def test_link(self):
hfs = HashFS(self.tmp_dir)
key = hfs.put(self.test_dir / 'data/think-hires.jpg')
self.assertRaises(FileNotFoundError,
lambda: hfs.link(key, 'data/think.jpg', True))
def test_get_simple(self):
original_file = self.test_dir / 'data/think-hires.jpg'
dst_file = self.tmp_dir / 'think-hires.jpg'
hfs = HashFS(self.tmp_dir, blocksize=1024 * 1024)
objkey = hfs.put(original_file)
hfs.get(objkey, dst_file)
self.assertEqual(self.md5sum(original_file), self.md5sum(dst_file))
def test_update_log(self):
original_file = 'data/think-hires.jpg'
hfs = HashFS(self.tmp_dir, blocksize=1024 * 1024)
store_log = os.path.join(self.tmp_dir, 'hashfs', 'log', 'store.log')
open(store_log, 'a').close()
hfs.update_log([original_file])
with open(store_log, 'r') as log_file:
self.assertIn(original_file, log_file.read())
def test_put1024K_pathexistence_level2(self):
hfs = HashFS(self.tmp_dir)
hfs.put(self.test_dir / 'data/think-hires.jpg')
m = hashlib.md5()
m.update('think-hires.jpg'.encode())
h = m.hexdigest()
fullpath = os.path.join(self.tmp_dir, 'hashfs', h[:2], h[2:4], 'think-hires.jpg')
self.assertTrue(os.path.exists(fullpath))
def test_put1024K_toomany_levels(self):
hfs = HashFS(self.tmp_dir, levels=17)
hfs.put(self.test_dir / 'data/think-hires.jpg')
m = hashlib.md5()
m.update('think-hires.jpg'.encode())
m.hexdigest()
fullpath = os.path.join(self.tmp_dir, 'hashfs', '58', '41', 'de', '64',
'a3', '7b', 'a5', 'af', '5c', 'f3', '19', 'd6',
'50', 'c1', '4a', 'b3', 'think-hires.jpg')
self.assertTrue(os.path.exists(fullpath))
def test_remove_hash(self):
idx = MultihashIndex('dataset-spec', self.tmp_dir, self.tmp_dir)
data = str(self.test_dir / 'data')
idx.add(data, '')
idx.add(str(self.test_dir / 'data2'), '')
hfs = HashFS(self.tmp_dir, blocksize=1024 * 1024)
o = Objects('dataset-spec', self.tmp_dir)
o.commit_index(self.tmp_dir, data)
for h in hash_list:
with open(os.path.join(self.tmp_dir, 'hashfs', 'log', STORAGE_LOG)) as f:
self.assertTrue(h in f.read())
for h in hash_list:
hfs.remove_hash(h)
for h in hash_list:
with open(os.path.join(self.tmp_dir, 'hashfs', 'log', STORAGE_LOG)) as f:
self.assertFalse(h in f.read())
def test_reset_log(self):
hfs = HashFS(self.tmp_dir, blocksize=1024 * 1024)
store_log = os.path.join(self.tmp_dir, 'hashfs', 'log', 'store.log')
open(store_log, 'a').close()
hfs.reset_log()
self.assertFalse(os.path.exists(store_log))
def test_put(self):
hfs = HashFS(self.tmp_dir)
hfs.put(self.test_dir / 'data/think-hires.jpg')
for files in hfs.walk():
for file in files:
self.assertTrue(file in hfsfiles)