Make Ledger aggregate facts sum-only

arch/core.py

@@ -45,7 +45,6 @@
     "institutional_sector",
 }
 ALLOWED_AGGREGATIONS = {
-    "count",
     "sum",
     "mean",
     "median",
@@ -245,7 +244,7 @@ def build_fact_key(fact: AggregateFact) -> str:
 def build_label(fact: AggregateFact) -> str:
     """Build a human-readable label from fact metadata."""
     concept = _humanize(fact.measure.concept)
-    aggregation = _humanize(fact.aggregation.method)
+    aggregation = _aggregation_label(fact)
     entity = _humanize(fact.entity.name)
     period = f"{fact.period.value} {_humanize(fact.period.type)}"
     geography = fact.geography.name or fact.geography.id
@@ -259,6 +258,12 @@ def build_label(fact: AggregateFact) -> str:
     return label
 
 
+def _aggregation_label(fact: AggregateFact) -> str:
+    if fact.aggregation.method == "sum" and fact.measure.unit == "count":
+        return "count"
+    return _humanize(fact.aggregation.method)
+
+
 def build_aggregate_constraints(
     fact: AggregateFact,
 ) -> tuple[AggregateConstraint, ...]:

arch/jurisdictions/us/soi.py

@@ -40,8 +40,7 @@
 IRS_SOI_TOTAL_WAGES_SOURCE_CONCEPT = "irs_soi.total_wages"
 IRS_SOI_TOTAL_WAGES_RETURNS_SOURCE_CONCEPT = "irs_soi.returns_with_total_wages"
 AGI_CONCEPT_EVIDENCE_URL = (
-    "https://uscode.house.gov/view.xhtml?"
-    "req=(title:26%20section:62%20edition:prelim)"
+    "https://uscode.house.gov/view.xhtml?req=(title:26%20section:62%20edition:prelim)"
 )
 AGI_CONCEPT_EVIDENCE_NOTES = (
     "IRS SOI Table 1.1 reports adjusted gross income for individual income tax "
@@ -171,7 +170,7 @@ def _legacy_soi_table_1_1_source_record_specs(
                     measure_id="return_count",
                     concept="irs_soi.individual_income_tax_returns",
                     unit="count",
-                    aggregation="count",
+                    aggregation="sum",
                     filters=filters,
                 ),
                 _legacy_source_record_spec(
@@ -215,7 +214,7 @@ def _legacy_soi_table_1_1_source_record_specs(
                     measure_id="income_tax_after_credits_returns",
                     concept="irs_soi.returns_with_income_tax_after_credits",
                     unit="count",
-                    aggregation="count",
+                    aggregation="sum",
                     filters=filters,
                 ),
             ]
@@ -280,9 +279,7 @@ def _legacy_source_record_spec(
     legal_vintage: str | None = None,
     value_scale: int = 1,
 ) -> SourceRecordSpec:
-    source_record_id = (
-        f"irs_soi.ty{year}.table_1_1.{range_id}.{measure_id}"
-    )
+    source_record_id = f"irs_soi.ty{year}.table_1_1.{range_id}.{measure_id}"
     return SourceRecordSpec(
         source_record_id=source_record_id,
         selector=CellSelectorSpec(

arch/source_package.py

@@ -17,6 +17,7 @@
 import yaml
 
 from arch.core import (
+    ALLOWED_AGGREGATIONS,
     Aggregation,
     AggregateConstraint,
     EntityDimension,
@@ -1065,9 +1066,7 @@ def _validate_record_set_authoring(
             errors.append(
                 SourcePackageIssue(
                     code="malformed_row_column",
-                    message=(
-                        "Row column must be an Excel column name like B or AA."
-                    ),
+                    message=("Row column must be an Excel column name like B or AA."),
                     record_set_id=record_set.record_set_id,
                     row_id=row.value_id,
                 )
@@ -1100,6 +1099,18 @@ def _validate_record_set_authoring(
                     measure_id=measure.measure_id,
                 )
             )
+        if measure.aggregation not in ALLOWED_AGGREGATIONS:
+            errors.append(
+                SourcePackageIssue(
+                    code="malformed_measure_aggregation",
+                    message=(
+                        "Measure aggregation must be one of "
+                        f"{sorted(ALLOWED_AGGREGATIONS)!r}."
+                    ),
+                    record_set_id=record_set.record_set_id,
+                    measure_id=measure.measure_id,
+                )
+            )
         if measure.concept_relation == "exact" and not (
             measure.concept_evidence_url or measure.concept_evidence_notes
         ):
@@ -1835,5 +1846,5 @@ def _scaffold_template(
         column: B
         concept: TODO_concept
         unit: count
-        aggregation: count
+        aggregation: sum
 """