Magika

1.0.2 · active · verified Thu Apr 09

Magika is an AI-powered file type detection tool developed by Google. It leverages deep learning to accurately identify content types, supporting over 200 formats including binary and textual files. It boasts high accuracy (~99%) and fast inference times, making it suitable for security, data processing, and development workflows. The current version is 1.0.2, and it receives active development and regular updates.

Warnings

breaking Starting with version 1.0.2, Magika no longer automatically loads `.env` files. If your application relied on `python-dotenv` being implicitly managed by Magika, this behavior has changed.
Fix: Manually load `.env` files using `python-dotenv` if needed, before initializing Magika, or explicitly pass configuration to Magika if such options become available.
breaking Magika 1.0 represents a significant rewrite, with the core engine and CLI implemented in Rust, and the Python module being 'revamped.' While intended for easier integration, users migrating from pre-1.0 experimental versions (0.x.x) might encounter API changes or behavioral differences requiring code updates. The Python CLI itself also shifted from a pure Python implementation to a Rust binary wrapper around version 0.6.0.
Fix: Consult the official documentation and changelogs for specific migration paths and updated API usage when upgrading from 0.x versions. Review `Magika` class constructor arguments and identification methods.
gotcha For large files, using `magika.identify_bytes()` can lead to high memory consumption as it loads the entire file content into memory. This is not efficient for very large files.
Fix: Prefer `magika.identify_path(path)` or `magika.identify_stream(file_stream)` for large files, as these methods are optimized to read only necessary portions of the file by seeking, reducing memory footprint.
gotcha The `onnxruntime` dependency, while crucial for Magika's performance, can be substantial in size and may introduce platform-specific installation complexities. While Magika 1.0.2 improved support for Python 3.14 and removed some Windows-specific pins, `onnxruntime` itself has had historical issues with specific Python versions or architectures.
Fix: Ensure you have a compatible `onnxruntime` version for your Python environment and operating system. In case of installation issues, check `onnxruntime`'s official documentation for troubleshooting.

Install

pip install magika Install latest stable version

Imports

Magika
```
from magika import Magika
```
PredictionMode
```
from magika import Magika, PredictionMode
```
Used for custom instantiation of the Magika class to control prediction confidence.

Quickstart

Instantiate the `Magika` class and use `identify_bytes` or `identify_path` to determine the content type of a file or byte string. The `output.label` field provides the identified content type. For optimal performance with large files, `identify_path` or `identify_stream` are recommended as they avoid loading the entire content into memory.

from magika import Magika
import os

m = Magika()

# Example 1: Identify from bytes
file_content_bytes = b"console.log('Hello, Magika!');"
result_bytes = m.identify_bytes(file_content_bytes)
print(f"Content type (bytes): {result_bytes.output.label}")

# Example 2: Identify from a dummy file path
# Create a dummy file for demonstration
dummy_file_path = "./dummy_script.js"
with open(dummy_file_path, "wb") as f:
    f.write(file_content_bytes)

result_path = m.identify_path(dummy_file_path)
print(f"Content type (path): {result_path.output.label}")

os.remove(dummy_file_path)

view raw JSON →