Dump all weave docs into a single MD file

weave_docs_dump.py

from dataclasses import dataclass
from pathlib import Path
import simple_parsing


# clone weave:
# git clone https://github.com/wandb/weave
WEAVE_DOCS_PATH = Path("weave/docs/docs")

@dataclass
class ScriptArgs:
    weave_docs_path: Path = WEAVE_DOCS_PATH
    append_file_path: bool = True
    outfile: Path = Path("all_weave_docs.md")

# let's iterate over all markdown files in the WEAVE_DOCS_PATH

def all_md(path: Path) -> list[Path]:
    all_md_files = [f for f in path.rglob("**/*.md")]
    print(f"Found {len(all_md_files)} markdown files")
    return all_md_files
            
def read_md(path: Path) -> str:
    with open(path, "r") as f:
        return f.read()


def all_weave_docs(weave_docs_path: Path, append_file_path: bool) -> str:
    all_md_files = all_md(weave_docs_path)
    all_content = ""
    for file in all_md_files:
        content = read_md(file)
        if append_file_path:
            all_content += f"filepath: docs/{file.relative_to(weave_docs_path)}\n\n"
        all_content += f"{content}\n\n"
    return all_content

if __name__=="__main__":
    args = simple_parsing.parse(ScriptArgs)
    all_content = all_weave_docs(args.weave_docs_path, args.append_file_path)
    with open(args.outfile, "w") as f:
        f.write(all_content)