[Notes] [Git][BuildStream/buildstream][valentindavid/git_shallow_fetch] Fetch git shallow clone when possible

[Valentin David <gitlab mg gitlab com>]

Title: GitLab

Valentin David pushed to branch valentindavid/git_shallow_fetch at BuildStream / buildstream

Commits:

• 11908d3a
  by Valentin David at 2018-12-05T11:37:05Z

  Fetch git shallow clone when possible
  
  When the requested ref is advertised by remote and that no tag is
  required, then we shallow clone that requested ref. Otherwise we
  fallback on full clone.
  
  Workspace opening and tracking opeerations still get a full clone.
  
  Fixes #261
  

2 changed files:

• buildstream/plugins/sources/git.py
• tests/sources/git.py

Changes:

buildstream/plugins/sources/git.py

---

...	...	@@ -175,11 +175,81 @@ class GitMirror(SourceFetcher):
175	175	self.ref = ref
176	176	self.tags = tags
177	177	self.primary = primary
	178	+ dirname = utils.url_directory_name(url)
178	179	self.mirror = os.path.join(source.get_mirror_directory(), utils.url_directory_name(url))
	180	+ self.fetch_mirror = os.path.join(source.get_mirror_directory(), '{}-{}'.format(dirname, ref))
179	181	self.mark_download_url(url)
180	182
	183	+ def ensure_fetchable(self, alias_override=None):
	184	+
	185	+ if os.path.exists(self.mirror):
	186	+ return
	187	+
	188	+ if self.tags:
	189	+ for tag, commit, _ in self.tags:
	190	+ if commit != self.ref:
	191	+ self.source.status("{}: tag '{}' is not on commit '{}' {}, so a full clone is required"
	192	+ .format(self.source, tag, commit))
	193	+ return ensure_trackable(alias_override=alias_override)
	194	+
	195	+ if os.path.exists(self.fetch_mirror):
	196	+ return
	197	+
	198	+ with self.source.tempdir() as tmpdir:
	199	+ self.source.call([self.source.host_git, 'init', '--bare', tmpdir],
	200	+ fail="Failed to init git repository",
	201	+ fail_temporarily=True)
	202	+
	203	+ url = self.source.translate_url(self.url, alias_override=alias_override,
	204	+ primary=self.primary)
	205	+
	206	+ self.source.call([self.source.host_git, 'remote', 'add', '--mirror=fetch', 'origin', url],
	207	+ cwd=tmpdir,
	208	+ fail="Failed to init git repository",
	209	+ fail_temporarily=True)
	210	+
	211	+ exit_code, refs = self.source.check_output([self.source.host_git, 'ls-remote', 'origin'],
	212	+ cwd=tmpdir,
	213	+ fail="Failed to clone git repository {}".format(url),
	214	+ fail_temporarily=True)
	215	+
	216	+ advertised = False
	217	+ for ref_line in refs.splitlines():
	218	+ commit, ref = ref_line.split('\t', 1)
	219	+ print(commit, self.ref)
	220	+ if self.ref == commit:
	221	+ advertised = True
	222	+ break
	223	+ if not advertised:
	224	+ self.source.status("{}: {} is not advertised on {}, so a full clone is required"
	225	+ .format(self.source, self.ref, url))
	226	+
	227	+ return self.ensure_trackable(alias_override=alias_override)
	228	+
	229	+ self.source.call([self.source.host_git, 'fetch', '--depth=1', 'origin', self.ref],
	230	+ cwd=tmpdir,
	231	+ fail="Failed to fetch repository",
	232	+ fail_temporarily=True)
	233	+
	234	+ # We need to have a ref to make it clonable
	235	+ self.source.call([self.source.host_git, 'update-ref', 'HEAD', self.ref],
	236	+ cwd=tmpdir,
	237	+ fail="Failed to tag HEAD",
	238	+ fail_temporarily=True)
	239	+
	240	+ try:
	241	+ move_atomic(tmpdir, self.fetch_mirror)
	242	+ except DirectoryExistsError:
	243	+ # Another process was quicker to download this repository.
	244	+ # Let's discard our own
	245	+ self.source.status("{}: Discarding duplicate clone of {}"
	246	+ .format(self.source, url))
	247	+ except OSError as e:
	248	+ raise SourceError("{}: Failed to move cloned git repository {} from '{}' to '{}': {}"
	249	+ .format(self.source, url, tmpdir, self.fetch_mirror, e)) from e
	250	+
181	251	# Ensures that the mirror exists
182		- def ensure(self, alias_override=None):
	252	+ def ensure_trackable(self, alias_override=None):
183	253
184	254	# Unfortunately, git does not know how to only clone just a specific ref,
185	255	# so we have to download all of those gigs even if we only need a couple
...	...	@@ -214,18 +284,23 @@ class GitMirror(SourceFetcher):
214	284	alias_override=alias_override,
215	285	primary=self.primary)
216	286
	287	+ if os.path.exists(self.mirror):
	288	+ mirror = self.mirror
	289	+ else:
	290	+ mirror = self.fetch_mirror
	291	+
217	292	if alias_override:
218	293	remote_name = utils.url_directory_name(alias_override)
219	294	_, remotes = self.source.check_output(
220	295	[self.source.host_git, 'remote'],
221		- fail="Failed to retrieve list of remotes in {}".format(self.mirror),
	296	+ fail="Failed to retrieve list of remotes in {}".format(mirror),
222	297	cwd=self.mirror
223	298	)
224	299	if remote_name not in remotes:
225	300	self.source.call(
226	301	[self.source.host_git, 'remote', 'add', remote_name, url],
227	302	fail="Failed to add remote {} with url {}".format(remote_name, url),
228		- cwd=self.mirror
	303	+ cwd=mirror
229	304	)
230	305	else:
231	306	remote_name = "origin"
...	...	@@ -233,7 +308,7 @@ class GitMirror(SourceFetcher):
233	308	self.source.call([self.source.host_git, 'fetch', remote_name, '--prune'],
234	309	fail="Failed to fetch from remote git repository: {}".format(url),
235	310	fail_temporarily=True,
236		- cwd=self.mirror)
	311	+ cwd=mirror)
237	312
238	313	def fetch(self, alias_override=None):
239	314	# Resolve the URL for the message
...	...	@@ -244,7 +319,7 @@ class GitMirror(SourceFetcher):
244	319	with self.source.timed_activity("Fetching from {}"
245	320	.format(resolved_url),
246	321	silent_nested=True):
247		- self.ensure(alias_override)
	322	+ self.ensure_fetchable(alias_override)
248	323	if not self.has_ref():
249	324	self._fetch(alias_override)
250	325	self.assert_ref()
...	...	@@ -253,12 +328,16 @@ class GitMirror(SourceFetcher):
253	328	if not self.ref:
254	329	return False
255	330
256		- # If the mirror doesnt exist, we also dont have the ref
257		- if not os.path.exists(self.mirror):
258		- return False
	331	+ if os.path.exists(self.mirror):
	332	+ mirror = self.mirror
	333	+ else:
	334	+ # If the mirror doesnt exist, we also dont have the ref
	335	+ if not os.path.exists(self.fetch_mirror):
	336	+ return False
	337	+ mirror = self.fetch_mirror
259	338
260	339	# Check if the ref is really there
261		- rc = self.source.call([self.source.host_git, 'cat-file', '-t', self.ref], cwd=self.mirror)
	340	+ rc = self.source.call([self.source.host_git, 'cat-file', '-t', self.ref], cwd=mirror)
262	341	return rc == 0
263	342
264	343	def assert_ref(self):
...	...	@@ -308,11 +387,16 @@ class GitMirror(SourceFetcher):
308	387	def stage(self, directory, track=None):
309	388	fullpath = os.path.join(directory, self.path)
310	389
	390	+ if os.path.exists(self.mirror):
	391	+ mirror = self.mirror
	392	+ else:
	393	+ mirror = self.fetch_mirror
	394	+
311	395	# Using --shared here avoids copying the objects into the checkout, in any
312	396	# case we're just checking out a specific commit and then removing the .git/
313	397	# directory.
314		- self.source.call([self.source.host_git, 'clone', '--no-checkout', '--shared', self.mirror, fullpath],
315		- fail="Failed to create git mirror {} in directory: {}".format(self.mirror, fullpath),
	398	+ self.source.call([self.source.host_git, 'clone', '--no-checkout', '--shared', mirror, fullpath],
	399	+ fail="Failed to create git mirror {} in directory: {}".format(mirror, fullpath),
316	400	fail_temporarily=True)
317	401
318	402	self.source.call([self.source.host_git, 'checkout', '--force', self.ref],
...	...	@@ -350,9 +434,14 @@ class GitMirror(SourceFetcher):
350	434
351	435	# List the submodules (path/url tuples) present at the given ref of this repo
352	436	def submodule_list(self):
	437	+ if os.path.exists(self.mirror):
	438	+ mirror = self.mirror
	439	+ else:
	440	+ mirror = self.fetch_mirror
	441	+
353	442	modules = "{}:{}".format(self.ref, GIT_MODULES)
354	443	exit_code, output = self.source.check_output(
355		- [self.source.host_git, 'show', modules], cwd=self.mirror)
	444	+ [self.source.host_git, 'show', modules], cwd=mirror)
356	445
357	446	# If git show reports error code 128 here, we take it to mean there is
358	447	# no .gitmodules file to display for the given revision.
...	...	@@ -380,6 +469,11 @@ class GitMirror(SourceFetcher):
380	469	# Fetch the ref which this mirror requires its submodule to have,
381	470	# at the given ref of this mirror.
382	471	def submodule_ref(self, submodule, ref=None):
	472	+ if os.path.exists(self.mirror):
	473	+ mirror = self.mirror
	474	+ else:
	475	+ mirror = self.fetch_mirror
	476	+
383	477	if not ref:
384	478	ref = self.ref
385	479
...	...	@@ -388,7 +482,7 @@ class GitMirror(SourceFetcher):
388	482	_, output = self.source.check_output([self.source.host_git, 'ls-tree', ref, submodule],
389	483	fail="ls-tree failed for commit {} and submodule: {}".format(
390	484	ref, submodule),
391		- cwd=self.mirror)
	485	+ cwd=mirror)
392	486
393	487	# read the commit hash from the output
394	488	fields = output.split()
...	...	@@ -646,7 +740,7 @@ class GitSource(Source):
646	740	with self.timed_activity("Tracking {} from {}"
647	741	.format(self.tracking, resolved_url),
648	742	silent_nested=True):
649		- self.mirror.ensure()
	743	+ self.mirror.ensure_trackable()
650	744	self.mirror._fetch()
651	745
652	746	# Update self.mirror.ref and node.ref from the self.tracking branch
...	...	@@ -658,6 +752,7 @@ class GitSource(Source):
658	752
659	753	def init_workspace(self, directory):
660	754	# XXX: may wish to refactor this as some code dupe with stage()
	755	+ self.mirror.ensure_trackable()
661	756	self.refresh_submodules()
662	757
663	758	with self.timed_activity('Setting up workspace "{}"'.format(directory), silent_nested=True):
...	...	@@ -702,7 +797,7 @@ class GitSource(Source):
702	797
703	798	self.refresh_submodules()
704	799	for mirror in self.submodules:
705		- if not os.path.exists(mirror.mirror):
	800	+ if not os.path.exists(mirror.mirror) and not os.path.exists(mirror.fetch_mirror):
706	801	return False
707	802	if not mirror.has_ref():
708	803	return False
...	...	@@ -714,7 +809,7 @@ class GitSource(Source):
714	809	# Assumes that we have our mirror and we have the ref which we point to
715	810	#
716	811	def refresh_submodules(self):
717		- self.mirror.ensure()
	812	+ self.mirror.ensure_fetchable()
718	813	submodules = []
719	814
720	815	# XXX Here we should issue a warning if either:

tests/sources/git.py

---

...	...	@@ -27,6 +27,7 @@ import subprocess
27	27	from buildstream._exceptions import ErrorDomain
28	28	from buildstream import _yaml
29	29	from buildstream.plugin import CoreWarnings
	30	+from buildstream.utils import url_directory_name
30	31
31	32	from tests.testutils import cli, create_repo
32	33	from tests.testutils.site import HAVE_GIT
...	...	@@ -676,3 +677,193 @@ def test_default_do_not_track_tags(cli, tmpdir, datafiles):
676	677
677	678	element = _yaml.load(element_path)
678	679	assert 'tags' not in element['sources'][0]
	680	+
	681	+
	682	+@pytest.mark.skipif(HAVE_GIT is False, reason="git is not available")
	683	+@pytest.mark.datafiles(os.path.join(DATA_DIR, 'template'))
	684	+def test_fetch_shallow(cli, tmpdir, datafiles):
	685	+ project = str(datafiles)
	686	+
	687	+ repo = create_repo('git', str(tmpdir))
	688	+ previous_ref = repo.create(os.path.join(project, 'repofiles'))
	689	+
	690	+ file1 = os.path.join(str(tmpdir), 'file1')
	691	+ with open(file1, 'w') as f:
	692	+ f.write('test\n')
	693	+ ref = repo.add_file(file1)
	694	+
	695	+ source_config = repo.source_config(ref=ref)
	696	+
	697	+ # Write out our test target with a bad ref
	698	+ element = {
	699	+ 'kind': 'import',
	700	+ 'sources': [
	701	+ source_config
	702	+ ]
	703	+ }
	704	+ _yaml.dump(element, os.path.join(project, 'target.bst'))
	705	+
	706	+ sources_dir = os.path.join(str(tmpdir), 'sources')
	707	+ os.makedirs(sources_dir, exist_ok=True)
	708	+ config = {
	709	+ 'sourcedir': sources_dir
	710	+ }
	711	+ cli.configure(config)
	712	+
	713	+ result = cli.run(project=project, args=[
	714	+ 'fetch', 'target.bst'
	715	+ ])
	716	+ result.assert_success()
	717	+
	718	+ cache_dir_name = url_directory_name(source_config['url'])
	719	+ full_cache_path = os.path.join(sources_dir, 'git', cache_dir_name)
	720	+ shallow_cache_path = os.path.join(sources_dir, 'git', '{}-{}'.format(cache_dir_name, ref))
	721	+
	722	+ assert os.path.exists(shallow_cache_path)
	723	+ assert not os.path.exists(full_cache_path)
	724	+
	725	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	726	+ cwd=shallow_cache_path,
	727	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	728	+ assert output.splitlines() == [ref]
	729	+
	730	+ result = cli.run(project=project, args=[
	731	+ 'build', 'target.bst'
	732	+ ])
	733	+ result.assert_success()
	734	+
	735	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	736	+ cwd=shallow_cache_path,
	737	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	738	+ assert output.splitlines() == [ref]
	739	+
	740	+ assert os.path.exists(shallow_cache_path)
	741	+ assert not os.path.exists(full_cache_path)
	742	+
	743	+ result = cli.run(project=project, args=[
	744	+ 'track', 'target.bst'
	745	+ ])
	746	+ result.assert_success()
	747	+
	748	+ assert os.path.exists(full_cache_path)
	749	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	750	+ cwd=full_cache_path,
	751	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	752	+ assert output.splitlines() == [ref, previous_ref]
	753	+
	754	+
	755	+@pytest.mark.skipif(HAVE_GIT is False, reason="git is not available")
	756	+@pytest.mark.datafiles(os.path.join(DATA_DIR, 'template'))
	757	+def test_fetch_shallow_not_tagged(cli, tmpdir, datafiles):
	758	+ """When a ref is not tagged and not head of branch on remote we cannot
	759	+ get a shallow clone. It should automatically get a full clone.
	760	+ """
	761	+
	762	+ project = str(datafiles)
	763	+
	764	+ repo = create_repo('git', str(tmpdir))
	765	+ previous_ref = repo.create(os.path.join(project, 'repofiles'))
	766	+
	767	+ file1 = os.path.join(str(tmpdir), 'file1')
	768	+ with open(file1, 'w') as f:
	769	+ f.write('test\n')
	770	+ ref = repo.add_file(file1)
	771	+
	772	+ source_config = repo.source_config(ref=previous_ref)
	773	+
	774	+ # Write out our test target with a bad ref
	775	+ element = {
	776	+ 'kind': 'import',
	777	+ 'sources': [
	778	+ source_config
	779	+ ]
	780	+ }
	781	+ _yaml.dump(element, os.path.join(project, 'target.bst'))
	782	+
	783	+ sources_dir = os.path.join(str(tmpdir), 'sources')
	784	+ os.makedirs(sources_dir, exist_ok=True)
	785	+ config = {
	786	+ 'sourcedir': sources_dir
	787	+ }
	788	+ cli.configure(config)
	789	+
	790	+ result = cli.run(project=project, args=[
	791	+ 'fetch', 'target.bst'
	792	+ ])
	793	+ result.assert_success()
	794	+
	795	+ cache_dir_name = url_directory_name(source_config['url'])
	796	+ full_cache_path = os.path.join(sources_dir, 'git', cache_dir_name)
	797	+ shallow_cache_path = os.path.join(sources_dir, 'git', '{}-{}'.format(cache_dir_name, previous_ref))
	798	+
	799	+ assert not os.path.exists(shallow_cache_path)
	800	+ assert os.path.exists(full_cache_path)
	801	+
	802	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	803	+ cwd=full_cache_path,
	804	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	805	+ assert output.splitlines() == [ref, previous_ref]
	806	+
	807	+
	808	+@pytest.mark.skipif(HAVE_GIT is False, reason="git is not available")
	809	+@pytest.mark.datafiles(os.path.join(DATA_DIR, 'template'))
	810	+def test_fetch_shallow_workspace_open(cli, tmpdir, datafiles):
	811	+ """
	812	+ Workspaces should get a full clone.
	813	+ """
	814	+ project = str(datafiles)
	815	+
	816	+ repo = create_repo('git', str(tmpdir))
	817	+ previous_ref = repo.create(os.path.join(project, 'repofiles'))
	818	+
	819	+ file1 = os.path.join(str(tmpdir), 'file1')
	820	+ with open(file1, 'w') as f:
	821	+ f.write('test\n')
	822	+ ref = repo.add_file(file1)
	823	+
	824	+ source_config = repo.source_config(ref=ref)
	825	+
	826	+ # Write out our test target with a bad ref
	827	+ element = {
	828	+ 'kind': 'import',
	829	+ 'sources': [
	830	+ source_config
	831	+ ]
	832	+ }
	833	+ _yaml.dump(element, os.path.join(project, 'target.bst'))
	834	+
	835	+ sources_dir = os.path.join(str(tmpdir), 'sources')
	836	+ os.makedirs(sources_dir, exist_ok=True)
	837	+ config = {
	838	+ 'sourcedir': sources_dir
	839	+ }
	840	+ cli.configure(config)
	841	+
	842	+ result = cli.run(project=project, args=[
	843	+ 'fetch', 'target.bst'
	844	+ ])
	845	+ result.assert_success()
	846	+
	847	+ cache_dir_name = url_directory_name(source_config['url'])
	848	+ full_cache_path = os.path.join(sources_dir, 'git', cache_dir_name)
	849	+ shallow_cache_path = os.path.join(sources_dir, 'git', '{}-{}'.format(cache_dir_name, ref))
	850	+
	851	+ assert os.path.exists(shallow_cache_path)
	852	+ assert not os.path.exists(full_cache_path)
	853	+
	854	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	855	+ cwd=shallow_cache_path,
	856	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	857	+ assert output.splitlines() == [ref]
	858	+
	859	+ workspace = os.path.join(tmpdir, 'workspace')
	860	+
	861	+ result = cli.run(project=project, args=[
	862	+ 'workspace', 'open', 'target.bst', '--directory', workspace
	863	+ ])
	864	+ result.assert_success()
	865	+
	866	+ output = subprocess.run(['git', 'log', '--format=format:%H'],
	867	+ cwd=workspace,
	868	+ stdout=subprocess.PIPE).stdout.decode('ascii')
	869	+ assert output.splitlines() == [ref, previous_ref]

—
View it on GitLab.
You're receiving this email because of your account on gitlab.com. If you'd like to receive fewer emails, you can adjust your notification settings.