This is a production-ready AI-powered tool for summarizing and comparing tabular logs across multiple files (Excel, CSV, PDF). Designed for fast local use, it supports both file uploads and entire folders (ZIPs) and outputs intelligent summaries along with downloadable Excel reports.
- 🧠 AI-Based Summary using FLAN-T5
- 🧾 Handles Excel, CSV, PDF formats
- 📂 Accepts ZIP folders with multiple logs
- 🧮 Smart grouping of primary + secondary columns
- 🎨 Color-coded groups with legend support
- 📊 Generates comparison tables across files
- 📥 One-click Excel export with multiple sheets
- ⚙️ Built using Flask, Transformers, Pandas
| Layer | Tools/Packages |
|---|---|
| Frontend | HTML, CSS (custom), Flask Templates |
| Backend | Flask, Pandas, XlsxWriter, pdfplumber |
| AI/NLP | HuggingFace Transformers, FLAN-T5, SentenceTransformers |
| File Support | .xlsx, .xls, .csv, .pdf, .zip |
| Export Format | .xlsx with multiple sheets |
pip install -r requirements.txt
### 📁 2. Folder Structure
project/
├── app.py
├── templates/
├── static/
│ ├── styles.css
│ ├── demo_upload.png
│ ├── demo_columns.png
│ ├── demo_summary.png
│ └── demo_excel.png
├── uploads/
├── requirements.txt
├── README.md
## 🚀 3. Run the App
```bash
python app.pyVisit http://127.0.0.1:5000 in your browser.
When you click Download, the Excel file will contain:
- Sheet 1:
Supporting Summary Table - Sheet 2:
Comparison Across Files - Sheet 3+: Raw data from each uploaded file
(Each sheet is named after the original filename)
- 🔍 Model Used:
google/flan-t5-large - 🔗 Powered by Hugging Face Transformers
- 🧠 Embedded logic for summarizing patterns, relationships, and grouping insights
To reset session and delete uploaded files:
Open to suggestions and pull requests.
This tool was built to assist in internal audit and QA log analysis, but can be adapted for any structured tabular summarization use case.
MIT License — free to use, distribute, and modify.