def __attrs_post_init__(self):
self._cache = stdpopsim.CachedData(
namespace=f"annotations/{self.species.id}",
url=self.zarr_url,
sha256=self.zarr_sha256,
extract=False,
)
def __init__(
self,
species,
id=None,
url=None,
sha256=None,
file_pattern=None,
description=None,
long_description=None,
citations=None,
):
self.id = id
self.species = species
self.description = description
self.long_description = long_description
self.url = url
self.sha256 = sha256
self.file_pattern = file_pattern
self.description = description
self.citations = citations
self._cache = stdpopsim.CachedData(
namespace=f"genetic_maps/{self.species.id}/{id}",
url=url,
sha256=sha256,
extract=True,
)
def test_caching(self):
for extract in (True, False):
with tempfile.TemporaryDirectory() as tmpdir:
tmpdir = pathlib.Path(tmpdir)
with utils.cd(tmpdir):
filename = "test.foo"
with open(filename, "w") as f:
print("foo", file=f)
tar = tmpdir / "test.tgz"
with tarfile.open(tar, "w:gz") as tf:
tf.add(filename)
sha256 = utils.sha256(tar)
cache = stdpopsim.CachedData(
namespace="test",
url=tar.resolve().as_uri(),
sha256=sha256,
extract=extract,
)
self.assertFalse(cache.is_cached())
self.assertFalse(cache.is_valid())
cache.download()
self.assertTrue(cache.is_cached())
self.assertTrue(cache.is_valid())
# try to download with incorrect checksum
cache.sha256 = "1234"
self.assertTrue(cache.is_cached())
self.assertFalse(cache.is_valid())
with self.assertRaises(ValueError):
# checksum mismatch
cache.download()
self.assertFalse(cache.is_cached())
self.assertFalse(cache.is_valid())
# fix the checksum and download again
cache.sha256 = sha256
cache.download()
self.assertTrue(cache.is_cached())
self.assertTrue(cache.is_valid())
def test_multiple_threads_downloading(self):
with tempfile.TemporaryDirectory() as tmpdir:
tmpdir = pathlib.Path(tmpdir)
with utils.cd(tmpdir):
filename = "test.foo"
with open(filename, "w") as f:
print("foo", file=f)
tar = tmpdir / "test.tgz"
with tarfile.open(tar, "w:gz") as tf:
tf.add(filename)
cache = stdpopsim.CachedData(
namespace="test",
url=tar.resolve().as_uri(),
sha256=utils.sha256(tar),
extract=True,
)
cache.download()
# Trick the download code into thinking there's several happening
# concurrently
cache.is_cached = lambda: False
with self.assertWarns(UserWarning):
cache.download()
