Improve .ply parsing speed using compiled parser function (#150)

* Compile a parse function per element type in PlyReader

* Parse f_rest properties as array for faster .ply file parsing

src/defines.ts

@@ -36,3 +36,8 @@ export const SPLAT_TEX_DEPTH_MASK = SPLAT_TEX_DEPTH - 1;
 // time between pure JS and WASM are minimal and don't make a big difference.
 
 export const WASM_SPLAT_SORT = true;
+
+// Enable/disable compiling a dedicated parse function per element type
+// in the plyReader.
+
+export const USE_COMPILED_PARSER_FUNCTION = true;

src/ply.ts

@@ -1,14 +1,18 @@
 // PLY file format reader
 
-export type PlyPropertyType =
+import { USE_COMPILED_PARSER_FUNCTION } from "./defines";
-  | "char"
+
-  | "uchar"
+const PLY_PROPERTY_TYPES = [
-  | "short"
+  "char",
-  | "ushort"
+  "uchar",
-  | "int"
+  "short",
-  | "uint"
+  "ushort",
-  | "float"
+  "int",
-  | "double";
+  "uint",
+  "float",
+  "double",
+] as const;
+export type PlyPropertyType = (typeof PLY_PROPERTY_TYPES)[number];
 
 export type PlyElement = {
   name: string;
@@ -199,49 +203,14 @@ export class PlyReader {
     for (const elementName in this.elements) {
       const element = this.elements[elementName];
       const { count, properties } = element;
-      const item: Record<string, number | number[]> = {};
+      const item = createEmptyItem(properties);
-      // Construct an array of parser function to parse each property in an item
+      // Construct a parse function
-      const parsers = [];
+      const parseFn = createParseFn(properties, this.littleEndian);
-      for (const [propertyName, property] of Object.entries(properties)) {
-        if (!property.isList) {
-          item[propertyName] = 0;
-          parsers.push(() => {
-            item[propertyName] = PARSE_FIELD[property.type](
-              data,
-              offset,
-              this.littleEndian,
-            );
-            offset += FIELD_BYTES[property.type];
-          });
-        } else {
-          // Property is a list, so parse the count first
-          item[propertyName] = [];
-          parsers.push(() => {
-            const list = item[propertyName] as number[];
-            list.length = PARSE_FIELD[property.countType as PlyPropertyType](
-              data,
-              offset,
-              this.littleEndian,
-            );
-            offset += FIELD_BYTES[property.countType as PlyPropertyType];
-            for (let i = 0; i < list.length; i++) {
-              list[i] = PARSE_FIELD[property.type](
-                data,
-                offset,
-                this.littleEndian,
-              );
-              offset += FIELD_BYTES[property.type];
-            }
-          });
-        }
-      }
 
       // Parse all the items in the element
       const callback = elementCallback(element) ?? (() => {});
       for (let index = 0; index < count; index++) {
-        for (let parserIndex = 0; parserIndex < parsers.length; parserIndex++) {
+        offset = parseFn(data, offset, item);
-          parsers[parserIndex]();
-        }
         callback(index, item);
       }
     }
@@ -258,9 +227,9 @@ export class PlyReader {
     const ssChunks: SSChunk[] = [];
 
     let numSh = 0;
-    let sh1Props: string[] = [];
+    let sh1Props: number[] = [];
-    let sh2Props: string[] = [];
+    let sh2Props: number[] = [];
-    let sh3Props: string[] = [];
+    let sh3Props: number[] = [];
     let sh1: Float32Array | undefined = undefined;
     let sh2: Float32Array | undefined = undefined;
     let sh3: Float32Array | undefined = undefined;
@@ -270,18 +239,16 @@ export class PlyReader {
       const num_f_rest = NUM_SH_TO_NUM_F_REST[numSh];
       sh1Props = new Array(3)
         .fill(null)
-        .flatMap((_, k) =>
+        .flatMap((_, k) => [0, 1, 2].map((_, d) => k + (d * num_f_rest) / 3));
-          [0, 1, 2].map((_, d) => `f_rest_${k + (d * num_f_rest) / 3}`),
-        );
       sh2Props = new Array(5)
         .fill(null)
         .flatMap((_, k) =>
-          [0, 1, 2].map((_, d) => `f_rest_${3 + k + (d * num_f_rest) / 3}`),
+          [0, 1, 2].map((_, d) => 3 + k + (d * num_f_rest) / 3),
         );
       sh3Props = new Array(7)
         .fill(null)
         .flatMap((_, k) =>
-          [0, 1, 2].map((_, d) => `f_rest_${8 + k + (d * num_f_rest) / 3}`),
+          [0, 1, 2].map((_, d) => 8 + k + (d * num_f_rest) / 3),
         );
       sh1 = numSh >= 1 ? new Float32Array(3 * 3) : undefined;
       sh2 = numSh >= 2 ? new Float32Array(5 * 3) : undefined;
@@ -296,17 +263,19 @@ export class PlyReader {
       if (!sh1) {
         throw new Error("Missing sh1");
       }
+      const sh = item.f_rest as number[];
+
       for (let i = 0; i < sh1Props.length; i++) {
-        sh1[i] = ((item[sh1Props[i]] as number) * 8) / 255 - 4;
+        sh1[i] = (sh[sh1Props[i]] * 8) / 255 - 4;
       }
       if (sh2) {
         for (let i = 0; i < sh2Props.length; i++) {
-          sh2[i] = ((item[sh2Props[i]] as number) * 8) / 255 - 4;
+          sh2[i] = (sh[sh2Props[i]] * 8) / 255 - 4;
         }
       }
       if (sh3) {
         for (let i = 0; i < sh3Props.length; i++) {
-          sh3[i] = ((item[sh3Props[i]] as number) * 8) / 255 - 4;
+          sh3[i] = (sh[sh3Props[i]] * 8) / 255 - 4;
         }
       }
       shCallback?.(index, sh1, sh2, sh3);
@@ -623,19 +592,20 @@ export class PlyReader {
         );
 
         if (shCallback && sh1) {
+          const sh = item.f_rest as number[];
           if (sh1) {
-            for (const [i, key] of sh1Props.entries()) {
+            for (let i = 0; i < sh1Props.length; i++) {
-              sh1[i] = item[key] as number;
+              sh1[i] = sh[sh1Props[i]];
             }
           }
           if (sh2) {
-            for (const [i, key] of sh2Props.entries()) {
+            for (let i = 0; i < sh2Props.length; i++) {
-              sh2[i] = item[key] as number;
+              sh2[i] = sh[sh2Props[i]];
             }
           }
           if (sh3) {
-            for (const [i, key] of sh3Props.entries()) {
+            for (let i = 0; i < sh3Props.length; i++) {
-              sh3[i] = item[key] as number;
+              sh3[i] = sh[sh3Props[i]];
             }
           }
           shCallback(index, sh1, sh2, sh3);
@@ -887,6 +857,204 @@ const NUM_SH_TO_NUM_F_REST: Record<number, number> = {
   3: 45,
 };
 
+const F_REST_REGEX = /^f_rest_([0-9]{1,2})$/;
+
+function createEmptyItem(
+  properties: Record<string, PlyProperty>,
+): Record<string, number | number[]> {
+  const item: Record<string, number | number[]> = {};
+  for (const [propertyName, property] of Object.entries(properties)) {
+    // Treat f_rest properties as a single array for performance
+    if (F_REST_REGEX.test(propertyName)) {
+      item.f_rest = new Array(getNumSh(properties));
+    } else {
+      item[propertyName] = property.isList ? [] : 0;
+    }
+  }
+  return item;
+}
+
+function createParseFn(
+  properties: Record<string, PlyProperty>,
+  littleEndian: boolean,
+) {
+  if (USE_COMPILED_PARSER_FUNCTION && safeToCompile(properties)) {
+    return createCompiledParserFn(properties, littleEndian);
+  }
+  return createDynamicParserFn(properties, littleEndian);
+}
+
+// Detect if unsafe eval is allowed in the current execution context
+const UNSAFE_EVAL_ALLOWED = (() => {
+  try {
+    new Function("return 42;");
+  } catch (e) {
+    return false;
+  }
+  return true;
+})();
+const PROPERTY_NAME_REGEX = /^[a-zA-Z0-9_]+$/;
+
+function safeToCompile(properties: Record<string, PlyProperty>) {
+  if (!UNSAFE_EVAL_ALLOWED) {
+    return false;
+  }
+
+  for (const [propertyName, property] of Object.entries(properties)) {
+    if (!PROPERTY_NAME_REGEX.test(propertyName)) {
+      return false;
+    }
+
+    if (
+      property.isList &&
+      !PLY_PROPERTY_TYPES.includes(property.countType as PlyPropertyType)
+    ) {
+      return false;
+    }
+
+    if (!PLY_PROPERTY_TYPES.includes(property.type)) {
+      return false;
+    }
+  }
+  return true;
+}
+
+function createCompiledParserFn(
+  properties: Record<string, PlyProperty>,
+  littleEndian: boolean,
+) {
+  // Construct the parser function source.
+  const parserSrc: string[] = ["let list;"];
+  for (const [propertyName, property] of Object.entries(properties)) {
+    const fRestMatch = propertyName.match(F_REST_REGEX);
+    if (fRestMatch) {
+      const fRestIndex = +fRestMatch[1];
+      parserSrc.push(/*js*/ `
+        item.f_rest[${fRestIndex}] = PARSE_FIELD['${property.type}'](data, offset, ${littleEndian});
+        offset += ${FIELD_BYTES[property.type]};
+      `);
+    } else if (!property.isList) {
+      parserSrc.push(/*js*/ `
+        item['${propertyName}'] = PARSE_FIELD['${property.type}'](data, offset, ${littleEndian});
+        offset += ${FIELD_BYTES[property.type]};
+      `);
+    } else {
+      // Property is a list, so parse the count first
+      parserSrc.push(/*js*/ `
+        list = item['${propertyName}'];
+        list.length = PARSE_FIELD['${property.countType}'](data, offset, ${littleEndian});
+        offset += ${FIELD_BYTES[property.countType as PlyPropertyType]};
+        for (let i = 0; i < list.length; i++) {
+          list[i] = PARSE_FIELD['${property.type}'](data, offset, ${littleEndian});
+          offset += ${FIELD_BYTES[property.type]};
+        }
+      `);
+    }
+  }
+  parserSrc.push("return offset;");
+
+  const fn = new Function(
+    "data",
+    "offset",
+    "item",
+    "PARSE_FIELD",
+    parserSrc.join("\n"),
+  );
+  return (
+    data: DataView,
+    offset: number,
+    item: Record<string, number | number[]>,
+  ) => fn(data, offset, item, PARSE_FIELD);
+}
+
+function createDynamicParserFn(
+  properties: Record<string, PlyProperty>,
+  littleEndian: boolean,
+) {
+  // Construct an array of parser function to parse each property in an item
+  const parsers: Array<
+    (
+      data: DataView,
+      offset: number,
+      item: Record<string, number | number[]>,
+    ) => number
+  > = [];
+  for (const [propertyName, property] of Object.entries(properties)) {
+    const fRestMatch = propertyName.match(F_REST_REGEX);
+    if (fRestMatch) {
+      const fRestIndex = +fRestMatch[1];
+      parsers.push(
+        (
+          data: DataView,
+          offset: number,
+          item: Record<string, number | number[]>,
+        ) => {
+          (item.f_rest as number[])[fRestIndex] = PARSE_FIELD[property.type](
+            data,
+            offset,
+            littleEndian,
+          );
+          return offset + FIELD_BYTES[property.type];
+        },
+      );
+    } else if (!property.isList) {
+      parsers.push(
+        (
+          data: DataView,
+          offset: number,
+          item: Record<string, number | number[]>,
+        ) => {
+          item[propertyName] = PARSE_FIELD[property.type](
+            data,
+            offset,
+            littleEndian,
+          );
+          return offset + FIELD_BYTES[property.type];
+        },
+      );
+    } else {
+      // Property is a list, so parse the count first
+      parsers.push(
+        (
+          data: DataView,
+          offset: number,
+          item: Record<string, number | number[]>,
+        ) => {
+          const list = item[propertyName] as number[];
+          list.length = PARSE_FIELD[property.countType as PlyPropertyType](
+            data,
+            offset,
+            littleEndian,
+          );
+          let currentOffset =
+            offset + FIELD_BYTES[property.countType as PlyPropertyType];
+          for (let i = 0; i < list.length; i++) {
+            list[i] = PARSE_FIELD[property.type](
+              data,
+              currentOffset,
+              littleEndian,
+            );
+            currentOffset += FIELD_BYTES[property.type];
+          }
+          return currentOffset;
+        },
+      );
+    }
+  }
+
+  return (
+    data: DataView,
+    offset: number,
+    item: Record<string, number | number[]>,
+  ) => {
+    let currentOffset = offset;
+    for (let parserIndex = 0; parserIndex < parsers.length; parserIndex++) {
+      currentOffset = parsers[parserIndex](data, currentOffset, item);
+    }
+    return currentOffset;
+  };
+}
+
 function getNumSh(properties: Record<string, PlyProperty>) {
   let num_f_rest = 0;
   while (properties[`f_rest_${num_f_rest}`]) {