Python ChatGPT Scraper with Proxies

I need a lightweight Python script that automatically submits a list of prompts to chatgpt.com, captures the full response for each prompt, and saves everything to a structured JSON file. The JSON should clearly separate:

• main text response
• any product listings with their image URLs
• sources or reference links
• tables or other rich-text blocks

Key requirements
• Proxy rotation: the script must cycle through a plain-text list of HTTP/S proxies so I can run several thousand prompts without rate-limit issues. (you can propose how/what proxy provider you prefer to use)
• Headless reliability: choose a stable approach—Playwright, Selenium, or another headless browser—so the page renders exactly as a normal user would see it (needed for product cards and tables).
• Prompt input: accept either a CSV or plain list; return one JSON object per prompt.
• Resilience: retry logic for failed loads, graceful handling of timeouts or blocked proxies.
• Clear README: quick-start instructions and dependencies.

I’ll be using the data for text-based analysis, so accuracy of the captured text responses is critical. Medium-sized scope is fine: a functional prototype with clean, well-commented code and basic error reporting meets my needs.