title	FinTabula
emoji	📊
colorFrom	green
colorTo	green
sdk	gradio
sdk_version	6.16.0
python_version	3.11
app_file	app.py
pinned	false
short_description	Modern financial statement table extractor

FinTabula - Financial Statement Table Extractor

FinTabula is a modern, light, and beautiful web application designed to extract tabular data from financial statements (like Balance Sheets, Income Statements, and Cash Flow Statements) in PDF format and convert them into beautifully styled Excel workbooks or CSV files.

---

🚀 Key Features

• Dual Extraction Engines:
  • Local Engine (Offline & Free): Powered by pdfplumber to instantly parse vector-text PDF statements locally.
  • Gemini AI Engine (Highly Intelligent): Uses Google Gemini 1.5 Flash to intelligently scan and rebuild complex, multi-page, or scanned (image-only) tables into structured formats.
• Interactive Spreadsheet Editor:
  • Preview all extracted sheets inside a sleek, dark-themed browser grid.
  • Double-click to edit cell values directly in the browser.
  • Add/remove rows, columns, rename sheets, and delete sheets before exporting.
• Premium Excel Exporting:
  • Zebra striping for alternate row readability.
  • Bold, navy-filled headers with white text.
  • Auto-fitted column widths (no text truncations!).
  • Intelligent cell alignment (numbers align right, text aligns left).
  • Gridlines displayed by default.

---

🛠️ Quick Start

1. Requirements

• Python 3.10+
• Google Chrome, Safari, or Firefox browser

2. Launching the App

The dependencies have already been set up in your virtual environment. Simply open your terminal and run:

# 1. Navigate to the project directory
cd "/Users/mangalam/Desktop/Data Scraper"

# 2. Activate the virtual environment
source .venv/bin/activate

# 3. Start the FastAPI server
python -m uvicorn main:app --reload

After launching, open your browser and navigate to: 👉 http://127.0.0.1:8000

---

📖 How to Use

1. Upload a PDF: Drag and drop your financial statement PDF into the upload zone or click to select.
2. Choose Engine:
   • Select Local Engine for standard digital statements.
   • Select Gemini AI for complex, multi-page, or scanned statements.
3. Authentication for Gemini AI:
   • To unlock the Gemini AI mode, enter the Admin credentials:
     • Admin Username: ****
     • Admin Password: *******
   • Get a free Gemini API Key from Google AI Studio and enter it in the API Key field.
4. Specify Pages (Optional): Input page numbers (e.g. 1-3, 5) to parse only specific pages.
5. Extract & Edit: Click Extract Tables. Once loaded, edit in the spreadsheet, and download as Excel or CSV.

---

💡 Gemini AI vs. Local Engine: Why Gemini AI is Superior

For high-volume, complex, or official corporate disclosures (like SEC 10-K filings), Gemini AI is highly recommended over the Local Engine:

Feature	Local Engine	Gemini AI Engine
Grid Rebuilding	Uses strict coordinates (susceptible to misalignment if borders are absent)	Semantically parses structured cells based on reading flow
Scanned Documents (OCR)	Fails or returns garbage text on scanned/image-only PDFs	Extracts text and structure perfectly using multi-modal AI vision
Split Word Prevention	Minimizes word cuts but can split text on highly dense columns	Word-splitting is impossible since text elements are understood semantically
Hierarchical Headers	Places multi-level column titles across separate cells	Intelligently handles spanned column headers and merges cells correctly
Footnotes & Disclosures	Extracts footnotes into disjointed cells across the grid	Reconstructs footnote sentences across margins back into a clean footer

To transition from Local to Gemini AI, toggle the engine selector in the application, input the Admin credentials (Admin / Kukku404#), paste your API key, and run the extraction.