fix: set _drive_id_endpoint before early return in SharePointReader._get_drive_id

llama-index-integrations/readers/llama-index-readers-microsoft-sharepoint/llama_index/readers/microsoft_sharepoint/base.py

@@ -312,14 +312,14 @@ def _get_drive_id(self) -> str:
             ValueError: If there is an error in obtaining the drive ID.
 
         """
+        self._drive_id_endpoint = f"https://graph.microsoft.com/v1.0/sites/{self._site_id_with_host_name}/drives"
+
         if hasattr(self, "_drive_id"):
             return self._drive_id
 
         if self.drive_id:
             return self.drive_id
 
-        self._drive_id_endpoint = f"https://graph.microsoft.com/v1.0/sites/{self._site_id_with_host_name}/drives"
-
         # Use generic pagination helper to get all drives
         drives = self._get_all_items_with_pagination(self._drive_id_endpoint)
 

llama-index-integrations/readers/llama-index-readers-microsoft-sharepoint/pyproject.toml

@@ -26,7 +26,7 @@ dev = [
 
 [project]
 name = "llama-index-readers-microsoft-sharepoint"
-version = "0.8.0"
+version = "0.8.1"
 description = "llama-index readers microsoft_sharepoint integration"
 authors = [{name = "Your Name", email = "you@example.com"}]
 requires-python = ">=3.9,<4.0"

llama-index-integrations/readers/llama-index-readers-microsoft-sharepoint/tests/test_readers_microsoft_sharepoint.py

@@ -577,3 +577,32 @@ def test_get_all_items_with_pagination_empty_result():
         )
 
     assert len(items) == 0
+
+
+def test_drive_id_endpoint_set_when_drive_id_provided():
+    """
+    Test that _drive_id_endpoint is set even when drive_id is provided directly.
+
+    Regression test for https://github.com/run-llama/llama_index/issues/18753
+    """
+    reader = SharePointReader(
+        client_id="dummy_client_id",
+        client_secret="dummy_client_secret",
+        tenant_id="dummy_tenant_id",
+        drive_id="my_drive_id",
+    )
+    reader._site_id_with_host_name = "dummy_site_id"
+
+    # Bypass the autouse _get_drive_id mock by calling the unpatched method directly
+    # via the descriptor protocol on the original class
+    import importlib
+
+    mod = importlib.import_module("llama_index.readers.microsoft_sharepoint.base")
+    importlib.reload(mod)
+    orig_get_drive_id = mod.SharePointReader._get_drive_id
+    result = orig_get_drive_id(reader)
+
+    assert result == "my_drive_id"
+    assert reader._drive_id_endpoint == (
+        "https://graph.microsoft.com/v1.0/sites/dummy_site_id/drives"
+    )