MoFin/venv/lib/python3.12/site-packages/newspaper/urls.py

# -*- coding: utf-8 -*-
"""
Newspaper treats urls for news articles as critical components.
Hence, we have an entire module dedicated to them.
"""
__title__ = 'newspaper'
__author__ = 'Lucas Ou-Yang'
__license__ = 'MIT'
__copyright__ = 'Copyright 2014, Lucas Ou-Yang'

import logging
import re

from urllib.parse import parse_qs, urljoin, urlparse, urlsplit, urlunsplit

from tldextract import tldextract

log = logging.getLogger(__name__)


MAX_FILE_MEMO = 20000

_STRICT_DATE_REGEX_PREFIX = r'(?<=\W)'
DATE_REGEX = r'([\./\-_]{0,1}(19|20)\d{2})[\./\-_]{0,1}(([0-3]{0,1}[0-9][\./\-_])|(\w{3,5}[\./\-_]))([0-3]{0,1}[0-9][\./\-]{0,1})?'
STRICT_DATE_REGEX = _STRICT_DATE_REGEX_PREFIX + DATE_REGEX

ALLOWED_TYPES = ['html', 'htm', 'md', 'rst', 'aspx', 'jsp', 'rhtml', 'cgi',
                 'xhtml', 'jhtml', 'asp', 'shtml']

GOOD_PATHS = ['story', 'article', 'feature', 'featured', 'slides',
              'slideshow', 'gallery', 'news', 'video', 'media',
              'v', 'radio', 'press']

BAD_CHUNKS = ['careers', 'contact', 'about', 'faq', 'terms', 'privacy',
              'advert', 'preferences', 'feedback', 'info', 'browse', 'howto',
              'account', 'subscribe', 'donate', 'shop', 'admin']

BAD_DOMAINS = ['amazon', 'doubleclick', 'twitter']


def remove_args(url, keep_params=(), frags=False):
    """
    Remove all param arguments from a url.
    """
    parsed = urlsplit(url)
    filtered_query = '&'.join(
        qry_item for qry_item in parsed.query.split('&')
        if qry_item.startswith(keep_params)
    )
    if frags:
        frag = parsed[4:]
    else:
        frag = ('',)

    return urlunsplit(parsed[:3] + (filtered_query,) + frag)


def redirect_back(url, source_domain):
    """
    Some sites like Pinterest have api's that cause news
    args to direct to their site with the real news url as a
    GET param. This method catches that and returns our param.
    """
    parse_data = urlparse(url)
    domain = parse_data.netloc
    query = parse_data.query

    # If our url is even from a remotely similar domain or
    # sub domain, we don't need to redirect.
    if source_domain in domain or domain in source_domain:
        return url

    query_item = parse_qs(query)
    if query_item.get('url'):
        # log.debug('caught redirect %s into %s' % (url, query_item['url'][0]))
        return query_item['url'][0]

    return url


def prepare_url(url, source_url=None):
    """
    Operations that purify a url, removes arguments,
    redirects, and merges relatives with absolutes.
    """
    try:
        if source_url is not None:
            source_domain = urlparse(source_url).netloc
            proper_url = urljoin(source_url, url)
            proper_url = redirect_back(proper_url, source_domain)
            # proper_url = remove_args(proper_url)
        else:
            # proper_url = remove_args(url)
            proper_url = url
    except ValueError as e:
        log.critical('url %s failed on err %s' % (url, str(e)))
        proper_url = ''

    return proper_url


def valid_url(url, verbose=False, test=False):
    """
    Is this URL a valid news-article url?

    Perform a regex check on an absolute url.

    First, perform a few basic checks like making sure the format of the url
    is right, (scheme, domain, tld).

    Second, make sure that the url isn't some static resource, check the
    file type.

    Then, search of a YYYY/MM/DD pattern in the url. News sites
    love to use this pattern, this is a very safe bet.

    Separators can be [\.-/_]. Years can be 2 or 4 digits, must
    have proper digits 1900-2099. Months and days can be
    ambiguous 2 digit numbers, one is even optional, some sites are
    liberal with their formatting also matches snippets of GET
    queries with keywords inside them. ex: asdf.php?topic_id=blahlbah
    We permit alphanumeric, _ and -.

    Our next check makes sure that a keyword is within one of the
    separators in a url (subdomain or early path separator).
    cnn.com/story/blah-blah-blah would pass due to "story".

    We filter out articles in this stage by aggressively checking to
    see if any resemblance of the source& domain's name or tld is
    present within the article title. If it is, that's bad. It must
    be a company link, like 'cnn is hiring new interns'.

    We also filter out articles with a subdomain or first degree path
    on a registered bad keyword.
    """
    # If we are testing this method in the testing suite, we actually
    # need to preprocess the url like we do in the article's constructor!
    if test:
        url = prepare_url(url)

    # 11 chars is shortest valid url length, eg: http://x.co
    if url is None or len(url) < 11:
        if verbose: print('\t%s rejected because len of url is less than 11' % url)
        return False

    r1 = ('mailto:' in url)  # TODO not sure if these rules are redundant
    r2 = ('http://' not in url) and ('https://' not in url)

    if r1 or r2:
        if verbose: print('\t%s rejected because len of url structure' % url)
        return False

    path = urlparse(url).path

    # input url is not in valid form (scheme, netloc, tld)
    if not path.startswith('/'):
        return False

    # the '/' which may exist at the end of the url provides us no information
    if path.endswith('/'):
        path = path[:-1]

    # '/story/cnn/blahblah/index.html' --> ['story', 'cnn', 'blahblah', 'index.html']
    path_chunks = [x for x in path.split('/') if len(x) > 0]

    # siphon out the file type. eg: .html, .htm, .md
    if len(path_chunks) > 0:
        file_type = url_to_filetype(url)

        # if the file type is a media type, reject instantly
        if file_type and file_type not in ALLOWED_TYPES:
            if verbose: print('\t%s rejected due to bad filetype' % url)
            return False

        last_chunk = path_chunks[-1].split('.')
        # the file type is not of use to use anymore, remove from url
        if len(last_chunk) > 1:
            path_chunks[-1] = last_chunk[-2]

    # Index gives us no information
    if 'index' in path_chunks:
        path_chunks.remove('index')

    # extract the tld (top level domain)
    tld_dat = tldextract.extract(url)
    subd = tld_dat.subdomain
    tld = tld_dat.domain.lower()

    url_slug = path_chunks[-1] if path_chunks else ''

    if tld in BAD_DOMAINS:
        if verbose: print('%s caught for a bad tld' % url)
        return False

    if len(path_chunks) == 0:
        dash_count, underscore_count = 0, 0
    else:
        dash_count = url_slug.count('-')
        underscore_count = url_slug.count('_')

    # If the url has a news slug title
    if url_slug and (dash_count > 4 or underscore_count > 4):

        if dash_count >= underscore_count:
            if tld not in [x.lower() for x in url_slug.split('-')]:
                if verbose: print('%s verified for being a slug' % url)
                return True

        if underscore_count > dash_count:
            if tld not in [x.lower() for x in url_slug.split('_')]:
                if verbose: print('%s verified for being a slug' % url)
                return True

    # There must be at least 2 subpaths
    if len(path_chunks) <= 1:
        if verbose: print('%s caught for path chunks too small' % url)
        return False

    # Check for subdomain & path red flags
    # Eg: http://cnn.com/careers.html or careers.cnn.com --> BAD
    for b in BAD_CHUNKS:
        if b in path_chunks or b == subd:
            if verbose: print('%s caught for bad chunks' % url)
            return False

    match_date = re.search(DATE_REGEX, url)

    # if we caught the verified date above, it's an article
    if match_date is not None:
        if verbose: print('%s verified for date' % url)
        return True

    for GOOD in GOOD_PATHS:
        if GOOD.lower() in [p.lower() for p in path_chunks]:
            if verbose: print('%s verified for good path' % url)
            return True

    if verbose: print('%s caught for default false' % url)
    return False


def url_to_filetype(abs_url):
    """
    Input a URL and output the filetype of the file
    specified by the url. Returns None for no filetype.
    'http://blahblah/images/car.jpg' -> 'jpg'
    'http://yahoo.com'               -> None
    """
    path = urlparse(abs_url).path
    # Eliminate the trailing '/', we are extracting the file
    if path.endswith('/'):
        path = path[:-1]
    path_chunks = [x for x in path.split('/') if len(x) > 0]
    last_chunk = path_chunks[-1].split('.')  # last chunk == file usually
    if len(last_chunk) < 2:
        return None
    file_type = last_chunk[-1]
    # Assume that file extension is maximum 5 characters long
    if len(file_type) <= 5 or file_type.lower() in ALLOWED_TYPES:
        return file_type.lower()
    return None


def get_domain(abs_url, **kwargs):
    """
    returns a url's domain, this method exists to
    encapsulate all url code into this file
    """
    if abs_url is None:
        return None
    return urlparse(abs_url, **kwargs).netloc


def get_scheme(abs_url, **kwargs):
    """
    """
    if abs_url is None:
        return None
    return urlparse(abs_url, **kwargs).scheme


def get_path(abs_url, **kwargs):
    """
    """
    if abs_url is None:
        return None
    return urlparse(abs_url, **kwargs).path


def is_abs_url(url):
    """
    this regex was brought to you by django!
    """
    regex = re.compile(
        r'^(?:http|ftp)s?://'                                                                 # http:// or https://
        r'(?:(?:[A-Z0-9](?:[A-Z0-9-]{0,61}[A-Z0-9])?\.)+(?:[A-Z]{2,6}\.?|[A-Z0-9-]{2,}\.?)|'  # domain...
        r'localhost|'                                                                         # localhost...
        r'\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}|'                                                # ...or ipv4
        r'\[?[A-F0-9]*:[A-F0-9:]+\]?)'                                                        # ...or ipv6
        r'(?::\d+)?'                                                                          # optional port
        r'(?:/?|[/?]\S+)$', re.IGNORECASE)

    c_regex = re.compile(regex)
    return (c_regex.search(url) is not None)