agentcourse_unit4/tools/image_text_extractor.py

from smolagents import Tool
import easyocr


class ImageTextExtractorTool(Tool):
    name = "image_text_extractor"
    description = """
        Multilingual OCR tool to extract key information or presented text from any image.
        Supported image extensions: .png, .jpg, .jpeg, .bmp, .svg.
    """
    inputs = {
        "image_path": {
            "type": "string",
            "description": "The path to the image file",
        }
    }
    output_type = "array"

    def __init__(self):
        super().__init__()
        self.reader = easyocr.Reader(['ch_sim', 'en'])

    def forward(self, image_path: str) -> list[str]:
        result = self.reader.readtext(image_path, detail=False)
        return result