fix(db): add LIMIT to unbounded message history queries (#692) (#735)

getNewMessages() and getMessagesSince() loaded all rows after a
checkpoint with no cap, causing growing memory and token costs.
Both queries now use a DESC LIMIT subquery to return only the
most recent N messages, re-sorted chronologically.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

src/db.test.ts

@@ -391,6 +391,64 @@ describe('task CRUD', () => {
   });
 });
 
+// --- LIMIT behavior ---
+
+describe('message query LIMIT', () => {
+  beforeEach(() => {
+    storeChatMetadata('group@g.us', '2024-01-01T00:00:00.000Z');
+
+    for (let i = 1; i <= 10; i++) {
+      store({
+        id: `lim-${i}`,
+        chat_jid: 'group@g.us',
+        sender: 'user@s.whatsapp.net',
+        sender_name: 'User',
+        content: `message ${i}`,
+        timestamp: `2024-01-01T00:00:${String(i).padStart(2, '0')}.000Z`,
+      });
+    }
+  });
+
+  it('getNewMessages caps to limit and returns most recent in chronological order', () => {
+    const { messages, newTimestamp } = getNewMessages(
+      ['group@g.us'],
+      '2024-01-01T00:00:00.000Z',
+      'Andy',
+      3,
+    );
+    expect(messages).toHaveLength(3);
+    expect(messages[0].content).toBe('message 8');
+    expect(messages[2].content).toBe('message 10');
+    // Chronological order preserved
+    expect(messages[1].timestamp > messages[0].timestamp).toBe(true);
+    // newTimestamp reflects latest returned row
+    expect(newTimestamp).toBe('2024-01-01T00:00:10.000Z');
+  });
+
+  it('getMessagesSince caps to limit and returns most recent in chronological order', () => {
+    const messages = getMessagesSince(
+      'group@g.us',
+      '2024-01-01T00:00:00.000Z',
+      'Andy',
+      3,
+    );
+    expect(messages).toHaveLength(3);
+    expect(messages[0].content).toBe('message 8');
+    expect(messages[2].content).toBe('message 10');
+    expect(messages[1].timestamp > messages[0].timestamp).toBe(true);
+  });
+
+  it('returns all messages when count is under the limit', () => {
+    const { messages } = getNewMessages(
+      ['group@g.us'],
+      '2024-01-01T00:00:00.000Z',
+      'Andy',
+      50,
+    );
+    expect(messages).toHaveLength(10);
+  });
+});
+
 // --- RegisteredGroup isMain round-trip ---
 
 describe('registered group isMain', () => {

src/db.ts

@@ -306,24 +306,29 @@ export function getNewMessages(
   jids: string[],
   lastTimestamp: string,
   botPrefix: string,
+  limit: number = 200,
 ): { messages: NewMessage[]; newTimestamp: string } {
   if (jids.length === 0) return { messages: [], newTimestamp: lastTimestamp };
 
   const placeholders = jids.map(() => '?').join(',');
   // Filter bot messages using both the is_bot_message flag AND the content
   // prefix as a backstop for messages written before the migration ran.
+  // Subquery takes the N most recent, outer query re-sorts chronologically.
   const sql = `
-    SELECT id, chat_jid, sender, sender_name, content, timestamp, is_from_me
+    SELECT * FROM (
-    FROM messages
+      SELECT id, chat_jid, sender, sender_name, content, timestamp, is_from_me
-    WHERE timestamp > ? AND chat_jid IN (${placeholders})
+      FROM messages
-      AND is_bot_message = 0 AND content NOT LIKE ?
+      WHERE timestamp > ? AND chat_jid IN (${placeholders})
-      AND content != '' AND content IS NOT NULL
+        AND is_bot_message = 0 AND content NOT LIKE ?
-    ORDER BY timestamp
+        AND content != '' AND content IS NOT NULL
+      ORDER BY timestamp DESC
+      LIMIT ?
+    ) ORDER BY timestamp
   `;
 
   const rows = db
     .prepare(sql)
-    .all(lastTimestamp, ...jids, `${botPrefix}:%`) as NewMessage[];
+    .all(lastTimestamp, ...jids, `${botPrefix}:%`, limit) as NewMessage[];
 
   let newTimestamp = lastTimestamp;
   for (const row of rows) {
@@ -337,20 +342,25 @@ export function getMessagesSince(
   chatJid: string,
   sinceTimestamp: string,
   botPrefix: string,
+  limit: number = 200,
 ): NewMessage[] {
   // Filter bot messages using both the is_bot_message flag AND the content
   // prefix as a backstop for messages written before the migration ran.
+  // Subquery takes the N most recent, outer query re-sorts chronologically.
   const sql = `
-    SELECT id, chat_jid, sender, sender_name, content, timestamp, is_from_me
+    SELECT * FROM (
-    FROM messages
+      SELECT id, chat_jid, sender, sender_name, content, timestamp, is_from_me
-    WHERE chat_jid = ? AND timestamp > ?
+      FROM messages
-      AND is_bot_message = 0 AND content NOT LIKE ?
+      WHERE chat_jid = ? AND timestamp > ?
-      AND content != '' AND content IS NOT NULL
+        AND is_bot_message = 0 AND content NOT LIKE ?
-    ORDER BY timestamp
+        AND content != '' AND content IS NOT NULL
+      ORDER BY timestamp DESC
+      LIMIT ?
+    ) ORDER BY timestamp
   `;
   return db
     .prepare(sql)
-    .all(chatJid, sinceTimestamp, `${botPrefix}:%`) as NewMessage[];
+    .all(chatJid, sinceTimestamp, `${botPrefix}:%`, limit) as NewMessage[];
 }
 
 export function createTask(