Binn

Binn is a binary data serialization format designed to be compact, fast and easy to use.

Performance

The elements are stored with their sizes to increase the read performance.

The library uses zero-copy when reading strings, blobs and containers.

The strings are null terminated so when read the library returns a pointer to them inside the buffer, avoiding memory allocation and data copying.

Data Types

The Binn format supports all these:

Primitive data types:

• null
• boolean (true and false)
• integer (up to 64 bits signed or unsigned)
• floating point numbers (IEEE single and double precision)
• string
• blob (binary data)
• user defined

Containers:

• list
• map (numeric key associative array)
• object (text key associative array)

Format

The elements are stored in this way:

boolean, null:
[type]

int, float (storage: byte, word, dword or qword):
[type][data]

string, blob:
[type][size][data]

list, object, map:
[type][size][count][data]

Example Structure

A json data such as {"hello":"world"} is serialized in binn as:

  \xE2           // type = object (container)
  \x11           // container total size
  \x01           // key/value pairs count
  \x05hello      // key
  \xA0           // type = string
  \x05world\x00  // value (null terminated)

You can check the complete specification

Usage Example

Writing

binn *obj;

// create a new object
obj = binn_object();

// add values to it
binn_object_set_int32(obj, "id", 123);
binn_object_set_str(obj, "name", "John");
binn_object_set_double(obj, "total", 2.55);

// send over the network or save to a file...
send(sock, binn_ptr(obj), binn_size(obj));

// release the buffer
binn_free(obj);

Reading

int id;
char *name;
double total;

id = binn_object_int32(obj, "id");
name = binn_object_str(obj, "name");
total = binn_object_double(obj, "total");

More examples

You can find more usage examples here and in the examples folder

Wrappers

• Javascript: liteserver/binn.js
• PHP: ET-NiK/binn-php
• Python: meeron/pybinn

Feel free to make a wrapper for your preferred language. Then inform us so we can list it here.

How to use

1. Including the binn.c file in your project; or
2. Linking to the binn library:

On Linux and MacOSX:

gcc myapp.c -lbinn

On Windows:

Include the binn-1.0.lib in your MSVC project or use MinGW:

gcc myapp.c -lbinn-1.0

Compiling the Library

On Linux and MacOSX:

git clone https://github.com/liteserver/binn
cd binn
make
sudo make install

It will create the file libbinn.so.1.0 on Linux and libbinn.1.dylib on MacOSX

On Windows:

Use the included Visual Studio project in the src/win32 folder or compile it using MinGW:

git clone https://github.com/liteserver/binn
cd binn
make

Both will create the file binn-1.0.dll

On Android:

Check for pre-compiled binaries in the android-binn-native project

Regression Tests

On Linux, MacOSX and Windows (MinGW):

cd binn
make test

On Windows (Visual Studio):

Use the included project in the test/win32 folder

Reliability

The current version (1.0) is stable and production ready

As it is cross-platform, data can be transferred between little-endian and big-endian devices

Licence

Apache 2.0

Contact

Questions, suggestions, support: contact AT litereplica DOT io