Make the reconstructable job not take any arguments at all. (#3996)

src/pudl/etl/cli.py

@@ -23,7 +23,10 @@
 
 
 def pudl_etl_job_factory(
-    logfile: str | None = None, loglevel: str = "INFO", process_epacems: bool = True
+    logfile: str | None = None,
+    loglevel: str = "INFO",
+    process_epacems: bool = True,
+    base_job: str = "etl_full",
 ) -> Callable[[], JobDefinition]:
     """Factory for parameterizing a reconstructable pudl_etl job.
 
@@ -36,11 +39,11 @@ def pudl_etl_job_factory(
         The job definition to be executed.
     """
 
-    def get_pudl_etl_job(job_name: str | None = None):
+    def get_pudl_etl_job():
         """Create an pudl_etl_job wrapped by to be wrapped by reconstructable."""
         pudl.logging_helpers.configure_root_logger(logfile=logfile, loglevel=loglevel)
-        if job_name is None:
-            job_name = "etl_full_no_cems" if not process_epacems else "etl_full"
+        cems_suffix = "" if process_epacems else "_no_cems"
+        job_name = f"{base_job}{cems_suffix}"
         return defs.get_job_def(job_name)
 
     return get_pudl_etl_job

src/pudl/transform/vcerare.py

@@ -365,6 +365,8 @@ def check_rows(context: AssetCheckExecutionContext) -> AssetCheckResult:
     logger.info("Check VCE RARE hourly table is the expected length")
 
     # Define row counts for fast/full etl
+    # TODO 2024-12-27: make this check row counts per year instead of having
+    # two different counts based on job name - less brittle.
     row_counts = {
         "etl_full": 136437000,
         "etl_fast": 27287400,

test/conftest.py

@@ -363,7 +363,7 @@ def pudl_io_manager(
         md = PUDL_PACKAGE.to_sql()
         md.create_all(engine)
         # Run the ETL and generate a new PUDL SQLite DB for testing:
-        execute_result = pudl_etl_job_factory()("etl_fast").execute_in_process(
+        execute_result = pudl_etl_job_factory(base_job="etl_fast")().execute_in_process(
             run_config={
                 "resources": {
                     "dataset_settings": {