CLI tool for packing multiple files into a single Binpack binary in order to save network requests in the browser.
You can use Binpacker
as an efficient replacement for spritesheets.
Inspired by GLB File Format Specification, MM.Loader, MM.Packer, Magipack.js and this Twitter thread.
Make sure you have Node.js installed.
$ npm install -g --save @timvanscherpenzeel/binpacker
$ node ./bin/binpacker.js -i ./input -o ./output/example.binpack -vb
$ node ./bin/binpacker.js -i ./input/manifest.json -o ./output/example.binpack -vb
{
"path": "./input/",
"manifest": [
{
"src": "icon-twitter.svg"
},
{
"src": "spritesheet (2).json"
},
{
"src": "texture.jpg"
}
]
}
A .binpack
file has the following structure (very similar to the GLB File Format Specification):
Figure from the GLB File Format Specification.
Binpack
is little endian.
The 12-byte header consists of three 4-byte entries:
uint32 magic
uint32 version
uint32 length
-
magic
equals0x504e4942
. It is ASCII stringBINP
, and can be used to identify data asBinpack
. -
version
indicates the version of theBinpack
. This specification defines version 1. -
length
is the total length of theBinpack
file, including Header and all Chunks, in bytes.
A single JSON chunk header
The JSON chunk header has a field that marks the length of the JSON chunk (uint32 chunkLength
) and a type (uint32 chunkType
) marked JSON
in ASCII.
A single JSON chunk
The JSON chunk (ubyte[] chunkData
) contains a stringified JSON description of the processed files: name
, bufferStart
, bufferEnd
and mimeType
. The difference between bufferStart
and bufferEnd
describe the length of the file. This length is used to extract the correct amount of bytes per file from the binary chunk that follows next.
A single binary chunk header
The binary chunk header has a field that marks the length of the binary chunk (uint32 chunkLength
) and a type (uint32 chunkType
) marked BIN
in ASCII.
A single binary chunk
The binary chunk (ubyte[] chunkData
) contains a single Uint8Array
typed array buffer that has been constructed out of concatenated files. Using the data described in the JSON chunk one can correctly extract the file from the binary chunk.
-i, --input [example: ./input (glob) or ./input/manifest.json (manifest)] [required]
-o, --output [example: ./output/example.binpack] [required]
-vb, --verbose [true / false, default: false] [not required]
My work is released under the MIT license.