Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.
Python ai-agentsai-toolsbrowser-agentbrowser-automationbrowser-usecdpcloud-browsercomputer-useheadless-browserllmmcpplaywrightpythonremote-browserstealth-browserweb-automationweb-scraping
feat: Add enhanced screenshot API
avocardio3336b9d55f1a
diff --git a/browser_use/browser/session.py b/browser_use/browser/session.pyindex ec0447efc..448705271 100644--- a/browser_use/browser/session.py+++ b/browser_use/browser/session.py@@ -2255,3 +2255,129 @@ class BrowserSession(BaseModel):self.logger.debug(f'Failed to get CDP client for target {node.target_id}: {e}, using main session')return await self.get_or_create_cdp_session()++ async def take_screenshot(self, path: str | None = None, full_page: bool = False, format: str = 'png', quality: int | None = None, clip: dict | None = None) -> bytes:+ """Take a screenshot using CDP.++ Args:+ path: Optional file path to save screenshot+ full_page: Capture entire scrollable page beyond viewport+ format: Image format ('png', 'jpeg', 'webp')+ quality: Quality 0-100 for JPEG format+ clip: Region to capture {'x': int, 'y': int, 'width': int, 'height': int}++ Returns:+ Screenshot data as bytes+ """+ import base64+ from cdp_use.cdp.page import CaptureScreenshotParameters++ cdp_session = await self.get_or_create_cdp_session()++ # Build parameters dict explicitly to satisfy TypedDict expectations+ params: CaptureScreenshotParameters = {+ 'format': format,+ 'captureBeyondViewport': full_page,+ }++ if quality is not None and format == 'jpeg':+ params['quality'] = quality++ if clip:+ params['clip'] = {+ 'x': clip['x'],+ 'y': clip['y'],+ 'width': clip['width'],+ 'height': clip['height'],+ 'scale': 1,+ }++ params = CaptureScreenshotParameters(**params)++ result = await cdp_session.cdp_client.send.Page.captureScreenshot(+ params=params,+ session_id=cdp_session.session_id+ )++ if not result or 'data' not in result:+ raise Exception('Screenshot failed - no data returned')++ screenshot_data = base64.b64decode(result['data'])++ if path:+ with open(path, 'wb') as f:+ f.write(screenshot_data)++ return screenshot_data++ async def screenshot_element(+ self,+ selector: str,+ path: str | None = None,+ format: str = 'png',+ quality: int | None = None,+ ) -> bytes:+ """Take a screenshot of a specific element.++ Args:+ selector: CSS selector for the element+ path: Optional file path to save screenshot+ format: Image format ('png', 'jpeg', 'webp')+ quality: Quality 0-100 for JPEG format++ Returns:+ Screenshot data as bytes+ """++ bounds = await self._get_element_bounds(selector)+ if not bounds:+ raise ValueError(f"Element '{selector}' not found or has no bounds")++ return await self.take_screenshot(+ path=path,+ format=format,+ quality=quality,+ clip=bounds,+ )++ async def _get_element_bounds(self, selector: str) -> dict | None:+ """Get element bounding box using CDP."""++ cdp_session = await self.get_or_create_cdp_session()++ # Get document+ doc = await cdp_session.cdp_client.send.DOM.getDocument(+ params={'depth': 1},+ session_id=cdp_session.session_id+ )++ # Query selector+ node_result = await cdp_session.cdp_client.send.DOM.querySelector(+ params={+ 'nodeId': doc['root']['nodeId'],+ 'selector': selector+ },+ session_id=cdp_session.session_id+ )++ node_id = node_result.get('nodeId')+ if not node_id:+ return None++ # Get bounding box+ box_result = await cdp_session.cdp_client.send.DOM.getBoxModel(+ params={'nodeId': node_id},+ session_id=cdp_session.session_id+ )++ box_model = box_result.get('model')+ if not box_model:+ return None++ content = box_model['content']+ return {+ 'x': min(content[0], content[2], content[4], content[6]),+ 'y': min(content[1], content[3], content[5], content[7]),+ 'width': max(content[0], content[2], content[4], content[6]) - min(content[0], content[2], content[4], content[6]),+ 'height': max(content[1], content[3], content[5], content[7]) - min(content[1], content[3], content[5], content[7])+ }\ No newline at end of filediff --git a/tests/ci/test_browser_session_take_screenshot.py b/tests/ci/test_browser_session_take_screenshot.pynew file mode 100644index 000000000..c108d76ff--- /dev/null+++ b/tests/ci/test_browser_session_take_screenshot.py@@ -0,0 +1,33 @@+import pytest++from browser_use.browser.events import NavigateToUrlEvent+from browser_use.browser.profile import BrowserProfile+from browser_use.browser.session import BrowserSession++[email protected](scope="function")+async def browser_session():+ session = BrowserSession(browser_profile=BrowserProfile(headless=True))+ await session.start()+ yield session+ await session.kill()+++async def test_basic_screenshots(browser_session: BrowserSession, httpserver):+ """Navigate to a local page and ensure screenshot helpers return bytes."""++ html = """+ <html><body><h1 id='title'>Hello</h1><p>Screenshot demo.</p></body></html>+ """+ httpserver.expect_request("/demo").respond_with_data(html, content_type="text/html")+ url = httpserver.url_for("/demo")++ nav = browser_session.event_bus.dispatch(NavigateToUrlEvent(url=url, new_tab=False))+ await nav++ data = await browser_session.take_screenshot(full_page=False)+ assert data, "Viewport screenshot returned no data"++ element = await browser_session.screenshot_element("h1")+ assert element, "Element screenshot returned no data"