed \(400\): Only one sdist may be uploaded per release\b'),
 108     ]
 109
 110     def __init__(self, glob_root, rel_globs):
 111         super().__init__(glob_root, rel_globs)
 112         self.seen_packages = set()
 113
 114     def upload_file(self, path):
 115         src_dir = os.path.dirname(os.path.dirname(path))
 116         if src_dir in self.seen_packages:
 117             return
 118         self.seen_packages.add(src_dir)
 119         # NOTE: If we ever start uploading Python 3 packages, we'll need to
 120         # figure out some way to adapt cmd to match.  It might be easiest
 121         # to give all our setup.py files the executable bit, and run that
 122         # directly.
 123         # We also must run `sdist` before `upload`: `upload` uploads any
 124         # distributions previously generated in the command.  It doesn't
 125         # know how to upload distributions already on disk.  We write the
 126         # result to a dedicated directory to avoid interfering with our
 127         # timestamp tracking.
 128         cmd = ['python2.7', 'setup.py']
 129         if not self.logger.isEnabledFor(logging.INFO):
 130             cmd.append('--quiet')
 131         cmd.extend(['sdist', '--dist-dir', '.upload_dist', 'upload'])
 132         upload_returncode, repushed = run_and_grep(
 133             cmd, 'stderr', *self.REUPLOAD_REGEXPS, cwd=src_dir)
 134         if (upload_returncode != 0) and not repushed:
 135             raise subprocess.CalledProcessError(upload_returncode, cmd)
 136         shutil.rmtree(os.path.join(src_dir, '.upload_dist'))
 137
 138
 139 class GemPackageSuite(PackageSuite):
 140     LOGGER_PART = 'gems'
 141     REUPLOAD_REGEXP = re.compile(r'^Repushing of gem versions is not allowed\.$')
 142
 143     def upload_file(self, path):
 144         cmd = ['gem', 'push', path]
 145         push_returncode, repushed = run_and_grep(cmd, 'stdout', self.REUPLOAD_REGEXP)
 146         if (push_returncode != 0) and not repushed:
 147             raise subprocess.CalledProcessError(push_returncode, cmd)
 148
 149
 150 class DistroPackageSuite(PackageSuite):
 151     NEED_SSH = True
 152     REMOTE_DEST_DIR = 'tmp'
 153
 154     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts):
 155         super().__init__(glob_root, rel_globs)
 156         self.target = target
 157         self.ssh_host = ssh_host
 158         self.ssh_opts = ['-o' + opt for opt in ssh_opts]
 159         if not self.logger.isEnabledFor(logging.INFO):
 160             self.ssh_opts.append('-q')
 161
 162     def _build_cmd(self, base_cmd, *args):
 163         cmd = [base_cmd]
 164         cmd.extend(self.ssh_opts)
 165         cmd.extend(args)
 166         return cmd
 167
 168     def _paths_basenames(self, paths):
 169         return (os.path.basename(path) for path in paths)
 170
 171     def _run_script(self, script, *args):
 172         # SSH will use a shell to run our bash command, so we have to
 173         # quote our arguments.
 174         # self.__class__.__name__ provides $0 for the script, which makes a
 175         # nicer message if there's an error.
 176         subprocess.check_call(self._build_cmd(
 177                 'ssh', self.ssh_host, 'bash', '-ec', pipes.quote(script),
 178                 self.__class__.__name__, *(pipes.quote(s) for s in args)))
 179
 180     def upload_files(self, paths):
 181         dest_dir = os.path.join(self.REMOTE_DEST_DIR, self.target)
 182         mkdir = self._build_cmd('ssh', self.ssh_host, 'install', '-d', dest_dir)
 183         subprocess.check_call(mkdir)
 184         cmd = self._build_cmd('scp', *paths)
 185         cmd.append('{}:{}'.format(self.ssh_host, dest_dir))
 186         subprocess.check_call(cmd)
 187
 188
 189 class DebianPackageSuite(DistroPackageSuite):
 190     FREIGHT_SCRIPT = """
 191 cd "$1"; shift
 192 DISTNAME=$1; shift
 193 freight add "$@" "apt/$DISTNAME"
 194 freight cache "apt/$DISTNAME"
 195 rm "$@"
 196 """
 197
 198     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts, repo):
 199         super().__init__(glob_root, rel_globs, target, ssh_host, ssh_opts)
 200         self.TARGET_DISTNAMES = {
 201             'debian8': 'jessie-'+repo,
 202             'debian9': 'stretch-'+repo,
 203             'debian10': 'buster-'+repo,
 204             'ubuntu1404': 'trusty-'+repo,
 205             'ubuntu1604': 'xenial-'+repo,
 206             'ubuntu1804': 'bionic-'+repo,
 207             'ubuntu2004': 'focal-'+repo,
 208             }
 209
 210     def post_uploads(self, paths):
 211         self._run_script(self.FREIGHT_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
 212                          self.TARGET_DISTNAMES[self.target],
 213                          *self._paths_basenames(paths))
 214
 215
 216 class RedHatPackageSuite(DistroPackageSuite):
 217     CREATEREPO_SCRIPT = """
 218 cd "$1"; shift
 219 REPODIR=$1; shift
 220 rpmsign --addsign "$@" </dev/null
 221 mv "$@" "$REPODIR"
 222 createrepo "$REPODIR"
 223 """
 224     REPO_ROOT = '/var/www/rpm.arvados.org/'
 225
 226     def __init__(self, glob_root, rel_globs, target, ssh_host, ssh_opts, repo):
 227         super().__init__(glob_root, rel_globs, target, ssh_host, ssh_opts)
 228         self.TARGET_REPODIRS = {
 229             'centos7': 'CentOS/7/%s/x86_64/' % repo,
 230         }
 231
 232     def post_uploads(self, paths):
 233         repo_dir = os.path.join(self.REPO_ROOT,
 234                                 self.TARGET_REPODIRS[self.target])
 235         self._run_script(self.CREATEREPO_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
 236                          repo_dir, *self._paths_basenames(paths))
 237
 238
 239 def _define_suite(suite_class, *rel_globs, **kwargs):
 240     return functools.partial(suite_class, rel_globs=rel_globs, **kwargs)
 241
 242 PACKAGE_SUITES = {
 243     'python': _define_suite(PythonPackageSuite,
 244                             'sdk/python/dist/*.tar.gz',
 245                             'sdk/cwl/dist/*.tar.gz',
 246                             'services/fuse/dist/*.tar.gz',
 247                         ),
 248     'gems': _define_suite(GemPackageSuite,
 249                           'sdk/ruby/*.gem',
 250                           'sdk/cli/*.gem',
 251                           'services/login-sync/*.gem',
 252                       ),
 253     }
 254
 255 def parse_arguments(arguments):
 256     parser = argparse.ArgumentParser(
 257         description="Upload Arvados packages to various repositories")
 258     parser.add_argument(
 259         '--workspace', '-W', default=os.environ.get('WORKSPACE'),
 260         help="Arvados source directory with built packages to upload")
 261     parser.add_argument(
 262         '--ssh-host', '-H',
 263         help="Host specification for distribution repository server")
 264     parser.add_argument('-o', action='append', default=[], dest='ssh_opts',
 265                          metavar='OPTION', help="Pass option to `ssh -o`")
 266     parser.add_argument('--verbose', '-v', action='count', default=0,
 267                         help="Log more information and subcommand output")
 268     parser.add_argument(
 269         '--repo', choices=['dev', 'testing'],
 270         help="Whether to upload to dev (nightly) or testing (release candidate) repository")
 271
 272     parser.add_argument(
 273         'targets', nargs='*', default=['all'], metavar='target',
 274         help="Upload packages to these targets (default all)\nAvailable targets: " +
 275         ', '.join(sorted(PACKAGE_SUITES.keys())))
 276     args = parser.parse_args(arguments)
 277     if 'all' in args.targets:
 278         args.targets = list(PACKAGE_SUITES.keys())
 279
 280     if args.workspace is None:
 281         parser.error("workspace not set from command line or environment")
 282
 283     for target in ['debian8', 'debian9', 'debian10', 'ubuntu1404', 'ubuntu1604', 'ubuntu1804', 'ubuntu2004']:
 284         PACKAGE_SUITES[target] = _define_suite(
 285             DebianPackageSuite, os.path.join('packages', target, '*.deb'),
 286             target=target, repo=args.repo)
 287     for target in ['centos7']:
 288         PACKAGE_SUITES[target] = _define_suite(
 289             RedHatPackageSuite, os.path.join('packages', target, '*.rpm'),
 290             target=target, repo=args.repo)
 291
 292     for target in args.targets:
 293         try:
 294             suite_class = PACKAGE_SUITES[target].func
 295         except KeyError:
 296             parser.error("unrecognized target {!r}".format(target))
 297         if suite_class.NEED_SSH and (args.ssh_host is None):
 298             parser.error(
 299                 "--ssh-host must be specified to upload distribution packages")
 300     return args
 301
 302 def setup_logger(stream_dest, args):
 303     log_handler = logging.StreamHandler(stream_dest)
 304     log_handler.setFormatter(logging.Formatter(
 305             '%(asctime)s %(name)s[%(process)d] %(levelname)s: %(message)s',
 306             '%Y-%m-%d %H:%M:%S'))
 307     logger = logging.getLogger('arvados-dev.upload')
 308     logger.addHandler(log_handler)
 309     logger.setLevel(max(1, logging.WARNING - (10 * args.verbose)))
 310
 311 def build_suite_and_upload(target, since_timestamp, args):
 312     suite_def = PACKAGE_SUITES[target]
 313     kwargs = {}
 314     if suite_def.func.NEED_SSH:
 315         kwargs.update(ssh_host=args.ssh_host, ssh_opts=args.ssh_opts)
 316     suite = suite_def(args.workspace, **kwargs)
 317     suite.update_packages(since_timestamp)
 318
 319 def main(arguments, stdout=sys.stdout, stderr=sys.stderr):
 320     args = parse_arguments(arguments)
 321     setup_logger(stderr, args)
 322
 323     for target in args.targets:
 324         ts_file = TimestampFile(os.path.join(args.workspace, 'packages',
 325                                              '.last_upload_%s' % target))
 326         last_upload_ts = ts_file.last_upload()
 327         build_suite_and_upload(target, last_upload_ts, args)
 328         ts_file.update()
 329
 330 if __name__ == '__main__':
 331     main(sys.argv[1:])