Files
tubestation/taskcluster/taskgraph/util/docker.py
Mike Hommey c7945d8ea1 Bug 1341214 - Add a small API to handle taskcluster queue and index requests. r=dustin
Various modules under taskcluster are doing ad-hoc url formatting or
requests to taskcluster services. While we could use the taskcluster
client python module, it's kind of overkill for the simple requests done
here. So instead of vendoring that module, create a smaller one with
a limited set of functions we need.

This changes the behavior of the get_artifact function to return a
file-like object when the file is neither a json nor a yaml, but that
branch was never used (and was actually returning an unassigned
variable, so it was broken anyways).

At the same time, make the function that does HTTP requests more
error-resistant, using urllib3's Retry with a backoff factor.

Also add a function that retrieves the list of artifacts, that while
currently unused, will be used by `mach artifact` shortly.
2017-02-17 12:04:48 +09:00

166 lines
5.5 KiB
Python

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
from __future__ import absolute_import, print_function, unicode_literals
import hashlib
import os
import shutil
import subprocess
import tarfile
import tempfile
from mozpack.archive import (
create_tar_gz_from_files,
)
from .. import GECKO
IMAGE_DIR = os.path.join(GECKO, 'taskcluster', 'docker')
INDEX_PREFIX = 'docker.images.v2'
def docker_image(name, by_tag=False):
'''
Resolve in-tree prebuilt docker image to ``<registry>/<repository>@sha256:<digest>``,
or ``<registry>/<repository>:<tag>`` if `by_tag` is `True`.
'''
try:
with open(os.path.join(IMAGE_DIR, name, 'REGISTRY')) as f:
registry = f.read().strip()
except IOError:
with open(os.path.join(IMAGE_DIR, 'REGISTRY')) as f:
registry = f.read().strip()
if not by_tag:
hashfile = os.path.join(IMAGE_DIR, name, 'HASH')
try:
with open(hashfile) as f:
return '{}/{}@{}'.format(registry, name, f.read().strip())
except IOError:
raise Exception('Failed to read HASH file {}'.format(hashfile))
try:
with open(os.path.join(IMAGE_DIR, name, 'VERSION')) as f:
tag = f.read().strip()
except IOError:
tag = 'latest'
return '{}/{}:{}'.format(registry, name, tag)
def generate_context_hash(topsrcdir, image_path, image_name):
"""Generates a sha256 hash for context directory used to build an image."""
# It is a bit unfortunate we have to create a temp file here - it would
# be nicer to use an in-memory buffer.
fd, p = tempfile.mkstemp()
os.close(fd)
try:
return create_context_tar(topsrcdir, image_path, p, image_name)
finally:
os.unlink(p)
def create_context_tar(topsrcdir, context_dir, out_path, prefix):
"""Create a context tarball.
A directory ``context_dir`` containing a Dockerfile will be assembled into
a gzipped tar file at ``out_path``. Files inside the archive will be
prefixed by directory ``prefix``.
We also scan the source Dockerfile for special syntax that influences
context generation.
If a line in the Dockerfile has the form ``# %include <path>``,
the relative path specified on that line will be matched against
files in the source repository and added to the context under the
path ``topsrcdir/``. If an entry is a directory, we add all files
under that directory.
Returns the SHA-256 hex digest of the created archive.
"""
archive_files = {}
for root, dirs, files in os.walk(context_dir):
for f in files:
source_path = os.path.join(root, f)
rel = source_path[len(context_dir) + 1:]
archive_path = os.path.join(prefix, rel)
archive_files[archive_path] = source_path
# Parse Dockerfile for special syntax of extra files to include.
with open(os.path.join(context_dir, 'Dockerfile'), 'rb') as fh:
for line in fh:
line = line.rstrip()
if not line.startswith('# %include'):
continue
p = line[len('# %include '):].strip()
if os.path.isabs(p):
raise Exception('extra include path cannot be absolute: %s' % p)
fs_path = os.path.normpath(os.path.join(topsrcdir, p))
# Check for filesystem traversal exploits.
if not fs_path.startswith(topsrcdir):
raise Exception('extra include path outside topsrcdir: %s' % p)
if not os.path.exists(fs_path):
raise Exception('extra include path does not exist: %s' % p)
if os.path.isdir(fs_path):
for root, dirs, files in os.walk(fs_path):
for f in files:
source_path = os.path.join(root, f)
archive_path = os.path.join(prefix, 'topsrcdir', p, f)
archive_files[archive_path] = source_path
else:
archive_path = os.path.join(prefix, 'topsrcdir', p)
archive_files[archive_path] = fs_path
with open(out_path, 'wb') as fh:
create_tar_gz_from_files(fh, archive_files, '%s.tar.gz' % prefix)
h = hashlib.sha256()
with open(out_path, 'rb') as fh:
while True:
data = fh.read(32768)
if not data:
break
h.update(data)
return h.hexdigest()
def build_from_context(docker_bin, context_path, prefix, tag=None):
"""Build a Docker image from a context archive.
Given the path to a `docker` binary, a image build tar.gz (produced with
``create_context_tar()``, a prefix in that context containing files, and
an optional ``tag`` for the produced image, build that Docker image.
"""
d = tempfile.mkdtemp()
try:
with tarfile.open(context_path, 'r:gz') as tf:
tf.extractall(d)
# If we wanted to do post-processing of the Dockerfile, this is
# where we'd do it.
args = [
docker_bin,
'build',
# Use --no-cache so we always get the latest package updates.
'--no-cache',
]
if tag:
args.extend(['-t', tag])
args.append('.')
res = subprocess.call(args, cwd=os.path.join(d, prefix))
if res:
raise Exception('error building image')
finally:
shutil.rmtree(d)