# Python 文件
def hello():
    print("world")

# JavaScript 文件
function greet() {
    console.log("hi");
}

# JSON 文件
{"name": "value"}

# 扩展名被改后都长得一样：
# script.txt vs script.py vs script.js

# 扩展名 vs 内容
"malware.exe" (扩展名是 .exe)
→ 内容是 Python 脚本 → Magika 识别为: python

"document.pdf" (扩展名是 .pdf)
→ 内容是文本 → Magika 识别为: generic text

# 检测单个文件
magika ./script.py

# 检测多个文件
magika file1.py file2.js file3.json

# 从标准输入读取
cat script.py | magika -

# 递归检测目录
magika -r ./my_project/

# 不跟随符号链接
magika -r --no-dereference ./my_project/

# JSON 输出
magika ./script.py --json

# JSONL 输出（适合批量处理）
magika -r ./files/ --jsonl > results.jsonl

# 仅输出标签
magika ./script.py --label
# 输出: python

# 仅输出 MIME 类型
magika ./script.py --mime-type
# 输出: text/x-python

# 输出预测分数
magika ./script.py --output-score
# 输出: Python source (score: 0.997)

# 使用占位符自定义输出
magika ./script.py --format "%l: %d (score: %S%)"

# 支持的占位符：
# %p - 文件路径
# %l - 标签 (python)
# %d - 描述 (Python source)
# %g - 分组 (code)
# %m - MIME type
# %e - 扩展名
# %s - 分数 (小数)
# %S - 分数 (百分比)
# %% - 字面 %

from magika import Magika

# 初始化（模型加载，一次性开销）
m = Magika()

# 检测字节内容
res = m.identify_bytes(b'function log(msg) {console.log(msg);}')
print(res.output.label)  # javascript

# 检测文件路径
res = m.identify_path('./config.json')
print(res.output.label)  # json

# 检测流（大型文件推荐）
with open('./large_file.py', 'rb') as f:
    res = m.identify_stream(f)
print(res.output.label)  # python

res = m.identify_path('./script.py')

# 获取所有信息
info = res.output
print(info.label)       # python
print(info.description)  # Python source
print(info.mime_type)    # text/x-python
print(info.extensions)   # ['py', 'pyi']
print(info.group)        # code
print(info.is_text)      # True

# 获取预测分数
print(res.score)  # 0.996999979019165

from pathlib import Path

# 批量检测多个文件
files = [
    './script.py',
    './config.json',
    './data.csv',
    './image.png'
]

for file_path in files:
    res = m.identify_path(file_path)
    print(f"{file_path}: {res.output.label}")

res = m.identify_path('./script.py')
print(res.score)  # 0.997

# 分数 > 阈值 → 返回具体类型
# 分数 < 阈值 → 返回泛型标签
#   - "Generic text document"
#   - "Unknown binary data"

# 核心依赖
tensorflow>=2.10  # 或 tensorflow-cpu
tensorflow-hub>=0.12

# 工具依赖
tqdm  # 进度条
rich  # 格式化输出

pipx install magika

brew install magika

# Linux/macOS
curl -LsSf https://securityresearch.google/magika/install.sh | sh

# Windows PowerShell
powershell -ExecutionPolicy Bypass -c "irm https://securityresearch.google/magika/install.ps1 | iex"

cargo install --locked magika-cli

pip install magika

# 或使用 uv（更快）
uv pip install magika

npm install magika

# 检查版本
magika --version

# 快速测试
echo 'print("hello")' | magika -
# 输出: Python source

from magika import Magika
from pathlib import Path

def scan_directory(path: str) -> dict:
    """扫描目录，标记异常文件类型"""
    m = Magika()
    results = {}

    for file_path in Path(path).rglob('*'):
        if file_path.is_file():
            res = m.identify_path(str(file_path))
            results[str(file_path)] = {
                'type': res.output.label,
                'score': res.score,
                'is_text': res.output.is_text
            }

    return results

# 检测可疑的 "图片" 文件
results = scan_directory('./uploads')
for path, info in results.items():
    if 'image' in info['type'] and not info['is_text']:
        print(f"⚠️ 可能的伪装文件: {path}")

from magika import Magika
from pathlib import Path
import shutil

def organize_by_type(source_dir: str, target_dir: str):
    """按文件类型自动整理"""
    m = Magika()
    source = Path(source_dir)
    target = Path(target_dir)

    for file_path in source.rglob('*'):
        if not file_path.is_file():
            continue

        res = m.identify_path(str(file_path))
        file_type = res.output.group or 'unknown'

        # 创建类型目录
        type_dir = target / file_type
        type_dir.mkdir(parents=True, exist_ok=True)

        # 移动文件
        dest = type_dir / file_path.name
        shutil.copy2(file_path, dest)
        print(f"✓ {file_path.name} → {file_type}/")

# .github/workflows/security-scan.yml
name: File Type Check

on: [push, pull_request]

jobs:
  check:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Install Magika
        run: pipx install magika

      - name: Scan uploads directory
        run: |
          magika -r --jsonl ./uploads/ > scan-results.jsonl

      - name: Check for executable files
        run: |
          cat scan-results.jsonl | jq -r 'select(.result.value.output.label == "pe" or .result.value.output.label == "elf")' | wc -l
          # 如果 > 0 则失败

from magika import Magika
from typing import List

ALLOWED_TYPES = {'python', 'javascript', 'typescript', 'json', 'yaml', 'markdown'}
BLOCKED_TYPES = {'exe', 'elf', 'pe', 'dll', 'shell'}

def validate_upload(file_path: str) -> tuple[bool, str]:
    """验证上传文件类型"""
    m = Magika()
    res = m.identify_path(file_path)

    file_type = res.output.label

    if file_type in BLOCKED_TYPES:
        return False, f"不允许的文件类型: {file_type}"

    if file_type not in ALLOWED_TYPES:
        return False, f"未授权的文件类型: {file_type}"

    return True, f"通过: {file_type}"

try:
    res = m.identify_path(file_path)
except FileNotFoundError:
    return None  # 文件不存在
except PermissionError:
    return None  # 权限不足