added envs for unstructured to control OCR quality and OCR languages

2025-10-04 05:21:02 +02:00
parent df5f85e0c6
commit c9a687171a
5 changed files with 96 additions and 6 deletions
@@ -86,11 +86,12 @@ async def parse_document(
        try:
            from nextcloud_mcp_server.client.unstructured_client import UnstructuredClient
            client = UnstructuredClient()
+            # The client will automatically use environment configuration
+            # (UNSTRUCTURED_STRATEGY and UNSTRUCTURED_LANGUAGES)
            return await client.partition_document(
                content=content,
                filename=filename or f"document.{doc_type}",
                content_type=content_type,
-                strategy="auto"
            )
        except Exception as e:
            logger.error(f"Unstructured API parsing failed: {e}")