fix: permissive geoparquet extent reading

src/utils/stac-geoparquet.ts

@@ -9,6 +9,7 @@ import {
   vectorFromArray,
 } from "apache-arrow";
 import * as stacWasm from "stac-wasm";
+import { toaster } from "../components/ui/toaster";
 import type { StacItemCollection } from "../types/stac";
 
 const SUPPORTED_GEOMETRY_TYPES = ["point", "polygon", "linestring"] as const;
@@ -43,14 +44,11 @@ export async function fetchStacGeoparquetValue({
   connection: AsyncDuckDBConnection;
   hivePartitioning: boolean;
 }): Promise<StacItemCollection> {
-  const result = await executeDuckdbQuery({
-    connection,
+  const { count, bbox } = await fetchStacGeoparquetSummary({
     href,
+    connection,
     hivePartitioning,
-    select:
-      "COUNT(*) as count, MIN(bbox.xmin) as xmin, MIN(bbox.ymin) as ymin, MAX(bbox.xmax) as xmax, MAX(bbox.ymax) as ymax",
   });
-  const row = result.toArray().map((row) => row.toJSON())[0];
   const datetimeExtent = await fetchStacGeoparquetDatetimeExtent({
     href,
     connection,
@@ -59,8 +57,8 @@ export async function fetchStacGeoparquetValue({
   return {
     type: "FeatureCollection",
     id: href.split("/").pop(),
-    description: `A stac-geoparquet file with ${row.count} item${row.count === 1 ? "" : "s"}`,
-    bbox: [row.xmin, row.ymin, row.xmax, row.ymax],
+    description: `A stac-geoparquet file with ${count} item${count === 1 ? "" : "s"}`,
+    bbox,
     features: [],
     datetimeExtent,
     assets: {
@@ -72,6 +70,98 @@ export async function fetchStacGeoparquetValue({
   };
 }
 
+async function fetchStacGeoparquetSummary({
+  href,
+  connection,
+  hivePartitioning,
+}: {
+  href: string;
+  connection: AsyncDuckDBConnection;
+  hivePartitioning: boolean;
+}): Promise<{ count: number; bbox: [number, number, number, number] }> {
+  const bboxType = await getBboxColumnType({
+    href,
+    connection,
+    hivePartitioning,
+  });
+
+  if (bboxType?.startsWith("STRUCT")) {
+    const result = await executeDuckdbQuery({
+      connection,
+      href,
+      hivePartitioning,
+      select:
+        "COUNT(*) as count, MIN(bbox.xmin) as xmin, MIN(bbox.ymin) as ymin, MAX(bbox.xmax) as xmax, MAX(bbox.ymax) as ymax",
+    });
+    const row = result.toArray().map((row) => row.toJSON())[0];
+    return {
+      count: row.count,
+      bbox: [row.xmin, row.ymin, row.xmax, row.ymax],
+    };
+  }
+
+  if (bboxType === "DOUBLE[]") {
+    toaster.create({
+      title: "Non-spec stac-geoparquet",
+      description:
+        "The 'bbox' column is a list of doubles, but the stac-geoparquet spec requires a struct of {xmin, ymin, xmax, ymax}. Reading bbox values from the list instead.",
+      type: "warning",
+    });
+    const result = await executeDuckdbQuery({
+      connection,
+      href,
+      hivePartitioning,
+      select:
+        "COUNT(*) as count, MIN(bbox[1]) as xmin, MIN(bbox[2]) as ymin, MAX(bbox[3]) as xmax, MAX(bbox[4]) as ymax",
+    });
+    const row = result.toArray().map((row) => row.toJSON())[0];
+    return {
+      count: row.count,
+      bbox: [row.xmin, row.ymin, row.xmax, row.ymax],
+    };
+  }
+
+  if (bboxType) {
+    toaster.create({
+      title: "Non-spec stac-geoparquet",
+      description: `The 'bbox' column has type '${bboxType}', but the stac-geoparquet spec requires a struct of {xmin, ymin, xmax, ymax}. Computing the extent from the geometry column instead.`,
+      type: "warning",
+    });
+  }
+
+  const result = await executeDuckdbQuery({
+    connection,
+    href,
+    hivePartitioning,
+    select: "COUNT(*) as count, ST_Extent_Agg(geometry) as extent",
+  });
+  const row = result.toArray().map((row) => row.toJSON())[0];
+  const extent = row.extent;
+  return {
+    count: row.count,
+    bbox: [extent.min_x, extent.min_y, extent.max_x, extent.max_y],
+  };
+}
+
+async function getBboxColumnType({
+  href,
+  connection,
+  hivePartitioning,
+}: {
+  href: string;
+  connection: AsyncDuckDBConnection;
+  hivePartitioning: boolean;
+}): Promise<string | null> {
+  try {
+    const query = `SELECT typeof(bbox) as bbox_type FROM read_parquet('${href}', hive_partitioning = ${hivePartitioning}) LIMIT 1`;
+    const result = (await connection.query(query)) as unknown as Table;
+    const row = result.toArray().map((row) => row.toJSON())[0];
+    return (row?.bbox_type as string | undefined) ?? null;
+  } catch {
+    return null;
+  }
+}
+
 async function fetchStacGeoparquetDatetimeExtent({
   href,
   connection,