faster scientific notation parsing

JairusSW · Jun 10, 2024 · 53c444f · 53c444f
1 parent d05d107
commit 53c444f
Show file tree

Hide file tree

Showing 5 changed files with 23 additions and 14 deletions.
diff --git a/assembly/deserialize/number.ts b/assembly/deserialize/number.ts
@@ -3,10 +3,9 @@ import { snip_fast } from "../src/util";
 
 // @ts-ignore: Decorator
 @inline export function deserializeNumber<T>(data: string, start: i32 = 0, end: i32 = 0): T {
-  end = end || data.length - 1;
   if (isInteger<T>()) {
     // @ts-ignore
-    return snip_fast<T>(data.slice(start, end));
+    return snip_fast<T>(data, start, end || data.length << 1);
   }
   // @ts-ignore
   const type: T = 0;

diff --git a/assembly/deserialize/string.ts b/assembly/deserialize/string.ts
@@ -24,14 +24,15 @@ import { Sink } from "../src/sink";
   let result = Sink.withCapacity(end - start - 1);
   const firstChar = unsafeCharCodeAt(data, start);
   const lastChar = unsafeCharCodeAt(data, end);
+
   if (firstChar !== quoteCode || lastChar !== quoteCode) {
     throw new Error(`Expected string to start and end with ", but got ${data.slice(0, 100)} instead!`);
   }
+
   let last = start + 1;
   for (let i = last; i < end; i++) {
-    if (unsafeCharCodeAt(data, i) !== backSlashCode) {
-      continue;
-    }
+    if (unsafeCharCodeAt(data, i) !== backSlashCode) continue;
+
     const char = unsafeCharCodeAt(data, ++i);
     result.write(data, last, i - 1);
     switch (char) {
@@ -87,8 +88,10 @@ import { Sink } from "../src/sink";
       }
     }
   }
+
   if (end > last) {
     result.write(data, last, end);
   }
+
   return result.toString();
 }
diff --git a/assembly/deserialize/unknown.ts b/assembly/deserialize/unknown.ts
@@ -26,7 +26,7 @@ import { deserializeString } from "./string";
         return Result.Ok<JSON.Value, string>(JSON.Value.from<boolean>(true));
     } else if (
         firstChar === fCode
-        && load<u64>(changetype<usize>(data), 2) === 28429466576093281 // "alse"
+        && load<u64>(changetype<usize>(data), 2) === 28429466576093281 // "false"
     ) {
         return Result.Ok<JSON.Value, string>(JSON.Value.from<boolean>(false));
     }

diff --git a/assembly/src/util.ts b/assembly/src/util.ts
@@ -88,15 +88,15 @@ import { Product } from "../product";
  * @param str - Any number. Can include scientific notation.
 */
 // @ts-ignore: Decorator
-@inline export function snip_fast<T extends number>(str: string, len: u32 = 0, offset: u32 = 0): T {
+@inline export function snip_fast<T extends number>(str: string, start: u32 = 0, end: u32 = str.length << 1): T {
+  let val: T = 0 as T;
+  const len: u32 = end - start;
   if (isSigned<T>()) {
     const firstChar: u32 = load<u16>(changetype<usize>(str));
     if (firstChar === 48) return 0 as T;
     const isNegative = firstChar === 45; // Check if the number is negative
-    let val: T = 0 as T;
-    if (len == 0) len = u32(str.length << 1);
     if (isNegative) {
-      offset += 2;
+      let offset: u32 = 2;
       if (len >= 4) {
         // 32-bit route
         for (; offset < (len - 3); offset += 4) {
@@ -146,6 +146,7 @@ import { Product } from "../product";
       }
       return -val as T;
     } else {
+      let offset: u32 = 0;
       if (len >= 4) {
         // Duplet 16 bit lane load
         for (; offset < (len - 3); offset += 4) {
@@ -198,7 +199,7 @@ import { Product } from "../product";
     const firstChar: u32 = load<u16>(changetype<usize>(str));
     if (firstChar === 48) return 0 as T;
     let val: T = 0 as T;
-    if (len == 0) len = u32(str.length << 1);
+    let offset: u32 = 0;
     if (len >= 4) {
       // Duplet 16 bit lane load
       for (; offset < (len - 3); offset += 4) {

diff --git a/assembly/test.ts b/assembly/test.ts
@@ -3,10 +3,12 @@ import { deserializeArray } from "./deserialize/array/array";
 import { deserializeBooleanArray } from "./deserialize/array/boolean";
 import { deserializeNumberArray } from "./deserialize/array/number";
 import { deserializeStringArray } from "./deserialize/array/string";
+import { deserializeNumber } from "./deserialize/number";
 import { utoa32 } from "./serialize/integer";
 import { Sink } from "./src/sink";
 import { describe, expect } from "as-test/assembly";
-/*
+import { __atoi_fast } from "./src/util";
+
 describe("Serialize String", () => {
     expect(JSON.serialize("hello world").toString()).toBe("\"hello world\"");
 });
@@ -128,7 +130,11 @@ map.set("z", JSON.Value.from<f64>(-5.6));
 describe("Serialize Maps", () => {
     expect(JSON.serialize(map).toString()).toBe("{\"x\":3.4,\"y\":1.2,\"z\":-5.6}");
 });
-*/
+
+describe("Deserialize Number", () => {
+    expect(JSON.serialize(deserializeNumber<i32>("1234e3")).toString()).toBe("1234000");
+})
+/*
 
 describe("Deserialize String[]", () => {
     expect(JSON.serialize(deserializeStringArray("[\"hello\",\"world\"]")).toString()).toBe("[\"hello\",\"world\"]");
@@ -144,4 +150,4 @@ describe("Deserialize Boolean[]", () => {
 
 describe("Deserialize [][]", () => {
     deserializeArray<JSON.Value[]>("[[[]],[[[]]]]")
-});
+});*/