From 07b016bff9efd1e1ecb2036942c05be0ee187921 Mon Sep 17 00:00:00 2001
From: vrubel <vrubel@chesscalc.local>
Date: Sat, 20 Jun 2026 09:06:11 +0000
Subject: [PATCH] perf: SQL LIKE query instead of Python loop; skip warmup if
 cache >1000

---
 bots/tracker.py | 65 ++++++++++++++++++++++++++-----------------------
 1 file changed, 35 insertions(+), 30 deletions(-)

diff --git a/bots/tracker.py b/bots/tracker.py
index 75957b4..fbaaf8e 100644
--- a/bots/tracker.py
+++ b/bots/tracker.py
@@ -525,35 +525,34 @@ def scan_for_player(fide_id: int, max_tnr: int = None,
     results = []
     seen_tnrs = set()
 
-    # Phase 1: check tnr_cache (instant SQL, no HTTP). Return immediately if found.
-    for cluster_lo, cluster_hi in [(1434000, 1450000)]:
-        conn = _get_conn()
-        cached_tnrs = [
-            r[0] for r in conn.execute(
-                'SELECT tnr FROM tnr_cache WHERE tnr >= ? AND tnr <= ? ORDER BY tnr DESC',
-                (cluster_lo, cluster_hi)).fetchall()
-        ]
-        conn.close()
-        for tnr in cached_tnrs:
-            cdata = _get_cached_tnr(tnr)
-            if cdata is None:
-                continue
-            player_info = cdata['players'].get(fide_id)
-            if player_info is None:
-                continue
-            results.append({
-                'tnr': tnr,
-                'url': f'https://chess-results.com/tnr{tnr}.aspx?lan=11',
-                'name': cdata['name'] or f'Tournament {tnr}',
-                'start_date': cdata['start_date'],
-                'end_date': cdata['end_date'],
-                'sno': player_info['sno'],
-                'player_name': player_info['name'],
-                'fide_id': fide_id,
-            })
-        if results:
-            set_tnr_state('max_tnr_seen', str(max(r['tnr'] for r in results)))
-            return results
+    # Phase 1: check tnr_cache via SQL (single query, instant). Return if found.
+    conn = _get_conn()
+    pattern = f'%"{fide_id}"%'
+    rows = conn.execute(
+        'SELECT tnr, name, start_date, end_date, players_json FROM tnr_cache '
+        'WHERE tnr >= ? AND tnr <= ? AND players_json LIKE ? '
+        'ORDER BY tnr DESC',
+        (1434000, 1450000, pattern)).fetchall()
+    conn.close()
+    for row in rows:
+        tnr, tname, tstart, tend, players_json = row
+        players = {int(k): v for k, v in json.loads(players_json).items()}
+        player_info = players.get(fide_id)
+        if player_info is None:
+            continue
+        results.append({
+            'tnr': tnr,
+            'url': f'https://chess-results.com/tnr{tnr}.aspx?lan=11',
+            'name': tname or f'Tournament {tnr}',
+            'start_date': tstart,
+            'end_date': tend,
+            'sno': player_info['sno'],
+            'player_name': player_info['name'],
+            'fide_id': fide_id,
+        })
+    if results:
+        set_tnr_state('max_tnr_seen', str(max(r['tnr'] for r in results)))
+        return results
 
     # Phase 2: cache miss — fall back to sequential TNR scan
     saved = int(get_tnr_state('max_tnr_seen', '0'))
@@ -634,10 +633,16 @@ async def warmup_cache(context):
     """Build full tnr_cache index (runs once at startup, then every 6h)."""
     import asyncio, sys, traceback
 
-    # Skip if warmup already completed
+    # Skip if warmup already completed or cache already populated
     status = get_tnr_state('warmup_tnr', '')
     if status == 'done':
         return
+    conn = _get_conn()
+    cache_count = conn.execute('SELECT COUNT(*) FROM tnr_cache').fetchone()[0]
+    conn.close()
+    if cache_count > 1000:
+        set_tnr_state('warmup_tnr', 'done')
+        return
 
     print('warmup_cache: starting cluster scan (step=1, 1445000→1434000)...', file=sys.stderr)
     try: