Finishing touches on Push to Huggingface Task

src/core/tasks/scheduled/impl/huggingface/queries/get/mappings.py

@@ -44,6 +44,7 @@
     RecordType.COURT_CASES: RecordTypeCoarse.JAILS_AND_COURTS,
     RecordType.INCARCERATION_RECORDS: RecordTypeCoarse.JAILS_AND_COURTS,
     # Other
+    RecordType.OTHER: RecordTypeCoarse.OTHER,
     None: RecordTypeCoarse.NOT_RELEVANT
 }
 

src/external/huggingface/hub/format.py

@@ -16,8 +16,8 @@ def format_as_huggingface_dataset(outputs: list[GetForLoadingToHuggingFaceOutput
         d['url_id'].append(output.url_id)
         d['url'].append(output.url)
         d['relevant'].append(output.relevant)
-        d['record_type_fine'].append(output.record_type_fine)
-        d['record_type_coarse'].append(output.record_type_coarse)
+        d['record_type_fine'].append(output.record_type_fine.value)
+        d['record_type_coarse'].append(output.record_type_coarse.value)
         d['html'].append(output.html)
     return Dataset.from_dict(d)
 

tests/manual/core/tasks/scheduled/test_push_to_huggingface.py

@@ -0,0 +1,26 @@
+import pytest
+
+from environs import Env
+
+from src.core.env_var_manager import EnvVarManager
+from src.core.tasks.scheduled.impl.huggingface.operator import PushToHuggingFaceTaskOperator
+from src.db.client.async_ import AsyncDatabaseClient
+from src.external.huggingface.hub.client import HuggingFaceHubClient
+
+env = Env()
+env.read_env()
+
+@pytest.mark.asyncio
+@pytest.mark.manual
+async def test_push_to_huggingface():
+    operator = PushToHuggingFaceTaskOperator(
+        adb_client=AsyncDatabaseClient(
+            db_url=env.str("PROD_DATABASE_URL")
+        ),
+        hf_client=HuggingFaceHubClient(
+            env.str("HUGGINGFACE_HUB_TOKEN")
+        )
+    )
+
+    await operator.inner_task_logic()
+