fix: retranslate when glossary aliases change

Co-authored-by: Rainman <deusyu@users.noreply.github.com>

Author: cursoragent

scripts/glossary.py

@@ -74,9 +74,14 @@ def glossary_hash(glossary):
 
 
 def term_hash(term):
-    """SHA-256 of a single term's identifying fields."""
-    payload = f"{term.get('source', '')}→{term.get('target', '')}|{term.get('category', '')}"
-    return hashlib.sha256(payload.encode('utf-8')).hexdigest()
+    """SHA-256 of fields that affect how a term is injected into prompts."""
+    payload = {
+        'source': term.get('source', ''),
+        'target': term.get('target', ''),
+        'category': term.get('category', ''),
+        'aliases': sorted(term.get('aliases', []) or []),
+    }
+    return hashlib.sha256(_canonical_json(payload).encode('utf-8')).hexdigest()
 
 
 def _v2_term_defaults(term):

tests/test_glossary.py

@@ -625,6 +625,11 @@ def test_term_hash_changes_when_category_changes(self):
         t2 = {'source': 'Apple', 'target': '苹果', 'category': 'company'}
         self.assertNotEqual(glossary.term_hash(t1), glossary.term_hash(t2))
 
+    def test_term_hash_changes_when_aliases_change(self):
+        t1 = {'source': 'Tai', 'target': '太一', 'category': 'person', 'aliases': []}
+        t2 = {'source': 'Tai', 'target': '太一', 'category': 'person', 'aliases': ['Taichi']}
+        self.assertNotEqual(glossary.term_hash(t1), glossary.term_hash(t2))
+
 
 class FormatTermsForPromptTests(unittest.TestCase):
     def test_empty_terms_returns_empty_string(self):

tests/test_run_state.py

@@ -131,6 +131,25 @@ def test_new_alias_that_hits_chunk_changes_term_selection(self):
 
         self.assertEqual(plan["translation_chunk_ids"], ["chunk0001"])
 
+    def test_new_alias_retranslates_chunk_when_source_already_selected_term(self):
+        tmp = tempfile.TemporaryDirectory()
+        with tmp:
+            temp_dir = Path(tmp.name)
+            self._write(temp_dir / "input.md", "Tai and Taichi appear.\n")
+            self._write(temp_dir / "chunk0001.md", "Tai and Taichi appear.\n")
+            self._write(temp_dir / "output_chunk0001.md", "太一和 Taichi 出现。\n")
+            create_manifest(str(temp_dir), ["chunk0001.md"], str(temp_dir / "input.md"))
+            self._write(temp_dir / "glossary.json", json.dumps(glossary_doc(), ensure_ascii=False))
+            run_state.record_chunks(str(temp_dir), ["chunk0001"])
+
+            self._write(
+                temp_dir / "glossary.json",
+                json.dumps(glossary_doc(aliases=["Taichi"]), ensure_ascii=False),
+            )
+            plan = run_state.plan(str(temp_dir))
+
+        self.assertEqual(plan["translation_chunk_ids"], ["chunk0001"])
+
     def test_output_hash_change_is_record_only(self):
         tmp, temp_dir = self._workspace()
         with tmp: