Document Portal

A FastAPI-based Document Portal for uploading, analyzing, comparing, and chatting with documents using Retrieval-Augmented Generation (RAG). The project supports multiple document formats, multimodal retrieval (text, tables, images) via Unstructured.io + LangChain MultiVector, and includes authentication, pre-commit testing, and CI.

See the companion assignment write-up: Assignment.md

---

Features

• Multi-document ingestion with FAISS vector index per session
• Chat with your documents (RAG) using LangChain
• Multimodal chat for PDFs with text/table/image extraction (Unstructured.io) and MultiVectorRetriever
• Document analysis (metadata and summarization)
• Document comparison
• Authentication (signup/login) and protected routes
• LangChain InMemory cache to speed up repeated LLM calls
• Tests validated via pre-commit and GitHub Actions CI

---

Tech Stack

• API: FastAPI
• RAG: LangChain, FAISS
• Multimodal extraction: Unstructured.io (PDF partition), Pillow/OpenCV
• LLMs: OpenAI / Groq / Google (configurable)
• Auth: fastapi-users, SQLAlchemy (SQLite by default)
• PDF parsing: PyMuPDF (fitz)

---

Project Structure (high-level)

• api/main.py: FastAPI app, endpoints, startup, auth wiring
• src/document_ingestion/: Ingestion for text and multimodal
  • data_ingestion.py: ChatIngestor, DocHandler for many file types
  • mm_ingestion.py: Multimodal ingestion using Unstructured + MultiVector
• src/document_chat/: RAG logic
  • retrieval.py: Conversational RAG (text-only)
  • multimodal/: Indexer & retrieval chain for multimodal
• auth/: fastapi-users integration and models
• utils/: LLM cache, model loader, document ops, config loader
• templates/ and static/: Simple HTML UI
• tests/: Unit tests for endpoints, ingestion, analyzer, and flags

---

Setup

Prerequisites

• Python 3.10+
• API keys for one or more LLM providers (choose at least one):
  • OPENAI_API_KEY
  • GROQ_API_KEY
  • GOOGLE_API_KEY

Install

python -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install --upgrade pip
pip install -r requirements.txt

Configuration

• Environment variables (examples):
  • LLM_PROVIDER (default: openai) one of: openai, groq, google
  • FAISS_BASE (default: faiss_index)
  • UPLOAD_BASE (default: data)
  • FAISS_INDEX_NAME (default: index)
  • LLM_CACHE_ENABLED (default: true)
• Additional configuration lives in config/config.yaml (models, multimodal settings).
• The app also supports a combined JSON secret in apikeyliveclass env var; otherwise reads provider-specific env vars.

---

Run

uvicorn api.main:app --host 0.0.0.0 --port 8080 --reload

Open http://localhost:8080 for the login page. Use the Signup link to create an account, then login to access /app.

Optional Streamlit UI (if used):

streamlit run streamlit_ui.py

---

Key Endpoints

• Public UI: /, /login, /signup
• Protected example: /protected
• App UI (requires auth): /app
• Analyze: POST /analyze (upload file)
• Compare: POST /compare (upload reference, actual)
• Chat index: POST /chat/index
  • Form fields: files (upload), multimodal (true/false), use_session_dirs (true/false), k, chunk_size, chunk_overlap
• Chat query: POST /chat/query
  • Form fields: question, session_id (if use_session_dirs=true), multimodal (true/false), k

Example cURL

# Analyze
curl -X POST -F "file=@sample.pdf" http://localhost:8080/analyze

# Build chat index (text-only)
curl -X POST \
  -F "files=@sample.pdf" \
  -F "multimodal=false" \
  -F "use_session_dirs=true" \
  http://localhost:8080/chat/index

# Build chat index (multimodal)
curl -X POST \
  -F "files=@sample.pdf" \
  -F "multimodal=true" \
  -F "use_session_dirs=true" \
  http://localhost:8080/chat/index

# Query chat (text-only)
curl -X POST \
  -F "question=What is this document about?" \
  -F "use_session_dirs=true" \
  -F "session_id=<returned_session_id>" \
  -F "multimodal=false" \
  http://localhost:8080/chat/query

# Query chat (multimodal)
curl -X POST \
  -F "question=Summarize the figures and tables" \
  -F "use_session_dirs=true" \
  -F "session_id=<returned_session_id>" \
  -F "multimodal=true" \
  http://localhost:8080/chat/query

---

Supported File Types

Text and ingestion:

• .pdf, .docx, .pptx, .md, .txt, .csv, .xlsx, .xls, .db, .sqlite, .sqlite3

Multimodal (current focus):

• PDFs via unstructured.partition.pdf (extract texts, tables, and images)

---

Testing & Pre-commit

# Run tests
pytest tests/

# Install pre-commit hooks
pre-commit install
# Optionally run on all files once
pre-commit run --all-files

• Pre-commit is configured to run pytest before commits (.pre-commit-config.yaml).

---

Continuous Integration

• GitHub Actions workflow: .github/workflows/ci.yml
  • Unit tests on push/pull_request to dev
  • DeepEval jobs (text-only and multimodal) run in non-blocking mode to report metrics

---

Notes

• FastAPI lifespan initializes LLM cache and creates auth tables on startup.
• Session-based FAISS indices and assets are stored under faiss_index/<session> and data/<session>.
• Configure LLM provider and keys via environment variables; see Configuration above.

---

License

This project is for educational purposes. Review dependencies for their respective licenses.