async def signup(user: UserSignup):
try:
if MongoDB.user_lookup(user.email) is not True:
new_user = MongoDB.signup(user.username, user.password, user.email)
token = jwt.encode(
{
'user': new_user,
'exp':
datetime.datetime.utcnow() + datetime.timedelta(days=1)
}, SECRET_KEY, 'HS512')
return jsonable_encoder({'token': token.decode('UTF-8')})
else:
return Response(json.dumps({"message": "User already Exists"}),
409)
except KeyError:
return Response(json.dumps({"message": "missing fields"}), 409)
class ArticleService:
mysql = MySQL()
mongodb = MongoDB()
category_service = CategoryService()
def get_articles_by_current_user_id(self, current_user_id: int):
categories = self.category_service.get_category_by_current_user_id(
user_id=current_user_id)
category_names = [i.name for i in categories]
valid_session_id = self.mysql.get_valid_session_id()
number_of_articles_per_category = get_number_of_articles_per_category(
Config.NUMBER_OF_ARTICLES, len(categories))
article_scores = []
for i in range(len(categories)):
articles = self.mysql.fetch_articles_ranking(
session_id=valid_session_id,
category=category_names[i],
limit=number_of_articles_per_category[i])
article_scores.extend(articles)
result = []
for i in article_scores:
article_mongo = self.mongodb.get_article_by_uuid(i.article_id)
article_show = ArticleShow(id=article_mongo.id,
url=article_mongo.url,
domain=article_mongo.domain,
title=article_mongo.title,
category=article_mongo.category,
time=article_mongo.time,
content=article_mongo.content,
audio_path=i.audio_path)
result.append(article_show)
return result
def get_articles_no_login(self):
category_names = Config.CATEGORIES
valid_session_id = self.mysql.get_valid_session_id()
number_of_articles_per_category = get_number_of_articles_per_category(
Config.NUMBER_OF_ARTICLES, len(category_names))
article_scores = []
for i in range(len(category_names)):
articles = self.mysql.fetch_articles_ranking(
session_id=valid_session_id,
category=category_names[i],
limit=number_of_articles_per_category[i])
article_scores.extend(articles)
result = []
for i in article_scores:
article_mongo = self.mongodb.get_article_by_uuid(i.article_id)
article_show = ArticleShow(id=article_mongo.id,
url=article_mongo.url,
domain=article_mongo.domain,
title=article_mongo.title,
category=article_mongo.category,
time=article_mongo.time,
content=article_mongo.content,
audio_path=i.audio_path)
result.append(article_show)
return result
def getBookCollections(self):
results = self.service.GetBookCollections(self.api.url)
if results:
collections, count, user = results
else:
return False
if not user or not collections:
return False
collection_datas = []
for collection in collections:
collection_data = collection.__dict__.get('_data')
for key, value in collection_data.items():
if value is None:
collection_data.pop(key)
collection_data['author'] = DBRef('user', user.id)
if collection.book:
#添加该书到book_tag集合
db = MongoDB.getConnection('mining')
for tag in collection.tags:
db.book_tag.update({'title':tag},
{'$addToSet':{'books':collection.book.bid},
'$inc':{'count':1}},
upsert=True)
collection_data['book'] = DBRef('book', collection.book.id)
collection_datas.append(collection_data)
if collection_datas:
ids = self.db.book_collection.insert(collection_datas, check_keys=False)
for index, object_id in enumerate(ids):
collections[index].id = object_id
try:
user_book_related = None
user_book_related, created = UserBookRelated.objects.get_or_create(\
uid=self.api.api_id, auto_save=False, defaults=\
{'uid':self.api.api_id, 'book_collections':collections})
if not created:
user_book_related.book_collections.extend(collections)
try:
user_book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), user_book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserBookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
if user_book_related and created:
try:
user.user_book_related = user_book_related
user.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), user, self.api.url))
async def login(credentials: HTTPBasicCredentials = Depends(security)):
username = credentials.username
password = credentials.password
check_user = MongoDB.login(username, password)
if check_user["authenticated"]:
token = jwt.encode(
{
'user': check_user,
'exp': datetime.datetime.utcnow() + datetime.timedelta(days=1)
},
SECRET_KEY,
algorithm='HS512')
return jsonable_encoder({'token': token.decode('UTF-8')})
else:
return Response(json.dumps({"message": 'login failed'}), 401,
{'WWW-Authenticate': 'Basic realm="Login Required"'})
def getBookTags(self):
results = self.service.GetBookTags(self.api.url)
if results:
book_tags, count = results
else:
return False
book = None
books = Book.objects(bid=self.api.api_id)
if books:
book = books[0]
else:
return False
if not book_tags:
return False
else:
db = MongoDB.getConnection('mining')
for tag in book_tags:
db.book_tag.update({'title':tag.title},
{'$addToSet':{'books':book.bid},
'$inc':{'count':tag.count}},
upsert=True)
try:
book_related = None
book_related, created = BookRelated.objects.get_or_create(\
bid=book.bid, auto_save=False, defaults=\
{'bid':book.bid, 'book':book, 'tags':book_tags})
if not created:
book_related.tags.extend(book_tags)
try:
book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: BookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
def __init__(self, api, key='book'):
self.api = api
self.api_operation = ApiOperation(key, api)
self.service = self.api_operation.getService()
self.db = MongoDB.getConnection()
def __init__(self, api_type):
self.thread_stop = False
self.sleep_count = 0
self.db = MongoDB.getConnection()
self.api_type = api_type
def __init__(self, uid):
threading.Thread.__init__(self)
self.uid = uid
self.mining_db = MongoDB.getConnection('mining')
self.recommend_db = MongoDB.getConnection('recommend')
class ScoreService:
mongo = MongoDB()
mysql = MySQL()
keyword = Config.SCORE_KEYWORD
lsh = LSH()
def check_contains_keyword(self, text: str, key_list: List[str]):
for i in key_list:
if i in text:
return True
return False
def score_by_category(self, category: str):
print(
'\n------------------------------------------------------------------------'
)
print('category: ', category)
articles_by_category = self.mongo.get_articles_by_category(
category=category)
print('size: ', len(articles_by_category))
article_contents = [
article.content for article in articles_by_category
]
matrix = self.lsh.init_matrix(list_docs=article_contents)
min_hash = self.lsh.min_hashing(matrix=matrix, n_permutation=100)
results = []
invalid_len_articles = []
for article in articles_by_category:
print('article: ', article)
score = 0
if len(article.content) < 500 or len(article.content) > 5000:
score = -99999
invalid_len_articles.append(article.id)
else:
time_second = (datetime.datetime.now() -
article.time).total_seconds()
time_second_max = (datetime.datetime.now() -
datetime.datetime.now().replace(
day=datetime.datetime.now().day - 1,
hour=17,
minute=0,
second=0)).total_seconds()
time_score = (time_second_max -
time_second) / time_second_max * 50
score = score + time_score
if self.check_contains_keyword(article.title,
self.keyword[category]):
# print(article.title)
score = score + 10
index = self.lsh.get_index_of_doc(article.content,
article_contents)
for i in range(len(articles_by_category)):
if i != index:
sim = self.lsh.jaccard_signature(
min_hash[:, index], min_hash[:, i])
# sim = self.lsh.jaccard_signature(matrix[:, index], matrix[:, i])
if sim > Config.PROBABILITY_MIN_HASHING:
if articles_by_category[
index].time < articles_by_category[i].time:
score = score + 20
elif articles_by_category[index].time > articles_by_category[i].time and \
articles_by_category[i].id in invalid_len_articles:
score = score + 20
else:
score = score - 99999
print('------------------------')
print('sim = ', sim)
print(articles_by_category[index])
print(articles_by_category[i])
print('------------------------')
score_insert = ScoreInsert(article_id=article.id,
url=article.url,
category=article.category,
domain=article.domain,
score=score)
results.append(score_insert)
return results
def check_audio(self, audio: Audio):
if audio.result == "Success":
print("create gtts successfully with uuid: ", audio.uuid)
path = Config.BASE_AUDIO_DIR + audio.uuid + ".mp3"
self.mysql.add_audio_path(audio.uuid, path)
else:
print("create gtts failed with uuid: ", audio.uuid)
