def retrieve(self, url):
"""Public method to retrieve content for an article URL"""
pages = self._pages(url)
response = ParserResponse()
for page in pages:
article = Article(page, keep_article_html=True)
article.download()
article.parse()
response.content = response.content + article.article_html
if self._config.strip_images:
response.content = re.sub('<img [^<]+?>', '', response.content)
if not response.content:
response.error = "Parser could not get content for URL '%s'" % (url)
return response
def retrieve(self, url):
"""Public method to retrieve content for an article URL"""
request = requests.get(Parser.Endpoint, params={'token': self._config.readability.token, 'url': url})
article = json.loads(request.text)
response = ParserResponse()
if 'error' in article:
response.error = "Readability parser error: %s" % (article['messages'])
if 'content' in article:
response.content = article['content']
if self._config.strip_images:
response.content = re.sub('<img [^<]+?>', '', response.content)
if not response.content:
response.error = "Parser could not get content for URL '%s'" % (url)
return response
