def test_cached_path_offline(self, monkeypatch):
# Ensures `cached_path` just returns the path to the latest cached version
# of the resource when there's no internet connection.
# First we mock the `_http_etag` method so that it raises a `ConnectionError`,
# like it would if there was no internet connection.
def mocked_http_etag(url: str):
raise ConnectionError
monkeypatch.setattr(file_utils, "_http_etag", mocked_http_etag)
url = "https://github.com/allenai/allennlp/blob/master/some-fake-resource"
# We'll create two cached versions of this fake resource using two different etags.
etags = [
'W/"3e5885bfcbf4c47bc4ee9e2f6e5ea916"',
'W/"3e5885bfcbf4c47bc4ee9e2f6e5ea918"'
]
filenames = [
os.path.join(self.TEST_DIR, _resource_to_filename(url, etag))
for etag in etags
]
for filename, etag in zip(filenames, etags):
meta = _Meta(resource=url,
cached_path=filename,
creation_time=time.time(),
etag=etag,
size=2341)
meta.to_file()
with open(filename, "w") as f:
f.write("some random data")
# os.path.getmtime is only accurate to the second.
time.sleep(1.1)
# Should know to ignore lock files and extraction directories.
with open(filenames[-1] + ".lock", "w") as f:
f.write("")
os.mkdir(filenames[-1] + "-extracted")
# The version corresponding to the last etag should be returned, since
# that one has the latest "last modified" time.
assert get_from_cache(url, cache_dir=self.TEST_DIR) == filenames[-1]
# We also want to make sure this works when the latest cached version doesn't
# have a corresponding etag.
filename = os.path.join(self.TEST_DIR, _resource_to_filename(url))
meta = _Meta(resource=url,
cached_path=filename,
creation_time=time.time(),
size=2341)
with open(filename, "w") as f:
f.write("some random data")
assert get_from_cache(url, cache_dir=self.TEST_DIR) == filename
def create_cache_entry(self,
url: str,
etag: str,
as_extraction_dir: bool = False):
filename = os.path.join(self.TEST_DIR,
_resource_to_filename(url, etag))
cache_path = filename
if as_extraction_dir:
cache_path = filename + "-extracted"
filename = filename + "-extracted/glove.txt"
os.mkdir(cache_path)
with open(filename, "wb") as f:
f.write(self.glove_bytes)
open(cache_path + ".lock", "a").close()
meta = _Meta(
resource=url,
cached_path=cache_path,
etag=etag,
creation_time=time.time(),
size=len(self.glove_bytes),
extraction_dir=as_extraction_dir,
)
meta.to_file()