feat(arrow): Triangulate on worker

modules/arrow/src/geoarrow/convert-geoarrow-to-binary-geometry.ts

@@ -45,7 +45,7 @@ type BinaryGeometryContent = {
   geomOffset: Int32Array;
   /** Array of geometry indicies: the start index of each geometry */
   geometryIndicies: Uint16Array;
-  /** (Optional) indices of triangels returned from polygon tessellation (Polygon only) */
+  /** (Optional) indices of triangels returned from polygon triangulation (Polygon only) */
   triangles?: Uint32Array;
   /** (Optional) array of mean center of each geometry */
   meanCenters?: Float64Array;
@@ -54,17 +54,21 @@ type BinaryGeometryContent = {
 /**
  * binary geometry template, see deck.gl BinaryGeometry
  */
-export const BINARY_GEOMETRY_TEMPLATE = {
-  globalFeatureIds: {value: new Uint32Array(0), size: 1},
-  positions: {value: new Float32Array(0), size: 2},
-  properties: [],
-  numericProps: {},
-  featureIds: {value: new Uint32Array(0), size: 1}
-};
+export function getBinaryGeometryTemplate() {
+  return {
+    globalFeatureIds: {value: new Uint32Array(0), size: 1},
+    positions: {value: new Float32Array(0), size: 2},
+    properties: [],
+    numericProps: {},
+    featureIds: {value: new Uint32Array(0), size: 1}
+  };
+}
 
 export type BinaryGeometriesFromArrowOptions = {
   /** option to specify which chunk to get binary geometries from, for progressive rendering */
   chunkIndex?: number;
+  /** The offset (beginning index of rows) of input chunk. Used for reconstructing globalFeatureIds in web workers */
+  chunkOffset?: number;
   /** option to get mean centers from geometries, for polygon filtering */
   calculateMeanCenters?: boolean;
   /** option to compute the triangle indices by tesselating polygons */
@@ -90,9 +94,12 @@ export function getBinaryGeometriesFromArrow(
     line: geoEncoding === 'geoarrow.multilinestring' || geoEncoding === 'geoarrow.linestring'
   };
 
-  const chunks = options?.chunkIndex ? [geoColumn.data[options?.chunkIndex]] : geoColumn.data;
+  const chunks =
+    options?.chunkIndex !== undefined && options?.chunkIndex >= 0
+      ? [geoColumn.data[options?.chunkIndex]]
+      : geoColumn.data;
   let bounds: [number, number, number, number] = [Infinity, Infinity, -Infinity, -Infinity];
-  let globalFeatureIdOffset = 0;
+  let globalFeatureIdOffset = options?.chunkOffset || 0;
   const binaryGeometries: BinaryFeatures[] = [];
 
   chunks.forEach((chunk) => {
@@ -111,6 +118,7 @@ export function getBinaryGeometriesFromArrow(
         size: nDim
       },
       featureIds: {value: featureIds, size: 1},
+      // eslint-disable-next-line no-loop-func
       properties: [...Array(chunk.length).keys()].map((i) => ({
         index: i + globalFeatureIdOffset
       }))
@@ -123,18 +131,18 @@ export function getBinaryGeometriesFromArrow(
       shape: 'binary-feature-collection',
       points: {
         type: 'Point',
-        ...BINARY_GEOMETRY_TEMPLATE,
+        ...getBinaryGeometryTemplate(),
         ...(featureTypes.point ? binaryContent : {})
       },
       lines: {
         type: 'LineString',
-        ...BINARY_GEOMETRY_TEMPLATE,
+        ...getBinaryGeometryTemplate(),
         ...(featureTypes.line ? binaryContent : {}),
         pathIndices: {value: featureTypes.line ? geomOffset : new Uint16Array(0), size: 1}
       },
       polygons: {
         type: 'Polygon',
-        ...BINARY_GEOMETRY_TEMPLATE,
+        ...getBinaryGeometryTemplate(),
         ...(featureTypes.polygon ? binaryContent : {}),
         polygonIndices: {
           // use geomOffset as polygonIndices same as primitivePolygonIndices since we are using earcut to get triangule indices
@@ -281,7 +289,7 @@ function getBinaryGeometriesFromChunk(
  * @param primitivePolygonIndices Indices within positions of the start of each primitive Polygon/ring
  * @param flatCoordinateArray Array of x, y or x, y, z positions
  * @param nDim - number of dimensions per position
- * @returns
+ * @returns triangle indices or null if invalid polygon and earcut fails
  */
 export function getTriangleIndices(
   polygonIndices: Uint16Array,
@@ -306,13 +314,14 @@ export function getTriangleIndices(
         }
         primitiveIndex++;
       }
+      // TODO check if each ring is closed
       const triangleIndices = earcut(
         slicedFlatCoords,
         holeIndices.length > 0 ? holeIndices : undefined,
         nDim
       );
       if (triangleIndices.length === 0) {
-        throw Error('can not tesselate invalid polygon');
+        throw Error('earcut failed e.g. invalid polygon');
       }
       for (let j = 0; j < triangleIndices.length; j++) {
         triangles.push(triangleIndices[j] + startIdx);
@@ -325,9 +334,7 @@ export function getTriangleIndices(
     }
     return trianglesUint32;
   } catch (error) {
-    // TODO - add logging
-    // there is an expection when tesselating invalid polygon, e.g. polygon with self-intersection
-    // return null to skip tesselating
+    // if earcut fails, return null
     return null;
   }
 }
@@ -379,8 +386,8 @@ function getBinaryPolygonsFromChunk(
 
   return {
     featureIds,
-    flatCoordinateArray,
     nDim,
+    flatCoordinateArray,
     geomOffset,
     geometryIndicies,
     ...(options?.triangulate && triangles ? {triangles} : {})

modules/arrow/src/index.ts

@@ -52,7 +52,7 @@ export type {
   BinaryGeometriesFromArrowOptions
 } from './geoarrow/convert-geoarrow-to-binary-geometry';
 export {
-  BINARY_GEOMETRY_TEMPLATE,
+  getBinaryGeometryTemplate,
   getBinaryGeometriesFromArrow,
   getTriangleIndices,
   getMeanCentersFromBinaryGeometries
@@ -65,5 +65,11 @@ export {parseGeometryFromArrow} from './geoarrow/convert-geoarrow-to-geojson-geo
 export {convertArrowToGeoJSONTable} from './tables/convert-arrow-to-geojson-table';
 
 // EXPERIMENTAL WORKER
+export {hardClone} from './workers/hard-clone';
 
-export {TriangulationWorker, triangulateOnWorker} from './triangulate-on-worker';
+export type {ParseGeoArrowInput, ParseGeoArrowResult} from './triangulate-on-worker';
+export {
+  TriangulationWorker,
+  triangulateOnWorker,
+  parseGeoArrowOnWorker
+} from './triangulate-on-worker';

modules/arrow/src/triangulate-on-worker.ts

@@ -1,19 +1,52 @@
 // loaders.gl, MIT license
 // Copyright (c) vis.gl contributors
 
+import * as arrow from 'apache-arrow';
 import type {WorkerOptions} from '@loaders.gl/worker-utils';
 import {processOnWorker} from '@loaders.gl/worker-utils';
+import {BinaryDataFromGeoArrow, GeoArrowEncoding} from '@loaders.gl/arrow';
 
 // __VERSION__ is injected by babel-plugin-version-inline
 // @ts-ignore TS2304: Cannot find name '__VERSION__'.
 const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
 
-export type TriangulationWorkerInput = TriangulateInput | {operation: 'test'; data: any};
-export type TriangulationWorkerOutput = TriangulateResult | {operation: 'test'; data: any};
+export type TriangulationWorkerInput =
+  | ({operation: 'triangulate'} & TriangulateInput)
+  | ParseGeoArrowInput
+  | {operation: 'test'; data: any};
+
+export type TriangulationWorkerOutput =
+  | ({operation: 'triangulate'} & TriangulateResult)
+  | ({operation: 'parse-geoarrow'} & ParseGeoArrowResult)
+  | {operation: 'test'; data: any};
+
+type GeoArrowChunkData = {
+  type: arrow.DataType;
+  offset: number;
+  length: number;
+  nullCount: number;
+  buffers: any;
+  children: arrow.Data[];
+  dictionary?: arrow.Vector;
+};
+
+export type ParseGeoArrowInput = {
+  operation: 'parse-geoarrow';
+  chunkData: GeoArrowChunkData;
+  chunkIndex: number;
+  chunkOffset: number;
+  geometryEncoding: GeoArrowEncoding;
+  calculateMeanCenters: boolean;
+  triangle: boolean;
+};
+
+export type ParseGeoArrowResult = {
+  chunkIndex: number;
+  binaryDataFromGeoArrow: BinaryDataFromGeoArrow | null;
+};
 
 /** Input data for operation: 'triangulate' */
 export type TriangulateInput = {
-  operation: 'triangulate';
   polygonIndices: Uint16Array;
   primitivePolygonIndices: Int32Array;
   flatCoordinateArray: Float64Array;
@@ -37,11 +70,21 @@ export const TriangulationWorker = {
 };
 
 /**
- * Provide type safety
+ * Triangulate a set of polygons on worker, type safe API
  */
 export function triangulateOnWorker(
-  data: TriangulationWorkerInput,
+  data: TriangulateInput,
+  options: WorkerOptions = {}
+): Promise<TriangulateResult> {
+  return processOnWorker(TriangulationWorker, {...data, operation: 'triangulate'}, options);
+}
+
+/**
+ * Parse GeoArrow geometry colum on worker, type safe API
+ */
+export function parseGeoArrowOnWorker(
+  data: ParseGeoArrowInput,
   options: WorkerOptions = {}
-): Promise<TriangulationWorkerOutput> {
-  return processOnWorker(TriangulationWorker, data, options);
+): Promise<ParseGeoArrowResult> {
+  return processOnWorker(TriangulationWorker, {...data, operation: 'parse-geoarrow'}, options);
 }

modules/arrow/src/workers/hard-clone.ts

@@ -0,0 +1,162 @@
+import * as arrow from 'apache-arrow';
+import type {Buffers} from 'apache-arrow/data';
+
+type TypedArray =
+  | Uint8Array
+  | Uint8ClampedArray
+  | Uint16Array
+  | Uint32Array
+  | Int8Array
+  | Int16Array
+  | Int32Array
+  | Float32Array
+  | Float64Array;
+
+/**
+ * Clone an Arrow JS Data or Vector, detaching from an existing ArrayBuffer if
+ * it is shared with other.
+ *
+ * The purpose of this function is to enable transferring a `Data` instance,
+ * e.g. to a web worker, without neutering any other data.
+ *
+ * Any internal buffers that are a slice of a larger `ArrayBuffer` (i.e. where
+ * the typed array's `byteOffset` is not `0` and where its `byteLength` does not
+ * match its `array.buffer.byteLength`) are copied into new `ArrayBuffers`.
+ *
+ * If `force` is `true`, always clone internal buffers, even if not shared. If
+ * the default, `false`, any internal buffers that are **not** a slice of a
+ * larger `ArrayBuffer` will not be copied.
+ */
+export function hardClone<T extends arrow.DataType>(
+  input: arrow.Data<T>,
+  force?: boolean
+): arrow.Data<T>;
+export function hardClone<T extends arrow.DataType>(
+  input: arrow.Vector<T>,
+  force?: boolean
+): arrow.Vector<T>;
+
+export function hardClone<T extends arrow.DataType>(
+  data: arrow.Data<T> | arrow.Vector<T>,
+  force: boolean = false
+): arrow.Data<T> | arrow.Vector<T> {
+  // Check if `data` is an arrow.Vector
+  if ('data' in data) {
+    return new arrow.Vector(data.data.map((data) => hardClone(data, force)));
+  }
+
+  // Clone each of the children, recursively
+  const clonedChildren: arrow.Data[] = [];
+  for (const childData of data.children) {
+    clonedChildren.push(hardClone(childData, force));
+  }
+
+  // Clone the dictionary if there is one
+  let clonedDictionary: arrow.Vector | undefined;
+  if (data.dictionary !== undefined) {
+    clonedDictionary = hardClone(data.dictionary, force);
+  }
+
+  // Buffers can have up to four entries. Each of these can be `undefined` for
+  // one or more array types.
+  //
+  // - OFFSET: value offsets for variable size list types
+  // - DATA: the underlying data
+  // - VALIDITY: the null buffer. This may be empty or undefined if all elements
+  //   are non-null/valid.
+  // - TYPE: type ids for a union type.
+  const clonedBuffers: Buffers<T> = {
+    [arrow.BufferType.OFFSET]: cloneBuffer(data.buffers[arrow.BufferType.OFFSET], force),
+    [arrow.BufferType.DATA]: cloneBuffer(data.buffers[arrow.BufferType.DATA], force),
+    [arrow.BufferType.VALIDITY]: cloneBuffer(data.buffers[arrow.BufferType.VALIDITY], force),
+    [arrow.BufferType.TYPE]: cloneBuffer(data.buffers[arrow.BufferType.TYPE], force)
+  };
+
+  // Note: the data.offset is passed on so that a sliced Data instance will not
+  // be "un-sliced". However keep in mind that this means we're cloning the
+  // _original backing buffer_, not only the portion of the Data that was
+  // sliced.
+  return new arrow.Data(
+    data.type,
+    data.offset,
+    data.length,
+    // @ts-expect-error _nullCount is protected. We're using it here to mimic
+    // `Data.clone`
+    data._nullCount,
+    clonedBuffers,
+    clonedChildren,
+    clonedDictionary
+  );
+}
+
+/**
+ * Test whether an arrow.Data instance is a slice of a larger `ArrayBuffer`.
+ */
+export function isShared<T extends arrow.DataType>(data: arrow.Data<T> | arrow.Vector<T>): boolean {
+  // Loop over arrow.Vector
+  if ('data' in data) {
+    return data.data.some((data) => isShared(data));
+  }
+
+  // Check child data
+  for (const childData of data.children) {
+    if (isShared(childData)) {
+      return true;
+    }
+  }
+
+  // Check dictionary
+  if (data.dictionary !== undefined) {
+    if (isShared(data.dictionary)) {
+      return true;
+    }
+  }
+
+  const bufferTypes = [
+    arrow.BufferType.OFFSET,
+    arrow.BufferType.DATA,
+    arrow.BufferType.VALIDITY,
+    arrow.BufferType.TYPE
+  ];
+  for (const bufferType of bufferTypes) {
+    if (data.buffers[bufferType] !== undefined && isTypedArraySliced(data.buffers[bufferType])) {
+      return true;
+    }
+  }
+
+  return false;
+}
+
+/**
+ * Returns true if the current typed array is a partial slice on a larger
+ * ArrayBuffer
+ */
+function isTypedArraySliced(arr: TypedArray): boolean {
+  return !(arr.byteOffset === 0 && arr.byteLength === arr.buffer.byteLength);
+}
+
+/**
+ * If a slice of a larger ArrayBuffer, clone to a fresh `ArrayBuffer`.
+ *
+ * If `force` is `true`, always clone the array, even if not shared.
+ */
+function cloneBuffer<A extends TypedArray | undefined>(arr: A, force: boolean): A {
+  // Not all buffer types are defined for every type of Arrow array. E.g.
+  // `arrow.BufferType.TYPE` is only defined for the Union type.
+  if (arr === undefined) {
+    return arr;
+  }
+
+  // The current array is not a part of a larger ArrayBuffer, don't clone it
+  if (!force && !isTypedArraySliced(arr)) {
+    return arr;
+  }
+
+  // Note: TypedArray.slice() **copies** into a new ArrayBuffer
+
+  // @ts-expect-error 'Uint8Array' is assignable to the constraint of type 'A',
+  // but 'A' could be instantiated with a different subtype of constraint
+  // 'TypedArray'
+  // We know from arr.slice that it will always return the same
+  return arr.slice();
+}