205 lines
7.5 KiB
Python
Executable File
205 lines
7.5 KiB
Python
Executable File
#!/usr/local/bin/python3
|
||
|
||
# tdp.py - tilde data in tilde data protocol format.
|
||
# Copyright 2015 Michael F. Lamb <http://datagrok.org>
|
||
# License: GPLv3+
|
||
|
||
"""
|
||
Outputs JSON data conforming to "~dp (Tilde Description Protocol)" as defined
|
||
at: http://protocol.club/~datagrok/beta-wiki/tdp.html
|
||
|
||
It is a JSON structure of the form:
|
||
|
||
{
|
||
'name': (string) the name of the server.
|
||
'url': (string) the URL of the server.
|
||
'signup_url': (string) the URL of a page describing the process required to request an account on the server.
|
||
'user_count': (number) the number of users currently registered on the server.
|
||
'want_users': (boolean) whether the server is currently accepting new user requests.
|
||
'admin_email': (string) the email address of the primary server administrator.
|
||
'description': (string) a free-form description for the server.
|
||
'users': [ (array) an array of users on the server.
|
||
{
|
||
'username': (string) the username of the user.
|
||
'title': (string) the HTML title of the user’s index.html page.
|
||
'mtime': (number) a timestamp representing the last time the user’s index.html was modified.
|
||
},
|
||
...
|
||
]
|
||
}
|
||
|
||
We also overload this with the preexisting data format we were using in
|
||
/var/local/tildetown/tildetown-py/stats.py, which is of the form:
|
||
|
||
{
|
||
'all_users': [ (array) of users on the server.
|
||
{
|
||
'username': (string) the username of the user.
|
||
'default': (boolean) Is the user still using their unmodified default index.html?
|
||
'favicon': (string) a url to an image representing the user
|
||
},
|
||
...
|
||
]
|
||
'num_users': (number) count of all_users
|
||
'live_users': [ (array) an array of live users, same format as all_users. Users may appear in both arrays.
|
||
...
|
||
],
|
||
'num_live_users': (number) count of live users
|
||
'active_user_count': (number) count of currently logged in users
|
||
'generated_at': (string) the time this JSON was generated in '%Y-%m-%d %H:%M:%S' format.
|
||
'generated_at_msec': (number) the time this JSON was generated, in milliseconds since the epoch.
|
||
'site_name': (same as 'name' above)
|
||
'site_url': (same as 'url' above)
|
||
'uptime': (string) output of `uptime -p`
|
||
|
||
}
|
||
Usage: tdp.py > /var/www/html/tilde.json
|
||
"""
|
||
|
||
# I suppose I could import /var/local/tildetown/tildetown-py/stats.py which
|
||
# does much of the same work, but I wanted to try to make one that needs no
|
||
# venv nor 'sh' module. (Success.) Bonus: this runs in 0.127s, vs 5.2s
|
||
# for 'stats'
|
||
|
||
# FIXME: unlike stats.py, we calculate last modified only on index.html.
|
||
|
||
# FIXME: we output quite a bit of redundant data. I think we should lose
|
||
# 'live_users' and do that filtering on the client side.
|
||
|
||
# FIXME: If we're the only consumer of the stats.py data, let's change the
|
||
# client side to use 'users' and drop 'all_users'.
|
||
|
||
import datetime
|
||
import hashlib
|
||
import json
|
||
import os
|
||
import pwd
|
||
import re
|
||
import struct
|
||
import subprocess
|
||
|
||
title_re = re.compile(r'<title[^>]*>(.*)</title>', re.DOTALL)
|
||
defaultindex_hash = None
|
||
|
||
# modified from https://gist.github.com/likexian/f9da722585036d372dca
|
||
XTMP_STRUCT_FMT = 'hi32s4s32s256shhiii4i20x'
|
||
XTMP_STRUCT_SIZE = struct.calcsize(XTMP_STRUCT_FMT)
|
||
XTMP_STRUCT_KEYS = [
|
||
'type', 'pid', 'line', 'id', 'user', 'host', 'e_termination', 'e_exit',
|
||
'session', 'sec', 'usec', 'addr_v6', 'unused',
|
||
]
|
||
|
||
def read_xtmp(filename):
|
||
"""Pure-python replacement for who(1) and w(1); parses the data structure
|
||
in /var/run/utmp or /var/run/wtmp, generating a dict for each entry. See
|
||
man 5 utmp for meaning of fields.
|
||
|
||
"""
|
||
# This was fun but probably not worth the trouble, since we end up having
|
||
# to use subprocess.check_output() elsewhere anyway.
|
||
with open(filename, 'rb') as fp:
|
||
for entry in iter((lambda: fp.read(XTMP_STRUCT_SIZE)), b''):
|
||
yield dict(zip(
|
||
XTMP_STRUCT_KEYS,
|
||
(i.decode('UTF-8').partition('\x00')[0] if hasattr(i, 'partition') else i
|
||
for i in struct.unpack(XTMP_STRUCT_FMT, entry))))
|
||
|
||
def active_user_count():
|
||
"""Return the count of unique usernames logged in."""
|
||
return len(set(r['user'] for r in read_xtmp('/var/run/utmp') if r['type'] == 7))
|
||
|
||
def md5sum(filename):
|
||
"""Return the md5 hash of the contents of filename as a hexidecimal string."""
|
||
# This doesn't slurp the whole file in; it reads 4k at a time.
|
||
h = hashlib.md5()
|
||
with open(filename, 'rb') as fp:
|
||
for data in iter((lambda: fp.read(4096)), b''):
|
||
h.update(data)
|
||
return h.hexdigest()
|
||
|
||
def get_title(indexhtml):
|
||
"""Given an html file, return the content of its <title>"""
|
||
print(indexhtml)
|
||
fp = open(indexhtml, 'rt', errors='ignore')
|
||
title = title_re.search(fp.read())
|
||
if title:
|
||
return title.group(1)
|
||
|
||
def get_users():
|
||
"""Generate tuples of the form (username, homedir) for all normal
|
||
users on this system.
|
||
|
||
"""
|
||
return ((p.pw_name, p.pw_dir) for p in pwd.getpwall() if
|
||
p.pw_uid >= 1000 and
|
||
p.pw_shell != '/bin/false' and
|
||
p.pw_name not in ['nobody', 'ubuntu', 'poetry'])
|
||
|
||
def tdp_user(username, homedir):
|
||
"""Given a unix username, and their home directory, return a TDP format
|
||
dict with information about that user.
|
||
|
||
"""
|
||
indexhtml = os.path.join(homedir, 'public_html', 'index.html')
|
||
return {
|
||
'username': username,
|
||
'title': get_title(indexhtml),
|
||
'mtime': int(os.path.getmtime(indexhtml) * 1000),
|
||
# tilde.town extensions and backward compatibility
|
||
# FIXME: just shelling out to diff -q might be way faster than all
|
||
# these hashes.
|
||
'default': md5sum(indexhtml) == defaultindex_hash,
|
||
'favicon': 'TODO',
|
||
}
|
||
|
||
def tdp():
|
||
now = datetime.datetime.now()
|
||
users = [tdp_user(username, homedir) for username, homedir in get_users()]
|
||
|
||
# TDP format data
|
||
data = {
|
||
'name': 'tilde.town',
|
||
'url': 'http://tilde.town',
|
||
'signup_url': 'http://goo.gl/forms/8IvQFTDjlo',
|
||
'want_users': True,
|
||
'admin_email': 'nks@lambdaphil.es',
|
||
'description': " ".join(l.strip() for l in """
|
||
an intentional digital community for creating and sharing works of
|
||
art, educating peers, and technological anachronism. we are a
|
||
completely non-commercial, donation supported, and committed to
|
||
rejecting false technological progress in favor of empathy and
|
||
sustainable computing.
|
||
""".splitlines()),
|
||
'user_count': len(users),
|
||
'users': users,
|
||
}
|
||
|
||
# tilde.town extensions and backward compatibility
|
||
data.update({
|
||
'active_user_count': active_user_count(),
|
||
'generated_at': now.strftime('%Y-%m-%d %H:%M:%S'),
|
||
'generated_at_msec': int(now.timestamp() * 1000),
|
||
'uptime': subprocess.check_output(['uptime', '-p'], universal_newlines=True),
|
||
})
|
||
# redundant entries we should drop after changing homepage template
|
||
data.update({
|
||
'all_users': data['users'],
|
||
'num_users': data['user_count'],
|
||
'live_users': [u for u in data['users'] if not u['default']],
|
||
'site_name': data['name'],
|
||
'site_url': data['url'],
|
||
})
|
||
data.update({
|
||
'num_live_users': len(data['live_users']),
|
||
})
|
||
|
||
return data
|
||
|
||
def main():
|
||
global defaultindex_hash
|
||
defaultindex_hash = md5sum("/etc/skel/public_html/index.html")
|
||
print(json.dumps(tdp(), sort_keys=True, indent=2))
|
||
|
||
if __name__ == '__main__':
|
||
raise SystemExit(main())
|