Fix some find_first follow-up queries in QG builder.

Find-first follow-up queries caused crash in some conditions in
QG builder due to a missing postprocessing parameter. The fix is
one-liner, and this patch adds a unit test which reproduces the crash
with more than one run collection.

python/lsst/daf/butler/_topology.py

@@ -76,7 +76,7 @@ class TopologicalFamily(ABC):
     another's in a predefined way.
 
     This hierarchy means that endpoints in the same family do not generally
-    have to be have to be related using (e.g.) overlaps; instead, the regions
+    have to be related using (e.g.) overlaps; instead, the regions
     from one "best" endpoint from each family are related to the best endpoint
     from each other family in a query.
 

python/lsst/daf/butler/direct_query_driver/_sql_builders.py

@@ -231,7 +231,7 @@ def into_from_builder(
                 self.select(postprocessing).cte() if cte else self.select(postprocessing).subquery()
             )
             return SqlJoinsBuilder(db=self.joins.db, from_clause=sql_from_clause).extract_columns(
-                self.columns, special=self.joins.special.keys()
+                self.columns, postprocessing, special=self.joins.special.keys()
             )
         return self.joins
 

python/lsst/daf/butler/tests/butler_queries.py

@@ -1222,6 +1222,43 @@ def test_materialization(self) -> None:
                 [1, 2, 3],
             )
 
+    def test_materialization_find_first(self) -> None:
+        """Test querying for datasets with find_first against a materialized
+        query.
+        """
+        butler = self.make_butler("ci_hsc-subset.yaml", "ci_hsc-subset-skymap.yaml")
+
+        run = "HSC/runs/ci_hsc/20240806T180642Z"
+        extra_run = "HSC/runs/ci_hsc/20240806T180642Z-extra"
+
+        # Find few datasets to duplicate.
+        refs = butler.query_datasets("calexp", run, limit=3)
+        data_ids = [ref.dataId for ref in refs]
+
+        butler.collections.register(extra_run)
+        butler.registry.insertDatasets("calexp", data_ids, extra_run)
+
+        collections = [run, extra_run, "skymaps"]
+        with butler.query() as query:
+            query = query.join_dimensions(
+                [
+                    "instrument",
+                    "physical_filter",
+                    "band",
+                    "visit",
+                    "detector",
+                    "day_obs",
+                    "skymap",
+                    "tract",
+                ]
+            )
+            query = query.join_dataset_search("skyMap", collections)
+            query = query.join_dataset_search("calexp", collections)
+            query = query.where({}, "instrument='HSC' AND skymap='discrete/ci_hsc'", bind=None)
+            m_query = query.materialize()
+            _ = list(m_query.datasets("skyMap", collections))
+            _ = list(m_query.datasets("calexp", collections))
+
     def test_timespan_results(self) -> None:
         """Test returning dimension records that include timespans."""
         butler = self.make_butler("base.yaml", "spatial.yaml")

tests/data/registry/ci_hsc-subset-skymap.yaml

@@ -0,0 +1,45 @@
+# Skymap definition from ci_hsc_gen3, for our tests we only need tracts,
+# dimension records for patches are not included.
+description: Butler Data Repository Export
+version: 1.0.2
+universe_version: 7
+universe_namespace: daf_butler
+data:
+- type: dimension
+  element: skymap
+  records:
+  - name: discrete/ci_hsc
+    hash: !!binary |
+      Kt7swQ9e5zUif7fo1Onj3ezH+Ts=
+    tract_max: 1
+    patch_nx_max: 16
+    patch_ny_max: 16
+- type: dimension
+  element: tract
+  records:
+  - skymap: discrete/ci_hsc
+    id: 0
+    region: !<lsst.sphgeom.ConvexPolygon>
+      encoded: 70cb48dd838019e83fe79b0e4ace0ae5bfcae4291ac61395bf4884794db329e93f68b2ba9327c2e3bfa603a921c61395bf0e047a6ce727e93fc46da7b1aac0e3bf8779dd59c921a03f513c3fa3b417e83f6ffa86675109e5bf06a12054c921a03f
+- type: collection
+  collection_type: RUN
+  name: skymaps
+  host: null
+  timespan_begin: null
+  timespan_end: null
+- type: dataset_type
+  name: skyMap
+  dimensions:
+  - skymap
+  storage_class: SkyMap
+  is_calibration: false
+- type: dataset
+  dataset_type: skyMap
+  run: skymaps
+  records:
+  - dataset_id:
+    - !uuid '557e15c6-0529-4fc9-998b-b90a4750315e'
+    data_id:
+    - skymap: discrete/ci_hsc
+    path: skymaps/skyMap/skyMap_discrete_ci_hsc_skymaps.pickle
+    formatter: lsst.daf.butler.formatters.pickle.PickleFormatter