Faster Bun.hash.crc32

bench/snippets/bun-hash.mjs

@@ -0,0 +1,26 @@
+import { bench, group, run } from "./runner.mjs";
+
+const hashes = ["wyhash", "adler32", "crc32", "cityHash32", "cityHash64", "murmur32v3", "murmur32v2", "murmur64v2"];
+
+group("hello world", () => {
+  for (const name of hashes) {
+    const fn = Bun.hash[name];
+
+    bench(`${name}`, () => {
+      return fn("hello world");
+    });
+  }
+});
+
+group("hello world (x 1024)", () => {
+  for (const name of hashes) {
+    const fn = Bun.hash[name];
+
+    const repeated = Buffer.alloc("hello world".length * 1024, "hello world").toString();
+    bench(`${name}`, () => {
+      return fn(repeated);
+    });
+  }
+});
+
+await run();

src/bun.js/api/BunObject.zig

@@ -3613,11 +3613,10 @@ pub fn getHashObject(
 ) callconv(.C) JSC.JSValue {
     return HashObject.create(globalThis);
 }
-
 const HashObject = struct {
     pub const wyhash = hashWrap(std.hash.Wyhash).hash;
-    pub const adler32 = hashWrap(std.hash.Adler32).hash;
-    pub const crc32 = hashWrap(std.hash.Crc32).hash;
+    pub const adler32 = hashWrap(bun.zlib.Adler32).hash;
+    pub const crc32 = hashWrap(bun.zlib.Crc32).hash;
     pub const cityHash32 = hashWrap(std.hash.CityHash32).hash;
     pub const cityHash64 = hashWrap(std.hash.CityHash64).hash;
     pub const murmur32v2 = hashWrap(std.hash.murmur.Murmur2_32).hash;

src/deps/zlib.posix.zig

@@ -87,3 +87,6 @@ pub inline fn inflateInit2(strm: anytype, windowBits: anytype) ReturnCode {
 pub inline fn inflateBackInit(strm: anytype, windowBits: anytype, window: anytype) ReturnCode {
     return inflateBackInit_(strm, windowBits, window, zlibVersion(), @import("std").zig.c_translation.cast(c_int, @import("std").zig.c_translation.sizeof(z_stream)));
 }
+
+pub extern fn adler32(uLong, ptr: ?[*]const u8, len: uInt) uLong;
+pub extern fn crc32(uLong, ptr: ?[*]const u8, len: uInt) uLong;

src/deps/zlib.win32.zig

@@ -119,10 +119,10 @@ pub extern fn gzclose_r(file: gzFile) ReturnCode;
 pub extern fn gzclose_w(file: gzFile) ReturnCode;
 pub extern fn gzerror(file: gzFile, errnum: [*c]c_int) [*c]const u8;
 pub extern fn gzclearerr(file: gzFile) void;
-pub extern fn adler32(adler: uLong, buf: [*c]const Bytef, len: uInt) uLong;
+pub extern fn adler32(adler: uLong, buf: ?[*]const Bytef, len: uInt) uLong;
 pub extern fn adler32_z(adler: uLong, buf: [*c]const Bytef, len: z_size_t) uLong;
-pub extern fn crc32(crc: uLong, buf: [*c]const Bytef, len: uInt) uLong;
-pub extern fn crc32_z(crc: uLong, buf: [*c]const Bytef, len: z_size_t) uLong;
+pub extern fn crc32(crc: uLong, buf: ?[*]const Bytef, len: uInt) uLong;
+pub extern fn crc32_z(crc: uLong, buf: ?[*]const Bytef, len: z_size_t) uLong;
 pub extern fn crc32_combine_op(crc1: uLong, crc2: uLong, op: uLong) uLong;
 pub extern fn deflateInit_(strm: z_streamp, level: c_int, version: [*c]const u8, stream_size: c_int) ReturnCode;
 pub extern fn inflateInit_(strm: z_streamp, version: [*c]const u8, stream_size: c_int) ReturnCode;

src/zlib.zig

@@ -901,3 +901,42 @@ pub const ZlibCompressorArrayList = struct {
         }
     }
 };
+
+pub const adler32 = @import("zlib-internal").adler32;
+pub const crc32 = @import("zlib-internal").crc32;
+
+fn HasherFn(comptime function: anytype) type {
+    return struct {
+        state: uLong = 0,
+
+        const Hasher = @This();
+
+        pub fn init(this: *Hasher) void {
+            this.state = function(0, null, 0);
+        }
+
+        pub fn update(this: *Hasher, data: []const u8) void {
+            this.state = function(this.state, data.ptr, @intCast(data.len));
+        }
+
+        pub fn finish(this: *Hasher) uLong {
+            return this.state;
+        }
+
+        pub fn hash(data: []const u8) uLong {
+            var hasher = @This(){};
+            hasher.init();
+            hasher.update(data);
+            return hasher.finish();
+        }
+
+        pub fn hashWithSeed(data: []const u8, seed: uLong) uLong {
+            var hasher = @This(){ .state = seed };
+            hasher.update(data);
+            return hasher.finish();
+        }
+    };
+}
+
+pub const Adler32 = HasherFn(adler32);
+pub const Crc32 = HasherFn(crc32);