def find_or_create(cls, session, user, _origin: str, _origin_lang: str,
_translation: str, _translation_lang: str,
_context: str, _url: str, _url_title: str,
article_id: int):
"""
if the bookmark does not exist, it creates it and returns it
if it exists, it ** updates the translation** and returns the bookmark object
:param _origin:
:param _context:
:param _url:
:return:
"""
origin_lang = Language.find_or_create(_origin_lang)
translation_lang = Language.find_or_create(_translation_lang)
origin = UserWord.find_or_create(session, _origin, origin_lang)
article = Article.query.filter_by(id=article_id).one()
url = Url.find_or_create(session, article.url.as_string(), _url_title)
context = Text.find_or_create(session, _context, origin_lang, url,
article)
translation = UserWord.find_or_create(session, _translation,
translation_lang)
now = datetime.now()
try:
# try to find this bookmark
bookmark = Bookmark.find_by_user_word_and_text(
user, origin, context)
# update the translation
bookmark.translation = translation
except sqlalchemy.orm.exc.NoResultFound as e:
bookmark = cls(origin, translation, user, context, now)
except Exception as e:
raise e
session.add(bookmark)
session.commit()
return bookmark
def create_anonymous(cls,
uuid,
password,
learned_language_code=None,
native_language_code=None):
"""
:param uuid:
:param password:
:param learned_language_code:
:param native_language_code:
:return:
"""
# since the DB must have an email we generate a fake one
fake_email = uuid + cls.ANONYMOUS_EMAIL_DOMAIN
if learned_language_code is not None:
try:
learned_language = Language.find_or_create(
learned_language_code)
except NoResultFound as e:
learned_language = None
else:
learned_language = None
if native_language_code is not None:
try:
native_language = Language.find_or_create(native_language_code)
except NoResultFound as e:
native_language = None
else:
native_language = None
new_user = cls(fake_email,
uuid,
password,
learned_language=learned_language,
native_language=native_language)
# # Until we find_or_create a better way of adding exercises for anonymous and new users... we simply
# from zeeguu.temporary.default_words import default_bookmarks
# default_bookmarks(new_user, learned_language_code)
return new_user
def find_or_create(cls, session, _url:str, language=None, sleep_a_bit=False):
"""
If not found, download and extract all
the required info for this article.
:param url:
:return:
"""
from zeeguu.model import Url, Article, Language
import newspaper
url = Url.extract_canonical_url(_url)
try:
found = cls.find(url)
if found:
return found
art = newspaper.Article(url=url)
art.download()
art.parse()
if art.text == '':
raise Exception("Newspaper got empty article from: " + url)
if sleep_a_bit:
import time
from random import randint
print("GOT: " + url)
sleep_time = randint(3, 33)
print(f"sleeping for {sleep_time}s... so we don't annoy our friendly servers")
time.sleep(sleep_time)
if not language:
if art.meta_lang == '':
art.meta_lang = detect(art.text)
zeeguu.log(f"langdetect: {art.meta_lang} for {url}")
language = Language.find_or_create(art.meta_lang)
# Create new article and save it to DB
url_object = Url.find_or_create(session, url)
new_article = Article(
url_object,
art.title,
', '.join(art.authors),
art.text[0:32000], # any article longer than this will be truncated...
art.summary,
None,
None,
language
)
session.add(new_article)
session.commit()
return new_article
except sqlalchemy.exc.IntegrityError or sqlalchemy.exc.DatabaseError:
for i in range(10):
try:
session.rollback()
u = cls.find(url)
print("Found article by url after recovering from race")
return u
except:
print("Exception of second degree in article..." + str(i))
time.sleep(0.3)
continue
break