Architecture independent C library for packing and unpacking integers and floats of varying precision into and from arbitrary bit-wise locations within a byte array, supporting various types of byte and bit ordering.
Note: This library is in no way finished, the description here is more of a wishlist.
Example use case: you have been given a device which talks in a weird binary protocol where values are packed tightly into bit fields in big endian byte order. After successfully connecting to the device you are receiving messages which end up in a byte array on your Intel (little endian) machine and you want to pluck out the values in a logical manner.
How it works:
- The library contains a set of functions which can be used to pick out various size integers and floats from a given byte array.
- Each function notably takes a pointer to a byte array, a bit offset and a bit length of the value in question as well as byte ordering information.
- The passed buffer is treated as a large arbitrary precision integer which means bit offsets larger than 64 can be used.
- For Motorola (big endian) byte order, the bit offset is counted starting at the LSbit of the last byte in the array, counting forward through the bytes.
- For Intel (little endian) byte order, the bit offset is counted starting at the LSbit of the first byte in the array, counting backward through the bytes.
This library was originally designed for packing and parsing CAN messages to and from weird automotive devices. It's not meant to be fast or efficient, just flexible.
- install cmocka
# apt-get install libcmocka-dev(tested on Ubuntu 14.04) - make && make test
- Floating point values, float and double
- 8,16,32 bit precision integers
- See what gcc optimizes out by looking at the generated assembly
- Performance tests & improvements