From c44955845aaba85b4c9130b8e4e4cbce44939db2 Mon Sep 17 00:00:00 2001
From: Eric Hare <ericrhare@gmail.com>
Date: Fri, 19 Sep 2025 12:58:52 -0700
Subject: [PATCH] Fix onedrive and sharepoint ingest

---
 src/connectors/onedrive/connector.py   | 11 ++++++++++-
 src/connectors/sharepoint/connector.py | 11 ++++++++++-
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/src/connectors/onedrive/connector.py b/src/connectors/onedrive/connector.py
index 0664e874..9a7b6760 100644
--- a/src/connectors/onedrive/connector.py
+++ b/src/connectors/onedrive/connector.py
@@ -131,9 +131,18 @@ class OneDriveConnector(BaseConnector):
             content_resp = await client.get(
                 f"{self.base_url}/me/drive/items/{file_id}/content", headers=headers
             )
-            content_resp.raise_for_status()
             content = content_resp.content
 
+            # Handle the possibility of this being a redirect
+            if content_resp.status_code in (301, 302, 303, 307, 308):
+                redirect_url = content_resp.headers.get("Location")
+                if redirect_url:
+                    content_resp = await client.get(redirect_url)
+                    content_resp.raise_for_status()
+                    content = content_resp.content
+            else:
+                content_resp.raise_for_status()
+
             perm_resp = await client.get(
                 f"{self.base_url}/me/drive/items/{file_id}/permissions", headers=headers
             )
diff --git a/src/connectors/sharepoint/connector.py b/src/connectors/sharepoint/connector.py
index 8282f891..c31b9acd 100644
--- a/src/connectors/sharepoint/connector.py
+++ b/src/connectors/sharepoint/connector.py
@@ -136,9 +136,18 @@ class SharePointConnector(BaseConnector):
                 f"{self.base_url}/sites/{self.site_id}/drive/items/{file_id}/content",
                 headers=headers,
             )
-            content_resp.raise_for_status()
             content = content_resp.content
 
+            # Handle the possibility of this being a redirect
+            if content_resp.status_code in (301, 302, 303, 307, 308):
+                redirect_url = content_resp.headers.get("Location")
+                if redirect_url:
+                    content_resp = await client.get(redirect_url)
+                    content_resp.raise_for_status()
+                    content = content_resp.content
+            else:
+                content_resp.raise_for_status()
+
             perm_resp = await client.get(
                 f"{self.base_url}/sites/{self.site_id}/drive/items/{file_id}/permissions",
                 headers=headers,