Reorder run_id and action partitions

Why these changes are being introduced:

Discussions around retrieving subsets of records from the dataset suggested
that it would be beneficial to have the run_id partition before the
action partition.  This will allow using a prefix approach of partition
names and values when loading a dataset, that will pinpoint a particular
run even before the dataset is fully loaded.

This ordering was originally proposed in the engineering plan for this
library, but it switched somewhere along the way; so moving back to
agreed upon ordering.

How this addresses that need:
* Moves run_id before action in ordered partition columns

Side effects of this change:
* Omitting the action partition, and using everything until the
run_id partition, is sufficient for getting all records from a run.

Relevant ticket(s):
* https://mitlibraries.atlassian.net/browse/TIMX-424

Author: ghukill

tests/test_dataset_write.py

@@ -169,14 +169,14 @@ def test_dataset_write_schema_partitions_correctly_ordered(
             "source": "alma",
             "run_date": "2024-12-01",
             "run_type": "daily",
-            "action": "index",
             "run_id": "000-111-aaa-bbb",
+            "action": "index",
         },
     )
     file = written_files[0]
     assert (
         "/source=alma/run_date=2024-12-01/run_type=daily"
-        "/action=index/run_id=000-111-aaa-bbb" in file.path
+        "/run_id=000-111-aaa-bbb/action=index/" in file.path
     )
 
 

timdex_dataset_api/dataset.py

@@ -28,17 +28,17 @@
         pa.field("source", pa.string()),
         pa.field("run_date", pa.date32()),
         pa.field("run_type", pa.string()),
-        pa.field("action", pa.string()),
         pa.field("run_id", pa.string()),
+        pa.field("action", pa.string()),
     )
 )
 
 TIMDEX_DATASET_PARTITION_COLUMNS = [
     "source",
     "run_date",
     "run_type",
-    "action",
     "run_id",
+    "action",
 ]
 
 DEFAULT_BATCH_SIZE = 1_000

timdex_dataset_api/record.py

@@ -24,8 +24,8 @@ class DatasetRecord:
     source: str | None = None
     run_date: str | datetime.datetime | None = None
     run_type: str | None = None
-    action: str | None = None
     run_id: str | None = None
+    action: str | None = None
 
     def to_dict(
         self,
@@ -46,7 +46,7 @@ def validate(self) -> None:
         # ensure all partition columns are set
         missing_partition_values = [
             field
-            for field in ["source", "run_date", "run_type", "action", "run_id"]
+            for field in ["source", "run_date", "run_type", "run_id", "action"]
             if getattr(self, field) is None
         ]
         if missing_partition_values: