def test_special_cases(fuse_mntdir):
snp_dir = Path(fuse_mntdir / "archive" / FAKE_SNP_SPECIAL_CASES_SWHID)
snp_meta = get_data_from_web_archive(FAKE_SNP_SPECIAL_CASES_SWHID)
for branch_name, branch_meta in snp_meta.items():
curr = snp_dir / branch_name
assert curr.is_symlink()
if "expected_symlink" in branch_meta:
assert os.readlink(curr) == branch_meta["expected_symlink"]
def test_list_parents(fuse_mntdir):
rev_meta = get_data_from_web_archive(ROOT_REV)
dir_path = fuse_mntdir / "archive" / ROOT_REV / "parents"
for i, parent in enumerate(rev_meta["parents"]):
parent_path = dir_path / str(i + 1)
parent_swhid = f"swh:1:rev:{parent['id']}"
assert parent_path.is_symlink()
assert os.readlink(parent_path) == f"../../../archive/{parent_swhid}"
def test_list_history(fuse_mntdir):
dir_path = fuse_mntdir / "archive" / REV_SMALL_HISTORY / "history"
assert os.listdir(dir_path) == ["by-date", "by-hash", "by-page"]
history_meta = get_data_from_graph_archive(REV_SMALL_HISTORY,
GRAPH_API_REQUEST.HISTORY)
history = history_meta.strip()
# Only keep second node in the edge because first node is redundant
# information or the root node (hence not an ancestor)
expected = set(
map(CoreSWHID.from_string,
[edge.split(" ")[1] for edge in history.split("\n")]))
dir_by_hash = dir_path / "by-hash"
for swhid in expected:
depth1 = hash_to_hex(swhid.object_id)[:2]
depth2 = str(swhid)
assert (dir_by_hash / depth1).exists()
assert depth2 in (os.listdir(dir_by_hash / depth1))
dir_by_page = dir_path / "by-page"
for idx, swhid in enumerate(expected):
page_number = idx // RevisionHistoryShardByPage.PAGE_SIZE
depth1 = RevisionHistoryShardByPage.PAGE_FMT.format(
page_number=page_number)
depth2 = str(swhid)
assert (dir_by_page / depth1).exists()
assert depth2 in (os.listdir(dir_by_page / depth1))
dir_by_date = dir_path / "by-date"
# Wait max 2 seconds to populate by-date/ dir
for i in range(200):
entries = os.listdir(dir_by_date)
if entries and ".status" not in entries:
break
time.sleep(0.1)
assert not (dir_by_date / ".status").exists()
for swhid in expected:
meta = get_data_from_web_archive(str(swhid))
date = dateutil.parser.parse(meta["date"])
depth1 = RevisionHistoryShardByDate.DATE_FMT.format(year=date.year,
month=date.month,
day=date.day)
depth2 = str(swhid)
assert (dir_by_date / depth1).exists()
assert depth2 in (os.listdir(dir_by_date / depth1))
def test_access_meta(fuse_mntdir):
file_path = fuse_mntdir / "archive" / ROOT_REV / "meta.json"
expected = json.dumps(get_data_from_web_archive(ROOT_REV))
assert file_path.read_text().strip() == expected.strip()
def test_access_cnt_target(fuse_mntdir):
target_path = fuse_mntdir / "archive" / REL_TARGET_CNT / "target"
expected = get_data_from_web_archive(TARGET_CNT, raw=True)
assert target_path.read_text() == expected
def test_cat_file(fuse_mntdir):
file_path = fuse_mntdir / "archive" / REGULAR_FILE
expected = get_data_from_web_archive(REGULAR_FILE, raw=True)
assert file_path.read_text() == expected
def test_list_branches(fuse_mntdir):
snp_dir = Path(fuse_mntdir / "archive" / ROOT_SNP)
snp_meta = get_data_from_web_archive(ROOT_SNP)
for branch_name in snp_meta.keys():
assert (snp_dir / branch_name).is_symlink()
def test_access_meta_file(fuse_mntdir):
for swhid in ALL_ENTRIES:
# On the fly mounting
file_path_meta = fuse_mntdir / f"archive/{swhid}.json"
expected = json.dumps(get_data_from_web_archive(swhid))
assert file_path_meta.read_text().strip() == expected.strip()