def _fetch_article(news_site_id, url, link):
logger.info(f'Starting to fetch article at {link}')
article = None
try:
article = news.ArticlePage(news_site_id, _build_link(url, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body in it')
return None
return article
def _fetch_article(new_site_uid, host, link):
logging.info('Start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(new_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logging.warning('Error while feching the article', exc_info=False)
if article and not article.body:
logging.warning('Invalid article. There is not body.')
return None
return article
def _fetch_article(new_site_uid, host, link):
logger.info('iniciando busqueda del articulo'+ link)
article = None
try:
article = news.ArticlePage(new_site_uid,_build_link(host,link))
except (HTTPError,MaxRetryError) as e:
logger.warning('Error encontrando el error',exc_info=False)
if article and not article.body:
logger.warning('articulo invalido no tiene cuerpo')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info("Start fetching article at {}".format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTOError, MaxRetryError):
logger.warning("Error while fetching the article", exc_info=False)
if article and not article.body:
logger.warning("Invalid article. There is no body")
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f'Start fetching article at {link}')
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site, host, link):
article = None
try:
url = build_link(host, link)
article = news.ArticlePage(news_site, url)
except (HTTPError, MaxRetryError):
print('Error while fetching article')
if article and not article.body:
print('Invalid Article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError, requests.ConnectionError) as e:
logger.warning('error while fechting the article!', exc_info=False)
if article and not article.body:
logger.warning('invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
# logger.info(f'Start fetching article at {link}')
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
# logger.warning('Error while fetching the article', exc_info=False)
pass
if article and not article.body:
# logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_sites_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_sites_uid, link)
except (HTTPError, MaxRetryError) as e:
logger.error('The article coudn\'t be fetched')
if article and not article.body:
logger.warning('There isn\'t a body in this page. ')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError, NewConnectionError, ConnectionError,
gaierror, ContentDecodingError, SSLError, TooManyRedirects) as e:
logger.warning('Error while fechting the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at %s' % link)
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError):
# MaxRetryError: eliminar la posibilidad de que se vaya al infinito
# trantando de acceder a la url
logger.warning('Error while fetching the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_sites_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching the article', exc_info = false)
if article and not article.body:
logger.warning('Articulo Invalido')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('esta buscando el archivo en {}'.format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError) as e:
logger.warning('Error mientras se buscaba el articulo', exc_info=False)
if article and not article.body:
logger.warning('articulo invalido, no hay cuerpo en el articulo')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
article_page = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching the article', exc_info=False)
if article_page and not article_page.content or not article_page.title or not article_page.date:
logger.warning(
'Invalid article. There is no valid article (no content, no title)'
)
return None
return article_page
def _fetch_article(news_site_uid, host, link):
logger.info(f'Start fetching article at {link}')
article = None
try: # Aquí comienza la sección de estrategia de programación defensiva
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body.')
return None
return article
def _fetch_article(news_site_id, host, link):
logger.info(f"Start fetching article at {link=}")
article = None
try:
article = news.ArticlePage(news_site_id, _build_link(host, link))
except (HTTPError, MaxRetryError, ConnectionError) as e:
logger.warning("Error while fetching the article", exc_info=False)
if article and not article.body:
logger.warning("Invalid article. There is no body")
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f'Obteniendo articulos de: {link}')
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warn('Error mientras se obtenia el articulo', exc_info=False)
if article and not article.body:
logger.warning('Articulo invalido. No hay cuerpo')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('ERROR fetching article: {}'.format(e), exc_info=False)
except Exception as e:
logger.error('ERROR fetching article: {}'.format(e), exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f'Start fetching article at {link}')
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError, DecodeError, ContentDecodingError,
TimeoutError, NewConnectionError, ConnectionError) as e:
logger.warning('Error while fechting the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
#create an article page object with the link built with the host link and the link from the homepage object
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article, there is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f'Iniciando extracion de articulo en: {link}')
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error al extraer el articulo', exc_info=False)
if article and not article.body:
logger.warning('Articulo Invalido. No hay un body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start Fetching artitle at {}'.format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching article', exc_info=False) #No mostramos log para no ensuciar consola
print("\n\n\n\n Article, ", article)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f'Fetching article {link}...')
article = None
try:
article = news.ArticlePage(news_site_uid, link)
except (HTTPError, MaxRetryError) as e:
logger.warning(f'Error while fetching article {link}', exc_info=False)
except:
logger.warning(f'Unknown Error while fetching article {link}',
exc_info=True)
if article and not article.body:
logger.warn('Invalid article. No body present')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Srart fetching article at {}'.format(link))
article = None
try:
formated_link = _build_link(host, link)
article = news.ArticlePage(news_site_uid, formated_link)
article.url = formated_link
except (HTTPError, MaxRetryError) as e:
logger.warning('Error while fetching the articke', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at %s' % link)
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except Exception as e:
logger.warning('Error while fetching the article', exc_info=False)
print('Exception: ', e)
return None
if article and not article.body:
logger.warning('Invalid article. There is no body.')
return None
return article
def _fetch_article(news_site_uid, host, link):
logger.info(f"Start fetching article at {link}")
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(host, link))
except (HTTPError, MaxRetryError) as e:
logger.warning(f'Error while fetching the article: {e}',
exc_info=False)
if article and not article.body:
logger.warning(Fore.RED + 'Invalid article. There is no body')
#Colorama
print(Style.RESET_ALL)
return None
return article
async def _fetch_article(news_site_uid, link, session):
article = None
error = 0
logger.info('Start fetching article at {}'.format(link))
try:
article = news.ArticlePage(news_site_uid, link)
await article.visit(session)
if article and not article.body_csv:
logger.warning('Invalid article. There is no body')
error = 'Invalid article. There is no body'
article = None
except Exception as e:
logger.error('ERROR fetching article: {}'.format(e), exc_info=False)
error = 'ERROR fetching article {}: {}'.format(link, e)
return (error, article, news_site_uid)
def _fetch_article(news_site_uid, host, link):
logger.info('Start fetching article at {}'.format(link))
article = None
try:
#Envia cada articulo a la clase ArticlePage
article = news.ArticlePage(news_site_uid, _build_link(
host, link)) #revisa que los vinculos esten bien construidos
except (HTTPError, MaxRetryError) as e:
#si ocurre un error, invalida el articulo
logger.warning('Error while fechting the article', exc_info=False)
#si el articulo no tiene cuerpo, queda invalidado
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
return article
def _fetch_article(news_site_uid, host_url, link):
logger.info("Start fetching article at {}".format(link))
articles = None
print(_build_link(host_url, link))
try:
article = news.ArticlePage(news_site_uid, _build_link(host_url, link))
except (HTTPError, MaxRetryError) as e:
logger.warning('Error fetching the article', exc_info=False)
if article and not article.body:
logger.warning('Invalid article. There is no body')
return None
#print(article.title)
#print('\n')
#print(article.body)
return article
def _fetch_article(news_site_uid, link):
logger.info("Start fetching article at {}".format(link))
article = None
try:
article = news.ArticlePage(news_site_uid, _build_link(link))
# except (HTTPError, MaxRetryError) as e:
except:
# HTTPErrorr --> cuando no se ha encontrado la página
# MaxRetryError --> estoy eliminadno la posibildad de que se vaya al infinito tratando de seguir la URL
logger.warning("Error while fetching the article", exc_info=False)
# exc_info=False --> para que no me muestre el error
if article and not article.body and not article.title:
logger.warning("Invalid article. There is no body")
return None
return article
