This module helps you work with binary structures in Buffers. You first define the layout of your binary data and then use that definition to convert objects to/from their binary representation or you can wrap a buffer with an object where getters/setters for fields update and read the buffer in place.
- Use the same definition for reading and writing objects
- Wrap a buffer to use getters/setters to operate directly on Buffer contents
- Also able to read/write to/from objects
- "Fluid API" structure definition is easy on the eyes
- Customizable 64-bit integer support
- Easy big-endian and little-endian support
- No external dependencies or native extensions required (works in Windows!)
This module requires node 0.6 or better, as it uses the binary type read/write methods on Buffer introduced with node 0.6.
Adding numeric fields to the structure definition requires calling a method with the appropriate type name.
If a string is passed, it is used as the name of the field, otherwise the field is basically skipped (padding).
- int8, int16, int32, int64: Signed integers
- byte, uint16, uint32, uint64: Unsigned integers
- float: 32-bit IEEE floating point number
- double: 64-bit IEEE floating point number
When defining a structure, the default is to use "big endian" byte order when reading/writing numbers. You can change the default by specifying options to the constructor. For example:
var littleEndianStruct = require('binstruct')
.def({littleEndian:true})
.float('littleEndianFloat')
.double('littleEndianDouble')
.int32('littleEndianInteger');
You can also specify littleEndian by adding "le" to the type names as you declare the structure. For example:
var littleEndianStruct = require('binstruct')
.def()
.floatle('littleEndianFloat')
.doublele('littleEndianDouble')
.int32le('littleEndianInteger');
This could conceivably allow you to have a mix of little and big endian numbers in the same structure although in practice I doubt that would ever happen. Rather this may turn out be a shorter syntax when there are relatively few fields - adding 'le' a few times could be less painful than {littleEndian:true}.
Javascript doesn't normally support 64-bit numbers - all numbers are represented as a 64-bit floating point value. This can only represent a 53-bit signed integer, or a 52-bit unsigned integer correctly.
When reading 64-bit integers you can choose whether to convert to a Javascript number or to store the number as a buffer. If converting to a number, you can choose whether an overflow should throw an error or just set the field to Infinity or -Infinity. If storing as a buffer you can choose whether it should be a "slice" of the original buffer or a copy.
- strict: Convert to js number, throw Error on overflow. This is the default mode.
- lossy: Convert to js number, use Infinity/-Infinity on overflow
- copy: Copy the literal bytes into a Buffer for the field value
- slice: Field value is a "slice" containing the original bytes
To specify a mode, pass an options object to the initial call to def() or as a parameter to a field definition with a property 'int64mode' mode set to a value from binstruct.int64modes. Examples:
var binstruct = require('binstruct');
binstruct.def({int64mode:binstruct.int64modes.lossy});
binstruct.def().uint64('a', {int64mode:binstruct.int64modes.copy});
binstruct.def().int64('b', {int64mode:binstruct.int64modes.slice});
The library may create a "wrapper" around a buffer which allows you to read and write the binary fields from the buffer on the fly using property getters and setters on the wrapper. This allows for a pleasant looking syntax for editing fields, and may save some time if relatively few fields are actually used by the application.
Note, however, that this access isn't generally very fast, so it is best to minimize the number of property reads and writes you do.
To create a wrapper call wrap() on your structure definition and pass a buffer and an optional offset into that buffer at which to read the structure.
var binstruct = require('binstruct');
var buf = new Buffer('ab');
var twoBytes = binstruct.def()
.byte('a')
.byte('b')
.wrap(buf);
assert.equal(String.fromCharCode(twoBytes.a), 'a');
assert.equal(String.fromCharCode(twoBytes.b), 'b');
twoBytes.a = 'x';
twoBytes.b = 'y';
assert.equal(String.fromCharCode(buf[0]), 'x');
assert.equal(String.fromCharCode(buf[1]), 'y');
var offsetBy1 = binstruct.def().byte('b').wrap(buf, 1);
assert.equal(String.fromCharCode(offsetBy1.b), 'y');
The library allows you to 'pack' an object into a Buffer and 'unpack' an object from a Buffer.
The 'read' operation goes through all the fields defined and populates them into a new object and returns it. Unlike with a wrapper, changes to that object will not affect the underlying buffer.
The 'write' operation goes through all the fields defined and encodes the values for those fields from the provided object into the target buffer.
iobuf = new Buffer([1,2,3,4,5,6,7]);
var ledef = binstruct.def().uint32le('val').uint16le('short').byte('b');
var ledata = ledef.read(iobuf);
assert.equal(0x04030201, ledata.val);
ledata.val = 0x05060708;
ledata.short = 0x090a;
ledata.b = 0xb;
assert.equal('01020304050607', iobuf.toString('hex'));
ledef.write(ledata, iobuf);
assert.equal('080706050a090b', iobuf.toString('hex'));
// Read/write at an offset
var iobuf2 = new Buffer([1,2,3,4,5,6,7,8,9,10,11]);
ledef.write(ledata, iobuf2, 2);
assert.equal(iobuf2.toString('hex'), '0102'+iobuf.toString('hex')+'0a0b');
// write() with no parameters creates a new buffer
assert.equal(ledef.write().toString('hex'), iobuf.toString('hex'));
When adding a numeric field you can specify a parameter with a "default" for that field. This will be applied during a write() operation if that field is not provided.
assert.equal(require('binstruct').def()
.byte(1)
.byte(2)
.write().toString('hex'),
'0102');
When using a wrapper you can request that the default values be written by calling writeValues() on the result of wrap();
var buf = new Buffer(2);
require('binstruct').def()
.byte(1)
.byte(2)
.wrap(buf)
.writeValues();
assert.equal(buf.toString('hex'), '0102');
When using a wrapper you can check whether the values in the buffer are equal to the default values specified. This can be helpful if you are looking for a certain "signature" in the file that identifies the file format or data structure you are working with.
assert.doesNotThrow(function() {
binstruct.def().uint16(0x0102).wrap(new Buffer([1,2])).checkValues();
});
assert.throws(function() {
binstruct.def().uint16(0x0102).wrap(new Buffer([5,6])).checkValues();
});
When defining a structure you are probably implementing something from a third party. As a sanity check, it's helpful to verify the structure you've defined is the right size. Do this by adding checkSize(size) to the end of your definition. For example:
require('binstruct').def()
.uint64('x')
.uint32('y') // Simulated copy/paste error using 32 instead of 64
.checkSize(16); // throws an error!
Install using npm:
npm install binstruct
- String and buffer fields
- Sub-structures
- Arrays of types
- Dynamically sized arrays, strings, buffers
- Computed values / sizes (using functions instead of constants)