Fetch Hacker News threads by ID and recursively find my comments.

hn.py

#!/usr/bin/env python3

import html
import json
import os
import re
import ssl
import sys
from urllib.request import Request, urlopen

USERNAMES = {"modinfo"}
CA_FILE = "/etc/ssl/certs/ca-certificates.crt"
SSL_CONTEXT = ssl.create_default_context(
    cafile=CA_FILE if os.path.exists(CA_FILE) else None
)


def fetch(ident):
    request = Request(
        f"https://hn.algolia.com/api/v1/items/{ident}",
        headers={"User-Agent": "hn.py/1.0"},
    )
    with urlopen(request, timeout=30, context=SSL_CONTEXT) as response:
        return json.load(response)


def clean(text):
    text = html.unescape(re.sub(r"<[^>]+>", " ", text or ""))
    return re.sub(r"\s+", " ", text).strip()


def print_user_comments(comments):
    for comment in comments:
        if comment.get("author") in USERNAMES:
            print(
                "USER COMMENT",
                comment["author"],
                comment.get("id"),
                ":",
                clean(comment.get("text")),
            )
        print_user_comments(comment.get("children", []))


def main(ids):
    if not ids or any(not ident.isdigit() for ident in ids):
        raise SystemExit(f"Usage: {sys.argv[0]} <hn_id> [hn_id ...]")

    for ident in ids:
        item = fetch(ident)
        print(f"\n=== {ident} ===")
        print("title:", item.get("title"))
        print("author:", item.get("author"))
        print("points:", item.get("points"))
        print("url:", item.get("url"))
        print("top comments:", len(item.get("children", [])))
        print_user_comments(item.get("children", []))


if __name__ == "__main__":
    main(sys.argv[1:])