A small program that chunks up program files to be loaded into chatgpt. 1/2 written by chatgpt. Inspired by https://github.com/mpoon/gpt-repository-loader. Note that this has limitations as chatgpt doesn't seem to remember past 32KB-ish of data. This is probably documented somewhere but I haven't looked it up.
Ideally you could load up all the chunk{num}.txts and query chatgpt questions in plain english as a tool to learn a new codebase.
git clone https://github.com/zricethezav/go2chatgpt.git
cd go2chatgpt
go build (install go dependencies)
Usage: go2chatgpt [options] <source> <output_folder>
-chunksize int
Chunk size in KB (GPT-4 max is 16KB) (default 13)
-exclude string
Comma-separated list of glob patterns to exclude
-include string
Comma-separated list of glob patterns to include
See /examples
- ignores non-text files (binaries, jpgs, etc)
- ignores
{report-path}/chunk{num}.txtfiles