DocuSearch

• 

  DocuSearch

Inspiration

-- We wanted to understand how search engines actually work instead of blindly using libraries that hide the core logic. The goal was to challenge ourselves to build something real, minimal, and technically honest from scratch.

What it does

-- DocuSearch indexes and searches TXT, PDF, and DOC files using custom-built parsing and exact-match search logic. It highlights results in real time and shows clean document previews with basic performance metrics.

How we built it

-- We built the backend with FastAPI and implemented our own parsing, indexing, and search pipeline without external search libraries. The frontend was written using pure HTML, CSS, and vanilla JavaScript to keep full control over behavior and performance.

Challenges we ran into

-- Handling different document formats consistently was difficult, especially PDFs with complex layouts and embedded elements. Maintaining search speed and accurate highlighting without prebuilt tools required multiple iterations of optimization.

Accomplishments that we're proud of

-- We successfully built a functional document search engine completely from scratch with no frameworks or search libraries. The project clearly demonstrates our understanding of backend APIs, frontend rendering, and core search mechanics.

What we learned

-- We learned how real-world search systems manage parsing, indexing, and performance trade-offs. More importantly, we learned when simplicity is a strength and when abstraction becomes a liability.

What's next for DocuSearch

-- We plan to improve indexing efficiency and add optional fuzzy search while keeping the core logic transparent. Future versions may support richer document formats using modular extensions without compromising simplicity.

Built With

• css
• fastapi
• html
• javascript
• python