Add support for Debian 10 (aka 'Buster') to run_upload_packages.py
[arvados-dev.git] / 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     TARGET_DISTNAMES = {
198         'debian8': 'jessie-dev',
199         'debian9': 'stretch-dev',
200         'debian10': 'buster-dev',
201         'ubuntu1404': 'trusty-dev',
202         'ubuntu1604': 'xenial-dev',
203         'ubuntu1804': 'bionic-dev',
204         }
205
206     def post_uploads(self, paths):
207         self._run_script(self.FREIGHT_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
208                          self.TARGET_DISTNAMES[self.target],
209                          *self._paths_basenames(paths))
210
211
212 class RedHatPackageSuite(DistroPackageSuite):
213     CREATEREPO_SCRIPT = """
214 cd "$1"; shift
215 REPODIR=$1; shift
216 rpmsign --addsign "$@" </dev/null
217 mv "$@" "$REPODIR"
218 createrepo "$REPODIR"
219 """
220     REPO_ROOT = '/var/www/rpm.arvados.org/'
221     TARGET_REPODIRS = {
222         'centos7': 'CentOS/7/dev/x86_64/',
223         }
224
225     def post_uploads(self, paths):
226         repo_dir = os.path.join(self.REPO_ROOT,
227                                 self.TARGET_REPODIRS[self.target])
228         self._run_script(self.CREATEREPO_SCRIPT, self.REMOTE_DEST_DIR + '/' + self.target,
229                          repo_dir, *self._paths_basenames(paths))
230
231
232 def _define_suite(suite_class, *rel_globs, **kwargs):
233     return functools.partial(suite_class, rel_globs=rel_globs, **kwargs)
234
235 PACKAGE_SUITES = {
236     'python': _define_suite(PythonPackageSuite,
237                             'sdk/pam/dist/*.tar.gz',
238                             'sdk/python/dist/*.tar.gz',
239                             'sdk/cwl/dist/*.tar.gz',
240                             'services/nodemanager/dist/*.tar.gz',
241                             'services/fuse/dist/*.tar.gz',
242                         ),
243     'gems': _define_suite(GemPackageSuite,
244                           'sdk/ruby/*.gem',
245                           'sdk/cli/*.gem',
246                           'services/login-sync/*.gem',
247                       ),
248     }
249 for target in ['debian8', 'debian9', 'debian10', 'ubuntu1404', 'ubuntu1604', 'ubuntu1804']:
250     PACKAGE_SUITES[target] = _define_suite(
251         DebianPackageSuite, os.path.join('packages', target, '*.deb'),
252         target=target)
253 for target in ['centos7']:
254     PACKAGE_SUITES[target] = _define_suite(
255         RedHatPackageSuite, os.path.join('packages', target, '*.rpm'),
256         target=target)
257
258 def parse_arguments(arguments):
259     parser = argparse.ArgumentParser(
260         prog="run_upload_packages.py",
261         description="Upload Arvados packages to various repositories")
262     parser.add_argument(
263         '--workspace', '-W', default=os.environ.get('WORKSPACE'),
264         help="Arvados source directory with built packages to upload")
265     parser.add_argument(
266         '--ssh-host', '-H',
267         help="Host specification for distribution repository server")
268     parser.add_argument('-o', action='append', default=[], dest='ssh_opts',
269                          metavar='OPTION', help="Pass option to `ssh -o`")
270     parser.add_argument('--verbose', '-v', action='count', default=0,
271                         help="Log more information and subcommand output")
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     for target in args.targets:
283         try:
284             suite_class = PACKAGE_SUITES[target].func
285         except KeyError:
286             parser.error("unrecognized target {!r}".format(target))
287         if suite_class.NEED_SSH and (args.ssh_host is None):
288             parser.error(
289                 "--ssh-host must be specified to upload distribution packages")
290     return args
291
292 def setup_logger(stream_dest, args):
293     log_handler = logging.StreamHandler(stream_dest)
294     log_handler.setFormatter(logging.Formatter(
295             '%(asctime)s %(name)s[%(process)d] %(levelname)s: %(message)s',
296             '%Y-%m-%d %H:%M:%S'))
297     logger = logging.getLogger('arvados-dev.upload')
298     logger.addHandler(log_handler)
299     logger.setLevel(max(1, logging.WARNING - (10 * args.verbose)))
300
301 def build_suite_and_upload(target, since_timestamp, args):
302     suite_def = PACKAGE_SUITES[target]
303     kwargs = {}
304     if suite_def.func.NEED_SSH:
305         kwargs.update(ssh_host=args.ssh_host, ssh_opts=args.ssh_opts)
306     suite = suite_def(args.workspace, **kwargs)
307     suite.update_packages(since_timestamp)
308
309 def main(arguments, stdout=sys.stdout, stderr=sys.stderr):
310     args = parse_arguments(arguments)
311     setup_logger(stderr, args)
312     for target in args.targets:
313         ts_file = TimestampFile(os.path.join(args.workspace, 'packages',
314                                              '.last_upload_%s' % target))
315         last_upload_ts = ts_file.last_upload()
316         build_suite_and_upload(target, last_upload_ts, args)
317         ts_file.update()
318
319 if __name__ == '__main__':
320     main(sys.argv[1:])