I’ve added human.json to my website
Evan Hahn recently blogged about adding support for the “human.json protocol” to his website. I read the specification and thought this seemed like a straightforward protocol to implement. I've followed along, and added a /human.json file to my website and rel="human-json" in a <link> element to the <head> section of my HTML. Easy!
Vouching was a little more involved because I wanted to auto-discover who uses human.json from everyone that I follow using RSS. Then my vouches can be kept up-to-date as more people implement the protocol. I updated the script to parse the <link> element properly using BeautifulSoup (thanks Evan!). The script below is what I ended up with:
import contextlib import pathlib import json import datetime import re import urllib import opml import urllib3 import bs4 http = urllib3.PoolManager( headers={ # Be a good internet netizen: # always set a User-Agent with # your domain and email. "User-Agent": "sethmlarson.dev/1.0 (sethmichaellarson@gmail.com)", "Accept": "application/json", }, retries=0, timeout=5, ) today = datetime.date.today().strftime("%Y-%m-%d") human_json_path = pathlib.Path("app/static/human.json") human_json = json.loads(human_json_path.read_text()) vouched_urls = {vouch["url"] for vouch in human_json["vouches"]} doc = opml.OpmlDocument.load("archive/feeds.opml") for outline in doc.outlines: mat = re.search(r"^(https?://[^/]+)(?:/|$)", outline.html_url or "") if not mat: continue base_url = mat.group(1) urls_to_try = [f"{base_url}/human.json"] with contextlib.suppress(Exception): # Find <link rel="human-json" ...> resp = http.request("GET", base_url) html = bs4.BeautifulSoup(resp.data, "html.parser") for el in html.find_all( name="link", attrs={"rel": "human-json"}, recursive=True ): if href := el.attrs.get("href", None): urls_to_try.append(urllib.parse.urljoin(base_url, href)) # Try all candidate URLs. for url in urls_to_try: with contextlib.suppress(Exception): resp = http.request("GET", url) if resp.status < 300 and set(resp.json()) == {"version", "vouches", "url"}: # Use the URL that the domain specifies, not # our own. This helps with canonicalization, # (ie do we use 'www.example.com' or 'example.com'?) human_json_url = resp.json()["url"] if human_json_url in vouched_urls: continue vouched_urls.add(human_json_url) human_json["vouches"].append({"url": human_json_url, "date": today}) continue human_json_path.write_text(json.dumps(human_json, indent=2))
Running this script turned up with two websites that I follow already supporting the protocol:
- https://evanhahn.com (human.json)
- https://foosel.net (human.json)
Maybe this post will inspire you to add support to your own website? If you do and I follow you via RSS, your website will appear automatically after I regenerate my OPML files. If we're mutuals on Mastodon or Bluesky give me a ping and I'll add you sooner.
I'm probably not going to use the browser extension myself, but if others are using the data from this "network" then that is a win.
Thanks for keeping RSS alive! ♥
Discussion in the ATmosphere