def test_dangerously_convert_cache_to_gzip(self):
html_string = "<html></html>".encode("utf-8")
example_url = "http://www.example.com/testpage.html"
cache = Cache(CONVERT_DIRECTORY)
cache.save(example_url, html_string)
self.assertTrue(cache.has(example_url))
dangerously_convert_cache_to_gzip(CONVERT_DIRECTORY)
gcache = GzipCache(CONVERT_DIRECTORY)
self.assertTrue(gcache.has(example_url))
for name in os.listdir(CONVERT_DIRECTORY):
shutil.rmtree(os.path.join(CONVERT_DIRECTORY, name))
def test_cache_get(self):
c = GzipCache(COMPRESS_CACHE_DIRECTORY)
exists = [
"http://www.example.com/somepage.html",
"http://www.example.com/otherpage"
]
for filename in exists:
self.assertIsNotNone(c.get(filename))
does_not_exist = [
"http://www.example.com/doesnotexist.html",
"https://en.wikipedia.org/wiki/Python_(programming_language)"
]
for filename in does_not_exist:
self.assertIsNone(c.get(filename))
def test_cache_setup(self):
c = GzipCache(COMPRESS_CACHE_DIRECTORY)
self.assertEqual(c.folder, COMPRESS_CACHE_DIRECTORY)
# verify that each folder has been loaded
for name in os.listdir(os.path.join(COMPRESS_CACHE_DIRECTORY)):
if os.path.isdir(os.path.join(COMPRESS_CACHE_DIRECTORY, name)):
self.assertIn(name, c.sites)
def test_cache_save_existing(self):
c = GzipCache(COMPRESS_CACHE_DIRECTORY)
# verify that it adds a file to a pre-existing cached site
html_string = "<html></html>".encode("utf-8")
example_url = "http://www.example.com/testpage.html"
d = dir_domain(example_url)
f = clean_url_hash(example_url)
full_save_name = os.path.join(COMPRESS_CACHE_DIRECTORY, d, f)
self.assertNotIn(full_save_name, c.sites[d])
c.save(example_url, html_string)
zip_name = "{}.gz".format(full_save_name)
self.assertTrue(os.path.exists(zip_name))
self.assertIn(full_save_name, c.sites[d])
# cleanup
os.remove(zip_name)
def test_cache_save_new(self):
c = GzipCache(COMPRESS_CACHE_DIRECTORY)
html_string = "<html></html>".encode("utf-8")
sample_url = "http://www.sample.com/testpage.html"
d = dir_domain(sample_url)
f = clean_url_hash(sample_url)
DIRECTORY = os.path.join(COMPRESS_CACHE_DIRECTORY, d)
# the www_sample_com directory should not exist until the file is cached
self.assertFalse(os.path.exists(DIRECTORY))
self.assertNotIn(d, c.sites)
c.save(sample_url, html_string)
full_save_name = os.path.join(DIRECTORY, f)
zip_name = "{}.gz".format(full_save_name)
self.assertIn(full_save_name, c.sites[d])
self.assertTrue(os.path.exists(zip_name))
# remove this after the test is done
shutil.rmtree(DIRECTORY)
