def _get_file_hash(self, path, method='ahash'):
func = imagehash.average_hash
if method == 'phash':
func = imagehash.phadh
elif method == 'dhash':
func = imagehash.dhash
try:
res = func(Image.open(path))
info("hash of %s is: %s" % (path, res))
return res
except IOError as e:
warning("load file %s error: %s" % (path, repr(e)))
def Find(self):
info("find duplicate images in %s." % self._path)
self.Started()
dir_hash = self._get_dir_hash(self._path)
for k, v in dir_hash.iteritems():
if len(v) > 1:
info("The next %d lines file are same: " % len(v))
v.sort()
[info(" %03d. %s" %
(ind + 1, unicode(v[ind]))) for ind in range(len(v))]
hashs = dir_hash.keys()
for i in range(len(hashs) - 1):
for j in range(i + 1, len(hashs)):
dist = hashs[i] - hashs[j]
if dist < self._distance:
info("The next 2 lines file maybe are same: ")
info(" %03d. %s" % (1, unicode(dir_hash[hashs[i]][0])))
info(" %03d. %s" % (2, unicode(dir_hash[hashs[j]][0])))
self.Completed()
def post(self, content, client, is_device):
core_response = core.info(content["message_id"], client, is_device)
response = jsonify(core_response['data'])
response.status_code = core_response['status_code']
return response
def get(self, message_id):
try:
client, is_device = get_any_from_header(request)
return core.info(message_id, client, is_device)
except InvalidAuthentication as ia:
return FAILED(ia)
def Find(self):
info("using [%s] find duplicate files in %s." %
(self._method, self._path))
self.Started()
self._path_files = {}
self._size_dict = {}
self._get_dir_size(self._path)
emptyfiles = self._size_dict.pop(0, [])
if len(emptyfiles) > 0:
info("The next %d lines file are empty: " % len(emptyfiles))
if self._auto_first:
info("auto keep: %s" % emptyfiles[0])
for p in emptyfiles[1:]:
info("auto delete: %s" % p)
try:
os.unlink(p)
except OSError:
pass
if self._is_cache:
self._get_cache_size()
for siz, files in self._size_dict.iteritems():
if self._is_cache is False and len(files) <= 1:
continue
hashs = self._get_filelist_hash(files)
for k, v in hashs.iteritems():
if len(v) < 2:
continue
info("The next %d lines file are same: " % len(v))
if self._auto_first:
info("auto keep: %s" % v[0])
for p in v[1:]:
info("auto delete: %s" % p)
try:
os.unlink(p)
except OSError:
pass
if self._is_cache:
self._cc.execute(
"delete from files where path=?", (p, ))
self._cache_conn.commit()
else:
[info(" %03d. %s" %
(ind + 1, v[ind])) for ind in range(len(v))]
self.Completed()
