From 33ec3d2a9f552ea0bd04a90bfa447a52c925c409 Mon Sep 17 00:00:00 2001
From: Victor <194116445+dodekapod@users.noreply.github.com>
Date: Wed, 29 Oct 2025 15:35:11 +0100
Subject: [PATCH] file_metadata: JSON -> JSONB, index on file_metadata

---
 openrag/components/indexer/vectordb/utils.py | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/openrag/components/indexer/vectordb/utils.py b/openrag/components/indexer/vectordb/utils.py
index 99c63c40..7d0ebd0d 100644
--- a/openrag/components/indexer/vectordb/utils.py
+++ b/openrag/components/indexer/vectordb/utils.py
@@ -5,7 +5,6 @@
 from typing import Dict, Optional
 
 from sqlalchemy import (
-    JSON,
     Boolean,
     CheckConstraint,
     Column,
@@ -17,6 +16,9 @@
     UniqueConstraint,
     create_engine,
 )
+from sqlalchemy.dialects.postgresql import (
+    JSONB
+)
 from sqlalchemy.orm import (
     declarative_base,
     relationship,
@@ -45,7 +47,7 @@ class File(Base):
     partition_name = Column(
         String, ForeignKey("partitions.partition"), nullable=False, index=True
     )  # Added index
-    file_metadata = Column(JSON, nullable=True, default={})
+    file_metadata = Column(JSONB, nullable=True, default={})
 
     # relationship to the Partition object
     partition = relationship("Partition", back_populates="files")
@@ -55,6 +57,8 @@ class File(Base):
         UniqueConstraint("file_id", "partition_name", name="uix_file_id_partition"),
         # Additional composite index for common query patterns (partition first for better selectivity)
         Index("ix_partition_file", "partition_name", "file_id"),
+        # Metadata GIN index
+        Index("ix_files_metadata_gin", "file_metadata", postgresql_using="gin")
     )
 
     def to_dict(self):