fix: max_tokens_per_request error when pushing many documents at once

Resolves timescale#481
kolaente · Feb 14, 2025 · 1e83b35 · 1e83b35
1 parent 073fa5b
commit 1e83b35
Showing 1 changed file with 8 additions and 0 deletions.
diff --git a/projects/pgai/pgai/vectorizer/embedders/openai.py b/projects/pgai/pgai/vectorizer/embedders/openai.py
@@ -124,6 +124,14 @@ async def embed(
             if not isinstance(msg, str):
                 raise e
 
+            error_type: Any = body["type"]
+            if isinstance(error_type, str) and error_type == "max_tokens_per_request":
+                mid = len(documents) // 2
+
+                await self.embed(documents[: mid + 1])
+
+                return await self.embed(documents[mid + 1 :])
+
             m = openai_token_length_regex.match(msg)
             if not m:
                 raise e