feat(GeoArrow): getBinaryGeometriesFromArrow enhancement (#2785)

modules/arrow/package.json

@@ -49,6 +49,7 @@
     "@loaders.gl/gis": "4.0.3",
     "@loaders.gl/loader-utils": "4.0.3",
     "@loaders.gl/schema": "4.0.3",
+    "@math.gl/polygon": "4.0.0",
     "apache-arrow": "^13.0.0"
   },
   "gitHead": "c95a4ff72512668a93d9041ce8636bac09333fd5"

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

@@ -2,28 +2,49 @@
 // Copyright (c) vis.gl contributors
 
 import * as arrow from 'apache-arrow';
+import {earcut} from '@math.gl/polygon';
 import {BinaryFeatureCollection as BinaryFeatures} from '@loaders.gl/schema';
 import {GeoArrowEncoding} from '@loaders.gl/gis';
 import {updateBoundsFromGeoArrowSamples} from './get-arrow-bounds';
+import {TypedArray} from '@loaders.gl/loader-utils';
 
 /**
  * Binary data from geoarrow column and can be used by e.g. deck.gl GeojsonLayer
  */
 export type BinaryDataFromGeoArrow = {
+  /** Binary format geometries, an array of BinaryFeatureCollection */
   binaryGeometries: BinaryFeatures[];
+  /** Boundary of the binary geometries */
   bounds: [number, number, number, number];
+  /** Feature types of the binary geometries */
   featureTypes: {polygon: boolean; point: boolean; line: boolean};
+  /** (Optional) mean centers of the binary geometries for e.g. polygon filtering */
+  meanCenters?: number[][];
 };
 
+/**
+ * Binary geometry content returned from getBinaryGeometriesFromChunk
+ */
 type BinaryGeometryContent = {
+  // Array of Point feature indexes by vertex
   featureIds: Uint32Array;
+  /** Flat coordinate array of e.g. x, y or x,y,z */
   flatCoordinateArray: Float64Array;
+  /** Dimention of each position */
   nDim: number;
+  /** Array of geometry offsets: the start index of primitive geometry */
   geomOffset: Int32Array;
+  /** Array of geometry indicies: the start index of each geometry */
   geometryIndicies: Uint16Array;
+  /** (Optional) indices of triangels returned from polygon tessellation (Polygon only) */
+  triangles?: Uint32Array;
+  /** (Optional) array of mean center of each geometry */
+  meanCenters?: Float64Array;
 };
 
-// binary geometry template, see deck.gl BinaryGeometry
+/**
+ * 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},
@@ -32,33 +53,40 @@ export const BINARY_GEOMETRY_TEMPLATE = {
   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;
+  /** option to get mean centers from geometries, for polygon filtering */
+  meanCenter?: boolean;
+};
+
 /**
  * get binary geometries from geoarrow column
  *
  * @param geoColumn the geoarrow column, e.g. arrowTable.getChildAt(geoColumnIndex)
  * @param geoEncoding the geo encoding of the geoarrow column, e.g. getGeoArrowEncoding(arrowTable.schema, geoColumnName)
+ * @param options options for getting binary geometries {meanCenter: boolean}
  * @returns BinaryDataFromGeoArrow
  */
 export function getBinaryGeometriesFromArrow(
   geoColumn: arrow.Vector,
-  geoEncoding: GeoArrowEncoding
+  geoEncoding: GeoArrowEncoding,
+  options?: BinaryGeometriesFromArrowOptions
 ): BinaryDataFromGeoArrow {
   const featureTypes = {
     polygon: geoEncoding === 'geoarrow.multipolygon' || geoEncoding === 'geoarrow.polygon',
     point: geoEncoding === 'geoarrow.multipoint' || geoEncoding === 'geoarrow.point',
     line: geoEncoding === 'geoarrow.multilinestring' || geoEncoding === 'geoarrow.linestring'
   };
 
-  const chunks = geoColumn.data;
+  const chunks = options?.chunkIndex ? [geoColumn.data[options?.chunkIndex]] : geoColumn.data;
   let bounds: [number, number, number, number] = [Infinity, Infinity, -Infinity, -Infinity];
   let globalFeatureIdOffset = 0;
   const binaryGeometries: BinaryFeatures[] = [];
 
   chunks.forEach((chunk) => {
-    const {featureIds, flatCoordinateArray, nDim, geomOffset} = getBinaryGeometriesFromChunk(
-      chunk,
-      geoEncoding
-    );
+    const {featureIds, flatCoordinateArray, nDim, geomOffset, triangles} =
+      getBinaryGeometriesFromChunk(chunk, geoEncoding);
 
     const globalFeatureIds = new Uint32Array(featureIds.length);
     for (let i = 0; i < featureIds.length; i++) {
@@ -79,7 +107,6 @@ export function getBinaryGeometriesFromArrow(
 
     // TODO: check if chunks are sequentially accessed
     globalFeatureIdOffset += chunk.length;
-
     // NOTE: deck.gl defines the BinaryFeatures structure must have points, lines, polygons even if they are empty
     binaryGeometries.push({
       shape: 'binary-feature-collection',
@@ -99,22 +126,111 @@ export function getBinaryGeometriesFromArrow(
         ...BINARY_GEOMETRY_TEMPLATE,
         ...(featureTypes.polygon ? binaryContent : {}),
         polygonIndices: {
-          // NOTE: polygonIndices and primitivePolygonIndices are not used together to render polygon (binary) with holes
-          // for GeoJsonLayer with binary geometries in deck.gl currently, so we pass geomOffset and triangles.
+          // use geomOffset as polygonIndices same as primitivePolygonIndices since we are using earcut to get triangule indices
           value: featureTypes.polygon ? geomOffset : new Uint16Array(0),
           size: 1
         },
         primitivePolygonIndices: {
           value: featureTypes.polygon ? geomOffset : new Uint16Array(0),
           size: 1
-        }
+        },
+        ...(triangles ? {triangles: {value: triangles, size: 1}} : {})
       }
     });
 
     bounds = updateBoundsFromGeoArrowSamples(flatCoordinateArray, nDim, bounds);
   });
 
-  return {binaryGeometries, bounds, featureTypes};
+  return {
+    binaryGeometries,
+    bounds,
+    featureTypes,
+    ...(options?.meanCenter
+      ? {meanCenters: getMeanCentersFromBinaryGeometries(binaryGeometries)}
+      : {})
+  };
+}
+
+/**
+ * Get mean centers from binary geometries
+ * @param binaryGeometries binary geometries from geoarrow column, an array of BinaryFeatureCollection
+ * @returns mean centers of the binary geometries
+ */
+export function getMeanCentersFromBinaryGeometries(binaryGeometries: BinaryFeatures[]): number[][] {
+  const globalMeanCenters: number[][] = [];
+  binaryGeometries.forEach((binaryGeometry: BinaryFeatures) => {
+    let binaryGeometryType: string | null = null;
+    if (binaryGeometry.points && binaryGeometry.points.positions.value.length > 0) {
+      binaryGeometryType = 'points';
+    } else if (binaryGeometry.lines && binaryGeometry.lines.positions.value.length > 0) {
+      binaryGeometryType = 'lines';
+    } else if (binaryGeometry.polygons && binaryGeometry.polygons.positions.value.length > 0) {
+      binaryGeometryType = 'polygons';
+    }
+
+    const binaryContent = binaryGeometryType ? binaryGeometry[binaryGeometryType] : null;
+    if (binaryContent && binaryGeometryType !== null) {
+      const featureIds = binaryContent.featureIds.value;
+      const flatCoordinateArray = binaryContent.positions.value;
+      const nDim = binaryContent.positions.size;
+      const primitivePolygonIndices = binaryContent.primitivePolygonIndices?.value;
+
+      const meanCenters = getMeanCentersFromGeometry(
+        featureIds,
+        flatCoordinateArray,
+        nDim,
+        binaryGeometryType,
+        primitivePolygonIndices
+      );
+      meanCenters.forEach((center) => {
+        globalMeanCenters.push(center);
+      });
+    }
+  });
+  return globalMeanCenters;
+}
+
+/**
+ * Get mean centers from raw coordinates and feature ids
+ * @param featureIds Array of feature ids indexes by vertex
+ * @param flatCoordinateArray  Array of vertex, e.g. x, y or x, y, z, positions
+ * @param nDim number of dimensions per position
+ * @returns - mean centers of each polygon
+ */
+function getMeanCentersFromGeometry(
+  featureIds: TypedArray,
+  flatCoordinateArray: TypedArray,
+  nDim: number,
+  geometryType: string,
+  primitivePolygonIndices?: TypedArray
+) {
+  const meanCenters: number[][] = [];
+  const vertexCount = flatCoordinateArray.length;
+  let vertexIndex = 0;
+  while (vertexIndex < vertexCount) {
+    const featureId = featureIds[vertexIndex / nDim];
+    const center = [0, 0];
+    let vertexCountInFeature = 0;
+    while (vertexIndex < vertexCount && featureIds[vertexIndex / nDim] === featureId) {
+      if (
+        geometryType === 'polygons' &&
+        primitivePolygonIndices &&
+        primitivePolygonIndices.indexOf(vertexIndex / nDim) >= 0
+      ) {
+        // skip the first point since it is the same as the last point in each ring for polygons
+        vertexIndex += nDim;
+      } else {
+        center[0] += flatCoordinateArray[vertexIndex];
+        center[1] += flatCoordinateArray[vertexIndex + 1];
+        vertexIndex += nDim;
+        vertexCountInFeature++;
+      }
+    }
+    center[0] /= vertexCountInFeature;
+    center[1] /= vertexCountInFeature;
+    meanCenters.push(center);
+  }
+  return meanCenters;
 }
 
 /**
@@ -142,6 +258,53 @@ function getBinaryGeometriesFromChunk(
   }
 }
 
+/**
+ * get triangle indices. Allows deck.gl to skip performing costly triangulation on main thread.
+ * @param polygonIndices Indices within positions of the start of each simple Polygon
+ * @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
+ */
+export function getTriangleIndices(
+  polygonIndices: Uint16Array,
+  primitivePolygonIndices: Int32Array,
+  flatCoordinateArray: Float64Array,
+  nDim: number
+): Uint32Array {
+  let primitiveIndex = 0;
+  const triangles: number[] = [];
+  // loop polygonIndices to get triangles
+  for (let i = 0; i < polygonIndices.length - 1; i++) {
+    const startIdx = polygonIndices[i];
+    const endIdx = polygonIndices[i + 1];
+    // get subarray of flatCoordinateArray
+    const slicedFlatCoords = flatCoordinateArray.subarray(startIdx * nDim, endIdx * nDim);
+    // get holeIndices for earcut
+    const holeIndices: number[] = [];
+    while (primitivePolygonIndices[primitiveIndex] < endIdx) {
+      if (primitivePolygonIndices[primitiveIndex] > startIdx) {
+        holeIndices.push(primitivePolygonIndices[primitiveIndex] - startIdx);
+      }
+      primitiveIndex++;
+    }
+    const triangleIndices = earcut(
+      slicedFlatCoords,
+      holeIndices.length > 0 ? holeIndices : undefined,
+      nDim
+    );
+    for (let j = 0; j < triangleIndices.length; j++) {
+      triangles.push(triangleIndices[j] + startIdx);
+    }
+  }
+  // convert traingles to Uint32Array
+  const trianglesUint32 = new Uint32Array(triangles.length);
+  for (let i = 0; i < triangles.length; i++) {
+    trianglesUint32[i] = triangles[i];
+  }
+  return trianglesUint32;
+}
+
 /**
  * get binary polygons from geoarrow polygon column
  * @param chunk one chunk of geoarrow polygon column
@@ -178,12 +341,14 @@ function getBinaryPolygonsFromChunk(chunk: arrow.Data, geoEncoding: string): Bin
     }
   }
 
+  const triangles = getTriangleIndices(geometryIndicies, geomOffset, flatCoordinateArray, nDim);
   return {
     featureIds,
     flatCoordinateArray,
     nDim,
     geomOffset,
-    geometryIndicies
+    geometryIndicies,
+    triangles
   };
 }
 
@@ -209,11 +374,23 @@ function getBinaryLinesFromChunk(chunk: arrow.Data, geoEncoding: string): Binary
 
   const numOfVertices = flatCoordinateArray.length / nDim;
   const featureIds = new Uint32Array(numOfVertices);
-  for (let i = 0; i < chunk.length; i++) {
-    const startIdx = geomOffset[i];
-    const endIdx = geomOffset[i + 1];
-    for (let j = startIdx; j < endIdx; j++) {
-      featureIds[j] = i;
+
+  if (isMultiLineString) {
+    const partData = chunk.valueOffsets;
+    for (let i = 0; i < partData.length - 1; i++) {
+      const startIdx = geomOffset[partData[i]];
+      const endIdx = geomOffset[partData[i + 1]];
+      for (let j = startIdx; j < endIdx; j++) {
+        featureIds[j] = i;
+      }
+    }
+  } else {
+    for (let i = 0; i < chunk.length; i++) {
+      const startIdx = geomOffset[i];
+      const endIdx = geomOffset[i + 1];
+      for (let j = startIdx; j < endIdx; j++) {
+        featureIds[j] = i;
+      }
     }
   }
 
@@ -248,8 +425,20 @@ function getBinaryPointsFromChunk(chunk: arrow.Data, geoEncoding: string): Binar
 
   const numOfVertices = flatCoordinateArray.length / nDim;
   const featureIds = new Uint32Array(numOfVertices);
-  for (let i = 0; i < chunk.length; i++) {
-    featureIds[i] = i;
+
+  if (isMultiPoint) {
+    const partData = chunk.valueOffsets;
+    for (let i = 0; i < partData.length - 1; i++) {
+      const startIdx = partData[i];
+      const endIdx = partData[i + 1];
+      for (let j = startIdx; j < endIdx; j++) {
+        featureIds[j] = i;
+      }
+    }
+  } else {
+    for (let i = 0; i < chunk.length; i++) {
+      featureIds[i] = i;
+    }
   }
 
   return {

modules/arrow/src/index.ts

@@ -64,10 +64,15 @@ export type {GeoArrowEncoding} from '@loaders.gl/gis';
 // getGeometryColumnsFromArrowTable,
 // getGeoArrowEncoding
 
-export type {BinaryDataFromGeoArrow} from './geoarrow/convert-geoarrow-to-binary-geometry';
+export type {
+  BinaryDataFromGeoArrow,
+  BinaryGeometriesFromArrowOptions
+} from './geoarrow/convert-geoarrow-to-binary-geometry';
 export {
   BINARY_GEOMETRY_TEMPLATE,
-  getBinaryGeometriesFromArrow
+  getBinaryGeometriesFromArrow,
+  getTriangleIndices,
+  getMeanCentersFromBinaryGeometries
 } from './geoarrow/convert-geoarrow-to-binary-geometry';
 
 export {parseGeometryFromArrow} from './geoarrow/convert-geoarrow-to-geojson';