Cohost Export

cohost.py

import json
from pathlib import Path
import argparse

"""
This is a very simple script to extract raw Markdown files from a Cohost export.
Feel free to modify as needed!

Usage:

Given that you've extracted your Cohost export to a directory like "export-123-12345"
you can run this script like this:

    python cohost.py export-123-12345

The script will output a "posts" folder in the same directory, containing the Markdown
of your posts (both drafts and published posts).
"""

parser = argparse.ArgumentParser(
    prog="Cohost export",
    description="Extracts the raw Markdown files from a Cohost export.",
)
parser.add_argument("source", help="The unzipped Cohost export directory")
args = parser.parse_args()

output_dir = Path("./posts")
export_dir = Path(args.source)
for user_dir in (export_dir / "project").iterdir():
    user_output_dir = output_dir / (user_dir.parts[-1])
    for subdir in ["published", "draft"]:
        posts_dir = Path(user_dir / "posts") / subdir
        user_output_subdir = user_output_dir / subdir
        user_output_subdir.mkdir(parents=True, exist_ok=True)
        for post in posts_dir.iterdir():
            post_json = post / "post.json"
            with open(post_json) as f:
                post_data = json.load(f)
            title = post_data["headline"] or "Untitled"
            slug = post.parts[-1]
            content = "\n\n".join(
                block["markdown"]["content"]
                for block in post_data["blocks"]
                if block["type"] == "markdown"
            )
            with open(str(user_output_subdir / slug) + ".md", "w") as f:
                f.writelines([f"# {title}", "\n\n"])
                f.write(content)