instant-data-scraper-utils/bin/stage-2/remove-utm.py
2024-07-16 18:08:36 -04:00

79 lines
2.7 KiB
Python
Executable file

#!/usr/bin/env python
# Script Name: remove-utm
import os
import csv
import sys
from pyfiglet import Figlet
from halo import Halo
# Define the paths based on the project details
PROJECT_ROOT = "/home/ld/mgk-scrapes"
CURRENT_DATASET = os.path.join(PROJECT_ROOT, "current-data")
DATA_DIRECTORY = os.path.join(CURRENT_DATASET, ".data")
STAGE_2_DIRECTORY = os.path.join(DATA_DIRECTORY, "stage-2")
def process_csv_file_remove_tracking_params(file_path):
"""Remove UTM, CID, and CMMMC parameters from URLs in the given CSV file."""
with open(file_path, 'r') as file:
reader = csv.reader(file)
rows = list(reader)
if not rows:
return
modified_rows = []
for row in rows:
modified_row = []
for cell in row:
if "?utm" in cell:
modified_cell = cell.split("?utm")[0]
elif "?cid" in cell:
modified_cell = cell.split("?cid")[0]
elif "?cm_mmc" in cell:
modified_cell = cell.split("?cm_mmc")[0]
else:
modified_cell = cell
modified_row.append(modified_cell)
modified_rows.append(modified_row)
with open(file_path, 'w', newline='') as file:
writer = csv.writer(file)
writer.writerows(modified_rows)
def process_csv_files(stage_directory):
"""Process all CSV files in the given stage directory to remove UTM, CID, and CMMMC parameters from URLs."""
for state_dir in os.listdir(stage_directory):
state_path = os.path.join(stage_directory, state_dir)
if os.path.isdir(state_path):
spinner = Halo(text=f'Processing {state_dir}', spinner='dots')
spinner.start()
for county_dir in os.listdir(state_path):
county_path = os.path.join(state_path, county_dir)
if os.path.isdir(county_path):
csv_files = [file for file in os.listdir(county_path) if file.endswith('.csv')]
for file in csv_files:
file_path = os.path.join(county_path, file)
process_csv_file_remove_tracking_params(file_path)
spinner.succeed(f'Finished processing {state_dir}')
def main():
figlet = Figlet(font='slant')
script_name = "remove-utm".replace("-", " ").title()
print(figlet.renderText(script_name))
if not os.path.exists(STAGE_2_DIRECTORY):
print(f"Error: The directory {STAGE_2_DIRECTORY} does not exist.")
sys.exit(1)
print(f"Removing tracking parameters in {STAGE_2_DIRECTORY}...")
process_csv_files(STAGE_2_DIRECTORY)
final_spinner = Halo(spinner='dots', color='green')
final_spinner.start()
final_spinner.succeed("Tracking parameters removed.")
if __name__ == "__main__":
main()