xnl-h4ck3r/xnldorker v2.3 on GitHub

v2.3
- New
  - -sb/--show-browser argument now accepts optional values tabs or windows. Use -sb windows to open each source in a separate browser window instead of tabs. Default is tabs if no value specified.
  - Tabs and windows now close automatically when each source completes (unless --debug is used). Use --debug flag to keep all tabs/windows open for inspection and debugging.
  - If --debug is passed then keep all tabs or windows open until the user presses Enter to end the program. This lets you see the page the source stopped on.
  - Add xnldorker_*.html to .gitignore to ignore any debug files created.
  - Check for a different google captcha that appears with the text detected unusual traffic from your computer.
  - Add some ant-bot measures to try and reduce detection.
    - Added browser launch arguments to disable automation detection features like AutomationControlled
    - Injected JavaScript to hide the navigator.webdriver property that browsers expose when controlled by automation
    - Added realistic HTTP headers including Accept-Language, DNT, and proper Accept headers
    - Set a consistent User-Agent across all contexts
    - Added a small random delay (0.5-1.5 seconds) before navigation to appear more human-like
    - More comprehensive JavaScript injections - hiding plugins, languages, chrome runtime, and permissions API
- Changed
  - BUG FIX: Fixed memory leak issue when getting Google endpoints. This often led to the program getting Killed in the terminal.
  - BUG FIX: If more than one source required a captcha to be submitted at the same time, it would only listen for the last source word in the terminal. Each source calling wait_for_word_or_sleep() would overwrite the previous stdin reader.
  - BUG FIX: Improved error handling when browser windows are manually closed during searches. All sources now properly detect and log when the browser is closed, displaying messages like "[ Source ] Search aborted - got X results" instead of throwing raw errors.
  - BUG FIX: Fixed inconsistent page creation across sources. All sources now correctly use context.new_page() instead of some using browser.new_page().
  - BUG FIX: Set explicit viewport size (1280x720) for browser context to prevent small black windows.
  - BUG FIX: Browser and context are no longer closed when using -sb/--show-browser. The browser instance now stays open until the user manually closes it, preventing all windows from disappearing when searches complete.
  - BUG FIX: Baidu was only checking for Captcha on the first page
  - BUG FIX: Fixed an issue with Seznam source where it was timing out after pressing the Next button for the first time.
  - BUG FIX: Sometimes Baidu threw the error ERROR getBaidu: Page.query_selector_all: Execution context was destroyed, most likely because of a navigation after clicking the next page.
  - BUG FIX: Endpoints weren't correctly retrieved from a Bing page because they changed to site.
  - Changed back from Firefox to Chromium browser engine for improved stability and reliability when using -sb/--show-browser option.
  - Change help text for --antibot-timeout to show default value, and add the parameter when showing the options using -v.
  - BUG FIX: Seznam was very very slow when not showing the browser. It is now the same speed regardless whether the browser is shown or not.
  - Seznam - Only check further for links and clicking next if the text "Bohužel jsem nic nenašel" (Unfortunately I didn't find anything) isn't shown.
  - Baidu - Only wait for the popup to disappear and check further for links and clicking next if the text "抱歉，未找到相关结果" (Sorry, no relevant results were found) isn't shown.
  - If --debug was passed and a source is complete with 0 results, write the contents of the page to file so it can be checked for any potential problems.