Skip to content

About

Scan an image of a cooking recipe to recipe json

Resources

Readme
Apache-2.0 license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository
BranchesTags

Repository files navigation

Scan To Recipe

A Python application that reads images from a WebDAV server or local folder, processes them with OpenAI-compatible vision models, and saves the results to another WebDAV folder.

This has been developed with help OpenCodeCI

The Webdav support is intended to be used with Nextcloud and the Nextcloud Recipe Application.

This has been tested with the model qwen3-vl-4b-instruct running in a LocalAI instance.

Features

  • WebDAV Integration: Connect to source and destination WebDAV servers
  • AI Vision Processing: Uses OpenAI GPT-4 Vision or compatible models to analyze images
  • Recipe Extraction: Optimized prompts for extracting recipe information from food images
  • Batch Processing: Process multiple images in a directory
  • Error Handling: Robust retry logic and error recovery
  • Flexible Configuration: Environment-based configuration management
  • Multiple Output Formats: Saves results as both JSON and text files
  • Self-hosted AI: Can be used with self-hosted AI platform like LocalAI or Open WebUI

Installation

  1. Clone the repository:
git clone <repository-url>
cd ScanToCookbook
  1. Install dependencies:
pip install -r requirements.txt
  1. Copy and configure environment variables:
cp .env.example .env

Configuration

Edit the .env file with your configuration:

WebDAV Source

WEBDAV_SOURCE_HOST=https://your-source-server.com/webdav
WEBDAV_SOURCE_USERNAME=your-username
WEBDAV_SOURCE_PASSWORD=your-password

WebDAV Destination

WEBDAV_DEST_HOST=https://your-destination-server.com/webdav
WEBDAV_DEST_USERNAME=your-username
WEBDAV_DEST_PASSWORD=your-password

OpenAI/LLM Configuration

OPENAI_API_KEY=your-openai-api-key
OPENAI_BASE_URL=https://api.openai.com/v1  # or custom endpoint
LLM_MODEL=gpt-4-vision-preview
VISION_PROMPT=Analyze this image and extract recipe information...

Optional Settings

MAX_IMAGE_SIZE=10485760  # 10MB
SUPPORTED_FORMATS=jpg,jpeg,png
TEMP_DIR=/tmp/scan_to_cookbook
MAX_RETRIES=3
RETRY_DELAY=2

Usage

WebDAV Mode (Default)

Process all images in a directory:

python src/main.py --source-dir /recipes --dest-dir /processed

Process Single Image

python src/main.py --single "/recipes/pasta.jpg" --dest-dir /processed

Test Connections

python src/main.py --test

Verbose Logging

python src/main.py --source-dir /recipes --dest-dir /processed --verbose

Local Test Mode

Single File:

python src/main.py --local-test --local-input ./test_image.jpg

Directory with Progress Bar:

python src/main.py --local-test --local-input ./recipes --local-output ./processed --progress

Custom Configuration:

# Using .env file
python src/main.py --local-test --local-config ./my_local.env --local-input ./images

# Using YAML file
python src/main.py --local-test --local-config ./local.yaml --local-input ./photos

Default Local Configuration:

# Uses local.env or local.yaml (or environment variables)
python src/main.py --local-test --local-input ./test_images

Command Line Options

WebDAV Mode

  • --source-dir: Source WebDAV directory (default: "/")
  • --dest-dir: Destination WebDAV directory (default: "/processed")
  • --test: Test connections only
  • --single: Process a single image file
  • --verbose, -v: Enable verbose logging

Local Test Mode

  • --local-test: Enable local file processing mode
  • --local-input: Local input file or directory path
  • --local-output: Local output directory (default: "./output")
  • --local-config: Local config file (.env or .yaml)
  • --progress: Show progress bar for batch operations (default: True)
  • --help: Show help message

Mode Selection

The application automatically detects the mode:

  • WebDAV Mode: Default (no --local-test flag)
  • Local Mode: When --local-test flag is used

Output Files

For each processed image, the application creates:

  1. JSON Analysis: {filename}_analysis.json

    • Contains structured analysis results
    • Includes metadata (timestamp, model, prompt)
    • Machine-readable format

  2. Text Recipe: recipe-name/recipe.json

    • Json format using schema.org recipe standard

Logging

The application logs to:

  • Console output
  • logs/scan_to_cookbook.log file

Log levels:

  • INFO: General operation information
  • WARNING: Non-critical issues
  • ERROR: Critical errors
  • DEBUG: Detailed debugging information (use --verbose)

Error Handling

  • Connection Failures: Automatic retry with configurable delays
  • Invalid Images: Skips corrupted or unsupported files
  • Large Images: Automatically resizes images over the size limit
  • API Errors: Handles rate limits and API failures gracefully

Supported Image Formats

  • JPEG (.jpg, .jpeg)
  • PNG (.png)
  • GIF (.gif)
  • WebP (.webp)
  • BMP (.bmp)

Examples

Recipe Processing Workflow

  1. Upload recipe images to /incoming on your WebDAV server
  2. Run the processor:
python src/main.py --source-dir /incoming --dest-dir /recipes
  1. Results appear in /recipes with:
    • pasta_analysis.json - Structured data
    • pasta_recipe.txt - Human-readable recipe

Custom Prompt Example

Set a custom vision prompt in .env:

VISION_PROMPT=Extract detailed cooking instructions, ingredients with measurements, cooking time, temperature, and serving size from this recipe image.

Nextcloud Cookbook App Workflow

Edit the .env file with your configuration:

Configure the source folder of the images to scan

WebDAV Source

WEBDAV_SOURCE_HOST=https://<your-nextcloud-server>/remote.php/dav/files/<username>/Bilder
WEBDAV_SOURCE_USERNAME=your-username
WEBDAV_SOURCE_PASSWORD=your-app-apssword

WebDAV Destination

WEBDAV_DEST_HOST=https://<your-nextcloud-server>/remote.php/dav/files/<username>/Rezepte
WEBDAV_DEST_USERNAME=your-username
WEBDAV_DEST_PASSWORD=your-app-password

Also you can use the local test mode and copy or upload the data manually into to the recipe folder.

Configuration

WebDAV Mode Configuration

Edit the .env file for WebDAV settings:

# WebDAV Source
WEBDAV_SOURCE_HOST=https://your-source-server.com/webdav
WEBDAV_SOURCE_USERNAME=your-username
WEBDAV_SOURCE_PASSWORD=your-password

# WebDAV Destination
WEBDAV_DEST_HOST=https://your-destination-server.com/webdav
WEBDAV_DEST_USERNAME=your-username
WEBDAV_DEST_PASSWORD=your-password

# OpenAI Configuration
OPENAI_API_KEY=your-openai-api-key
LLM_MODEL=gpt-4-vision-preview

Local Mode Configuration

For local testing, copy and configure:

# Copy template
cp local.env.example local.env
# Edit with your settings
nano local.env

Or use YAML format:

cp local.yaml.example local.yaml
# Edit with your settings
nano local.yaml

Required Local Settings:

  • OPENAI_API_KEY: Your OpenAI API key
  • LOCAL_INPUT_DIR: Directory containing test images
  • LOCAL_OUTPUT_DIR: Directory for processed results

Optional Local Settings:

  • MAX_IMAGE_SIZE: Maximum image size before resizing (default: 10MB)
  • SUPPORTED_FORMATS: Image formats to process (default: jpg,jpeg,png)
  • LOG_LEVEL: Logging level (default: INFO)

Troubleshooting

Common Issues

  1. WebDAV Connection Failed

    • Verify server URL and credentials
    • Check network connectivity
    • Ensure WebDAV is enabled on server

  2. OpenAI API Error

    • Verify API key is valid
    • Check API quota and rate limits
    • Ensure model supports vision (gpt-4-vision-preview)

  3. Local Mode Issues

    • Verify OpenAI API key is set in local config
    • Check input directory exists and contains images
    • Ensure output directory is writable

  4. Image Processing Errors

    • Check image file format is supported
    • Verify image isn't corrupted
    • Check file size limits

Debug Mode

Run with verbose logging to diagnose issues:

# WebDAV mode
python src/main.py --test --verbose

# Local mode
python src/main.py --local-test --local-input ./test --verbose

Development

Project Structure

ScanToCookbook/
├── src/
│   ├── __init__.py
│   ├── config.py             # WebDAV configuration management
│   ├── local_config.py       # Local configuration management
│   ├── webdav_handler.py     # WebDAV operations
│   ├── local_processor.py    # Local file processing
│   ├── llm_processor.py      # OpenAI API client
│   ├── image_utils.py        # Image processing utilities
│   ├── file_utils.py         # File handling utils
│   └── main.py               # Main application logic
├── logs/                     # Log files
├── temp/                     # Temporary files (WebDAV mode only)
├── output/                   # Local output directory
├── requirements.txt
├── .env.example             # WebDAV configuration template
├── local.env.example        # Local configuration template (.env)
├── local.yaml.example       # Local configuration template (YAML)
└── README.md

Dependencies

  • webdavclient3: WebDAV client library
  • openai: OpenAI API client
  • python-dotenv: Environment variable management
  • PyYAML: YAML configuration file support
  • Pillow: Image processing
  • requests: HTTP requests
  • tqdm: Progress bar for batch operations (included with OpenAI)

Local AI Platform

  • Install LocalAI, see LocalAI Docker Installation
  • Install the model qwen3-vl-4b-instruct fromthe model gallery of your local ai installation.
  • Use your local ai platform (e.g http://localhost:8080/v1) as OpenAI url.
  • Use qwen3-vl-4b-instruct as model nale in your configuration.

License

Apache License

About

Scan an image of a cooking recipe to recipe json

Resources

Readme
Apache-2.0 license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages