def process_checksum(self, _file, allow_duplicate):
db = Db()
checksum = db.checksum(_file)
if(checksum is None):
log.info('Could not get checksum for %s.' % _file)
return None
# If duplicates are not allowed then we check if we've seen this file
# before via checksum. We also check that the file exists at the
# location we believe it to be.
# If we find a checksum match but the file doesn't exist where we
# believe it to be then we write a debug log and proceed to import.
checksum_file = db.get_hash(checksum)
if(allow_duplicate is False and checksum_file is not None):
if(os.path.isfile(checksum_file)):
log.info('%s already at %s.' % (
_file,
checksum_file
))
return None
else:
log.info('%s matched checksum but file not found at %s.' % ( # noqa
_file,
checksum_file
))
return checksum
def test_get_hash_does_not_exist():
db = Db()
random_key = helper.random_string(10)
assert db.get_hash(
random_key
) is None, 'Lookup for hash that should not exist did not return None'
def test_get_hash_exists():
db = Db()
random_key = helper.random_string(10)
random_value = helper.random_string(12)
# Test with explicit False value as 3rd param
db.add_hash(random_key, random_value, False)
assert db.get_hash(random_key) == random_value, 'Lookup for hash that exists did not return value'
def process_file(self, _file, destination, media, **kwargs):
move = False
if('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if(not media.is_valid()):
print '%s is not a valid media file. Skipping...' % _file
return
metadata = media.get_metadata()
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if(checksum is None):
if(constants.debug is True):
print 'Could not get checksum for %s. Skipping...' % _file
return
# If duplicates are not allowed and this hash exists in the db then we
# return
if(allow_duplicate is False and db.check_hash(checksum) is True):
if(constants.debug is True):
print '%s already exists at %s. Skipping...' % (
_file,
db.get_hash(checksum)
)
return
self.create_directory(dest_directory)
if(move is True):
stat = os.stat(_file)
shutil.move(_file, dest_path)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
shutil.copy2(_file, dest_path)
db.add_hash(checksum, dest_path)
db.update_hash_db()
return dest_path
def process_file(self, _file, destination, media, **kwargs):
move = False
if ('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if ('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if (not media.is_valid()):
print '%s is not a valid media file. Skipping...' % _file
return
metadata = media.get_metadata()
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if (checksum is None):
if (constants.debug is True):
print 'Could not get checksum for %s. Skipping...' % _file
return
# If duplicates are not allowed and this hash exists in the db then we
# return
if (allow_duplicate is False and db.check_hash(checksum) is True):
if (constants.debug is True):
print '%s already exists at %s. Skipping...' % (
_file, db.get_hash(checksum))
return
self.create_directory(dest_directory)
if (move is True):
stat = os.stat(_file)
shutil.move(_file, dest_path)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
shutil.copy2(_file, dest_path)
db.add_hash(checksum, dest_path)
db.update_hash_db()
return dest_path
def process_file(self, _file, destination, media, **kwargs):
move = False
if('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if(not media.is_valid()):
print('%s is not a valid media file. Skipping...' % _file)
return
media.set_original_name()
metadata = media.get_metadata()
directory_name = self.destination_folder.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if(checksum is None):
log.info('Could not get checksum for %s.' % _file)
return None
# If duplicates are not allowed then we check if we've seen this file
# before via checksum. We also check that the file exists at the
# location we believe it to be.
# If we find a checksum match but the file doesn't exist where we
# believe it to be then we write a debug log and proceed to import.
checksum_file = db.get_hash(checksum)
if(allow_duplicate is False and checksum_file is not None):
if(os.path.isfile(checksum_file)):
log.info('%s already at %s.' % (
_file,
checksum_file
))
return None
else:
log.info('%s matched checksum but file not found at %s.' % ( # noqa
_file,
checksum_file
))
return checksum
def process_file(self, _file, destination, media, **kwargs):
move = False
if ('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if ('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if (not media.is_valid()):
print('%s is not a valid media file. Skipping...' % _file)
return
metadata = media.get_metadata()
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if (checksum is None):
log.info('Could not get checksum for %s. Skipping...' % _file)
return
# If duplicates are not allowed then we check if we've seen this file
# before via checksum. We also check that the file exists at the
# location we believe it to be.
# If we find a checksum match but the file doesn't exist where we
# believe it to be then we write a debug log and proceed to import.
checksum_file = db.get_hash(checksum)
if (allow_duplicate is False and checksum_file is not None):
if (os.path.isfile(checksum_file)):
log.info('%s already exists at %s. Skipping...' %
(_file, checksum_file))
return
else:
log.info(
'%s matched checksum but file not found at %s. Importing again...'
% ( # noqa
_file, checksum_file))
self.create_directory(dest_directory)
if (move is True):
stat = os.stat(_file)
shutil.move(_file, dest_path)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
# Do not use copy2(), will have an issue when copying to a
# network/mounted drive using copy and manual
# set_date_from_filename gets the job done
shutil.copy(_file, dest_path)
self.set_utime(media)
db.add_hash(checksum, dest_path)
db.update_hash_db()
return dest_path
def process_file(self, _file, destination, media, **kwargs):
move = False
if ('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if ('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if (not media.is_valid()):
print('%s is not a valid media file. Skipping...' % _file)
return
media.set_original_name()
metadata = media.get_metadata()
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if (checksum is None):
log.info('Could not get checksum for %s. Skipping...' % _file)
return
# If duplicates are not allowed then we check if we've seen this file
# before via checksum. We also check that the file exists at the
# location we believe it to be.
# If we find a checksum match but the file doesn't exist where we
# believe it to be then we write a debug log and proceed to import.
checksum_file = db.get_hash(checksum)
if (allow_duplicate is False and checksum_file is not None):
if (os.path.isfile(checksum_file)):
log.info('%s already exists at %s. Skipping...' %
(_file, checksum_file))
return
else:
log.info(
'%s matched checksum but file not found at %s. Importing again...'
% ( # noqa
_file, checksum_file))
# If source and destination are identical then
# we should not write the file. gh-210
if (_file == dest_path):
print('Final source and destination path should not be identical')
return
self.create_directory(dest_directory)
if (move is True):
stat = os.stat(_file)
shutil.move(_file, dest_path)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
compatability._copyfile(_file, dest_path)
self.set_utime(media)
db.add_hash(checksum, dest_path)
db.update_hash_db()
return dest_path
def test_get_hash_does_not_exist():
db = Db()
random_key = helper.random_string(10)
assert db.get_hash(random_key) is None, 'Lookup for hash that should not exist did not return None'
def process_file(self, _file, destination, media, **kwargs):
move = False
if('move' in kwargs):
move = kwargs['move']
allow_duplicate = False
if('allowDuplicate' in kwargs):
allow_duplicate = kwargs['allowDuplicate']
if(not media.is_valid()):
print('%s is not a valid media file. Skipping...' % _file)
return
metadata = media.get_metadata()
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(media)
dest_path = os.path.join(dest_directory, file_name)
db = Db()
checksum = db.checksum(_file)
if(checksum is None):
log.info('Could not get checksum for %s. Skipping...' % _file)
return
# If duplicates are not allowed then we check if we've seen this file
# before via checksum. We also check that the file exists at the
# location we believe it to be.
# If we find a checksum match but the file doesn't exist where we
# believe it to be then we write a debug log and proceed to import.
checksum_file = db.get_hash(checksum)
if(allow_duplicate is False and checksum_file is not None):
if(os.path.isfile(checksum_file)):
log.info('%s already exists at %s. Skipping...' % (
_file,
checksum_file
))
return
else:
log.info('%s matched checksum but file not found at %s. Importing again...' % ( # noqa
_file,
checksum_file
))
self.create_directory(dest_directory)
if(move is True):
stat = os.stat(_file)
shutil.move(_file, dest_path)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
# Do not use copy2(), will have an issue when copying to a
# network/mounted drive using copy and manual
# set_date_from_filename gets the job done
shutil.copy(_file, dest_path)
self.set_utime(media)
db.add_hash(checksum, dest_path)
db.update_hash_db()
return dest_path
