#!/usr/bin/env python3
#this file = selenium2.py
# Run it: ./selenium2.py "https://example.com"
# Based on HTML title tag being or not being "Just a moment...",
# not a 140-second wait or however long

## Bash:
#read -p "url: " url; ./selenium2.py $url | tail -n1 | xargs -d "\n" sh -c 'for args do path=$(echo "$args" | sed "s/.* to .//g" | sed "s/.$//g"); path2=$(echo "$path" | sed "s/^...............//g"); path3=$(echo $path2 | sed "s/\/[^\/]*$//g"); ssh user@10.0.0.2 "mkdir -p /path/to/$path3"; rsync -a --info=progress2 "$path" user@10.0.0.2:/path/to/$path2; done' _
## rsync -a --info=progress2 /src/path/ user@10.0.0.2:/path/to/memento/src/
## ssh user@10.0.0.2
## find /path/to/memento/src -maxdepth 1 -type d | tail -n+2 | xargs -d "\n" sh -c 'for args do echo -n "$args"; echo /memento/; done' _ | xargs -d "\n" sh -c 'for args do rsync -a --info=progress2 "$args" /path/to/memento/; done' _

import sys
import time
import re
import io
import os
import shutil
from datetime import datetime, timezone
from selenium import webdriver
from selenium.webdriver.common.proxy import *

def wait_for_title_change(driver, initial_title, timeout=300):
    end_time = time.time() + timeout
    while time.time() < end_time:
        current_title = driver.title
        if current_title != initial_title:
            print(f"Title changed to: {current_title}")
            return current_title
        time.sleep(1)
    print("Timeout waiting for title to change.")
    return None

def main():
    options = webdriver.ChromeOptions()
    options.binary_location = "/usr/bin/brave-browser"
    options.accept_untrusted_certs = True
    options.add_argument('--ignore-certificate-errors')
    driver = webdriver.Chrome(options=options)
#    url = sys.argv[2]
    url = sys.argv[1]
    urlsafe = re.sub(r'[:/?=@&(),+*%#]', '-', url)
    driver.get(url)

#    time.sleep(int(sys.argv[1]))
    # Wait for the title to change from "Just a moment..."
    new_title = wait_for_title_change(driver, "Just a moment...")
    if new_title:
        now = datetime.now(timezone.utc)
        formatted_time = now.strftime("%Y%m%d%H%M%S")
        file_time_url = formatted_time + "-" + urlsafe
        with open(file_time_url, "w", encoding="utf-8") as w:
            w.write(driver.page_source)
        with open(file_time_url + ".txt", "w", encoding="utf-8") as w:
            w.write(url + "\n")

        directory_path = formatted_time + "/memento/" + formatted_time + "/" + re.sub(r"https?://", "", url)
        os.makedirs(directory_path, exist_ok=True)
        source_file = file_time_url
        destination_dir = directory_path
        destination_file = os.path.join(destination_dir, "index.html")
        if not os.path.exists(destination_file):
            shutil.copy2(source_file, destination_file)
            print(f"Copied '{source_file}' to '{destination_file}'")
        else:
            print(f"'{destination_file}' already exists. No file copied.")

        driver.quit()

if __name__=="__main__":
    main()