def _generate_db(source):
"""Regenerate the hash.json database which contains all of the sha1 signatures of media files.
"""
result = Result()
source = os.path.abspath(os.path.expanduser(source))
extensions = set()
all_files = set()
valid_files = set()
if not os.path.isdir(source):
log.error('Source is not a valid directory %s' % source)
sys.exit(1)
subclasses = get_all_subclasses(Base)
for cls in subclasses:
extensions.update(cls.extensions)
all_files.update(FILESYSTEM.get_all_files(source, None))
db = Db()
db.backup_hash_db()
db.reset_hash_db()
for current_file in all_files:
if os.path.splitext(current_file)[1][1:].lower() not in extensions:
log.info('Skipping invalid file %s' % current_file)
result.append((current_file, False))
continue
result.append((current_file, True))
db.add_hash(db.checksum(current_file), current_file)
db.update_hash_db()
result.write()
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 test_reset_hash_db():
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 random_key in db.hash_db, random_key
db.reset_hash_db()
assert random_key not in db.hash_db, random_key
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 test_add_hash_explicit_write():
db = Db()
random_key = helper.random_string(10)
random_value = helper.random_string(12)
# Test with explicit True value as 3rd param
db.add_hash(random_key, random_value, True)
assert db.check_hash(random_key) == True, 'Lookup for hash did not return True'
# Instnatiate new db class to confirm random_key exists
db2 = Db()
assert db2.check_hash(random_key) == True
def test_add_hash_default_do_not_write():
db = Db()
random_key = helper.random_string(10)
random_value = helper.random_string(12)
# Test with default False value as 3rd param
db.add_hash(random_key, random_value)
assert db.check_hash(random_key) == True, 'Lookup for hash did not return True'
# Instnatiate new db class to confirm random_key does not exist
db2 = Db()
assert db2.check_hash(random_key) == False
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 test_get_all():
db = Db()
db.reset_hash_db()
random_keys = []
random_values = []
for _ in range(10):
random_keys.append(helper.random_string(10))
random_values.append(helper.random_string(12))
db.add_hash(random_keys[-1:][0], random_values[-1:][0], False)
counter = 0
for key, value in db.all():
assert key in random_keys, key
assert value in random_values, value
counter += 1
assert counter == 10, counter
def _generate_db(source):
"""Regenerate the hash.json database which contains all of the sha1 signatures of media files.
"""
result = Result()
source = os.path.abspath(os.path.expanduser(source))
if not os.path.isdir(source):
log.error('Source is not a valid directory %s' % source)
sys.exit(1)
db = Db()
db.backup_hash_db()
db.reset_hash_db()
for current_file in FILESYSTEM.get_all_files(source):
result.append((current_file, True))
db.add_hash(db.checksum(current_file), current_file)
log.progress()
db.update_hash_db()
log.progress('', True)
result.write()
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']
stat_info_original = os.stat(_file)
metadata = media.get_metadata()
if(not media.is_valid()):
print('%s is not a valid media file. Skipping...' % _file)
return
checksum = self.process_checksum(_file, allow_duplicate)
if(checksum is None):
log.info('Original checksum returned None for %s. Skipping...' %
_file)
return
# Run `before()` for every loaded plugin and if any of them raise an exception
# then we skip importing the file and log a message.
plugins_run_before_status = self.plugins.run_all_before(_file, destination)
if(plugins_run_before_status == False):
log.warn('At least one plugin pre-run failed for %s' % _file)
return
directory_name = self.get_folder_path(metadata)
dest_directory = os.path.join(destination, directory_name)
file_name = self.get_file_name(metadata)
dest_path = os.path.join(dest_directory, file_name)
media.set_original_name()
# 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)
# exiftool renames the original file by appending '_original' to the
# file name. A new file is written with new tags with the initial file
# name. See exiftool man page for more details.
exif_original_file = _file + '_original'
# Check if the source file was processed by exiftool and an _original
# file was created.
exif_original_file_exists = False
if(os.path.exists(exif_original_file)):
exif_original_file_exists = True
if(move is True):
stat = os.stat(_file)
# Move the processed file into the destination directory
shutil.move(_file, dest_path)
if(exif_original_file_exists is True):
# We can remove it as we don't need the initial file.
os.remove(exif_original_file)
os.utime(dest_path, (stat.st_atime, stat.st_mtime))
else:
if(exif_original_file_exists is True):
# Move the newly processed file with any updated tags to the
# destination directory
shutil.move(_file, dest_path)
# Move the exif _original back to the initial source file
shutil.move(exif_original_file, _file)
else:
compatability._copyfile(_file, dest_path)
# Set the utime based on what the original file contained
# before we made any changes.
# Then set the utime on the destination file based on metadata.
os.utime(_file, (stat_info_original.st_atime, stat_info_original.st_mtime))
self.set_utime_from_metadata(metadata, dest_path)
db = Db()
db.add_hash(checksum, dest_path)
db.update_hash_db()
# Run `after()` for every loaded plugin and if any of them raise an exception
# then we skip importing the file and log a message.
plugins_run_after_status = self.plugins.run_all_after(_file, destination, dest_path, metadata)
if(plugins_run_after_status == False):
log.warn('At least one plugin pre-run failed for %s' % _file)
return
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):
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 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