feat: add sdg training data bridge

2026-07-21 13:14:52 -05:00
parent f1a8bb0f35
commit 1d683ab380
13 changed files with 933 additions and 1 deletions
@@ -0,0 +1,101 @@
+import argparse
+import json
+import shutil
+import tempfile
+import unittest
+from pathlib import Path
+
+from scripts import generate_training_data as bridge
+
+
+class GenerateTrainingDataTests(unittest.TestCase):
+    def setUp(self):
+        self.repo = Path(__file__).resolve().parents[1]
+        self.temp = tempfile.TemporaryDirectory()
+        self.root = Path(self.temp.name)
+        self.dictionary_dir = self.root / "dictionary"
+        shutil.copytree(self.repo / "dictionary", self.dictionary_dir)
+        self.output_dir = self.root / "workspace_Data" / "data"
+        self.output_dir.mkdir(parents=True)
+        self.manifest = self.root / "workspace_Data" / "manifest.llm.json"
+        self.manifest.write_text(json.dumps({"manifest_version": "0.0.0", "datasets": {}}))
+
+    def tearDown(self):
+        self.temp.cleanup()
+
+    def args(self, **overrides):
+        values = {
+            "dictionary_dir": str(self.dictionary_dir),
+            "output_dir": str(self.output_dir),
+            "manifest": str(self.manifest),
+            "dry_run": False,
+            "sdg_host": None,
+            "sdg_user": None,
+            "sdg_key": None,
+            "num_records": 1000,
+            "timestamp": 1234567890,
+        }
+        values.update(overrides)
+        return argparse.Namespace(**values)
+
+    def test_seed_generation_covers_current_dictionary(self):
+        artifacts = bridge.load_dictionary_artifacts(self.dictionary_dir)
+        records = bridge.build_seed_records(artifacts, generated_at=123)
+        actions = {record["metadata"]["action_id"] for record in records}
+        self.assertEqual(len(actions), 13)
+        self.assertIn("calculate", actions)
+        self.assertIn("delete_file", actions)
+        self.assertEqual(len(records), 260)
+
+    def test_compact_output_uses_codebook_codes(self):
+        artifacts = bridge.load_dictionary_artifacts(self.dictionary_dir)
+        compact = bridge.compact_target(
+            artifacts,
+            "hash_string",
+            {"text": "hello", "algorithm": "sha256"},
+        )
+        self.assertEqual(compact["a"], artifacts["action_to_code"]["hash_string"])
+        self.assertIn(artifacts["field_to_code"]["text"], compact["p"])
+        self.assertIn(artifacts["field_to_code"]["algorithm"], compact["p"])
+
+    def test_empty_payload_action(self):
+        artifacts = bridge.load_dictionary_artifacts(self.dictionary_dir)
+        records = bridge.build_seed_records(artifacts, generated_at=123)
+        uuid_records = [
+            record for record in records if record["metadata"]["action_id"] == "generate_uuid"
+        ]
+        self.assertTrue(uuid_records)
+        output = json.loads(uuid_records[0]["output"])
+        compact = json.loads(uuid_records[0]["metadata"]["compact"])
+        self.assertEqual(output["payload"], {})
+        self.assertEqual(compact["p"], {})
+
+    def test_split_includes_every_action(self):
+        artifacts = bridge.load_dictionary_artifacts(self.dictionary_dir)
+        records = bridge.build_seed_records(artifacts, generated_at=123)
+        train, validation = bridge.split_records_by_action(records)
+        train_actions = {record["metadata"]["action_id"] for record in train}
+        validation_actions = {record["metadata"]["action_id"] for record in validation}
+        self.assertEqual(train_actions, validation_actions)
+        self.assertEqual(len(train_actions), 13)
+
+    def test_dry_run_writes_nothing(self):
+        summary = bridge.run_bridge(self.args(dry_run=True))
+        self.assertTrue(summary["dry_run"])
+        self.assertFalse((self.output_dir / "module_controller_intents_1234567890").exists())
+        manifest = json.loads(self.manifest.read_text())
+        self.assertEqual(manifest["datasets"], {})
+
+    def test_manifest_patch_adds_dataset_without_damaging_existing_keys(self):
+        bridge.run_bridge(self.args())
+        manifest = json.loads(self.manifest.read_text())
+        self.assertEqual(manifest["manifest_version"], "0.0.0")
+        dataset = manifest["datasets"]["module_controller_intents_1234567890"]
+        self.assertEqual(dataset["schema_profile"], "instruction")
+        self.assertEqual(dataset["codebook_version"], "1.0.0")
+        self.assertIn("codebook_checksum", dataset)
+        self.assertTrue((self.output_dir / "module_controller_intents_1234567890" / "train.jsonl").exists())
+
+
+if __name__ == "__main__":
+    unittest.main()