[8.x] Change dto for page blog documents to access fields in safe man…

…ner (#2984) (#2986) Co-authored-by: Artem Shelkovnikov <[email protected]> Co-authored-by: Jedr Blaszyk <[email protected]>
elastic · Nov 25, 2024 · fa53dcb · fa53dcb
1 parent 25acf7a
commit fa53dcb
Showing 1 changed file with 13 additions and 5 deletions.
diff --git a/connectors/sources/confluence.py b/connectors/sources/confluence.py
@@ -908,14 +908,22 @@ async def fetch_documents(self, api_query):
             doc = {
                 "_id": str(document["id"]),
                 "type": document["type"],
-                "_timestamp": document["history"]["lastUpdated"]["when"],
+                "_timestamp": nested_get_from_dict(
+                    document, ["history", "lastUpdated", "when"]
+                ),
                 "title": document.get("title"),
                 "ancestors": ancestor_title,
-                "space": document["space"]["name"],
-                "body": html_to_text(document["body"]["storage"]["value"]),
+                "space": nested_get_from_dict(document, ["space", "name"]),
+                "body": html_to_text(
+                    nested_get_from_dict(document, ["body", "storage", "value"])
+                ),
                 "url": document_url,
-                "author": document["history"]["createdBy"][self.authorkey],
-                "createdDate": document["history"]["createdDate"],
+                "author": nested_get_from_dict(
+                    document, ["history", "createdBy", self.authorkey]
+                ),
+                "createdDate": nested_get_from_dict(
+                    document, ["history", "createdDate"]
+                ),
             }
             if self.confluence_client.index_labels:
                 doc["labels"] = document["labels"]