feat: SWAR support all control codes

Author: JairusSW

assembly/serialize/swar/string.ts

@@ -3,9 +3,20 @@ import { BACK_SLASH } from "../../custom/chars";
 import { SERIALIZE_ESCAPE_TABLE } from "../../globals/tables";
 import { bytes } from "../../util/bytes";
 
+
+// @ts-ignore: decorator allowed
+@lazy const LANE_MASK_HIGH = 0xFF00_FF00_FF00_FF00;
+// @ts-ignore: decorator allowed
+@lazy const LANE_MASK_LOW = 0x00FF_00FF_00FF_00FF;
+// @ts-ignore: decorator allowed
+@lazy const UINT64_H8 = 0x8080808080808080;
 // @ts-ignore: decorator allowed
 @lazy const QUOTE_MASK = 0x0022_0022_0022_0022;
 // @ts-ignore: decorator allowed
+@lazy const BACKSLASH_MASK = 0x005C_005C_005C_005C;
+// @ts-ignore: decorator allowed
+@lazy const CONTROL_MASK = 0x0020_0020_0020_0020;
+// @ts-ignore: decorator allowed
 @lazy const U00_MARKER = 13511005048209500;
 
 export function serializeString_SWAR(src: string): void {
@@ -22,9 +33,11 @@ export function serializeString_SWAR(src: string): void {
     const block = load<u64>(srcStart);
     store<u64>(bs.offset, block);
 
-    const quotes = COMP(block, QUOTE_MASK);
-    let mask = quotes;
-
+    let mask = (
+      v64x4_eq(block, QUOTE_MASK) |
+      v64x4_eq(block, BACKSLASH_MASK) |
+      v64x4_ltu(block & LANE_MASK_LOW, CONTROL_MASK)
+    ) & UINT64_H8
 
     while (mask != 0) {
       const lane_index = usize(ctz(mask) >> 3);
@@ -81,9 +94,14 @@ export function serializeString_SWAR(src: string): void {
 }
 
 // @ts-ignore: decorators allowed
-@inline function COMP(x: u64, y: u64): u64 {
-  const LANE_MASK = 0xFF00_FF00_FF00_FF00;
-  const xored = (x ^ LANE_MASK) ^ y;
-  const mask = (((xored >> 1) | 0x8080808080808080) - xored) & 0x8080808080808080;
-  return (mask << 1) - (mask >> 7) & 0x8080808080808080;
+@inline function v64x4_eq(x: u64, y: u64): u64 {
+  const xored = (x ^ LANE_MASK_HIGH) ^ y;
+  const mask = (((xored >> 1) | UINT64_H8) - xored) & UINT64_H8;
+  return (mask << 1) - (mask >> 7);
 }
+
+// @ts-ignore: decorators allowed
+@inline function v64x4_ltu(a: u64, b: u64): u64 {
+  // Vigna's algorithm - fastest SWAR unsigned less-than
+  return (((a | UINT64_H8) - (b & ~UINT64_H8)) | (a ^ b)) ^ (a | ~b);
+}

assembly/test.tmp.ts

@@ -2,6 +2,6 @@ import { bs } from "../lib/as-bs";
 import { serializeString_SIMD } from "./serialize/simd/string";
 import { serializeString_SWAR } from "./serialize/swar/string";
 
-serializeString_SWAR("ab\"d");
-// serializeString_SIMD("ab\"defgh");
+serializeString_SWAR("a\0b\"\nd");
+serializeString_SIMD("a\0b\"\nd");
 console.log(bs.out<string>());

assembly/test.ts

@@ -9,12 +9,12 @@
   "main": "transform/lib/index.js",
   "devDependencies": {
     "@assemblyscript/wasi-shim": "^0.1.0",
-    "@types/node": "^24.2.1",
-    "assemblyscript": "^0.28.4",
+    "@types/node": "^25.0.3",
+    "assemblyscript": "^0.28.9",
     "assemblyscript-prettier": "^3.0.1",
-    "prettier": "^3.6.2",
-    "tinybench": "^5.0.1",
-    "typescript": "^5.9.2"
+    "prettier": "^3.7.4",
+    "tinybench": "^6.0.0",
+    "typescript": "^5.9.3"
   },
   "bugs": {
     "url": "https://github.com/JairusSW/json-as/issues"

package.json

@@ -1,5 +1,4 @@
 import { ArrayLiteralExpression, AssertionExpression, BinaryExpression, CallExpression, ElementAccessExpression, FloatLiteralExpression, FunctionTypeNode, IdentifierExpression, NamedTypeNode, Node, ObjectLiteralExpression, Source, TypeParameterNode, BlockStatement, BreakStatement, ClassDeclaration, ClassExpression, CommaExpression, ContinueStatement, DecoratorNode, DoStatement, EmptyStatement, EnumDeclaration, EnumValueDeclaration, ExportDefaultStatement, ExportImportStatement, ExportMember, ExportStatement, ExpressionStatement, FieldDeclaration, ForStatement, FunctionDeclaration, FunctionExpression, IfStatement, ImportDeclaration, ImportStatement, IndexSignatureNode, InstanceOfExpression, IntegerLiteralExpression, InterfaceDeclaration, LiteralExpression, MethodDeclaration, NamespaceDeclaration, NewExpression, ParameterNode, ParenthesizedExpression, PropertyAccessExpression, RegexpLiteralExpression, ReturnStatement, StringLiteralExpression, SwitchCase, SwitchStatement, TemplateLiteralExpression, TernaryExpression, ThrowStatement, TryStatement, TypeDeclaration, TypeName, UnaryPostfixExpression, UnaryPrefixExpression, VariableDeclaration, VariableStatement, WhileStatement, NodeKind, TypeNode, Expression, LiteralKind, UnaryExpression, SuperExpression, FalseExpression, TrueExpression, ThisExpression, NullExpression, ConstructorExpression, Statement, VoidStatement, CompiledExpression, CommentNode, Module, OmittedExpression, ForOfStatement, ModuleDeclaration } from "assemblyscript/dist/assemblyscript.js";
-import { toString } from "./util.js";
 
 export class Visitor {
   public currentSource: Source | null = null;