def find_dupes(path='tmp'):
dups = []
record = {}
d = DiskWalk(path)
files = d.paths
for file in files:
compound_key = (getsize(file), create_checksum(file))
if compound_key in record:
dups.append(file)
else:
record[compound_key] = file
return dups
def find_figure_data(self,directory):
directory = self.condition_path(directory)
dw = DiskWalk(directory)
paths_bag_ext = dw.enumerate_paths_with_ext(FILE_TOOLS.bag_ext)
figure_paths = []
for path in paths_bag_ext:
(dir,file) = os.path.split(path)
if file.startswith(FILE_TOOLS.bag_prefix):
dw_path = DiskWalk(dir)
figure_files_in_dir = dw_path.enumerate_paths_with_ext(FILE_TOOLS.figure_ext)
tracking_files_in_dir = dw_path.enumerate_paths_with_ext(FILE_TOOLS.tracking_ext)
if tracking_files_in_dir > 1 and (self.overwrite or len(figure_files_in_dir) == 0):
figure_paths.append(path)
# else:
# print(path + " tracking data already processed!")
return figure_paths
def find_video_data(self,directory):
directory = self.condition_path(directory)
dw = DiskWalk(directory)
paths_bag_ext = dw.enumerate_paths_with_ext(FILE_TOOLS.bag_ext)
video_paths = []
for path in paths_bag_ext:
(dir,file) = os.path.split(path)
if file.startswith(FILE_TOOLS.bag_prefix):
dw_path = DiskWalk(dir)
video_files_in_dir = dw_path.enumerate_paths_with_ext(FILE_TOOLS.video_ext)
if self.overwrite or len(video_files_in_dir) == 0:
video_paths.append(path)
# else:
# print(path + " bag data already processed!")
return video_paths
def find_tracking_data(self,directory):
directory = self.condition_path(directory)
dw = DiskWalk(directory)
paths_tracking_ext = dw.enumerate_paths_with_ext(FILE_TOOLS.tracking_ext)
tracking_paths = []
for path in paths_tracking_ext:
(dir,file) = os.path.split(path)
if file.startswith(FILE_TOOLS.tracking_prefix):
dw_path = DiskWalk(dir)
tracking_files_in_dir = dw_path.enumerate_paths_with_ext(FILE_TOOLS.tracking_ext)
# one tracking file should already exist, more are produced when analyzed
if self.overwrite or len(tracking_files_in_dir) == 1:
tracking_paths.append(path)
# else:
# print(path + " tracking data already processed!")
return tracking_paths
def get_image_paths(self,directory):
dw = DiskWalk(directory)
paths_images_ext = dw.enumerate_paths_with_ext(self.image_ext)
return paths_images_ext