feat: support PR-specific base resolution in recce instance

recce/state/cloud.py

@@ -233,7 +233,9 @@ def _load_state_from_session(self) -> RecceState:
         current_artifacts = self._download_session_artifacts(self.recce_cloud, org_id, project_id, self.session_id)
 
         logger.debug(f"Downloading base session artifacts for project {project_id}")
-        base_artifacts = self._download_base_session_artifacts(self.recce_cloud, org_id, project_id)
+        base_artifacts = self._download_base_session_artifacts(
+            self.recce_cloud, org_id, project_id, session_id=self.session_id
+        )
 
         # 3. Try to download existing recce_state, otherwise create new state
         try:
@@ -299,12 +301,17 @@ def _download_session_recce_state(self, recce_cloud, org_id: str, project_id: st
 
         return state
 
-    def _download_base_session_artifacts(self, recce_cloud, org_id: str, project_id: str) -> dict:
-        """Download manifest and catalog for the base session, return JSON data directly."""
+    def _download_base_session_artifacts(
+        self, recce_cloud, org_id: str, project_id: str, session_id: str = None
+    ) -> dict:
+        """Download manifest and catalog for the base session, return JSON data directly.
+
+        If session_id is provided, the server resolves PR-specific base if available.
+        """
         import requests
 
         # Get download URLs for base session
-        presigned_urls = recce_cloud.get_base_session_download_urls(org_id, project_id)
+        presigned_urls = recce_cloud.get_base_session_download_urls(org_id, project_id, session_id=session_id)
 
         artifacts = {}
 

recce/util/recce_cloud.py

@@ -341,9 +341,14 @@ def get_download_urls_by_session_id(self, org_id: str, project_id: str, session_
             presigned_urls[key] = self._replace_localhost_with_docker_internal(url)
         return presigned_urls
 
-    def get_base_session_download_urls(self, org_id: str, project_id: str) -> dict[str, str]:
-        """Get download URLs for the base session of a project."""
+    def get_base_session_download_urls(self, org_id: str, project_id: str, session_id: str = None) -> dict[str, str]:
+        """Get download URLs for the base session of a project.
+
+        If session_id is provided, the server resolves PR-specific base if available.
+        """
         api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/base-session/download-url"
+        if session_id:
+            api_url += f"?session_id={session_id}"
         response = self._request("GET", api_url)
         if response.status_code != 200:
             raise RecceCloudException(

tests/state/test_cloud.py

@@ -677,6 +677,79 @@ def test_export_state_to_session_upload_failure(self, mock_put):
             self.assertIn("Internal Server Error", result_message)
             self.assertIsNone(result_etag)
 
+    @patch.object(CloudStateLoader, "_download_session_recce_state")
+    @patch.object(CloudStateLoader, "_download_base_session_artifacts")
+    @patch.object(CloudStateLoader, "_download_session_artifacts")
+    def test_load_state_from_session_passes_session_id_to_base_download(
+        self,
+        mock_download_session_artifacts,
+        mock_download_base_session_artifacts,
+        mock_download_session_state,
+    ):
+        """Verify _load_state_from_session passes session_id to _download_base_session_artifacts."""
+        loader = CloudStateLoader(cloud_options={"api_token": "token", "session_id": "test_session"})
+        loader.catalog = "session"
+        loader.session_id = "test_session"
+
+        loader.recce_cloud = Mock()
+        loader.recce_cloud.get_session.return_value = {
+            "org_id": "org1",
+            "project_id": "proj1",
+        }
+
+        mock_download_session_artifacts.return_value = {"manifest": "current_manifest", "catalog": "current_catalog"}
+        mock_download_base_session_artifacts.return_value = {"manifest": "base_manifest", "catalog": "base_catalog"}
+        mock_download_session_state.return_value = RecceState()
+
+        loader._load_state_from_session()
+
+        # Verify session_id was passed to _download_base_session_artifacts
+        mock_download_base_session_artifacts.assert_called_once_with(
+            loader.recce_cloud, "org1", "proj1", session_id="test_session"
+        )
+
+    @patch("requests.get")
+    def test_download_base_session_artifacts_passes_session_id(self, mock_get):
+        """Verify _download_base_session_artifacts passes session_id to get_base_session_download_urls."""
+        loader = CloudStateLoader(cloud_options={"api_token": "token", "session_id": "test_session"})
+        loader.catalog = "session"
+
+        mock_cloud = Mock()
+        mock_cloud.get_base_session_download_urls.return_value = {
+            "manifest_url": "http://base_manifest.url",
+            "catalog_url": "http://base_catalog.url",
+        }
+
+        mock_response = Mock()
+        mock_response.status_code = 200
+        mock_response.json.side_effect = ["base_manifest_data", "base_catalog_data"]
+        mock_get.return_value = mock_response
+
+        loader._download_base_session_artifacts(mock_cloud, "org1", "proj1", session_id="test_session")
+
+        mock_cloud.get_base_session_download_urls.assert_called_once_with("org1", "proj1", session_id="test_session")
+
+    @patch("requests.get")
+    def test_download_base_session_artifacts_without_session_id(self, mock_get):
+        """Verify _download_base_session_artifacts works without session_id (backward compat)."""
+        loader = CloudStateLoader(cloud_options={"api_token": "token", "session_id": "test_session"})
+        loader.catalog = "session"
+
+        mock_cloud = Mock()
+        mock_cloud.get_base_session_download_urls.return_value = {
+            "manifest_url": "http://base_manifest.url",
+            "catalog_url": "http://base_catalog.url",
+        }
+
+        mock_response = Mock()
+        mock_response.status_code = 200
+        mock_response.json.side_effect = ["base_manifest_data", "base_catalog_data"]
+        mock_get.return_value = mock_response
+
+        loader._download_base_session_artifacts(mock_cloud, "org1", "proj1")
+
+        mock_cloud.get_base_session_download_urls.assert_called_once_with("org1", "proj1", session_id=None)
+
     @patch.object(CloudStateLoader, "_download_session_recce_state")
     @patch.object(CloudStateLoader, "_download_base_session_artifacts")
     @patch.object(CloudStateLoader, "_download_session_artifacts")

tests/util/test_recce_cloud.py

@@ -227,5 +227,72 @@ def test_list_sessions_missing_key(self, mock_request):
         self.assertEqual(result, [])
 
 
+class TestRecceCloudBaseSessionDownloadUrls(unittest.TestCase):
+    """Test cases for get_base_session_download_urls with optional session_id."""
+
+    def setUp(self):
+        self.token = "test-api-token"
+        self.cloud = RecceCloud(self.token)
+
+    @patch("recce.util.recce_cloud.RecceCloud._request")
+    def test_get_base_session_download_urls_without_session_id(self, mock_request):
+        """Without session_id, URL has no query params."""
+        mock_response = Mock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            "presigned_urls": {
+                "manifest_url": "http://manifest.url",
+                "catalog_url": "http://catalog.url",
+            }
+        }
+        mock_request.return_value = mock_response
+
+        result = self.cloud.get_base_session_download_urls("org1", "proj1")
+
+        expected_url = f"{self.cloud.base_url_v2}/organizations/org1/projects/proj1/base-session/download-url"
+        mock_request.assert_called_once_with("GET", expected_url)
+        self.assertIn("manifest_url", result)
+        self.assertIn("catalog_url", result)
+
+    @patch("recce.util.recce_cloud.RecceCloud._request")
+    def test_get_base_session_download_urls_with_session_id(self, mock_request):
+        """With session_id, URL includes ?session_id= query param."""
+        mock_response = Mock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            "presigned_urls": {
+                "manifest_url": "http://manifest.url",
+                "catalog_url": "http://catalog.url",
+            }
+        }
+        mock_request.return_value = mock_response
+
+        result = self.cloud.get_base_session_download_urls("org1", "proj1", session_id="abc-123")
+
+        expected_url = (
+            f"{self.cloud.base_url_v2}/organizations/org1/projects/proj1/base-session/download-url?session_id=abc-123"
+        )
+        mock_request.assert_called_once_with("GET", expected_url)
+        self.assertIn("manifest_url", result)
+
+    @patch("recce.util.recce_cloud.RecceCloud._request")
+    def test_get_base_session_download_urls_with_session_id_none(self, mock_request):
+        """Explicit session_id=None behaves same as no session_id."""
+        mock_response = Mock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            "presigned_urls": {
+                "manifest_url": "http://manifest.url",
+                "catalog_url": "http://catalog.url",
+            }
+        }
+        mock_request.return_value = mock_response
+
+        self.cloud.get_base_session_download_urls("org1", "proj1", session_id=None)
+
+        expected_url = f"{self.cloud.base_url_v2}/organizations/org1/projects/proj1/base-session/download-url"
+        mock_request.assert_called_once_with("GET", expected_url)
+
+
 if __name__ == "__main__":
     unittest.main()