Fixed file reading when Graph leaves out the download link

src/app.py

@@ -637,9 +637,9 @@ def resolve(note):
         item = note.get(chat_read.drive_item_path(ref["contentUrl"], "name,size,file,@microsoft.graph.downloadUrl"))
         if (item.get("size") or 0) > MAX_CHAT_FILE_BYTES:
             raise HTTPException(status_code=413, detail="The file is larger than 100 MB.")
-        url = item.get("@microsoft.graph.downloadUrl")
+        url = teams_files.download_url(note, f"shares/{chat_read.share_id(ref['contentUrl'])}/driveItem", item)
         if not url:
-            raise HTTPException(status_code=404, detail="The file has no download link.")
+            raise HTTPException(status_code=502, detail="Microsoft did not offer a download link for this file.")
         return item, url
 
     carrier = Response()
@@ -886,9 +886,9 @@ def resolve(note):
             raise HTTPException(status_code=404, detail="File not found.")
         if (item.get("size") or 0) > teams_files.MAX_DOWNLOAD_BYTES:
             raise HTTPException(status_code=413, detail="The file is larger than 100 MB.")
-        url = item.get("@microsoft.graph.downloadUrl")
+        url = teams_files.download_url(note, teams_files.item_base(drive_id, item_id), item)
         if not url:
-            raise HTTPException(status_code=404, detail="The file has no download link.")
+            raise HTTPException(status_code=502, detail="Microsoft did not offer a download link for this file.")
         if not teams_files.is_download_url(url):
             raise HTTPException(status_code=502, detail="The file has an unexpected download location.")
         return item, url
@@ -938,9 +938,9 @@ def _file_text(note, drive_id: str | None, item_id: str, offset: int, max_chars:
             name = posixpath.splitext(name)[0] + ".pdf"
             notes.append("Converted to PDF by Microsoft 365 before reading.")
         else:
-            url = item.get("@microsoft.graph.downloadUrl")
+            url = teams_files.download_url(note, teams_files.item_base(drive_id, item_id), item)
             if not url:
-                raise HTTPException(status_code=404, detail="The file has no download link.")
+                raise HTTPException(status_code=502, detail="Microsoft did not offer a download link for this file.")
             data = teams_files.fetch_bytes(url, file_text.MAX_SOURCE_BYTES)
         result = file_text.extract(name, data, offset, max_chars, sheet, row_offset, max_rows, notes)
     except file_text.TextError as ex:

src/teams_files.py

@@ -76,6 +76,21 @@ def fetch_bytes(url: str, cap: int) -> bytes:
     return bytes(out)
 
 
+def download_url(note, path: str, item: dict) -> str | None:
+    """The pre-authorized link of the item at `path`. Graph can leave it out of a `$select` answer, so ask again without one, then through `/content`."""
+    url = item.get("@microsoft.graph.downloadUrl")
+    if url:
+        return url
+    full = note.get(path)
+    url = full.get("@microsoft.graph.downloadUrl") if isinstance(full, dict) else None
+    if url:
+        return url
+    try:
+        return note.redirect(f"{path}/content")
+    except GraphError:
+        return None
+
+
 def item_base(drive_id: str | None, item_id: str) -> str:
     return f"drives/{seg(drive_id)}/items/{seg(item_id)}" if drive_id else f"me/drive/items/{seg(item_id)}"
 

tests/test_teams_files.py

@@ -279,3 +279,42 @@ def test_convert_text_body_cap(env, monkeypatch):
     monkeypatch.setattr(app_module, "MAX_TEXT_SOURCE_BYTES", 100)
     assert client.post("/api/convert/text?name=x.txt", content=b"a" * 101, headers=H).status_code == 413
     assert client.post("/api/convert/text?name=x.txt", content=b"a" * 100, headers=H).json()["text"] == "a" * 100
+
+
+class LinkNote:
+    def __init__(self, plain=None, redirect=None):
+        self.plain, self.location, self.paths = plain, redirect, []
+
+    def get(self, path):
+        self.paths.append(path)
+        return self.plain
+
+    def redirect(self, path):
+        self.paths.append(path)
+        if self.location is None:
+            raise teams_files.GraphError(404, "no content")
+        return self.location
+
+
+LINK = "https://contoso.sharepoint.com/download?tempauth=x"
+
+
+def test_download_url_uses_the_link_from_the_select_answer():
+    note = LinkNote()
+    assert teams_files.download_url(note, "me/drive/items/I", {"@microsoft.graph.downloadUrl": LINK}) == LINK
+    assert note.paths == []
+
+
+def test_download_url_asks_again_without_select():
+    note = LinkNote(plain={"@microsoft.graph.downloadUrl": LINK})
+    assert teams_files.download_url(note, "me/drive/items/I", {"name": "Plan.xlsx"}) == LINK
+
+
+def test_download_url_falls_back_to_the_content_redirect():
+    note = LinkNote(plain={"name": "Plan.xlsx"}, redirect=LINK)
+    assert teams_files.download_url(note, "me/drive/items/I", {}) == LINK
+    assert note.paths == ["me/drive/items/I", "me/drive/items/I/content"]
+
+
+def test_download_url_is_none_when_microsoft_offers_nothing():
+    assert teams_files.download_url(LinkNote(plain={}), "me/drive/items/I", {}) is None