Emit certified populace-us URI in US national Reproduce-in-Python snippets

app/src/constants.ts

@@ -2,6 +2,19 @@ export const FOREVER = '2100-12-31';
 export const BASE_URL = 'https://api.policyengine.org';
 export const CURRENT_YEAR = '2026';
 
+// Certified default US microdata dataset URI, pinned to the revision that
+// policyengine.py's bundle manifest resolves as data_releases.us.default_dataset_uri.
+// The legacy `policyengine-us-data` Hugging Face repo is deprecated/archived, so any
+// generated "Reproduce in Python" snippet for a US national run must point here rather
+// than at a `policyengine-us-data` path. Pinned (rather than tracking a floating branch)
+// so a copied snippet reproduces the exact certified dataset the app ran against.
+// Ported from policyengine-app v1 (POPULACE_US_DEFAULT_DATASET_URI, PR #2846).
+// NOTE: subnational (state/CD) and place fallbacks still reference `policyengine-us-data`
+// pending Populace place/geo scoping — tracked in policyengine-app-v2#1079 — and are
+// intentionally NOT switched to this national URI here.
+export const POPULACE_US_DEFAULT_DATASET_URI =
+  'hf://policyengine/populace-us/populace_us_2024.h5@populace-us-2024-sparse-l0-refit-57k-71a0887-national-only-20260701';
+
 // App URLs for the split website/calculator architecture
 // In dev mode, these are set via VITE_* env vars to localhost URLs
 // In production, they fall back to the prod URLs

app/src/tests/unit/utils/reproducibilityCode.test.ts

@@ -1,4 +1,5 @@
 import { describe, expect, test, vi } from 'vitest';
+import { POPULACE_US_DEFAULT_DATASET_URI } from '@/constants';
 import { TEST_COUNTRIES } from '@/tests/fixtures/constants';
 import {
   BASELINE_AND_REFORM_POLICY,
@@ -596,9 +597,10 @@ describe('reproducibilityCode', () => {
         );
         const code = lines.join('\n');
 
-        // Then
-        expect(code).toContain('enhanced_cps_2024.h5');
-        expect(code).toContain('dataset=');
+        // Then - a bare US national dataset name resolves to the certified
+        // populace-us URI, never the deprecated policyengine-us-data repo.
+        expect(code).toContain(`dataset="${POPULACE_US_DEFAULT_DATASET_URI}"`);
+        expect(code).not.toContain('policyengine-us-data');
       });
 
       test('given full dataset url then uses it verbatim', () => {

app/src/utils/reproducibilityCode.ts

@@ -6,7 +6,7 @@
  * https://github.com/PolicyEngine/policyengine-app/blob/main/src/data/reformDefinitionCode.js
  */
 
-import { CURRENT_YEAR } from '@/constants';
+import { CURRENT_YEAR, POPULACE_US_DEFAULT_DATASET_URI } from '@/constants';
 import type { Household } from '@/models/Household';
 import {
   addVariationAxesToPythonPackageHouseholdData,
@@ -38,15 +38,21 @@ function normalizeDatasetUrlForReproducibility(countryId: string, datasetName: s
 
 /**
  * Build a HuggingFace dataset URL for a US national-level dataset.
- * Dataset files follow the pattern: {name}_{year}.h5
+ *
+ * A fully-qualified URI (e.g. the exact `hf://...` the backend ran against) is
+ * returned verbatim. For a bare US national dataset name, emit the certified
+ * `populace-us` URI rather than a legacy `policyengine-us-data/{name}_{year}.h5`
+ * path: that repo is deprecated/archived and policyengine.py no longer resolves
+ * it as the default, so a copied snippet must point at the certified dataset.
+ * Ported from policyengine-app v1 PR #2846.
  */
-function getDatasetUrl(countryId: string, datasetName: string, year: number): string | null {
+function getDatasetUrl(countryId: string, datasetName: string): string | null {
   if (datasetName.includes('://')) {
     return normalizeDatasetUrlForReproducibility(countryId, datasetName);
   }
 
   if (countryId === 'us') {
-    return `hf://policyengine/policyengine-us-data/${datasetName}_${year}.h5`;
+    return POPULACE_US_DEFAULT_DATASET_URI;
   }
   return null;
 }
@@ -57,6 +63,12 @@ function getDatasetUrl(countryId: string, datasetName: string, year: number): st
  *   - "state/ca" → states/CA.h5
  *   - "congressional_district/CA-01" → districts/CA-01.h5
  * Note: place/ regions are handled separately via getPlaceStateDatasetUrl.
+ *
+ * These `policyengine-us-data/{states,districts}/*.h5` paths are the geography-
+ * scoped fallback used only when a non-default subnational dataset is selected.
+ * They intentionally still reference the legacy repo pending Populace geo scoping;
+ * see policyengine-app-v2#1079. Default state/CD runs already use national Populace
+ * scoping via getScopedUsRegionImplementationCode, so they do not hit this path.
  */
 function getSubnationalDatasetUrl(region: string): string | null {
   for (const [prefix, folder] of Object.entries(US_REGION_PREFIX_TO_FOLDER)) {
@@ -90,6 +102,11 @@ function isDefaultUsScopedRegion(
 /**
  * For place/ regions, get the parent state's dataset URL.
  * "place/NJ-57000" → states/NJ.h5
+ *
+ * Places load the parent state's `policyengine-us-data` H5 and filter by
+ * place_fips because the certified Populace export does not yet carry place
+ * geography. This legacy reference is deliberately retained pending Populace
+ * place scoping; see policyengine-app-v2#1079.
  */
 function getPlaceStateDatasetUrl(region: string): string | null {
   if (!region.startsWith('place/')) {
@@ -318,7 +335,7 @@ function getImplementationCode(
   const isNational = region === countryId;
   const year = timePeriod || DEFAULT_YEAR;
   const resolvedDatasetUrl =
-    dataset && !isDefaultDataset ? getDatasetUrl(countryId, dataset, year) : null;
+    dataset && !isDefaultDataset ? getDatasetUrl(countryId, dataset) : null;
 
   if (isDefaultUsScopedRegion(countryId, region, dataset, isDefaultDataset)) {
     return getScopedUsRegionImplementationCode(region, year, hasBaseline, hasReform);

changelog_entry.yaml

@@ -8,3 +8,5 @@
     changed:
       - Proxy UK Python package docs at /uk/python and UK API docs at /uk/api so the country-parameterized "Python" and "API" nav links work on UK pages
       - Align the calculator app header (e.g. /us/reports) with the main website header — hover-open dropdowns, per-item underline, Model dropdown with sub-items, Python link, and Events under About
+    fixed:
+      - Point "Reproduce in Python" snippets for a US national run at the certified populace-us dataset instead of the deprecated policyengine-us-data repo