Silicon Paddle Ocr

#!/bin/bash
# Example usage script for OCR skill

# Set API key (or load from environment)
export SILICONFLOW_API_KEY="your_api_key_here"

# Path to the OCR script
OCR_SCRIPT="$(dirname "$0")/../scripts/ocr_skill.py"

echo "=== OCR Skill Examples ==="
echo ""

# Example 1: Single image recognition
echo "Example 1: Single image"
# python3 "$OCR_SCRIPT" /path/to/test.jpg
echo "python3 $OCR_SCRIPT /path/to/test.jpg"
echo ""

# Example 2: Batch processing with glob pattern
echo "Example 2: Batch processing"
# python3 "$OCR_SCRIPT" /path/to/images/*.png
echo "python3 $OCR_SCRIPT /path/to/images/*.png"
echo ""

# Example 3: JSON output format
echo "Example 3: JSON format output"
# python3 "$OCR_SCRIPT" --json /path/to/image.jpg
echo "python3 $OCR_SCRIPT --json /path/to/image.jpg"
echo ""

# Example 4: Custom prompt for specific task
echo "Example 4: Custom prompt for table extraction"
# python3 "$OCR_SCRIPT" -p "Please extract and format as Markdown table" /path/to/table.jpg
echo "python3 $OCR_SCRIPT -p \"Please extract and format as Markdown table\" /path/to/table.jpg"
echo ""

# Example 5: Save results to file
echo "Example 5: Save results to file"
# python3 "$OCR_SCRIPT" --json --output results.json /path/to/images/*.jpg
echo "python3 $OCR_SCRIPT --json --output results.json /path/to/images/*.jpg"
echo ""


{
  "id": "silicon-paddle-ocr",
  "title": "Silicon PaddleOCR",
  "description": "OCR skill using PaddleOCR model via SiliconFlow API for text extraction from images",
  "author": "aotenjou",
  "version": "1.0.0",
  "license": "MIT",
  "tags": [
    "ocr",
    "text-recognition",
    "image-processing",
    "paddleocr",
    "siliconflow"
  ],
  "category": "text-processing",
  "home": "https://skills.sh/aotenjou/silicon-paddle-ocr",
  "repository": "https://github.com/aotenjou/silicon-PaddleOCR",
  "skills": {
    "ocr": {
      "name": "ocr",
      "description": "Extract text from images using PaddleOCR via SiliconFlow API",
      "whenToUse": "When the user wants to recognize text from an image, extract text from a photo, OCR a screenshot, or mentions PaddleOCR, image text recognition, or text extraction from images",
      "entry": "scripts/ocr_skill.py",
      "compatibleAgents": ["claude"]
    }
  },
  "capabilities": [
    "single-image-ocr",
    "batch-image-ocr",
    "json-output",
    "custom-prompts",
    "multiple-image-formats",
    "text-bounding-boxes",
    "structured-output"
  ],
  "dependencies": {
    "python": ">3.7",
    "packages": ["openai>=1.0.0", "Pillow>=8.0.0"]
  },
  "environment": {
    "required": ["SILICONFLOW_API_KEY"],
    "optional": []
  },
  "supportedFormats": ["jpg", "jpeg", "png", "webp", "bmp", "gif"]
}


# API Configuration

## SiliconFlow API

This skill uses the SiliconFlow API for OCR operations.

### API Endpoint

```
https://api.siliconflow.cn/v1
```

### Authentication

Use the SILICONFLOW_API_KEY environment variable:
```bash
export SILICONFLOW_API_KEY="sk-xxxxxxxxxxxxx"
```

### Default Model

```
PaddlePaddle/PaddleOCR-VL-1.5
```

### Supported Models

- `PaddlePaddle/PaddleOCR-VL-1.5` (default)
- Other multilingual VL models compatible with the API

### Rate Limits

- Requests per minute: 60 (configurable)
- Max concurrent requests: 5

### API Response Format

The API returns a chat completion JSON with the recognized text in the message content.


#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
OCR Skill - 使用 PaddleOCR 识别图片中的文字
"""

import base64
import json
import re
import sys
from pathlib import Path
from typing import Dict, Any, List, Optional, Tuple

try:
    from openai import OpenAI
except ImportError:
    print("错误: 需要安装 openai 库")
    print("运行: pip install openai")
    sys.exit(1)


def image_to_base64(image_path: str) -> str:
    """将图片文件转换为 base64 字符串"""
    with open(image_path, "rb") as f:
        return base64.b64encode(f.read()).decode("utf-8")


def get_image_size(image_path: str) -> Tuple[int, int]:
    """获取图片尺寸"""
    from PIL import Image
    with Imag