From 86315461423fbbc1eaae63f4b1a8bfb2962ecd91 Mon Sep 17 00:00:00 2001
From: htlee <htlee@gcsc.co.kr>
Date: Tue, 31 Mar 2026 09:03:33 +0900
Subject: [PATCH] =?UTF-8?q?fix:=20=ED=8A=B8=EB=9E=99=20API=20DB=20?=
 =?UTF-8?q?=EC=A0=91=EC=86=8D=20=EB=B2=84=EA=B7=B8=20=EC=88=98=EC=A0=95=20?=
 =?UTF-8?q?(context=20manager)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- kcgdb.get_conn()을 with문 없이 사용 → cursor 에러
- with kcgdb.get_conn() as conn: 으로 수정
- 디버그 로그 추가 (rows 수, track 매칭 수, vessel_store 크기)
- 결과: 47 vessels, 47 with track data (25567#1 그룹)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 prediction/main.py | 37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

diff --git a/prediction/main.py b/prediction/main.py
index 00c9b33..139912f 100644
--- a/prediction/main.py
+++ b/prediction/main.py
@@ -85,24 +85,26 @@ def get_correlation_tracks(
     from cache.vessel_store import vessel_store
 
     try:
-        conn = kcgdb.get_conn()
-        cur = conn.cursor()
+        with kcgdb.get_conn() as conn:
+            cur = conn.cursor()
 
-        # Get correlated vessels from ALL active models
-        cur.execute("""
-            SELECT s.target_mmsi, s.target_type, s.target_name,
-                   s.current_score, m.name AS model_name
-            FROM kcg.gear_correlation_scores s
-            JOIN kcg.correlation_param_models m ON s.model_id = m.id
-            WHERE s.group_key = %s
-              AND s.current_score >= %s
-              AND m.is_active = TRUE
-            ORDER BY s.current_score DESC
-        """, (group_key, min_score))
+            # Get correlated vessels from ALL active models
+            cur.execute("""
+                SELECT s.target_mmsi, s.target_type, s.target_name,
+                       s.current_score, m.name AS model_name
+                FROM kcg.gear_correlation_scores s
+                JOIN kcg.correlation_param_models m ON s.model_id = m.id
+                WHERE s.group_key = %s
+                  AND s.current_score >= %s
+                  AND m.is_active = TRUE
+                ORDER BY s.current_score DESC
+            """, (group_key, min_score))
 
-        rows = cur.fetchall()
-        cur.close()
-        conn.close()
+            rows = cur.fetchall()
+            cur.close()
+
+        logger.info('correlation tracks: group_key=%r, min_score=%s, rows=%d',
+                     group_key, min_score, len(rows))
 
         if not rows:
             return {'groupKey': group_key, 'vessels': []}
@@ -131,6 +133,9 @@ def get_correlation_tracks(
 
         # Get tracks from vessel_store
         tracks = vessel_store.get_vessel_tracks(mmsis, hours)
+        with_tracks = sum(1 for m in mmsis if m in tracks and len(tracks[m]) > 0)
+        logger.info('correlation tracks: %d unique mmsis, %d with track data, vessel_store._tracks has %d entries',
+                     len(mmsis), with_tracks, len(vessel_store._tracks))
 
         # Build response
         vessels = []