def summarise_one(url, title=True, keywords=True, summary=False, \
top_img_src=False):
'''
Get url and return summary
'''
article = Article(url)
# configuration for Newspaper to minimize processing time
configure = Config()
configure.fetch_images = False
configure.MAX_SUMMARY = 300
configure.MAX_SUMMARY_SENT = 3
try:
article.download()
article.parse()
except:
print(url)
title = article.title
if keywords or summary:
try:
article.nlp()
if keywords:
keywords = article.keywords
if summary:
summary = article.summary
except :
print('NEwspaper error with nlp() call')
if top_img_src:
top_img_src = article.top_image
return title, keywords, summary, top_img_src
def load_and_parse_full_article_text_and_image(url: str) -> Article:
config = Config()
config.MAX_SUMMARY_SENT = 8
article = Article(url, config=config)
article.set_html(load_page_safe(url)) # safer than article.download()
article.parse()
return article
def analysis():
articles = request.data
articles = json.loads(articles)
urls = articles['urls']
summaries = []
results = []
config = Config()
config.MAX_SUMMARY_SENT = 2
for pairs in urls:
url = pairs['url']
source = pairs['source']
print(url)
article = Article(url, config=config)
article.download()
article.parse()
summary, summary2 = generateSummary(article)
str_summary = ' '.join(summary)
blob = TextBlob(str_summary)
summaries.append(str_summary)
print(str_summary)
subjectivity = []
polarity = []
for sentence in blob.sentences:
subjectivity.append(sentence.sentiment.subjectivity)
polarity.append(sentence.sentiment.polarity)
subj_sum = 0.5 if len(subjectivity) == 0 else sum(subjectivity)/len(subjectivity)
pol_sum = 0.5 if len(polarity) == 0 else ((sum(polarity) / len(polarity)) + 1)/2
results.append({"headline": article.title, "summary": summary2, "_summary": str_summary, "image": article.top_image, "subjectivity": subj_sum, "p_group": pol_sum, "url": url, "source": source, "keywords": article.keywords})
clusters = clusterSentences(summaries, 3)
for cluster in range(3):
for i, article_summary in enumerate(clusters[cluster]):
for res in results:
if res['_summary'] == summaries[article_summary]:
print("FOUND")
res['group'] = convertNum(cluster)
return json.dumps({"results": results})
import pandas as pd
from tqdm import tqdm
import newspaper
from newspaper import Config
config = Config()
config.memoize_articles = False
config.fetch_images = False
config.verbose = True
config.MAX_SUMMARY_SENT = 10
config.language = "vi"
vnexpress = newspaper.build('https://vnexpress.net', config=config)
zingvn = newspaper.build('https://news.zing.vn', config=config)
kenh14 = newspaper.build('https://kenh14.vn/', config=config)
# list_url_slate_paper = []
# for article in slate_paper.articles:
# list_url_slate_paper.append(article.url)
def extract_data(news, file_name):
print("crawling: ", file_name)
data = []
for article in tqdm(news.articles):
# count += 1
# if count > 10:
# break
if "#box_comment" in article.url:
continue
temp = {"link": article.url}