Ejemplo n.º 1
0
def whoHasTimeToRead(url):
	is_article = valid_url(url, verbose=True)
	config = Config()
	config.MAX_KEYWORDS = 10
	if is_article:
		sumitup = {}
		b = Article(url=url,config=config)
		b.download()
		b.parse()
		b.nlp()
		sumNews = summary(b.title, b.text, b.keywords)
		sumTitle = b.title
		movies = b.movies[0] if len(b.movies) > 0 else "None"
		return sumNews,sumTitle,movies
	return "Nope"
import argparse
import csv
import sys
import threading
import warc

from newspaper import Article, Config
from tld import get_tld
configuration = Config()
configuration.fetch_images = False
configuration.memoize_articles = False
configuration.MAX_KEYWORDS = 10


def parse_args():
    """Parses command line arguments.

    Parameters
    ----------
    None

    Returns
    -------
    argparse args object (named-tuple)
    """
    argparser = argparse.ArgumentParser()
    argparser.add_argument("-f", "--file", default="", help="")
    return argparser.parse_args()


def build_csv_line(article):