1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
|
# 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/.
import hashlib
import io
import os
import re
from typing import Optional
from taskgraph.util.archive import create_tar_gz_from_files
from taskgraph.util.memoize import memoize
IMAGE_DIR = os.path.join(".", "taskcluster", "docker")
from .yaml import load_yaml
def docker_image(name: str, by_tag: bool = False) -> Optional[str]:
"""
Resolve in-tree prebuilt docker image to ``<registry>/<repository>@sha256:<digest>``,
or ``<registry>/<repository>:<tag>`` if `by_tag` is `True`.
Args:
name (str): The image to build.
by_tag (bool): If True, will apply a tag based on VERSION file.
Otherwise will apply a hash based on HASH file.
Returns:
Optional[str]: Image if it can be resolved, otherwise None.
"""
try:
with open(os.path.join(IMAGE_DIR, name, "REGISTRY")) as f:
registry = f.read().strip()
except OSError:
try:
with open(os.path.join(IMAGE_DIR, "REGISTRY")) as f:
registry = f.read().strip()
except OSError:
return None
if not by_tag:
hashfile = os.path.join(IMAGE_DIR, name, "HASH")
try:
with open(hashfile) as f:
return f"{registry}/{name}@{f.read().strip()}"
except OSError:
return None
try:
with open(os.path.join(IMAGE_DIR, name, "VERSION")) as f:
tag = f.read().strip()
except OSError:
tag = "latest"
return f"{registry}/{name}:{tag}"
class VoidWriter:
"""A file object with write capabilities that does nothing with the written
data."""
def write(self, buf):
pass
def generate_context_hash(topsrcdir, image_path, args=None):
"""Generates a sha256 hash for context directory used to build an image."""
return stream_context_tar(topsrcdir, image_path, VoidWriter(), args=args)
class HashingWriter:
"""A file object with write capabilities that hashes the written data at
the same time it passes down to a real file object."""
def __init__(self, writer):
self._hash = hashlib.sha256()
self._writer = writer
def write(self, buf):
self._hash.update(buf)
self._writer.write(buf)
def hexdigest(self):
return self._hash.hexdigest()
def create_context_tar(topsrcdir, context_dir, out_path, args=None):
"""Create a context tarball.
A directory ``context_dir`` containing a Dockerfile will be assembled into
a gzipped tar file at ``out_path``.
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.
If a line in the Dockerfile has the form ``# %ARG <name>``, occurrences of
the string ``$<name>`` in subsequent lines are replaced with the value
found in the ``args`` argument. Exception: this doesn't apply to VOLUME
definitions.
Returns the SHA-256 hex digest of the created archive.
"""
with open(out_path, "wb") as fh:
return stream_context_tar(
topsrcdir,
context_dir,
fh,
image_name=os.path.basename(out_path),
args=args,
)
RUN_TASK_ROOT = os.path.join(os.path.dirname(os.path.dirname(__file__)), "run-task")
RUN_TASK_FILES = {
f"run-task/{path}": os.path.join(RUN_TASK_ROOT, path)
for path in [
"run-task",
"fetch-content",
"hgrc",
"robustcheckout.py",
]
}
RUN_TASK_SNIPPET = [
"COPY run-task/run-task /usr/local/bin/run-task\n",
"COPY run-task/fetch-content /usr/local/bin/fetch-content\n",
"COPY run-task/robustcheckout.py /usr/local/mercurial/robustcheckout.py\n"
"COPY run-task/hgrc /etc/mercurial/hgrc.d/mozilla.rc\n",
]
def stream_context_tar(topsrcdir, context_dir, out_file, image_name=None, args=None):
"""Like create_context_tar, but streams the tar file to the `out_file` file
object."""
archive_files = {}
replace = []
content = []
topsrcdir = os.path.abspath(topsrcdir)
context_dir = os.path.join(topsrcdir, context_dir)
for root, dirs, files in os.walk(context_dir):
for f in files:
source_path = os.path.join(root, f)
archive_path = source_path[len(context_dir) + 1 :]
archive_files[archive_path] = open(source_path, "rb")
# Parse Dockerfile for special syntax of extra files to include.
content = []
with open(os.path.join(context_dir, "Dockerfile")) as fh:
for line in fh:
if line.startswith("# %ARG"):
p = line[len("# %ARG ") :].strip()
if not args or p not in args:
raise Exception(f"missing argument: {p}")
replace.append((re.compile(rf"\${p}\b"), args[p]))
continue
for regexp, s in replace:
line = re.sub(regexp, s, line)
content.append(line)
if not line.startswith("# %include"):
continue
if line.strip() == "# %include-run-task":
content.extend(RUN_TASK_SNIPPET)
archive_files.update(RUN_TASK_FILES)
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)
rel = source_path[len(fs_path) + 1 :]
archive_path = os.path.join("topsrcdir", p, rel)
archive_files[archive_path] = source_path
else:
archive_path = os.path.join("topsrcdir", p)
archive_files[archive_path] = fs_path
archive_files["Dockerfile"] = io.BytesIO("".join(content).encode("utf-8"))
writer = HashingWriter(out_file)
create_tar_gz_from_files(writer, archive_files, image_name)
return writer.hexdigest()
@memoize
def image_paths():
"""Return a map of image name to paths containing their Dockerfile."""
config = load_yaml("taskcluster", "kinds", "docker-image", "kind.yml")
return {
k: os.path.join(IMAGE_DIR, v.get("definition", k))
for k, v in config["tasks"].items()
}
def image_path(name):
paths = image_paths()
if name in paths:
return paths[name]
return os.path.join(IMAGE_DIR, name)
@memoize
def parse_volumes(image):
"""Parse VOLUME entries from a Dockerfile for an image."""
volumes = set()
path = image_path(image)
with open(os.path.join(path, "Dockerfile"), "rb") as fh:
for line in fh:
line = line.strip()
# We assume VOLUME definitions don't use %ARGS.
if not line.startswith(b"VOLUME "):
continue
v = line.split(None, 1)[1]
if v.startswith(b"["):
raise ValueError(
"cannot parse array syntax for VOLUME; "
"convert to multiple entries"
)
volumes |= {volume.decode("utf-8") for volume in v.split()}
return volumes
|