win_monitor/backend/tests/test_vision_node.py

from __future__ import annotations

import unittest
from unittest.mock import patch

from app.automation.context import WorkflowContext
from app.automation.nodes.vision import locate_element_node


class VisionLocateNodeTest(unittest.TestCase):
    def test_locate_element_converts_percent_to_screen_coordinates(self) -> None:
        context = WorkflowContext(workflow_id=1, provider_id=1, model_id=1, temperature=0.1)
        node = {
            "params": {
                "target_description": "点击第一个推荐视频",
                "screen_context": "视频首页",
                "save_screenshot": False,
            }
        }

        with (
            patch(
                "app.automation.nodes.vision.windows_automation.take_screenshot",
                return_value={
                    "image_base64": "abc",
                    "mime_type": "image/png",
                    "width": 800,
                    "height": 600,
                    "path": None,
                },
            ),
            patch(
                "app.automation.nodes.vision.ai_service.chat_with_images",
                return_value={
                    "content": (
                        '{"found": true, "x_percent": 25, "y_percent": 50, '
                        '"confidence": 0.9, "target_label": "推荐视频", "reason": "清晰可见"}'
                    )
                },
            ),
        ):
            result = locate_element_node(node, {}, context)

        self.assertTrue(result["located"])
        self.assertEqual(result["x"], 200)
        self.assertEqual(result["y"], 300)
        self.assertEqual(result["x_percent"], 25)
        self.assertEqual(result["y_percent"], 50)


if __name__ == "__main__":
    unittest.main()