Bump to apache-arrow 15 and simplify worker utils (#22)

* Bump to apache-arrow 15 and simplify worker utils

* update readme

README.md

@@ -29,7 +29,7 @@ Refer to the [`algorithm` namespace](https://geoarrow.github.io/geoarrow-js/modu
 
 ## Web Worker utilities
 
-Refer to the [`worker` namespace](https://geoarrow.github.io/geoarrow-js/modules/worker.html). Note that due to limitations in Arrow JS (as of v14) you **must** use `preparePostMessage` before a call to `structuredClone` or `postMessage`, to ensure it can correctly be rehydrated on the worker.
+Refer to the [`worker` namespace](https://geoarrow.github.io/geoarrow-js/modules/worker.html). Use `preparePostMessage` to obtain references to all underlying `ArrayBuffer` objects, so they can be transfered instead of copied.
 
 ```ts
 import * as arrow from "apache-arrow";

package.json

@@ -44,14 +44,14 @@
     "src/"
   ],
   "peerDependencies": {
-    "apache-arrow": ">=14"
+    "apache-arrow": ">=15"
   },
   "devDependencies": {
     "@rollup/plugin-terser": "^0.4.3",
     "@rollup/plugin-typescript": "^11.1.2",
     "@types/node": "^20.9.3",
     "@types/proj4": "^2",
-    "apache-arrow": "^14",
+    "apache-arrow": "^15",
     "esbuild": "^0.19.8",
     "gh-pages": "^6.1.0",
     "prettier": "^3.1.0",

src/worker/rehydrate.ts

@@ -24,20 +24,11 @@ import { Data } from "apache-arrow/data";
 import { Vector } from "apache-arrow/vector";
 import { Field } from "apache-arrow/schema";
 import type { Buffers } from "apache-arrow/data";
-import {
-  LineString,
-  Point,
-  Polygon,
-  MultiLineString,
-  MultiPoint,
-  MultiPolygon,
-} from "../type";
+import { Polygon, isPolygon } from "../type";
 import { PolygonData } from "../data";
 
 // Typedefs that include only the information kept from a structuredClone
-type PostMessageDataType = Pick<DataType, "children"> & {
-  __type: Type;
-};
+type PostMessageDataType = Pick<DataType, "children" | "typeId">;
 type PostMessageField = Pick<Field, "name" | "nullable" | "metadata"> & {
   type: PostMessageDataType;
 };
@@ -62,16 +53,7 @@ type PostMessageVector<T extends DataType> = Pick<
 > & { type: PostMessageDataType };
 
 function rehydrateType<T extends Type>(type: PostMessageDataType): DataType<T> {
-  // Note: by default in Arrow JS, the `DataType` is a class with no identifying
-  // attribute. Since a `structuredClone` is unable to maintain class
-  // information, the result of `structuredClone(new arrow.Utf8())` is an empty
-  // object `{}`.
-  //
-  // To get around this, in `preparePostMessage`, we manually assign the
-  // `typeId` (usually a getter) onto `__type`. Then when rehydrating the type,
-  // we can match on the `__type`, checking `arrow.Type` values, and
-  // reconstitute a full `arrow.DataType` object.
-  switch (type.__type) {
+  switch (type.typeId) {
     case Type.Null:
       return new Null() as DataType<T>;
     case Type.Int:
@@ -158,7 +140,6 @@ function rehydrateField(field: PostMessageField): Field {
 export function rehydrateData<T extends DataType>(
   data: PostMessageData<T>,
 ): Data<T> {
-  // @ts-expect-error
   const children = data.children.map((childData) => rehydrateData(childData));
   const dictionary = data.dictionary
     ? rehydrateVector(data.dictionary)
@@ -209,124 +190,3 @@ export function rehydratePolygonData(
   // on the JS side.
   return data;
 }
-
-// NOTE: these functions are copied from `type.ts` to work on __type
-
-/** Check that the given type is a Point data type */
-function isPoint(type: DataType): type is Point {
-  // @ts-expect-error
-  if (type.__type === Type.FixedSizeList) {
-    // Check list size
-    // @ts-expect-error
-    if (![2, 3, 4].includes(type.listSize)) {
-      return false;
-    }
-
-    // Check child of FixedSizeList is floating type
-    // @ts-expect-error
-    if (type.children[0].__type !== Type.Float) {
-      return false;
-    }
-
-    return true;
-  }
-
-  // @ts-expect-error
-  if (type.__type === Type.Struct) {
-    // Check number of children
-    if (![2, 3, 4].includes(type.children.length)) {
-      return false;
-    }
-
-    // Check that children have correct field names
-    if (
-      !type.children.every((field) => ["x", "y", "z", "m"].includes(field.name))
-    ) {
-      return false;
-    }
-
-    // @ts-expect-error
-    if (!type.children.every((field) => field.__type === Type.Float)) {
-      return false;
-    }
-
-    return true;
-  }
-
-  return false;
-}
-
-function isLineString(type: DataType): type is LineString {
-  // Check the outer type is a List
-  // @ts-expect-error
-  if (type.__type !== Type.List) {
-    return false;
-  }
-
-  // Check the child is a point type
-  if (!isPoint(type.children[0].type)) {
-    return false;
-  }
-
-  return true;
-}
-
-function isPolygon(type: DataType): type is Polygon {
-  // Check the outer vector is a List
-  // @ts-expect-error
-  if (type.__type !== Type.List) {
-    return false;
-  }
-
-  // Check the child is a linestring vector
-  if (!isLineString(type.children[0].type)) {
-    return false;
-  }
-
-  return true;
-}
-
-function isMultiPoint(type: DataType): type is MultiPoint {
-  // Check the outer vector is a List
-  // @ts-expect-error
-  if (type.__type !== Type.List) {
-    return false;
-  }
-
-  // Check the child is a point vector
-  if (!isPoint(type.children[0].type)) {
-    return false;
-  }
-
-  return true;
-}
-
-function isMultiLineString(type: DataType): type is MultiLineString {
-  // Check the outer vector is a List
-  // @ts-expect-error
-  if (type.__type !== Type.List) {
-    return false;
-  }
-
-  // Check the child is a linestring vector
-  if (!isLineString(type.children[0].type)) {
-    return false;
-  }
-
-  return true;
-}
-
-function isMultiPolygon(type: DataType): type is MultiPolygon {
-  // Check the outer vector is a List
-  // @ts-expect-error
-  if (type.__type !== Type.List) {
-    return false;
-  }
-
-  // Check the child is a polygon vector
-  if (!isPolygon(type.children[0].type)) {
-    return false;
-  }
-
-  return true;
-}

src/worker/transferable.ts

@@ -1,5 +1,5 @@
 import { DataType } from "apache-arrow/type";
-import { BufferType, Type } from "apache-arrow/enum";
+import { BufferType } from "apache-arrow/enum";
 import { Data } from "apache-arrow/data";
 import { Vector } from "apache-arrow/vector";
 import { hardClone } from "./hard-clone";
@@ -30,7 +30,6 @@ export function preparePostMessage<T extends DataType>(
       transferArrayBuffers.push(...arrayBuffers);
     }
     const vector = new Vector(postMessageDatas);
-    assignTypeIdOnType(vector.type);
     return [vector, transferArrayBuffers];
   }
 
@@ -74,18 +73,5 @@ export function preparePostMessage<T extends DataType>(
     transferArrayBuffers.push(input.buffers[BufferType.TYPE].buffer);
   }
 
-  assignTypeIdOnType(input.type);
-
   return [input, transferArrayBuffers];
 }
-
-function assignTypeIdOnType<T extends Type>(type: DataType<T>): void {
-  // @ts-expect-error __type does not exist
-  type.__type = type.typeId;
-
-  if (type.children && type.children.length > 0) {
-    for (const child of type.children) {
-      assignTypeIdOnType(child.type);
-    }
-  }
-}