Handle invalid ffprobe JSON during stage1

src/tgvideo_check/cli.py

@@ -435,6 +435,26 @@ def _execute_stage1_scan(config: ScanConfig) -> list[ReportRow]:
 
                 payload = _parse_probe_payload(probe_result.stdout_text)
                 if probe_result.stdout_text and payload is None:
+                    classification = classify_stage1_result(
+                        exit_code=probe_result.exit_code,
+                        stderr_text=probe_result.stderr_text,
+                        probe_payload=None,
+                        extension=stored_file.extension,
+                        high_risk_extensions=config.high_risk_extensions,
+                    )
+                    if classification.reason_code in {"ffprobe_exit_nonzero", "hard_error_keyword"}:
+                        final_status = classification.status
+                        store.record_stage1_status(
+                            stored_file.file_id,
+                            stage1_status=classification.status,
+                            final_status=final_status,
+                            reason_code=classification.reason_code,
+                        )
+                        reasons[stored_file.file_id] = classification.reason_code
+                        if stage1_progress is not None:
+                            stage1_progress.record(classification.status)
+                        continue
+
                     stage1_status = "command_error"
                     store.record_stage1_status(
                         stored_file.file_id,
@@ -656,7 +676,10 @@ def _reload_probe_payload(path: Path) -> dict[str, object] | None:
 def _parse_probe_payload(stdout_text: str) -> dict[str, object] | None:
     if not stdout_text.strip():
         return None
+    try:
         payload = orjson.loads(stdout_text)
+    except orjson.JSONDecodeError:
+        return None
     if isinstance(payload, dict):
         return payload
     return None

tests/test_cli.py

@@ -161,6 +161,88 @@ def test_scan_command_processes_stage1_results_and_exports_reports(
     assert csv_path.exists()
 
 
+def test_scan_command_treats_invalid_probe_json_with_nonzero_exit_as_confirmed_broken(
+    tmp_path: Path, monkeypatch
+) -> None:
+    root = tmp_path / "videos"
+    root.mkdir()
+    broken = root / "broken.mp4"
+    broken.write_bytes(b"broken")
+    db_path = tmp_path / "scan_state.db"
+
+    def fake_run_ffprobe(path, **kwargs):
+        return ProbeExecutionResult(
+            status="completed",
+            command=("ffprobe", str(path)),
+            exit_code=1,
+            stdout_text='{\n',
+            stderr_text="",
+            elapsed_ms=1,
+        )
+
+    monkeypatch.setattr("tgvideo_check.cli.run_ffprobe", fake_run_ffprobe)
+
+    result = runner.invoke(
+        app,
+        [
+            "scan",
+            str(root),
+            "--db",
+            str(db_path),
+        ],
+    )
+
+    assert result.exit_code == 0
+    assert "Scanned 1 file(s)" in result.stdout
+    assert "confirmed_broken=1" in result.stdout
+
+    store = ScanStore(db_path)
+    rows = store.list_files(1)
+    assert rows[0].final_status == "confirmed_broken"
+    assert rows[0].reason_code == "ffprobe_exit_nonzero"
+
+
+def test_scan_command_treats_invalid_probe_json_with_zero_exit_as_command_error(
+    tmp_path: Path, monkeypatch
+) -> None:
+    root = tmp_path / "videos"
+    root.mkdir()
+    broken = root / "broken.mp4"
+    broken.write_bytes(b"broken")
+    db_path = tmp_path / "scan_state.db"
+
+    def fake_run_ffprobe(path, **kwargs):
+        return ProbeExecutionResult(
+            status="completed",
+            command=("ffprobe", str(path)),
+            exit_code=0,
+            stdout_text='{\n',
+            stderr_text="",
+            elapsed_ms=1,
+        )
+
+    monkeypatch.setattr("tgvideo_check.cli.run_ffprobe", fake_run_ffprobe)
+
+    result = runner.invoke(
+        app,
+        [
+            "scan",
+            str(root),
+            "--db",
+            str(db_path),
+        ],
+    )
+
+    assert result.exit_code == 0
+    assert "Scanned 1 file(s)" in result.stdout
+    assert "command_error=1" in result.stdout
+
+    store = ScanStore(db_path)
+    rows = store.list_files(1)
+    assert rows[0].final_status == "command_error"
+    assert rows[0].reason_code == "invalid_probe_output"
+
+
 def test_scan_command_leaves_stage2_pending_by_default(tmp_path, monkeypatch) -> None:
     root = tmp_path / "videos"
     root.mkdir()