Replace instanceof arrow.* with isArrow* type guards

Signed-off-by: Ilya Boyandin <ilyabo@gmail.com>

Author: ilyabo

src/deckgl-arrow-layers/package.json

@@ -32,6 +32,7 @@
   "dependencies": {
     "@geoarrow/geoarrow-js": "^0.3.0",
     "@kepler.gl/constants": "^3.2.2",
+    "@kepler.gl/utils": "^3.2.2",
     "@math.gl/core": "^4.0.0",
     "@math.gl/polygon": "^4.0.0",
     "@math.gl/types": "^4.0.0",

src/deckgl-arrow-layers/src/utils/utils.ts

@@ -9,6 +9,7 @@ import {assert} from '@deck.gl/core/typed';
 import * as arrow from 'apache-arrow';
 import * as ga from '@geoarrow/geoarrow-js';
 import {AccessorContext, AccessorFunction, _InternalAccessorContext} from '../types';
+import {isArrowFixedSizeList, isArrowStruct, isArrowVector} from '@kepler.gl/utils';
 
 export type TypedArray =
   | Uint8Array
@@ -45,14 +46,14 @@ function isDataInterleavedCoords(
   data: arrow.Data
 ): data is arrow.Data<arrow.FixedSizeList<arrow.Float64>> {
   // TODO: also check 2 or 3d? Float64?
-  return data.type instanceof arrow.FixedSizeList;
+  return isArrowFixedSizeList(data.type);
 }
 
 function isDataSeparatedCoords(
   data: arrow.Data
 ): data is arrow.Data<arrow.Struct<{x: arrow.Float64; y: arrow.Float64}>> {
   // TODO: also check child names? Float64?
-  return data.type instanceof arrow.Struct;
+  return isArrowStruct(data.type);
 }
 
 /**
@@ -162,7 +163,7 @@ export function assignAccessor(args: AssignAccessorProps) {
     return;
   }
 
-  if (propInput instanceof arrow.Vector) {
+  if (isArrowVector(propInput)) {
     const columnData = propInput.data[chunkIdx];
 
     if (arrow.DataType.isFixedSizeList(columnData)) {

src/deckgl-arrow-layers/src/utils/validate.ts

@@ -6,6 +6,7 @@
 // Copyright (c) vis.gl contributors
 
 import {assert} from '@deck.gl/core/typed';
+import {isArrowVector} from '@kepler.gl/utils';
 import * as arrow from 'apache-arrow';
 
 export function validateAccessors(props: Record<string, any>, table: arrow.Table): void {
@@ -15,7 +16,7 @@ export function validateAccessors(props: Record<string, any>, table: arrow.Table
     // Is it an accessor
     if (accessorName.startsWith('get')) {
       // Is it a vector accessor
-      if (accessorValue instanceof arrow.Vector) {
+      if (isArrowVector(accessorValue)) {
         vectorAccessors.push(accessorValue);
 
         // Is it a color vector accessor

src/duckdb/package.json

@@ -31,6 +31,7 @@
     "@kepler.gl/processors": "3.2.2",
     "@kepler.gl/table": "3.2.2",
     "@kepler.gl/types": "3.2.2",
+    "@kepler.gl/utils": "3.2.2",
     "@monaco-editor/react": "^4.6.0",
     "@radix-ui/react-collapsible": "^1.1.0",
     "apache-arrow": ">=15.0.0",

src/duckdb/src/table/duckdb-table.ts

@@ -19,7 +19,13 @@ import {
 } from '@kepler.gl/processors';
 import {KeplerTable} from '@kepler.gl/table';
 import {Field} from '@kepler.gl/types';
-import {getApplicationConfig, DatabaseAdapter, DatabaseConnection} from '@kepler.gl/utils';
+import {
+  getApplicationConfig,
+  DatabaseAdapter,
+  DatabaseConnection,
+  isArrowTable,
+  isArrowVector
+} from '@kepler.gl/utils';
 
 import {
   processCsvRowObject,
@@ -161,10 +167,9 @@ export class KeplerGlDuckDbTable extends KeplerTable {
     try {
       // 1) data.rows contains an arrow table created by Add to Map data from DuckDb query.
       // 2) arrow table is in cols & fields when a file is dragged & dropped into Add Data To Map dialog.
-      const arrowTable =
-        data.rows instanceof arrow.Table
-          ? data.rows
-          : restoreArrowTable(data.cols || [], data.fields, data.arrowSchema);
+      const arrowTable = isArrowTable(data.rows)
+        ? data.rows
+        : restoreArrowTable(data.cols || [], data.fields, data.arrowSchema);
 
       // remove unsupported extensions from an arrow table that throw exceptions in DuckDB.
       adjustedMetadata = removeUnsupportedExtensions(arrowTable);
@@ -215,7 +220,7 @@ export class KeplerGlDuckDbTable extends KeplerTable {
         format = DATASET_FORMATS.row;
       } else if (data.rows?.type === 'FeatureCollection') {
         format = DATASET_FORMATS.geojson;
-      } else if (data.cols?.[0] instanceof arrow.Vector) {
+      } else if (isArrowVector(data.cols?.[0])) {
         format = DATASET_FORMATS.arrow;
       }
     }

src/layers/src/base-layer.ts

@@ -41,7 +41,8 @@ import {
   hexToRgb,
   isPlainObject,
   isDomainStops,
-  updateColorRangeByMatchingPalette
+  updateColorRangeByMatchingPalette,
+  isArrowTable
 } from '@kepler.gl/utils';
 import {generateHashId, toArray, notNullorUndefined} from '@kepler.gl/common-utils';
 import {Datasets, GpuFilter, KeplerTable} from '@kepler.gl/table';
@@ -1585,7 +1586,7 @@ class Layer implements KeplerLayer {
     const {data, mapState} = renderOpts;
     const {textLabel} = this.config;
 
-    const TextLayerClass = data.data instanceof arrow.Table ? GeoArrowTextLayer : TextLayer;
+    const TextLayerClass = isArrowTable(data.data) ? GeoArrowTextLayer : TextLayer;
 
     return data.textLabels.reduce((accu, d, i) => {
       if (d.getText) {

src/layers/src/layer-text-label.ts

@@ -3,7 +3,7 @@
 
 import * as arrow from 'apache-arrow';
 import {getDistanceScales} from 'viewport-mercator-project';
-import {DataContainerInterface, ArrowDataContainer} from '@kepler.gl/utils';
+import {DataContainerInterface, ArrowDataContainer, isArrowTable} from '@kepler.gl/utils';
 import {notNullorUndefined} from '@kepler.gl/common-utils';
 import uniq from 'lodash/uniq';
 
@@ -82,7 +82,7 @@ export const formatTextLabelData = ({
       getText = oldLayerData.textLabels[i].getText;
       rebuildArrowTextVector = false;
     } else {
-      if (data instanceof arrow.Table) {
+      if (isArrowTable(data)) {
         // we don't filter out arrow tables,
         // so we use filteredIndex array instead
         const allLabels: string[] = [];
@@ -108,7 +108,7 @@ export const formatTextLabelData = ({
     // For now check here for ArrowTable, not ArrowDataContainer.
     if (
       rebuildArrowTextVector &&
-      data instanceof arrow.Table &&
+      isArrowTable(data) &&
       dataContainer instanceof ArrowDataContainer
     ) {
       getText = dataContainer.getColumn(tl.field.fieldIdx);

src/utils/src/arrow-data-container.ts

@@ -40,6 +40,64 @@ export function isArrowTable(data: any): data is arrow.Table {
   );
 }
 
+/**
+ * Check if data is an ArrowVector object.
+ * Uses duck-typing instead of `instanceof` to handle DuckDB's bundled Arrow version.
+ *
+ * @param data - object to check
+ * @returns true if data is an ArrowVector object (type guarded)
+ */
+export function isArrowVector(data: any): data is arrow.Vector {
+  return (
+    typeof data === 'object' &&
+    data !== null &&
+    'type' in data &&
+    'length' in data &&
+    typeof data.length === 'number' &&
+    'get' in data &&
+    typeof data.get === 'function' &&
+    'data' in data &&
+    Array.isArray(data.data)
+  );
+}
+
+/**
+ * Check if data is an Arrow FixedSizeList DataType.
+ * Uses duck-typing instead of `instanceof` to handle DuckDB's bundled Arrow version.
+ *
+ * @param data - object to check
+ * @returns true if data is an Arrow FixedSizeList DataType (type guarded)
+ */
+export function isArrowFixedSizeList(data: any): data is arrow.FixedSizeList {
+  return (
+    typeof data === 'object' &&
+    data !== null &&
+    'typeId' in data &&
+    'listSize' in data &&
+    typeof data.listSize === 'number' &&
+    'children' in data &&
+    Array.isArray(data.children)
+  );
+}
+
+/**
+ * Check if data is an Arrow Struct DataType.
+ * Uses duck-typing instead of `instanceof` to handle DuckDB's bundled Arrow version.
+ *
+ * @param data - object to check
+ * @returns true if data is an Arrow Struct DataType (type guarded)
+ */
+export function isArrowStruct(data: any): data is arrow.Struct {
+  return (
+    typeof data === 'object' &&
+    data !== null &&
+    'typeId' in data &&
+    'children' in data &&
+    Array.isArray(data.children) &&
+    !('listSize' in data)
+  );
+}
+
 /**
  * @param dataContainer
  * @param sharedRow

src/utils/src/data-utils.ts

@@ -20,6 +20,7 @@ import {Field, Millisecond, ProtoDatasetField} from '@kepler.gl/types';
 
 import {snapToMarks} from './plot';
 import {isPlainObject} from './utils';
+import {isArrowVector} from './arrow-data-container';
 
 export type FieldFormatter = (value: any, field?: ProtoDatasetField) => string;
 
@@ -317,7 +318,7 @@ export const FIELD_DISPLAY_FORMAT: {
       ? `[${String(d)}]`
       : '',
   [ALL_FIELD_TYPES.geoarrow]: (data, field) => {
-    if (data instanceof arrow.Vector) {
+    if (isArrowVector(data)) {
       try {
         const encoding = field?.metadata?.get('ARROW:extension:name');
         if (encoding) {

src/utils/src/index.ts

@@ -128,7 +128,10 @@ export {
   ArrowDataContainer,
   arrowDataTypeToAnalyzerDataType,
   arrowDataTypeToFieldType,
-  isArrowTable
+  isArrowTable,
+  isArrowFixedSizeList,
+  isArrowStruct,
+  isArrowVector
 } from './arrow-data-container';
 export type {DataContainerInterface} from './data-container-interface';
 export {