fix: truncate rewritten training JSON

Fixes #1909

Coding-Agent: Codex

Codex-Version: codex-cli 0.149.0

Model: gpt-5.6-sol

Reasoning-Effort: xhigh

Author: njzjz-bot

dpgen/util.py

@@ -140,6 +140,9 @@ def convert_training_data_to_hdf5(input_files: list[str], h5_file: str):
                 systems.append(pp)
             f.seek(0)
             json.dump(jinput, f, indent=4)
+            # The rewritten configuration can be shorter than the original.
+            # Remove any bytes left after the new JSON document.
+            f.truncate()
     systems = list(set(systems))
 
     dlog.info("Combining %d training systems to %s...", len(systems), h5_file)

tests/test_util.py

@@ -0,0 +1,36 @@
+import json
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+
+from dpgen.util import convert_training_data_to_hdf5
+
+
+class TestConvertTrainingDataToHdf5(unittest.TestCase):
+    def test_rewritten_json_is_truncated(self):
+        with tempfile.TemporaryDirectory() as tmpdir:
+            tmp_path = Path(tmpdir)
+            input_file = tmp_path / "input.json"
+            input_file.write_text(
+                json.dumps(
+                    {
+                        "training": {"systems": []},
+                        "padding": "content that must not survive the rewrite",
+                    }
+                )
+            )
+
+            with patch(
+                "dpgen.util.json.dump",
+                side_effect=lambda _data, file, indent: file.write("{}"),
+            ):
+                convert_training_data_to_hdf5(
+                    [str(input_file)], str(tmp_path / "data.hdf5")
+                )
+
+            self.assertEqual(input_file.read_text(), "{}")
+
+
+if __name__ == "__main__":
+    unittest.main()