Set redirect meta

scrapy_playwright/handler.py

@@ -303,6 +303,7 @@ async def _download_request_with_page(self, request: Request, page: Page) -> Res
             )
             headers = Headers()
         else:
+            await self._set_redirect_meta(request=request, response=response)
             headers = Headers(await response.all_headers())
             headers.pop("Content-Encoding", None)
         await self._apply_page_methods(page, request)
@@ -334,6 +335,23 @@ async def _download_request_with_page(self, request: Request, page: Page) -> Res
             ip_address=server_ip_address,
         )
 
+    async def _set_redirect_meta(self, request: Request, response: PlaywrightResponse) -> None:
+        redirect_times: int = 0
+        redirect_urls: list = []
+        redirect_reasons: list = []
+        redirected = response.request.redirected_from
+        while redirected is not None:
+            redirect_times += 1
+            redirect_urls.append(redirected.url)
+            redirected_response = await redirected.response()
+            reason = None if redirected_response is None else redirected_response.status
+            redirect_reasons.append(reason)
+            redirected = redirected.redirected_from
+        if redirect_times:
+            request.meta["redirect_times"] = redirect_times
+            request.meta["redirect_urls"] = list(reversed(redirect_urls))
+            request.meta["redirect_reasons"] = list(reversed(redirect_reasons))
+
     async def _apply_page_methods(self, page: Page, request: Request) -> None:
         page_methods = request.meta.get("playwright_page_methods") or ()
         if isinstance(page_methods, dict):

tests/mockserver.py

@@ -40,7 +40,7 @@ def urljoin(self, url):
 
 
 class _RequestHandler(BaseHTTPRequestHandler):
-    def do_POST(self):
+    def do_POST(self) -> None:
         """Echo back the request body"""
         content_length = int(self.headers["Content-Length"])
         body = self.rfile.read(content_length)
@@ -49,21 +49,33 @@ def do_POST(self):
         self.wfile.write(b"Request body: ")
         self.wfile.write(body)
 
-    def do_GET(self):
-        body = "{}"
+    def do_GET(self) -> None:
         if self.path == "/headers":
-            body = json.dumps(dict(self.headers), indent=4)
+            self._send_json(dict(self.headers))
+        elif self.path == "/redirect2":
+            self.send_response(302)
+            self.send_header("Location", "/redirect")
+            self.end_headers()
+        elif self.path == "/redirect":
+            self.send_response(301)
+            self.send_header("Location", "/headers")
+            self.end_headers()
         else:
             delay_match = re.match(r"^/delay/(\d+)$", self.path)
             if delay_match:
                 delay = int(delay_match.group(1))
                 print(f"Sleeping {delay} seconds...")
                 time.sleep(delay)
-                body = json.dumps({"delay": delay})
-        self.send_response(200)
+                self._send_json({"delay": delay})
+            else:
+                self._send_json({"error": "unknown path"}, status=400)
+
+    def _send_json(self, body: dict, status: int = 200) -> None:
+        self.send_response(status)
         self.send_header("Content-Type", "application/json")
         self.end_headers()
-        self.wfile.write(body.encode())
+        body_bytes = json.dumps(body, indent=4).encode("utf8")
+        self.wfile.write(body_bytes)
 
 
 class MockServer:

tests/test_playwright_requests.py

@@ -341,6 +341,24 @@ async def init_page(page, request, unused_arg):
         assert f"[Context=default] Page init callback exception for {req!r}" in log_entry[2]
         assert "init_page() missing 1 required positional argument: 'unused_arg'" in log_entry[2]
 
+    @pytest.mark.asyncio
+    async def test_redirect(self):
+        async with make_handler({"PLAYWRIGHT_BROWSER_TYPE": self.browser_type}) as handler:
+            with MockServer() as server:
+                req = Request(
+                    url=server.urljoin("/redirect2"),
+                    meta={"playwright": True},
+                )
+                response = await handler._download_request(req, Spider("spider_name"))
+
+        assert response.url == server.urljoin("/headers")
+        assert response.meta["redirect_times"] == 2
+        assert response.meta["redirect_reasons"] == [302, 301]
+        assert response.meta["redirect_urls"] == [
+            server.urljoin("/redirect2"),
+            server.urljoin("/redirect"),
+        ]
+
 
 class TestCaseChromium(MixinTestCase):
     browser_type = "chromium"