nodeseek-cli

A command-line reader for NodeSeek forum threads.

NodeSeek serves each thread as fully server-rendered HTML, so nodeseek-cli fetches a post URL (no API key, no login) and converts it to Markdown, or to structured JSON with --json. It handles pagination, quoted replies, images, code blocks, tables, and NodeSeek's "magic tab" benchmark widgets.

Install / run

Run without installing (recommended):

uvx nodeseek-cli 355740

Or install:

uv tool install nodeseek-cli      # global CLI: `nodeseek` and `nodeseek-cli`
pip install nodeseek-cli          # into current environment

Usage

nodeseek <post> [options]

  <post>            bare id (355740), slug (post-355740-1), or full URL

  -p, --page N      fetch only page N (overrides page in the URL)
  -a, --all         fetch every page of the thread and merge
  -j, --json        emit structured JSON instead of markdown
      --ansi        keep ANSI colour codes inside code blocks (default: strip)
  -V, --version     print version
  -h, --help        show help

By default only one page is fetched (the page in the URL/slug, or page 1). The markdown header shows page X/N and a footer tells you how to reach the next page or the whole thread.

Examples

nodeseek 355740                                   # first page as markdown
nodeseek post-250906-2                             # a specific page
nodeseek 250906 --all                             # whole thread, merged
nodeseek 250906 --all --json | jq '.comments[].author'
nodeseek https://www.nodeseek.com/post-355740-1 --ansi

Formatting notes

• Stickers are meaningless out of context and render as [Sticker].
• Images render as markdown ![alt](url) with absolute URLs.
• Replies are detected from the leading @user #N marker and shown as a Reply to @user #N line (and a reply_to object in JSON); the marker is stripped from the body.
• Magic tab widgets (hardware/network benchmark reports, etc.) expand into one titled code block per tab. ANSI colour is reconstructed from NodeSeek's data-ansicode encoding and stripped by default (keep it with --ansi).

Library API

from nodeseek_cli import fetch_post, to_markdown, to_json

post = fetch_post("250906", all_pages=True)   # -> nodeseek_cli.models.Post
print(post.title, post.pages, len(post.comments))

for c in post.comments:
    print(c.floor, c.author, c.reply_to)

print(to_markdown(post))
print(to_json(post))

Parse already-downloaded HTML offline:

from nodeseek_cli import parse_page

title, max_page, comments = parse_page(html, ansi=False)

Errors

fetch_post / fetch_page raise PostNotFound (404), RateLimited (429), or the base NodeSeekError. parse_target raises ValueError on an unparseable reference.

Development

uv sync --extra dev
uv run pytest            # offline tests against saved fixtures
uv run nodeseek 355740   # run the CLI from source

Linting and formatting use ruff, wired up through prek (a drop-in pre-commit runner):

uv run ruff format .     # format
uv run ruff check .      # lint
uv run prek install      # install the git pre-commit hook
uv run prek run --all-files

License

MIT