From 2a59651fb9303bb43ffbce6e2984a5b5a914a95d Mon Sep 17 00:00:00 2001
From: perf3ct <jonfuller2012@gmail.com>
Date: Tue, 8 Jul 2025 20:16:33 +0000
Subject: [PATCH] fix(stats): try to fix stats export, again again

---
 ...708000001_fix_ocr_queue_stats_function.sql | 42 -------------------
 src/main.rs                                   |  2 +-
 2 files changed, 1 insertion(+), 43 deletions(-)
 delete mode 100644 migrations/20250708000001_fix_ocr_queue_stats_function.sql

diff --git a/migrations/20250708000001_fix_ocr_queue_stats_function.sql b/migrations/20250708000001_fix_ocr_queue_stats_function.sql
deleted file mode 100644
index 0370004..0000000
--- a/migrations/20250708000001_fix_ocr_queue_stats_function.sql
+++ /dev/null
@@ -1,42 +0,0 @@
--- Fix the get_ocr_queue_stats function to ensure it matches the expected structure
--- This migration ensures the function correctly gets completed_today from documents table
--- and handles the case where migration 20250620100019 may have failed silently
-
-CREATE OR REPLACE FUNCTION get_ocr_queue_stats()
-RETURNS TABLE (
-    pending_count BIGINT,
-    processing_count BIGINT,
-    failed_count BIGINT,
-    completed_today BIGINT,
-    avg_wait_time_minutes DOUBLE PRECISION,
-    oldest_pending_minutes DOUBLE PRECISION
-) AS $$
-BEGIN
-    RETURN QUERY
-    WITH queue_stats AS (
-        SELECT 
-            COUNT(*) FILTER (WHERE status = 'pending') as pending_count,
-            COUNT(*) FILTER (WHERE status = 'processing') as processing_count,
-            COUNT(*) FILTER (WHERE status = 'failed' AND attempts >= max_attempts) as failed_count,
-            CAST(AVG(EXTRACT(EPOCH FROM (COALESCE(started_at, NOW()) - created_at))/60) FILTER (WHERE status IN ('processing', 'completed')) AS DOUBLE PRECISION) as avg_wait_time_minutes,
-            CAST(MAX(EXTRACT(EPOCH FROM (NOW() - created_at))/60) FILTER (WHERE status = 'pending') AS DOUBLE PRECISION) as oldest_pending_minutes
-        FROM ocr_queue
-    ),
-    document_stats AS (
-        -- Count documents that completed OCR today (looking at documents table where actual completion is tracked)
-        SELECT COUNT(*) as completed_today
-        FROM documents
-        WHERE ocr_status = 'completed'
-        AND updated_at >= CURRENT_DATE
-        AND updated_at < CURRENT_DATE + INTERVAL '1 day'
-    )
-    SELECT 
-        queue_stats.pending_count,
-        queue_stats.processing_count,
-        queue_stats.failed_count,
-        document_stats.completed_today,
-        queue_stats.avg_wait_time_minutes,
-        queue_stats.oldest_pending_minutes
-    FROM queue_stats, document_stats;
-END;
-$$ LANGUAGE plpgsql;
\ No newline at end of file
diff --git a/src/main.rs b/src/main.rs
index 98d0e02..9071e58 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -251,7 +251,7 @@ async fn main() -> anyhow::Result<()> {
             match function_check {
                 Ok(Some(def)) => {
                     // Check if it contains the correct logic from our latest migration
-                    if def.contains("document_stats") && def.contains("documents.ocr_status = 'completed'") {
+                    if def.contains("document_stats") && def.contains("ocr_status = 'completed'") {
                         info!("✅ get_ocr_queue_stats function has correct logic (uses documents table for completed_today)");
                     } else {
                         error!("❌ get_ocr_queue_stats function still uses old logic - migration may have failed");