Retain bounded invalid ASR output and atomic failure evidence

tools/speech-eval/test_contract.py

@@ -10,9 +10,33 @@ import importlib.util
 import hashlib
 from run import segments, run_owned, recognized_window, decode_command, decoded_frames, primary_audio_range
 from score import score, require_eligible
+from run import parse_asr_output, AsrOutputError, MAX_OUTPUT, checkpoint_report
 
 
 class ContractTests(unittest.TestCase):
+    def test_atomic_checkpoint_failure_preserves_previous_evidence(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / "report.json"
+            checkpoint_report(path, {"complete": False, "rows": []})
+            old = path.read_bytes()
+            with self.assertRaises(TypeError):
+                checkpoint_report(path, {"invalid": object()})
+            self.assertEqual(path.read_bytes(), old)
+    def test_invalid_asr_utf8_and_json_retain_exact_bounded_bytes(self):
+        with tempfile.TemporaryDirectory() as directory:
+            parent = Path(directory)
+            path = parent / "output.json"
+            for raw in [b'{"text":"\xe2\x28\xa1"}', b'{"text":']:
+                path.write_bytes(raw)
+                with self.assertRaises(AsrOutputError) as caught:
+                    parse_asr_output(path, parent)
+                failure = caught.exception.failure
+                self.assertEqual(Path(failure["raw_output_path"]).read_bytes(), raw)
+                self.assertEqual(failure["raw_output_sha256"], hashlib.sha256(raw).hexdigest())
+                self.assertEqual(failure["raw_output_bytes"], len(raw))
+            path.write_bytes(b"x" * (MAX_OUTPUT + 1))
+            with self.assertRaises(ValueError):
+                parse_asr_output(path, parent)
     def test_harness_source_hash_is_captured_before_file_changes(self):
         with tempfile.TemporaryDirectory() as directory:
             path = Path(directory) / "run-copy.py"