diff options
Diffstat (limited to 'src/bun.js/webcore')
| -rw-r--r-- | src/bun.js/webcore/encoding.classes.ts | 3 | ||||
| -rw-r--r-- | src/bun.js/webcore/encoding.zig | 29 |
2 files changed, 26 insertions, 6 deletions
diff --git a/src/bun.js/webcore/encoding.classes.ts b/src/bun.js/webcore/encoding.classes.ts index 118dfd09e..7114f210e 100644 --- a/src/bun.js/webcore/encoding.classes.ts +++ b/src/bun.js/webcore/encoding.classes.ts @@ -16,6 +16,9 @@ export default [ fatal: { getter: "getFatal", }, + ignoreBOM: { + getter: "getIgnoreBOM", + }, decode: { fn: "decode", diff --git a/src/bun.js/webcore/encoding.zig b/src/bun.js/webcore/encoding.zig index 8ffbd3fd0..ca0f44e6a 100644 --- a/src/bun.js/webcore/encoding.zig +++ b/src/bun.js/webcore/encoding.zig @@ -559,6 +559,13 @@ pub const TextDecoder = struct { remainder = remainder[1..]; continue; }, + // BOM handling + 0xFEFF => { + buffer.ensureTotalCapacity(allocator, 1) catch unreachable; + buffer.items.ptr[buffer.items.len] = remainder[0]; + buffer.items.len += 1; + remainder = remainder[1..]; + }, // Is this an unpaired low surrogate or four-digit hex escape? else => { @@ -629,8 +636,13 @@ pub const TextDecoder = struct { }, EncodingLabel.@"UTF-8" => { const toUTF16 = if (stream) strings.toUTF16Alloc else strings.toUTF16AllocNoTrim; + const moved_buffer_slice_8 = if (!this.ignore_bom and buffer_slice.len > 3 and std.mem.eql(u8, &[_]u8{ '\xEF', '\xBB', '\xBF' }, buffer_slice[0..3])) + buffer_slice[3..] + else + buffer_slice; + if (this.fatal) { - if (toUTF16(default_allocator, buffer_slice, true)) |result_| { + if (toUTF16(default_allocator, moved_buffer_slice_8, true)) |result_| { if (result_) |result| { return ZigString.toExternalU16(result.ptr, result.len, globalThis); } @@ -649,7 +661,7 @@ pub const TextDecoder = struct { } } } else { - if (toUTF16(default_allocator, buffer_slice, false)) |result_| { + if (toUTF16(default_allocator, moved_buffer_slice_8, false)) |result_| { if (result_) |result| { return ZigString.toExternalU16(result.ptr, result.len, globalThis); } @@ -664,15 +676,20 @@ pub const TextDecoder = struct { } // Experiment: using mimalloc directly is slightly slower - return ZigString.init(buffer_slice).toValueGC(globalThis); + return ZigString.init(moved_buffer_slice_8).toValueGC(globalThis); }, EncodingLabel.@"UTF-16LE" => { - if (std.mem.isAligned(@intFromPtr(buffer_slice.ptr), @alignOf([*]const u16))) { - return this.decodeUTF16WithAlignment([]align(2) const u16, @as([]align(2) const u16, @alignCast(std.mem.bytesAsSlice(u16, buffer_slice))), globalThis); + const moved_buffer_slice_16 = if (!this.ignore_bom and buffer_slice.len > 2 and std.mem.eql(u8, &[_]u8{ '\xFF', '\xFE' }, buffer_slice[0..2])) + buffer_slice[2..] + else + buffer_slice; + + if (std.mem.isAligned(@intFromPtr(moved_buffer_slice_16.ptr), @alignOf([*]const u16))) { + return this.decodeUTF16WithAlignment([]align(2) const u16, @as([]align(2) const u16, @alignCast(std.mem.bytesAsSlice(u16, moved_buffer_slice_16))), globalThis); } - return this.decodeUTF16WithAlignment([]align(1) const u16, std.mem.bytesAsSlice(u16, buffer_slice), globalThis); + return this.decodeUTF16WithAlignment([]align(1) const u16, std.mem.bytesAsSlice(u16, moved_buffer_slice_16), globalThis); }, else => { globalThis.throwInvalidArguments("TextDecoder.decode set to unsupported encoding", .{}); |