def parse_content(self, response):
"""extract content of news by newspaper"""
item = response.meta['item']
is_special, content = self._handle_special_site(response)
if not is_special:
# 不是特殊网站
article = Article(item['url'], language='zh')
article.set_html(response.body)
article.is_downloaded = True
article.parse()
item['pic'] = article.top_image
item['content'] = str(article.text)
item['publish_date'] = article.publish_date
if publish_date:
item['publish_date'] = publish_date.strftime(
"%Y-%m-%d %H:%M:%S")
else:
item['publish_date'] = "null"
else:
item['pic'] = ""
item['content'] = content
# extract content failed
if item['content'] == '':
logging.error("empty content in: " + response.url)
yield item
# raw_content = response.xpath("//body//p/text()").extract()
# item['content'] = ''.join(raw_content)
item['content'] = item['content'].strip().replace(u"\xa0", "").replace(u"\u3000", "").replace("|", "")\
.replace("用微信扫码二维码分享至好友和朋友圈", "").strip("您当前的位置 ：").strip("您所在的位置：").strip("提示：点击上方").strip(">").strip()
yield item
def wrap_newspaper(self, web_page):
parser = NewspaperArticle(url=web_page.final_url)
parser.html = web_page.html
parser.is_downloaded = True
parser.parse()
return parser
def wrap_newspaper(self, web_page):
parser = NewspaperArticle(url=web_page.final_url)
parser.html = web_page.html
parser.is_downloaded = True
parser.parse()
return parser
def clean(html_content):
config = Configuration()
config.fetch_images = False
# TODO: allow URL passing
article = Article("http://example.com", config=config)
article.set_html(html_content)
article.is_downloaded = True
article.parse()
return article.text
from newspaper import Article
client = MongoClient()
db_articles = client.news.articles
db_web_cache = client.news.web_cache
docs = db_articles.find()
for doc in docs:
print doc['_id']
if not doc['page']:
continue
url = doc['page']['urls'][0]
web_cache_doc = db_web_cache.find_one({'url': url})
if 'html_compressed' in web_cache_doc:
article = Article(url=url)
article.html = bz2.decompress(web_cache_doc['html_compressed'])
article.is_downloaded = True
article.parse()
doc['page']['text'] = article.text
print len(doc['page']['text'])
db_articles.save(doc)
import re
from pymongo import MongoClient
from newspaper import Article
client = MongoClient()
db_articles = client.news.articles
db_web_cache = client.news.web_cache
docs = db_articles.find()
for doc in docs:
print doc['_id']
if not doc['page']:
continue
url = doc['page']['urls'][0]
web_cache_doc = db_web_cache.find_one({'url': url})
if 'html_compressed' in web_cache_doc:
article = Article(url=url)
article.html = bz2.decompress(web_cache_doc['html_compressed'])
article.is_downloaded = True
article.parse()
doc['page']['text'] = article.text
print len(doc['page']['text'])
db_articles.save(doc)
def parse(url=None, html=None, text=None, title=None,
sentences_count=5,
options={},
summarize_algo="luhn",
date_timezone="America/New_York"):
"""
Parse article to get relevant data
:param url:
:param html:
:param text:
:param title:
:param sentences_count:
:param options: {}
:param summarize_algo:
:param date_timezone: The timezone to convert the date to
:return:
"""
article = Article("")
if text and title:
article.is_parsed = True
article.is_downloaded = True
article.set_title(title)
article.set_text(text)
else:
if url:
r = requests.get(url.strip())
if r.status_code != 200:
raise Exception("Paper request failed '%s'" % url)
html = r.content
if html:
soup = get_soup(html)
else:
raise Exception("Paper missing HTML content")
article.set_html(remove_social_embeds(html))
article.parse()
article.nlp()
if options.get("title_selector"):
title = soup.select(options.get("title_selector"))
if title:
title = title[0].text
article.set_title(title)
if options.get("image_selector"):
img = soup.select(options.get("image_selector"))
if img:
img = img[0].text
article.set_top_img_no_check(img)
if options.get("content_selector"):
html = soup.select(options.get("content_selector"))
if html:
article.set_text(html[0].text)
summary = summarize(text=article.text,
title=article.title,
algo=summarize_algo,
sentences_count=sentences_count)
publish_date = article.publish_date
if not publish_date and html:
publish_date = extract_publish_date(html)
if not publish_date:
publish_date = datetime.datetime.now()
return {
"url": article.canonical_link,
"title": article.title,
"summary": summary,
"summaries": summary.split("\n\n"),
"text": article.text,
"html": article.html,
"top_image": article.top_image,
"images": article.images,
"videos": list(set(article.movies + extract_video_iframes(html))),
"social_media_content": extract_social_media_content(html),
"keywords": article.keywords,
"tags": article.tags,
"authors": article.authors,
"published_date": datetime_to_local_timezone(publish_date),
"md_text": ""
}
