tilde.chat/gen_stats

#!/usr/bin/env python3
import base64
import json
from lxml import etree

WORK_DIR = "/var/www/tilde.chat"
out = {}
parser = etree.XMLParser(strip_cdata=False)
root = etree.parse("http://localhost:8081/stats", parser)

assert root.getroot().tag == "inspircdstats"

with open(f"{WORK_DIR}/blacklist", "r") as f:
    BLACKLIST = f.read().splitlines()


def unsanitize(node, default=""):
    # workaround for weird behavior in insp's xml output
    # https://github.com/inspircd/inspircd/blob/v3.7.0/src/modules/m_httpd_stats.cpp#L55
    if node.text is None or node.text == "":
        return default
    elif str(etree.tostring(node)).startswith(f"b'<{node.tag}><![CDATA["):
        missing_padding = len(node.text) % 4
        if missing_padding:
            v = node.text + "=" * (4 - missing_padding)
        else:
            v = node.text
        return base64.b64decode(v).decode("utf-8")

    return node.text


out["usercount"] = int(root.find("./general/usercount").text)
out["channelcount"] = int(root.find("./general/channelcount").text)

out["channels"] = []
for c in root.findall("./channellist/channel"):
    name = unsanitize(c.find("channelname"))

    modes = c.find("./channelmodes").text
    if (
        # skip channels in the blacklist or with mode +s
        name in BLACKLIST
        or modes is None
        or "s" in modes.split()[0]
    ):
        continue

    out["channels"].append(
        dict(
            name=name,
            topic=unsanitize(c.find("./channeltopic/topictext"), "no topic set"),
            usercount=int(c.find("usercount").text),
            webchatlink="https://kiwi.tilde.chat/" + name,
        )
    )

out["channels"].sort(key=lambda x: x["name"].lower())

with open(f"{WORK_DIR}/stats.json", "w") as f:
    json.dump(out, f)