Send Virgin Media SuperHub 2 DOCSIS Stats to StatsD

vmparse.py

#!/usr/bin/env python
#
# Extracts channel information from Virgin Media Super Hub 2 devices
# and exports StatsD measurements.
#

import statsd
import urllib2
import re
from bs4 import BeautifulSoup

# Define the measurements root name
MEAS_ROOT = 'vm'

# Define the StatsD host
STATSD_HOST = 'mystatsdhost.example.com'
STATSD_PORT = 8125

# Convert a "nice" string into a measurement key
def name_to_meas_key(name):
    without_brackets = re.sub('\s?\(.*\)', '', name.strip() if name is not None else 'unknown')
    without_spaces = re.sub('\s', '_', without_brackets)
    return without_spaces.lower()

# Define URLs for the various information pages
stat_urls = {
    'ds': 'http://192.168.100.1/RouterStatus_downstream.html',
    'us': 'http://192.168.100.1/RouterStatus_upstream.html'
}

# Transmit measurements to StatsD
statsd_client = statsd.StatsClient(host=STATSD_HOST, port=STATSD_PORT)

# For each statistics page, generate measurements
for (stat_root, url) in stat_urls.iteritems():
    
    # Download the status page
    response = urllib2.urlopen(url)
    stat_soup = BeautifulSoup(response.read(), 'html.parser')

    # Each row is a measurement, each column (apart from the first) is a channel
    for row_idx, el_row in enumerate(stat_soup.find_all('tr')):
        for cell_idx, el_cell in enumerate(el_row.find_all('td')):
            cell_str = ' '.join(el_cell.stripped_strings)

            # The first cell contains the name of the measurement shown in this row
            if cell_idx == 0:
                meas_id = name_to_meas_key(cell_str)
                continue

            # Further cells contain the measurement values for the channels. Store each of them
            meas_key = MEAS_ROOT + '_' + stat_root + '_' + meas_id + '_' + str(cell_idx - 1)
            print 'Send: %28s -> %s' % (meas_key, cell_str)
            statsd_client.gauge(meas_key, cell_str)