Add LinkedIn scraper/chat CLI, Dockerfile, and docker-compose for RAG Gradio UI

[siva01c]

Motivation

• Convert the previous Jupyter notebook into a reusable CLI Python script to scrape LinkedIn posts and provide a RAG-powered chat UI.
• Provide a reproducible containerized environment so the scraper and Gradio UI can run consistently across systems.
• Integrate a retrieval pipeline so scraped posts can be indexed and queried with OpenAI + ChromaDB.
• Expose a simple developer workflow for running scraping in headless/headed modes and launching the chat UI.

Description

• Add linkedin_tool.py which implements login (cookie save/load), lazy-loading scrolling, post extraction, RagChat (Chromadb persistent collection and OpenAI chat completions), and a Gradio chat UI with CLI subcommands scrape, chat, and all.
• Add Dockerfile that uses python:3.11-slim, installs Chromium and required system libraries, installs Python dependencies from requirements.txt, and sets the default CMD to run the chat server with python linkedin_tool.py chat --host 0.0.0.0 --port 7860.
• Add docker-compose.yml which defines an osint service built from the repo and a selenium standalone-chrome service, maps ports (7860 for the app and 4444/7900 for Selenium), mounts the project, and injects SELENIUM_REMOTE_URL for remote browser usage.
• Update README.md with environment variable setup (cp .env.default .env), instructions for running scrape, chat, or all, Docker Compose usage, and notes about storage locations (cookies/, chromadb/, posts.json).

Testing

• No automated tests were added or executed for this change.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dbd49a00f3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Derive Chroma document IDs from post IDs instead of indices

The indexing logic uses chroma_id = f"topic_{idx}" and skips any existing ID, which makes the persistent linkedin_posts collection silently stale whenever posts.json changes order/content or a different profile is scraped later. In those cases, rows at the same index are treated as already indexed and never refreshed, so retrieval can return outdated or wrong-person context even though a new scrape was provided.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Avoid defaulting docker-compose command to chat-only mode

The compose service always launches python linkedin_tool.py chat ..., but run_chat hard-fails when posts.json is missing, so docker compose up --build on a fresh checkout exits immediately instead of bringing up a usable stack. This default command requires pre-scraped data that new environments do not have yet, making the advertised startup flow fail unless users run scrape manually first.

Useful? React with 👍 / 👎.