maps

Author: michielkallenberg

cybench/runs/analysis/collect_walk_forward_results.py

@@ -455,6 +455,11 @@ def write_model_comparison_dashboard(
     map_payload = build_region_map_payload(output_dir, summary_rows)
     html_dir = str(output_dir)
     if bundle_assets:
+        records = bundle_referenced_assets(
+            records=records,
+            output_dir=html_dir,
+            assets_dirname="assets",
+        )
         if map_payload.get("datasets"):
             map_payload = bundle_region_map_assets(
                 map_payload, output_dir, assets_dirname="assets"
@@ -463,11 +468,6 @@ def write_model_comparison_dashboard(
                 write_region_map_sidecar(output_dir, map_payload)
         if map_payload.get("datasets") and map_payload.get("geojson_by_country"):
             records = strip_map_pngs_from_records(records)
-        records = bundle_referenced_assets(
-            records=records,
-            output_dir=html_dir,
-            assets_dirname="assets",
-        )
     elif map_payload.get("datasets") and map_payload.get("geojson_by_country"):
         records = strip_map_pngs_from_records(records)
     html_path = output_dir / "compare_models.html"

cybench/runs/slurm/README.md

@@ -453,6 +453,12 @@ poetry run python data_preparation/fetch_zenodo_data.py --geometries
 # creates cybench/data/polygons/DE/DE.shp, NL/NL.shp, ...
 ```
 
+For map geometry only (no 6.2 GB cybench-data.zip):
+
+```bash
+poetry run python data_preparation/fetch_zenodo_data.py --geometries-only
+```
+
 **World outline** (grey background in map panels): bundled at
 ``data_preparation/ne_50m_admin_0_countries/`` (default). Falls back to 110m if missing.
 Override with ``CYBENCH_WORLD_MAP_SCALE=10|50|110``.

cybench/runs/viz/dashboard_template.html

@@ -324,13 +324,19 @@ <h3>Drill-down plots</h3>
       if (geoCache.has(country)) return geoCache.get(country);
       const href = (MAP_DATA.geojson_by_country || {})[country];
       if (!href) return null;
-      const geo = await d3.json(href);
-      const features = (geo.features || []).map(f => ({
-        ...f,
-        properties: { ...f.properties, loc: String(f.properties.loc ?? "") },
-      }));
-      geoCache.set(country, features);
-      return features;
+      try {
+        const geo = await d3.json(href);
+        if (!geo || !geo.features) return null;
+        const features = geo.features.map(f => ({
+          ...f,
+          properties: { ...f.properties, loc: String(f.properties.loc ?? "") },
+        }));
+        geoCache.set(country, features);
+        return features;
+      } catch (err) {
+        console.warn("GeoJSON load failed:", href, err);
+        return null;
+      }
     }
 
     function yieldColor(value, extent) {
@@ -822,9 +828,12 @@ <h3>Drill-down plots</h3>
           renderDynamicMaps(selectedCell.dataset, selectedCell.model).then(ok => {
             if (!ok) {
               showMapMode(false);
+              const href = ((MAP_DATA.geojson_by_country || {})[
+                (MAP_DATA.datasets[selectedCell.dataset] || {}).country
+              ]) || "assets/regions_*.geojson";
               missingNote.textContent =
-                "Could not render dynamic maps (missing geometry or preds). "
-                + "Re-run collect to export preds/ and ensure polygons exist.";
+                `Could not load map geometry (${href} missing or failed). `
+                + "Re-run collect + publish so regions_*.geojson is bundled in assets/.";
             }
           }).catch(err => {
             showMapMode(false);

pyproject.toml

@@ -42,6 +42,7 @@ transformers = ">=4.55.0,<5.0.0"
 tabpfn = ">=6.4.0,<7.0.0"
 tabicl = ">=2.1.0"
 tabdpt = ">=1.2.0"
+zenodo-get = "^1.6.1"
 
 [build-system]
 requires = ["poetry-core"]

tests/data_preparation/test_fetch_zenodo_data.py

@@ -0,0 +1,62 @@
+"""Tests for Zenodo dataset fetch helper."""
+
+from __future__ import annotations
+
+import zipfile
+from pathlib import Path
+
+import pytest
+
+from data_preparation.fetch_zenodo_data import (
+    DEFAULT_DOI,
+    extract_geometry_zip,
+    fetch,
+)
+
+
+def _write_geometry_zip(dest: Path, folder_name: str, country: str = "DE") -> None:
+    with zipfile.ZipFile(dest, "w") as zf:
+        zf.writestr(f"{folder_name}/{country}/readme.txt", "stub shapefile tree")
+
+
+def test_extract_geometry_zip(tmp_path: Path):
+    zip_path = tmp_path / "polygons.zip"
+    data_dir = tmp_path / "data"
+    _write_geometry_zip(zip_path, "polygons", "DE")
+    extract_geometry_zip(zip_path, data_dir, "polygons")
+    assert (data_dir / "polygons" / "DE" / "readme.txt").is_file()
+
+
+def test_fetch_geometries_only_skip_download(tmp_path: Path):
+    staging = tmp_path / "staging"
+    data_dir = tmp_path / "data"
+    staging.mkdir()
+    _write_geometry_zip(staging / "centroids.zip", "centroids", "NL")
+    _write_geometry_zip(staging / "polygons.zip", "polygons", "NL")
+
+    fetch(
+        data_dir,
+        staging_dir=staging,
+        include_geometries=True,
+        geometries_only=True,
+        skip_download=True,
+    )
+    assert (data_dir / "polygons" / "NL" / "readme.txt").is_file()
+    assert (data_dir / "centroids" / "NL" / "readme.txt").is_file()
+
+
+def test_geometries_only_requires_geometry_zips_when_skip_download(tmp_path: Path):
+    staging = tmp_path / "staging"
+    staging.mkdir()
+    with pytest.raises(FileNotFoundError, match="geometry zips missing"):
+        fetch(
+            tmp_path / "data",
+            staging_dir=staging,
+            include_geometries=True,
+            geometries_only=True,
+            skip_download=True,
+        )
+
+
+def test_default_doi_is_concept_record():
+    assert DEFAULT_DOI == "10.5281/zenodo.11502142"

tests/runs/test_region_map_lib.py

@@ -95,3 +95,67 @@ def test_strip_map_pngs_from_records():
     assert "map_actual" not in records[0]["images"]
     assert "map_pred" not in records[0]["images"]
     assert records[0]["images"]["scatter"] == "c.png"
+
+
+def test_bundle_region_map_survives_referenced_assets(tmp_path: Path, monkeypatch):
+    """GeoJSON must be written after bundle_referenced_assets (which wipes assets/)."""
+    from cybench.runs.analysis.collect_walk_forward_results import (
+        write_model_comparison_dashboard,
+    )
+
+    output_dir = tmp_path / "paper_walk_forward_us_eos_v2"
+    preds = output_dir / "preds" / "ridge_eos"
+    preds.mkdir(parents=True)
+    pd.DataFrame(
+        {
+            "adm_id": ["US-01-001"],
+            "year": [2020],
+            "yield": [10.0],
+            "Ridge": [9.5],
+        }
+    ).to_csv(preds / "maize_US_h10_year_2020.csv", index=False)
+
+    scatter_src = preds / "report_assets"
+    scatter_src.mkdir(parents=True)
+    scatter_src.joinpath("maize_US_scatter.png").write_bytes(b"png")
+
+    summary_rows = [
+        {
+            "dataset": "maize_US",
+            "model": "ridge",
+            "horizon": "eos",
+            "model_col": "Ridge",
+            "nrmse": 0.1,
+            "r2": 0.8,
+            "r_spatial": 0.5,
+            "r_spatial_agg": 0.5,
+            "r_temporal": 0.4,
+            "r_temporal_agg": 0.4,
+            "r_res": 0.3,
+            "r2_res": 0.2,
+            "n_regions": 1,
+            "n_years": 1,
+            "n_samples": 1,
+        }
+    ]
+
+    def _fake_export(country_code: str, dest: Path, **kwargs):
+        dest.parent.mkdir(parents=True, exist_ok=True)
+        dest.write_text(
+            '{"type":"FeatureCollection","features":[{"type":"Feature",'
+            '"properties":{"loc":"US-01-001"},"geometry":{"type":"Point","coordinates":[0,0]}}]}',
+            encoding="utf-8",
+        )
+        return dest
+
+    monkeypatch.setattr(
+        "cybench.runs.viz.region_map_lib.export_region_geojson",
+        _fake_export,
+    )
+
+    write_model_comparison_dashboard(output_dir, summary_rows, bundle_assets=True)
+    geojson = output_dir / "assets" / "regions_US.geojson"
+    assert geojson.is_file(), "regions_US.geojson must survive asset bundling"
+    html = (output_dir / "compare_models.html").read_text(encoding="utf-8")
+    assert "regions_US.geojson" in html
+    assert (output_dir / "assets" / "maize_US_scatter.png").is_file()