def output_lyrics():
content = request.get_json()
initial_string = content["initial_string"]
initial_string = clean(initial_string)
full_song = generate_text(model, initial_string)
song_final = clean_song(full_song)
response = {"id": str(uuid.uuid4()), "song":song_final}
return jsonify(response)
def get_token_list(self):
if not self.token_list:
self.token_list = []
for i, word in enumerate(self.nodes):
if i != 0 and word.get_form() != None:
token = tc.clean(word.get_form())
if token:
self.token_list.append(token)
return self.token_list
import pandas as pd
from text_clean import clean
conn = pymysql.connect(host='39.97.176.70',
port=3306,
user='******',
password='******',
database='test_timeline')
sql = f"""
SELECT news_text, news_attitude
FROM
news_info
WHERE news_attitude is not NULL;
"""
cur = conn.cursor()
cur.execute(sql)
rows = cur.fetchall()
rows = list(rows)
cur.close()
conn.close()
df = pd.DataFrame(rows, columns=['news_text', 'label'])
df['news_text'] = df['news_text'].apply(lambda x: ' '.join(clean(x)))
df = df[df['news_text'].apply(lambda x: len(x) > 61)]
df.reset_index()
df.to_csv('data/train_set.csv', index=False)
def draw_plot(res, start_date_in, end_date_in, stk):
try:
start_date_local = start_date_in
date_info_date = []
date_info_num_news = []
date_info_num_positive_news = []
date_info_percent = []
one_day = timedelta(days=1)
news_date = []
total_news = []
pos_news = []
try:
while (start_date_local <= end_date_in):
if (np.is_busday(start_date_local, holidays=KL_holidays)):
news_date.append(start_date_local)
total_news.append(0)
pos_news.append(0)
start_date_local = start_date_local + one_day
for doc in res['hits']['hits']:
doc_date = doc['_source']['published_date']
doc_date_refined = doc_date[:doc_date.find('+')]
doc_date_refined = doc_date_refined.replace('T', '-')
doc_date_refined = doc_date_refined.replace(':', '-')
doc_date_refined = datetime.strptime(
doc_date_refined,
'%Y-%m-%d-%H-%M-%S') #convert to datetime
doc_date_refined = doc_date_refined.replace(hour=0,
minute=0,
second=0)
if (doc_date_refined in news_date):
if (
np.is_busday(doc_date_refined,
holidays=KL_holidays)
): # need to be changed, if news of pubic holidays are also required to be considered
total_news[news_date.index(doc_date_refined)] += 1
doc_content = doc['_source']['content']
doc_content = doc_content.encode(
'ascii', 'ignore') # limits news to english only
cleaned_news = clean(
news_str=doc_content, configs=config_obj
) #writes in 'elastic_grabbed_news' folder
cleaned_news = cleaned_news.encode('ascii', 'ignore')
json_stock_prediction = stock_pred.run(
cleaned_news,
0.5) # TODO: decide what todo with output
if (json_stock_prediction['prediction'][stk]
['prediction'] == 1):
pos_news[news_date.index(doc_date_refined)] += 1
percentage = []
counter = 0
while (counter < len(news_date)):
if total_news[counter] != 0:
percentage.append(
float(pos_news[counter]) / float(total_news[counter]))
else:
percentage.append(0.0)
counter += 1
dates = matplotlib.dates.date2num(news_date)
ee = dict(zip(dates, percentage))
except Exception as e:
PrintException(e, exitStatus=True)
return ee
except:
return 0
def get_token(self):
return tc.clean(self.get_form())