Bun Bindgen

Bindgen

Bindgen for Bun

This document is for maintainers and contributors to Bun, and describes internal implementation details.

The new bindings generator, introduced to the codebase in Dec 2024, scans for *.bind.ts to find function and class definition, and generates glue code to interop between JavaScript and native code.

There are currently other code generators and systems that achieve similar purposes. The following will all eventually be completely phased out in favor of this one:

• "Classes generator", converting *.classes.ts for custom classes.
• "JS2Native", allowing ad-hoc calls from src/js to native code.

Creating JS Functions in Zig

Given a file implementing a simple function, such as add

pub fn add(global: *jsc.JSGlobalObject, a: i32, b: i32) !i32 {
    return std.math.add(i32, a, b) catch {
        // Binding functions can return `error.OutOfMemory` and `error.JSError`.
        // Others like `error.Overflow` from `std.math.add` must be converted.
        // Remember to be descriptive.
        return global.throwPretty("Integer overflow while adding", .{});
    };
}

const gen = bun.gen.math; // "math" being this file's basename

const std = @import("std");
const bun = @import("bun");
const jsc = bun.jsc;

Then describe the API schema using a .bind.ts function. The binding file goes next to the Zig file.

import { t, fn } from "bindgen";

export const add = fn({
  args: {
    global: t.globalObject,
    a: t.i32,
    b: t.i32.default(1),
  },
  ret: t.i32,
});

This function declaration is equivalent to:

/**
 * Throws if zero arguments are provided.
 * Wraps out of range numbers using modulo.
 */
declare function add(a: number, b: number = 1): number;

The code generator will provide bun.gen.math.jsAdd, which is the native function implementation. To pass to JavaScript, use bun.gen.math.createAddCallback(global). JS files in src/js/ may use $bindgenFn("math.bind.ts", "add") to get a handle to the implementation.

Strings

The type for receiving strings is one of t.DOMString, t.ByteString, and t.USVString. These map directly to their WebIDL counterparts, and have slightly different conversion logic. Bindgen will pass BunString to native code in all cases.

When in doubt, use DOMString.

t.UTF8String can be used in place of t.DOMString, but will call bun.String.toUTF8. The native callback gets []const u8 (WTF-8 data) passed to native code, freeing it after the function returns.

TLDRs from WebIDL spec:

• ByteString can only contain valid latin1 characters. It is not safe to assume bun.String is already in 8-bit format, but it is extremely likely.
• USVString will not contain invalid surrogate pairs, aka text that can be represented correctly in UTF-8.
• DOMString is the loosest but also most recommended strategy.

Function Variants

A variants can specify multiple variants (also known as overloads).

import { t, fn } from "bindgen";

export const action = fn({
  variants: [
    {
      args: {
        a: t.i32,
      },
      ret: t.i32,
    },
    {
      args: {
        a: t.DOMString,
      },
      ret: t.DOMString,
    },
  ],
});

In Zig, each variant gets a number, based on the order the schema defines.

fn action1(a: i32) i32 {
  return a;
}

fn action2(a: bun.String) bun.String {
  return a;
}

t.dictionary

A dictionary is a definition for a JavaScript object, typically as a function inputs. For function outputs, it is usually a smarter idea to declare a class type to add functions and destructuring.

Enumerations

To use WebIDL's enumeration type, use either:

• t.stringEnum: Create and codegen a new enum type.
• t.zigEnum: Derive a bindgen type off of an existing enum in the codebase.

An example of stringEnum as used in fmt.zig / bun:internal-for-testing

export const Formatter = t.stringEnum("highlight-javascript", "escape-powershell");

export const fmtString = fn({
  args: {
    global: t.globalObject,
    code: t.UTF8String,
    formatter: Formatter,
  },
  ret: t.DOMString,
});

WebIDL strongly encourages using kebab case for enumeration values, to be consistent with existing Web APIs.

Deriving enums from Zig code

TODO: zigEnum

t.oneOf

A oneOf is a union between two or more types. It is represented by union(enum) in Zig.

TODO:

Attributes

There are set of attributes that can be chained onto t.* types. On all types there are:

• .required, in dictionary parameters only
• .optional, in function arguments only
• .default(T)

When a value is optional, it is lowered to a Zig optional.

Depending on the type, there are more attributes available. See the type definitions in auto-complete for more details. Note that one of the above three can only be applied, and they must be applied at the end.

Integer Attributes

Integer types allow customizing the overflow behavior with clamp or enforceRange

import { t, fn } from "bindgen";

export const add = fn({
  args: {
    global: t.globalObject,
    // enforce in i32 range
    a: t.i32.enforceRange(),
    // clamp to u16 range
    b: t.u16,
    // enforce in arbitrary range, with a default if not provided
    c: t.i32.enforceRange(0, 1000).default(5),
    // clamp to arbitrary range, or null
    d: t.u16.clamp(0, 10).optional,
  },
  ret: t.i32,
});

Various Node.js validator functions such as validateInteger, validateNumber, and more are available. Use these when implementing Node.js APIs, so the error messages match 1:1 what Node would do.

Unlike enforceRange, which is taken from WebIDL, validate* functions are much more strict on the input they accept. For example, Node's numerical validator check typeof value === 'number', while WebIDL uses ToNumber for lossy conversion.

import { t, fn } from "bindgen";

export const add = fn({
  args: {
    global: t.globalObject,
    // throw if not given a number
    a: t.f64.validateNumber(),
    // valid in i32 range
    a: t.i32.validateInt32(),
    // f64 within safe integer range
    b: t.f64.validateInteger(),
    // f64 in given range
    c: t.f64.validateNumber(-10000, 10000),
  },
  ret: t.i32,
});

Callbacks

TODO

Classes

TODO