Working and up to date. WIP misc manual actions

This commit is contained in:
Remy Moll
2022-05-24 18:37:30 +02:00
parent 246729d376
commit 878a1dff5d
14 changed files with 142 additions and 182 deletions

View File

@@ -11,61 +11,52 @@ logging.getLogger('charset_normalizer').setLevel(logging.ERROR) #quieter logs
logger = logging.getLogger("fetch")
class NewspaperDummy():
title = "Error while running fetch"
summary = "Error while running fetch"
text = "Error while running fetch"
meta_lang = ""
authors = []
keywords = []
def get_description(article_object):
url = article_object.article_url
website = urlparse(url).netloc
article_object.source_name = website
try:
pub_date = datetime.datetime.strptime(find_date(url), '%Y-%d-%M')
article_object.pub_date = datetime.datetime.strptime(find_date(url), '%Y-%d-%M')
except: # other file types
pub_date = datetime.datetime(year=1900, month=1, day=1)
article_object.pub_date = pub_date
article_object.pub_date = datetime.datetime(year=1900, month=1, day=1)
fallback = NewspaperDummy()
try:
news_article = Article(url)
news_article.download()
news_article.parse()
except:
news_article = fallback
if news_article.title:
title = news_article.title
else:
title = fallback.title
if news_article.summary:
summary = news_article.summary
elif news_article.text:
ind = min(500, len(news_article.text))
summary = news_article.text[:ind] + "..."
else:
summary = fallback.summary
news_article = object() # fallback value
try:
print(f"lang: {news_article.meta_lang}")
except:
print("could not access meta_lang")
if news_article.meta_lang:
lang = news_article.meta_lang
else:
lang = ""
article_object.title = news_article.title
except AttributeError:
article_object.title = "Error while running fetch"
article_object.title = title
article_object.summary = summary
article_object.language = lang
article_object.set_authors(news_article.authors)
article_object.set_keywords(news_article.keywords)
try:
if article_object.summary:
article_object.summary = news_article.summary
elif news_article.text:
ind = min(500, len(news_article.text))
article_object.summary = news_article.text[:ind] + "..."
else:
article_object.summary = ""
except AttributeError:
article_object.summary = ""
try:
article_object.language = news_article.meta_lang
except AttributeError:
article_object.language = ""
try:
article_object.set_authors(news_article.authors)
except AttributeError:
pass # list would have been empty anyway
try:
article_object.set_keywords(news_article.keywords)
except AttributeError:
pass # list would have been empty anyway
return article_object