How to use the tldextract.TLDExtract function in tldextract

To help you get started, we’ve selected a few tldextract examples, based on popular ways it is used in public projects.

Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.

github istresearch / scrapy-cluster / kafka-monitor / plugins / scraper_handler.py View on Github external
def setup(self, settings):
        '''
        Setup redis and tldextract
        '''
        self.extract = tldextract.TLDExtract()
        self.redis_conn = redis.Redis(host=settings['REDIS_HOST'],
                                      port=settings['REDIS_PORT'],
                                      db=settings.get('REDIS_DB'))

        try:
            self.redis_conn.info()
            self.logger.debug("Connected to Redis in ScraperHandler")
        except ConnectionError:
            self.logger.error("Failed to connect to Redis in ScraperHandler")
            # plugin is essential to functionality
            sys.exit(1)
github vecna / trackmap / manyutils.py View on Github external
urldict = {}
    skipped = 0

    for urldir in os.listdir(outputdir):

        if urldir in INFOFILES:
            continue

        try:
            urlfile = os.path.join(outputdir, urldir, '__urls')
            related_urls = get_unique_urls(urldir, urlfile)
        except IOError or OSError as einfo:
            print "Unable to read", urldir, einfo, "skipping"
            continue

        TLDio = TLDExtract(cache_file='mozilla_tld_file.dat')
        for dirty_url in related_urls:
            # dirty_url because may contain ":"

            if dirty_url.split(':') != -1:
                url = dirty_url.split(':')[0]
            else:
                url = dirty_url

            if urldict.has_key(url):
                skipped +=1
                continue

            dnsplit= TLDio(url)
            urldict.update({url : {
                    'domain' : dnsplit.domain,
                    'tld' : dnsplit.suffix,
github mozilla / fjord / fjord / base / domain.py View on Github external
# from the host or the host from the port. If there's a scheme we
    # want to limit it to http or https.
    if ':' in url:
        scheme, host = url.split(':', 1)
        # If there's a . in the scheme, then there wasn't a scheme
        # and the : is delimiting the host from the port
        if '.' not in scheme and scheme not in ('http', 'https'):
            return u''

    # Get a thread-local extractor if there is one. If not, create it.
    extractor = getattr(_cached_tldextract, 'extractor', None)
    if extractor is None:
        # FIXME - This uses the tld set included with tldextract which
        # will age over time. We should fix this so that we get a new
        # file on deployment and use that file.
        extractor = tldextract.TLDExtract(
            suffix_list_url=None,  # disable fetching the file via http
        )
        _cached_tldextract.extractor = extractor

    res = extractor(url)

    # If there's no tld, then this is probably an ip address or
    # localhost. Also ignore .mil and .arpa addresses.
    if res.suffix in ('', 'mil', 'in-addr.arpa'):
        return u''

    # Suffix is the tld. We want that plus the next level up.
    return res.domain.decode('utf-8') + u'.' + res.suffix.decode('utf-8')
github golismero / golismero / golismero / api / net / web_utils.py View on Github external
>>> d = ParsedURL("http://www.example.com/")
    >>> d.split_hostname()
    ('www', 'example', 'com')
    >>> d = ParsedURL("http://some.subdomain.of.example.co.uk/")
    >>> d.split_hostname()
    ('some.subdomain.of', 'example', 'co.uk')
    >>> '.'.join(d.split_hostname())
    'some.subdomain.of.example.co.uk'

    :param hostname: Hostname to split.
    :type hostname: str

    :returns: Subdomain, domain and TLD.
    :rtype: tuple(str, str, str)
    """
    extract = TLDExtract(fetch = False)
    result  = extract( to_utf8(hostname) )
    return result.subdomain, result.domain, result.suffix
github fanmatics / metadoc / metadoc / domain / domaintools.py View on Github external
def get_domain(self, url):
    no_fetch_extract = tldextract.TLDExtract(suffix_list_urls=None)
    tld = no_fetch_extract(url)
    self.domain = "{}.{}".format(tld.domain, tld.suffix)
github shestera / django-multisite / multisite / management / commands / update_public_suffix_list.py View on Github external
def handle_noargs(self, **options):
        self.setup_logging(verbosity=options.get('verbosity', 1))

        filename = getattr(
            settings, 'MULTISITE_PUBLIC_SUFFIX_LIST_CACHE',
            os.path.join(tempfile.gettempdir(), 'multisite_tld.dat')
        )
        self.log("Updating {filename}".format(filename=filename))

        extract = tldextract.TLDExtract(cache_file=filename)
        extract.update(fetch_now=True)
        self.log("Done.")
github ctxis / CAPE / lib / cuckoo / common / abstracts.py View on Github external
def advanced_url_parse(self, url):
        if HAVE_TLDEXTRACT:
            EXTRA_SUFFIXES = ('bit',)
            parsed = False
            try:
                parsed = tldextract.TLDExtract(extra_suffixes=EXTRA_SUFFIXES, suffix_list_urls=None)(url)
            except Exception as e:
                log.error(e)
            return parsed
        else:
            log.info("missed tldextract dependency")
github jakeogh / dnsgate / dnsgate / global_vars.py View on Github external
OUTPUT_FILE_PATH_NAME
DEFAULT_REMOTE_BLACKLISTS = [
    'http://winhelp2002.mvps.org/hosts.txt',
    'http://someonewhocares.org/hosts/hosts',
    'https://raw.githubusercontent.com/jmdugan/blocklists/master/corporations/facebook/all']
ALL_REMOTE_BLACKLISTS = [
    'http://winhelp2002.mvps.org/hosts.txt',
    'http://someonewhocares.org/hosts/hosts',
    'https://adaway.org/hosts.txt',
    'https://raw.githubusercontent.com/StevenBlack/hosts/master/data/StevenBlack/hosts',
    'http://www.malwaredomainlist.com/hostslist/hosts.txt',
    'http://pgl.yoyo.org/adservers/serverlist.php?hostformat=hosts;showintro=0']
    # http://hosts-file.net/?s=Download

CACHE_EXPIRE = 3600*24*2 # 48 hours
TLD_EXTRACT = tldextract.TLDExtract(cache_file=TLDEXTRACT_CACHE)
github shmilylty / OneForAll / oneforall / common / domain.py View on Github external
def extract(self):
        """
        域名导出

        >>> d = Domain('www.example.com')
        
        >>> d.extract()
        ExtractResult(subdomain='www', domain='example', suffix='com')

        :return: 导出结果
        """
        extract_cache_file = config.data_storage_path.joinpath('public_suffix_list.dat')
        tldext = tldextract.TLDExtract(extract_cache_file)
        result = self.match()
        if result:
            return tldext(result)
        else:
            return None

tldextract

Accurately separates a URL's subdomain, domain, and public suffix, using the Public Suffix List (PSL). By default, this includes the public ICANN TLDs and their exceptions. You can optionally support the Public Suffix List's private domains as well.

BSD-3-Clause
Latest version published 17 days ago

Package Health Score

86 / 100
Full package analysis

Similar packages