entities extraction fix

test/entity-extraction.test.js

@@ -0,0 +1,49 @@
+// Entity extraction guards — tests the REAL mentionedIn and isIgnoredName filters
+// that keep the extractor from storing (a) greetings and (b) "known entities" hint
+// regurgitation. The mentionedIn cases are built from an actual observed failure:
+// a "good morning" turn where qwen echoed the hint list back as fake extractions.
+const { test } = require('node:test');
+const assert = require('node:assert');
+const { mentionedIn, isIgnoredName } = require('../packages/memory-service/src/entities/extraction');
+
+test('mentionedIn accepts names that actually appear in the exchange', () => {
+  const text = 'User: I love how Sanderson writes magic systems. Assistant: Brandon Sanderson is known for hard magic.';
+  assert.ok(mentionedIn('Sanderson', text));
+  assert.ok(mentionedIn('Brandon Sanderson', text));
+});
+
+test('mentionedIn is case- and whitespace-insensitive', () => {
+  assert.ok(mentionedIn('digimon world next order', 'talking about Digimon World Next Order today'));
+  assert.ok(mentionedIn('Digimon   World', 'the Digimon World game'));   // collapsed whitespace
+});
+
+test('mentionedIn matches a name inside a possessive', () => {
+  assert.ok(mentionedIn('Sanderson', "Sanderson's latest book"));
+});
+
+test('regurgitated hint entities are rejected on a greeting turn (the bug)', () => {
+  // The exchange that triggered the bug — a bare greeting, no entities in it.
+  const exchange = 'good morning again Morning! Good timing — anything on your mind today?';
+  // The names qwen wrongly "extracted" from the hint block, verbatim from the logs.
+  const regurgitated = [
+    'Sanderson', 'Baldree', 'Riordan',
+    'Digimon World Next Order',
+    'Digimon Adventure: Last Evolution Kizuna',
+    'Digimon Adventure (2020 reboot)',
+  ];
+  for (const name of regurgitated) {
+    assert.strictEqual(mentionedIn(name, exchange), false, `"${name}" should be rejected — not in the exchange`);
+  }
+});
+
+test('a real entity survives even when hint regurgitation is happening', () => {
+  const exchange = 'good morning — did you finish the Qdrant migration? Yes, the Qdrant collection is rebuilt.';
+  assert.ok(mentionedIn('Qdrant', exchange), 'genuinely-discussed entity is kept');
+});
+
+test('isIgnoredName catches greetings regardless of punctuation/case', () => {
+  assert.ok(isIgnoredName('Good morning'));
+  assert.ok(isIgnoredName('good morning!'));
+  assert.ok(isIgnoredName('  HELLO  '));
+  assert.strictEqual(isIgnoredName('Qdrant'), false);
+});