chrome-dns-lookup

#!/usr/bin/env python

# /etc/hosts format
# 127.0.0.1 www.google.com

import os
import re
import shutil
import socket
import sqlite3
import urllib.parse

# For easy portability...
BASE_DIR = os.path.expanduser("~/Library/Application Support/Google/Chrome/Default/")
BOOKMARKS_FILENAME = "Bookmarks"
HISTORY_FILENAME = "History"
ALLOWED_SCHEMAS = ["http", "https", "ftp"]
IGNORED_HOSTS = ["localhost"]
HOSTS = set()

# If possible, set the socket timeout, in case of DNS problems
if hasattr(socket, 'setdefaulttimeout'):
    socket.setdefaulttimeout(2)

# Some functions
def get_ip(hostname):
    return socket.gethostbyname(hostname)

def get_bookmarks():
    try:
        bookmarks = open(BASE_DIR + BOOKMAKRS_FILENAME).read()
    except:
        return
    for url in re.findall('"url": "(.*?)"', bookmarks):
        try:
            url = urllib.parse.urlsplit(url)
        except:
            continue
        else:
            if url.scheme in ALLOWED_SCHEMAS and ".".join(url.hostname.split(".")[-2:]) not in IGNORED_HOSTS:
                HOSTS.add(url.hostname)

def get_history():
    history_file = BASE_DIR + HISTORY_FILENAME
    history_tmp = os.path.expanduser("~/.chrome.history.sqlite3")
    # Make a temporary copy of the file
    shutil.copy(history_file, history_tmp)
    try:
        con = sqlite3.connect(history_tmp)
        urls = con.execute('select distinct url from urls').fetchall()
    except:
        return
    for url in urls:
        try:
            url = urllib.parse.urlsplit(url[0])
        except:
            continue
        else:
            if url.scheme in ALLOWED_SCHEMAS and ".".join(url.hostname.split(".")[-2:]) not in IGNORED_HOSTS:
                HOSTS.add(url.hostname)
    con.close()
    os.remove(history_tmp)

# Gather urls
get_bookmarks()
get_history()

# Cycle through everything and gather ips
if HOSTS:
    # Sort hostnames by domain name
    HOSTS = list(HOSTS)
    HOSTS.sort(key=lambda x: ".".join(x.split(".")[-2:]))

    for hostname in HOSTS:
        # Make sure the hostname isn't an IP address
        if re.findall('^\d+\.\d+\.\d+\.\d+$', hostname):
            continue
        # Make sure the hostname isn't a bonjour address
        if hostname.endswith(".local"):
            continue
        try:
            ip = get_ip(hostname)
        except:
            continue
        print("%s\t%s" % (ip, hostname))