add script to seperate dataset based on cells

scripts/seperatedataset.py

@@ -0,0 +1,37 @@
+#!/bin/python
+import tarfile
+import argparse
+
+if __name__ == "__main__":
+	parser = argparse.ArgumentParser("KissSplitDataset")
+	parser.add_argument('--data', '-d', required=True, help="Data input tar file")
+	parser.add_argument('--out', '-o', required=True, help="output file")
+	parser.add_argument('--extract_cells', '-e', required=True, help="The cells to extract from the input")
+	parser.add_argument('--create_inverse', '-i', action="store_true", help="The cells to extract from the input")
+	args = parser.parse_args()
+
+	cells = list[int]()
+	tokens = args.extract_cells.split(',')
+	for token in tokens:
+		cells.append(int(token))
+		print("will extract cell " + token)
+
+	with tarfile.open(args.out, mode="w") as outfile:
+		with tarfile.open(args.data, mode="r") as tar:
+			if args.create_inverse:
+				inverseoutfile = tarfile.open("inverse_" + args.out, mode="w")
+			for file_info in tar:
+				if file_info.isfile():
+					tokens = file_info.name.split("-")
+					if len(tokens) != 3:
+						print("Could not parse filename " + file_info.name)
+						continue
+					if int(tokens[1]) in cells:
+						print("Adding " + file_info.name)
+						outfile.addfile(file_info, tar.extractfile(file_info))
+					elif args.create_inverse:
+						print("Inverse " + file_info.name)
+						inverseoutfile.addfile(file_info, tar.extractfile(file_info))
+			if args.create_inverse:
+				inverseoutfile.close()
+