From aff7b907c7afed38675ab0ef19facf1687b70836 Mon Sep 17 00:00:00 2001
From: perf3ct <jonfuller2012@gmail.com>
Date: Tue, 29 Jul 2025 21:27:54 +0000
Subject: [PATCH] fix(db): backfill data for sources given missing counts

---
 ...0729000001_backfill_document_source_id.sql | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)
 create mode 100644 migrations/20250729000001_backfill_document_source_id.sql

diff --git a/migrations/20250729000001_backfill_document_source_id.sql b/migrations/20250729000001_backfill_document_source_id.sql
new file mode 100644
index 0000000..457b97c
--- /dev/null
+++ b/migrations/20250729000001_backfill_document_source_id.sql
@@ -0,0 +1,27 @@
+-- Backfill source_id and source_type for existing documents
+-- This migration fixes documents that were ingested before proper source tracking was implemented
+
+-- Update documents that have WebDAV source paths but missing source_id
+-- Link them to existing WebDAV sources based on user ownership
+UPDATE documents 
+SET 
+    source_id = sources.id,
+    source_type = 'webdav'
+FROM sources 
+WHERE 
+    documents.user_id = sources.user_id 
+    AND sources.source_type = 'webdav'
+    AND documents.source_metadata->>'source_path' IS NOT NULL
+    AND documents.source_id IS NULL;
+
+-- Update documents that have source paths but no source_metadata, likely from older ingestion
+-- This handles edge cases where source_path is populated but source_type is not
+UPDATE documents 
+SET source_type = 'webdav'
+WHERE 
+    source_path IS NOT NULL 
+    AND source_type IS NULL
+    AND source_id IN (SELECT id FROM sources WHERE source_type = 'webdav');
+
+-- Add helpful comment explaining the backfill
+COMMENT ON COLUMN documents.source_id IS 'References the source that this document was ingested from. Backfilled for existing documents on 2025-07-29.';
\ No newline at end of file