Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
177 changes: 177 additions & 0 deletions tests/test_release_snapshot.py
Original file line number Diff line number Diff line change
Expand Up @@ -209,4 +209,181 @@ def corrupt_second(oid,data):
self.assertFalse(out.exists(), 'failed snapshot must not publish a readable partial zip')
self.assertFalse(list(root.glob('.snapshot-*')), 'temporary snapshot was not cleaned')


def _identity_fixture(root):
fixture = root / 'fixture'; fixture.mkdir()
def git(*args):
return subprocess.check_output(['git', '-C', str(fixture), *args],
stderr=subprocess.DEVNULL).decode().strip()
git('init', '-q'); git('config', 'user.email', 'test@example.invalid')
git('config', 'user.name', 'test')
(fixture / 'proof.txt').write_bytes(b'unchanged source\n')
git('add', '.'); git('commit', '-qm', 'fixture')
return fixture, git('rev-parse', 'HEAD')


def _identity_collect(root, overrides):
"""Mocked collect(); `overrides` maps (repository, resource) to raw rows."""
fixture, sha = _identity_fixture(root)
calls = []
class API:
def get(self, path):
name = path.split('/')[2]
resource = '/'.join(path.split('/')[3:]).split('?')[0]
page = 2 if 'page=2&' in path else 1
calls.append((name, resource))
for (repo, wanted), value in overrides.items():
if repo == name and (resource == wanted or
(wanted.startswith('*') and resource.endswith(wanted[1:]))):
return value(page) if callable(value) else value
if not resource:
return {'private': False, 'full_name': f'{m.OWNER}/{name}', 'default_branch': 'main'}
if resource == 'commits/main':
return {'sha': sha}
if resource == 'pulls':
return [{'number': 7}] if name == 'main' else []
if resource == 'pulls/7':
return {'number': 7, 'head': {'sha': sha, 'repo': {'full_name': f'{m.OWNER}/main'}},
'base': {'sha': sha}, 'user': {'login': 'fixture-author'}}
if resource.endswith('check-runs'): return {'check_runs': [], 'total_count': 0}
if resource == 'actions/runs': return {'workflow_runs': [], 'total_count': 0}
if resource.endswith('/status'): return {'state': 'success'}
return []
out = root / 'out'
with mock.patch.object(m, 'REPOS', ('main', 'Math-')), \
mock.patch.object(m, 'Client', API), \
mock.patch.object(m, 'repo_url', return_value=str(fixture)):
result = m.collect(out)
return result, out, sha, calls


class IdentityPolicyTests(unittest.TestCase):
# (resource label in the error record, fake-API resource, raw output path, raw rows)
ROUTES = (
('/pulls/7/files', 'pulls/7/files', 'pr-7/files.json',
[{'filename': 'a.txt', 'sha': 'b1'}, {'filename': 'a.txt', 'sha': 'b2'}]),
('/pulls/7/reviews', 'pulls/7/reviews', 'pr-7/reviews.json',
[{'id': 11, 'state': 'COMMENTED'}, {'id': 11, 'state': 'COMMENTED'}]),
('/issues/7/comments', 'issues/7/comments', 'pr-7/comments.json',
[{'id': 12, 'body': 'x'}, {'id': 12, 'body': 'y'}]),
('/pulls/7/comments', 'pulls/7/comments', 'pr-7/inline-comments.json', [{'id': 13}, {'id': 13}]),
('/check-runs', '*/check-runs', 'pr-7/checks.json', [{'id': 14}, {'id': 14}]),
('/actions/runs', 'actions/runs', 'pr-7/runs.json', [{'id': 15}, {'id': 15}]),
('/rulesets', 'rulesets', 'rulesets.json', [{'id': 16}, {'id': 16}]),
)
WRAPPED = {'/check-runs': 'check_runs', '/actions/runs': 'workflow_runs'}

def test_duplicate_identity_is_refused_on_every_route_with_raw_rows_kept(self):
for label, resource, raw_path, rows in self.ROUTES:
with self.subTest(route=label), tempfile.TemporaryDirectory() as d:
value = rows
if label in self.WRAPPED:
value = {'total_count': len(rows), self.WRAPPED[label]: rows}
result, out, sha, _ = _identity_collect(pathlib.Path(d), {('main', resource): value})
errors = [e for e in result['errors'] if e['repository'] == 'main']
self.assertFalse(result['complete'])
self.assertEqual(len(errors), 1, errors)
self.assertIn(label, errors[0]['resource'])
self.assertIn('duplicate', errors[0]['error'])
self.assertEqual(json.loads((out / 'main' / raw_path).read_text()), rows)
later = result['repositories'][1]
self.assertEqual(later['default_commit'], sha)
self.assertEqual(len(later['snapshots']), 1)

def test_duplicate_valid_pr_numbers_refuse_the_open_pr_collection(self):
with tempfile.TemporaryDirectory() as d:
rows = [{'number': 7, 'v': 'first'}, {'number': 7, 'v': 'second'}]
result, out, sha, calls = _identity_collect(pathlib.Path(d), {('main', 'pulls'): rows})
errors = [e for e in result['errors'] if e['repository'] == 'main']
self.assertEqual([e['resource'] for e in errors], ['/pulls?state=open'])
self.assertIn('duplicate', errors[0]['error'])
self.assertIn('not processed', errors[0]['error'])
self.assertEqual(json.loads((out / 'main' / 'open-prs.json').read_text()), rows)
main = result['repositories'][0]
self.assertNotIn('open_pr_count', main)
self.assertEqual(main['pull_requests'], [])
self.assertNotIn(('main', 'pulls/7'), calls)
self.assertFalse((out / 'main' / 'pr-7').exists())
self.assertEqual(len(result['repositories'][1]['snapshots']), 1)

def test_invalid_then_valid_pr_keeps_partial_progress(self):
with tempfile.TemporaryDirectory() as d:
rows = [{'number': '../outside'}, {'number': 7}]
result, out, sha, calls = _identity_collect(pathlib.Path(d), {('main', 'pulls'): rows})
errors = [e for e in result['errors'] if e['repository'] == 'main']
self.assertEqual(len(errors), 1)
self.assertIn('positive integer PR number', errors[0]['error'])
main = result['repositories'][0]
self.assertEqual(main['open_pr_count'], 2)
self.assertEqual([p['number'] for p in main['pull_requests']], [7])
self.assertTrue((out / 'main' / 'pr-7' / 'detail.json').exists())

def test_same_blob_at_different_filenames_is_accepted(self):
with tempfile.TemporaryDirectory() as d:
rows = [{'filename': 'a.txt', 'sha': 'same'}, {'filename': 'b.txt', 'sha': 'same'},
{'filename': 'A.txt', 'sha': 'same'}, {'filename': 'a.txt ', 'sha': 'same'}]
result, out, sha, calls = _identity_collect(pathlib.Path(d), {('main', 'pulls/7/files'): rows})
self.assertTrue(result['complete'], result['errors'])
self.assertEqual(json.loads((out / 'main' / 'pr-7' / 'files.json').read_text()), rows)

def test_repeated_id_across_pages_with_matching_total_is_refused(self):
def runs(page):
if page == 1:
return {'total_count': 101, 'workflow_runs': [{'id': i} for i in range(1, 101)]}
return {'total_count': 101, 'workflow_runs': [{'id': 100}]}
with tempfile.TemporaryDirectory() as d:
result, out, sha, calls = _identity_collect(pathlib.Path(d), {('main', 'actions/runs'): runs})
errors = [e for e in result['errors'] if e['repository'] == 'main']
self.assertEqual(len(errors), 1)
self.assertIn('duplicate', errors[0]['error'])
self.assertEqual(len(json.loads((out / 'main' / 'pr-7' / 'runs.json').read_text())), 101)

def test_identity_values(self):
check = m.check_identity
check([{'id': 1}, {'id': 2}], m.ID_IDENTITY)
check([], m.ID_IDENTITY)
for rows in ([{'id': 1}, {'id': 1}], [{'id': 3, 'x': 1}, {'id': 3, 'x': 2}],
[{'id': True}], [{'id': '1'}], [{'id': 1.0}], [{'id': 0}], [{'id': -1}],
[{}], ['1'], [None]):
with self.subTest(rows=rows), self.assertRaises(ValueError):
check(rows, m.ID_IDENTITY)
check([{'filename': 'a'}, {'filename': 'A'}, {'filename': 'a '}], m.FILENAME_IDENTITY)
for rows in ([{'filename': 'a', 'sha': 'x'}, {'filename': 'a', 'sha': 'y'}],
[{'filename': ''}], [{'filename': 1}], [{'sha': 'x'}], ['a']):
with self.subTest(rows=rows), self.assertRaises(ValueError):
check(rows, m.FILENAME_IDENTITY)
# Malformed PR rows are left to the per-row path; only valid numbers must be unique.
check([{'number': 'x'}, {'number': 'x'}, {'number': 7}, None], m.PR_IDENTITY)
with self.assertRaises(ValueError):
check([{'number': 7}, {'number': 'x'}, {'number': 7}], m.PR_IDENTITY)

def test_missing_identity_policy_fails_closed(self):
for policy in (None, 'id', object()):
with self.subTest(policy=policy), self.assertRaises(ValueError):
m.check_identity([{'id': 1}], policy)

def test_every_paginated_read_names_its_identity_policy(self):
import ast
tree = ast.parse((ROOT / 'tools/release_snapshot.py').read_text())
expected = {'/pulls?state=open': 'PR_IDENTITY', "f'/pulls/{n}/files'": 'FILENAME_IDENTITY',
"f'/pulls/{n}/reviews'": 'ID_IDENTITY', "f'/issues/{n}/comments'": 'ID_IDENTITY',
"f'/pulls/{n}/comments'": 'ID_IDENTITY', "f'/commits/{sha}/check-runs'": 'ID_IDENTITY',
"f'/actions/runs?head_sha={sha}'": 'ID_IDENTITY', '/rulesets': 'ID_IDENTITY'}
seen = {}
for node in ast.walk(tree):
if not (isinstance(node, ast.Call) and isinstance(node.func, ast.Name) and node.func.id == 'read'):
continue
kw = {k.arg: k.value for k in node.keywords}
if not (isinstance(kw.get('paginated'), ast.Constant) and kw['paginated'].value is True):
self.assertNotIn('identity', kw)
continue
first = node.args[0]
label = first.value if isinstance(first, ast.Constant) else ast.get_source_segment(
(ROOT / 'tools/release_snapshot.py').read_text(), first)
self.assertIn('identity', kw, label)
self.assertIsInstance(kw['identity'], ast.Name, label)
seen[label] = kw['identity'].id
self.assertEqual(seen, expected)


if __name__ == '__main__': unittest.main()
60 changes: 51 additions & 9 deletions tools/release_snapshot.py
Original file line number Diff line number Diff line change
Expand Up @@ -179,6 +179,41 @@ def pr_number(value):
return value['number']


# Endpoint identity policies for paginated collections. Keys are collection-local.
PR_IDENTITY = 'pull request number'
FILENAME_IDENTITY = 'changed-file filename'
ID_IDENTITY = 'positive integer id'


def check_identity(rows, policy) -> None:
"""Refuse a repeated identity in one collection. Rows are never deduplicated or reordered."""
if policy not in (PR_IDENTITY, FILENAME_IDENTITY, ID_IDENTITY):
raise ValueError('paginated collection has no explicit identity policy')
if not isinstance(rows, list):
raise ValueError('expected a list collection')
seen = {}
for position, row in enumerate(rows):
if policy == PR_IDENTITY:
try:
key = pr_number(row)
except ValueError:
continue # Malformed PR rows keep the per-row error path in collect().
elif policy == FILENAME_IDENTITY:
if not isinstance(row, dict) or type(row.get('filename')) is not str or not row['filename']:
raise ValueError(f'expected nonempty string filename at row {position}')
key = row['filename']
else:
if not isinstance(row, dict) or type(row.get('id')) is not int or row['id'] <= 0:
raise ValueError(f'expected positive integer id at row {position}')
key = row['id']
if key in seen:
message = f'duplicate {policy} {key!r} at rows {seen[key]} and {position}'
if policy == PR_IDENTITY:
message += '; open pull requests not processed'
raise ValueError(message)
seen[key] = position


def validate_pr(value, number):
if pr_number(value) != number:
raise ValueError('PR identity mismatch')
Expand All @@ -204,11 +239,13 @@ def collect(out: Path, include_pr_sources: bool = True) -> dict:
summary['repositories'].append(row)
dest = out/name; dest.mkdir()
path = f'repos/{OWNER}/{name}'
def read(suffix, filename, key=None, paginated=False, validate=None):
def read(suffix, filename, key=None, paginated=False, validate=None, identity=None):
try:
value = pages(client.get, path+suffix, key) if paginated else client.get(path+suffix)
write_json(dest/filename, value)
# Retain raw evidence even when its target schema is unusable.
if paginated:
check_identity(value, identity)
if validate is not None:
validate(value)
return value
Expand All @@ -222,7 +259,7 @@ def read(suffix, filename, key=None, paginated=False, validate=None):
branch = meta['default_branch']
commit = read('/commits/'+urllib.parse.quote(branch,safe=''), 'default-commit.json',
validate=validate_commit)
prs = read('/pulls?state=open', 'open-prs.json', paginated=True)
prs = read('/pulls?state=open', 'open-prs.json', paginated=True, identity=PR_IDENTITY)
targets = {}
if commit is not None:
sha = checked_sha(commit['sha']); row['default_commit'] = sha; targets[sha] = ['default']
Expand All @@ -244,19 +281,24 @@ def read(suffix, filename, key=None, paginated=False, validate=None):
item = {k:detail.get(k) for k in ('number','title','html_url','draft','mergeable','mergeable_state','state')}
item.update(head=sha, base=detail['base']['sha'], author=detail['user']['login'])
row['pull_requests'].append(item)
read(f'/pulls/{n}/files', prefix+'files.json', paginated=True)
read(f'/pulls/{n}/reviews', prefix+'reviews.json', paginated=True)
read(f'/issues/{n}/comments', prefix+'comments.json', paginated=True)
read(f'/pulls/{n}/comments', prefix+'inline-comments.json', paginated=True)
read(f'/commits/{sha}/check-runs', prefix+'checks.json', key='check_runs', paginated=True)
read(f'/pulls/{n}/files', prefix+'files.json', paginated=True,
identity=FILENAME_IDENTITY)
read(f'/pulls/{n}/reviews', prefix+'reviews.json', paginated=True, identity=ID_IDENTITY)
read(f'/issues/{n}/comments', prefix+'comments.json', paginated=True,
identity=ID_IDENTITY)
read(f'/pulls/{n}/comments', prefix+'inline-comments.json', paginated=True,
identity=ID_IDENTITY)
read(f'/commits/{sha}/check-runs', prefix+'checks.json', key='check_runs', paginated=True,
identity=ID_IDENTITY)
read(f'/commits/{sha}/status', prefix+'combined-status.json')
read(f'/actions/runs?head_sha={sha}', prefix+'runs.json', key='workflow_runs', paginated=True)
read(f'/actions/runs?head_sha={sha}', prefix+'runs.json', key='workflow_runs',
paginated=True, identity=ID_IDENTITY)
if include_pr_sources:
if detail['head'].get('repo') and detail['head']['repo']['full_name'] == f'{OWNER}/{name}':
targets.setdefault(sha,[]).append(f'PR{n}')
else:
item['source_snapshot'] = 'not captured: fork outside same-repository scope'
read('/rulesets', 'rulesets.json', paginated=True)
read('/rulesets', 'rulesets.json', paginated=True, identity=ID_IDENTITY)
with tempfile.TemporaryDirectory(prefix='release-git-') as tmp:
repo = Path(tmp)
subprocess.run(['git','init','--bare','-q',str(repo)],check=True,timeout=30)
Expand Down
Loading