scott-griffiths / bitformat

A Python module for creating and parsing file formats, especially at the bit rather than byte level.

About me

Currently working at Graphcore in Bristol. We're making chips for AI. It's a cool place to be.

---

I'm still maintaining and improving the Python bitstring package.

   

---

My new project is a bitwise format parsing and building library called bitformat. I'm aiming for a first alpha release later in 2024.

   

---

I've not got much else on GitHub except my quixotic idea of a clock based on a Go board.

bitformat's Issues

Add a fieldtype that represents a series of fields of known length

This is really an efficiency issue, but looking at code like this:

    Repeat(1000, Format([
        'u8 <num1>',
        'uintle24 <num2>',
        'bool <bool1>',
        'u3 <num4>',
        'u4',
        'i8 * 3 <fixedarray1>',
        ]), 'items')

a lot of time is going to be spent parsing lots of small fields that we already know the lengths of.

Perhaps internally we could create a new type (let's call it Blob for now) that represents all the contiguous fields of known length. It also contains the internal lengths and names and can be transparently interrogated as if it's not present.

    Repeat(1000, Format([
        Blob(64, list_of_fields)
    ]), 'items')

but the Blob will print the same as its constituents so will look the same to the user. When being parsed it can just read its whole length for efficiency - it can later still be interrogated for its field information.