textanalysis

A Flask web application that analyzes texting conversations (and other text-dense screenshots) using chess-style move evaluations, Heavily vibe-coded so don't hope for the best readability

Installation

git clone https://github.com/ang-or-five/textanalysis.git
cd textanalysis
pip install -r requirements.txt

UV users:

git clone https://github.com/ang-or-five/textanalysis.git
cd textanalysis
uv pip install -r requirements.txt --system

Configuration

1. Copy .env.example to .env
2. Get a free API key from https://aistudio.google.com/api-keys
3. Paste your API key in .env

Usage

python app.py

Open http://localhost:5000 in your browser.

1. Drag and drop or click to upload one or more conversation screenshots
2. Select analysis options:
   • Concatenate - Treat multiple images as one continuous thread
   • OCR Only - Skip image analysis (faster)
   • Dense Analysis - Analyze every line in detail
3. Click "Analyze"
4. Browse results using the different view tabs

Demo

Upload a screenshot of a conversation and watch as each message gets evaluated with chess-style ratings:

• Brilliant - Exceptionally clever or witty
• Great Find - Great observation or comeback
• Best - The optimal response
• Excellent - Very well crafted
• Good - Solid, appropriate
• Book - Standard/expected
• Inaccuracy - Slightly off
• Mistake - Poor choice or timing
• Blunder - Major error or cringe
• Missed Win - Missed opportunity

Features

• Multiple View Modes

  • Append Replay - Watch messages appear one-by-one with animated evaluations
  • Semi-Annotated - Original image with evaluation icons overlaid
  • Full Annotated - Each message shown with evaluation and explanation
  • Interactive - Click through messages at your own pace

• Multi-Image Support

  • Upload multiple images at once
  • Concatenate mode - Combine multiple screenshots into one continuous analysis
  • Navigate between different analyzed images

• Analysis Options

  • Dense Mode - Analyze every single line/message exhaustively
  • OCR Only - Faster analysis without sending images to the model

• Notable Icons - Additional tags like "Winner", "Threat", "Sharp", "Forced", etc.

Tech Stack

• Backend: Flask (Python)
• OCR: EasyOCR
• LLM: OpenAI-compatible API (tested with Gemini)
• Image Processing: Pillow, cairosvg

API Endpoints

Endpoint	Method	Description
/	GET	Web interface
/api/upload	POST	Upload image file
/api/analyze	POST	Analyze single image
/api/analyze-multiple	POST	Analyze multiple images separately
/api/analyze-concatenated	POST	Analyze multiple images as one

Project Structure

./
├── app.py                 # Flask application
├── requirements.txt       # Python dependencies
├── .env                   # Environment variables (not tracked)
├── .env.example           # Example environment variables
├── static/
│   ├── js/
│   │   └── script.js      # Frontend JavaScript
│   ├── images/            # Evaluation icons (SVG)
│   ├── uploads/           # Uploaded images (generated)
│   └── analysis/          # Analysis outputs (generated)
└── templates/
    └── index.html         # Web interface

Extending the Analysis

The system prompt and evaluation schema can be customized in app.py:

• EVALUATION_ICONS - Available evaluation ratings
• NOTABLE_ICONS - Available notable tags
• SYSTEM_PROMPT - Instructions for the LLM

Credits

Evaluation icons are ripped from chess.com's analysis interface