merge-csv-lists: refactor

This commit is contained in:
Lukáš Kucharczyk 2023-12-29 11:29:57 +01:00
parent 7ff425cb2f
commit a7c1936acf
Signed by: lukas
SSH Key Fingerprint: SHA256:vMuSwvwAvcT6htVAioMP7rzzwMQNi3roESyhv+nAxeg
1 changed files with 16 additions and 24 deletions

View File

@ -2,60 +2,52 @@
import argparse
def split_and_flat(input_string):
if "\n" in input_string:
_without_newlines = input_string.split("\n")
_without_empty = filter(lambda x: x != "", _without_newlines)
input_string = ",".join(_without_empty)
comma_split = input_string.split(",")
flat_list = []
for item in comma_split:
dash_split = item.split("-")
flat_list.extend([value.strip().replace("\n", "") for value in dash_split])
return ",".join(flat_list)
def unique_combined_list(*inputs):
# Combine lists
combined_list = [
item.strip().title() for input_list in inputs for item in input_list.split(",")
]
# Create an empty list to store the final unique names
final_list = []
# Check for reversed names
for name in combined_list:
parts = name.split()
# If the name has two words, check for its reversed variant
if len(parts) == 2:
first, last = parts
reversed_name = f"{last} {first}"
# If neither the name nor its reversed variant is in the final list, add the name
if name not in final_list and reversed_name not in final_list:
final_list.append(name)
# If it's a single-word name, simply add it if it's not in the final list
else:
if name not in final_list:
final_list.append(name)
# Sort the list
sorted_list = sorted(final_list)
# Convert the list back to a comma-separated string
output = ",".join(sorted_list)
return output
def main():
# Create an argument parser
parser = argparse.ArgumentParser(
description="Combine multiple comma-separated lists into one unique sorted list."
)
# Add a variable number of input lists
parser.add_argument("lists", nargs="+", type=str, help="Comma-separated lists.")
# Parse the arguments
args = parser.parse_args()
# If only one list is provided, use it twice
if len(args.lists) == 1:
args.lists.append(args.lists[0])
# Get the unique combined list
result = unique_combined_list(*args.lists)
# Process each list with split_and_flat if it contains a dash
processed_lists = [split_and_flat(lst) if "-" in lst else lst for lst in args.lists]
result = unique_combined_list(*processed_lists)
print(result)