This issue occurs because the Structured Output Parser attempts to parse each streaming chunk separately. To handle this, you could try collecting the streamed chunks first using a Function node or Code node, then parse the complete output once all chunks are received.
If you’re using OpenAI’s streaming API, consider disabling streaming temporarily when using the Structured Output Parser, or implement a buffer to accumulate chunks before parsing.
This collects all chunks and outputs the complete text when the last chunk is received. You can then feed this complete text to the Structured Output Parser.