def fetchNewsHeadlines(ek, companyName):
"""
Mine the headlines for the given company
"""
# companyName = 'R:RDSa.L'
df = ek.get_news_headlines(companyName+' AND Language:LEN', date_to=\
"2018-05-22", count=100)
return df
def get_news(ric):
headlines = ek.get_news_headlines(ric + ' IN CHINESE',
date_from=from_date,
date_to=today,
count=100)
return [{
'time': datetime.date.strftime(a, '%Y-%m-%dT%H:%M:%S'),
'source': d,
'headline': b,
'storyid': c,
'content': ek.get_news_story(c)
} for a, b, c, d in zip(headlines['versionCreated'], headlines['text'],
headlines['storyId'], headlines['sourceCode'])]
def do_day(self, date, ric, do_clean):
pathtodir = "{}\\{}\\{}\\{}\\{}".format(self.outputdir, date.year,
date.month, date.day, ric)
with eventlet.Timeout(10, False) as event:
try:
news = ek.get_news_headlines(
ric + ' AND Language:LEN',
date_from=date.strftime("%Y-%m-%d"),
date_to=(date +
datetime.timedelta(days=1)).strftime("%Y-%m-%d"),
count=100)
self.process_headlines(news, pathtodir, date, ric, do_clean)
except Exception as e:
self._handle_time(e, event)
def news():
rqst = request.get_json()
try:
rst = eikon.get_news_headlines(query=rqst['query'],
count=100,
date_from=rqst['from'],
date_to=rqst['to'],
raw_output=True)
for idx, val in enumerate(rst['headlines']):
val['news'] = eikon.get_news_story(val['storyId'])
return jsonify(data=rst)
except Exception as err:
webapp.logger.warn(err)
return abort(500)
def news_occurance(ticker, regional_awards: list, international_awards: list,
date_to: str, date_from: str):
"""
We stick to refinitiv convention that date_to predeceeds date_from
"""
def award_check(awards, text):
return sum([award in text for award in awards])
regional_occurance = 0
international_occurance = 0
i = 0
n_df = 0
while date_from != date_to:
old_date_from = date_from
news = ek.get_news_headlines(query=f"R:{ticker}",
date_to=date_to,
date_from=date_from,
count=100)
if len(news) == 0:
break
if i % 25 == 0:
print(date_from)
news = news.sort_values("versionCreated",
ascending=False).reset_index(drop=True)
news["text"] = news["text"].apply(lambda x: x.lower())
date_from = str(news.loc[0, "versionCreated"].date())
if old_date_from == date_from:
if n_df >= 3:
n_df = 0
print(f"Stuck, breaking for ticker = {ticker}")
break
n_df += 1
regional_occurance += sum(
news["text"].apply(lambda x: award_check(regional_awards, x)))
international_occurance += sum(
news["text"].apply(lambda x: award_check(international_awards, x)))
i += 1
occurance = {
"Ticker": ticker,
"regional_occurance": regional_occurance,
"international_occurance": international_occurance
}
return occurance
def downloadNews(undlName, date, savePath):
if not checkFolderExist(savePath + formatDate(date)):
createFolder(savePath, formatDate(date))
# download data
df = ek.get_news_headlines(
"R:" + undlName + " and english",
date_from=formatDate(moveDate(date, -1)) + "T16:00:00",
date_to=formatDate(moveDate(date)) + "T16:00:00",
count=100)
# move date back to HK time
df.index = moveDate(np.array(list(df.index)), 0, 8)
df.versionCreated = moveDate(np.array(list(df.versionCreated)), 0, 8)
# save data
df.to_csv(savePath + formatDate(date) + "/" + undlName + "_headlines.csv")
def makeResponse(req):
if req.get("result").get("action") != "fetchNews":
return {}
result = req.get("result")
parameters = result.get("parameters")
ticker = parameters.get("ticker")
if ticker is None:
return None
tr.set_app_key('229dfa317f614c3c9cdfa2908c1ff66af4fdf1c6')
r = tr.get_news_headlines('ticker',
date_from='2020-03-09',
date_to='2020-03-10',
count=5)
r.head()
news = r("text")
print(news)
result = "The news on" + ticker + "are" + news
return {
"text": result,
}
def get_news(topic, source, start_date, end_date, n_articles):
'''
To search a company, Append 'R:' in front of RIC (ticker followed by a dot and Echange ID (eg: TSLA.O))
The following is a list of news sources and their respective code from Mexico.
mexican_sources = {
'Mexico Ledger': 'MEXLED',
'El Financiero': 'ELMEX',
'El Economista': 'ELECOP',
'La Jornada': 'LAMEX',
'Mega News': 'MEGNEW',
'Milenio': 'MILMEX',
'El Nacional': 'ELNACI',
'Expansion': 'EXPSPB',
'Excelsior': 'EXCMEX',
'Mural': 'MURMEX',
'El Norte': 'ELNORT',
'Estrategia': 'ESTMEX',
'La I': 'LAIMEX',
'Al Chile': 'ALCHIL',
'Publimetro': 'PUBMEX',
'Comunicae': 'COMMEX',
'Diario de Yucatan': 'DIADEY',
'Contexto Durango': 'CONDED',
}
'''
# formats query based on function inputs
#search = '{topic} and NS:{source}'.format(topic=topic,source=source)
search = topic
# getting headlines
headlines = ek.get_news_headlines(query=search,
date_from=start_date,
date_to=end_date,
count=n_articles)
return headlines
# Create list of all dates
datelist = []
for t in range(0, numdays):
datelist.append((today - datetime.timedelta(days=t)).date())
# Retrieve daily headlines
APIcode = 'R:AAPL.O AND Language:LEN'
frames = []
for date in datelist:
# Volontary slow down request (avoid too many requests error from eikon)
for i in range(0, 1000):
print('I love data science')
# Get headline
df = ek.get_news_headlines(APIcode,
date_from=date,
date_to=date,
count=100)
frames.append(df)
# Create news dataframe
news = pd.concat(frames)
# Drop duplicates
news = news.drop_duplicates(subset='text', keep='last')
# Retrieve content of news
story = []
for storyId in news['storyId'].values:
# Volontary slow down request (avoid too many requests error from eikon)
for i in range(0, 1000):
print('I love data science')
import numpy
import pandas
import eikon
eikon.set_app_key('638fa3bbb90349d5a97dc60c1c0cc4b0b5646846')
eikon.get_news_headlines('R:LHAG.DE', date_from='2018-01-01T00:00:00', date_to='2018-12-13T18:00:00')
#df = eikon.get_timeseries("SPY",
# ['TR.IndexConstituentRIC',
# 'TR.IndexConstituentName',
# 'TR.IndexConstituentWeightPercent'],
data_grid, err = eikon.get_data("SPY",
['TR.IndexConstituentRIC',
{'TR.IndexConstituentWeightPercent':{'sort_dir':'desc'}}],
{'SDate':'2018-12-14'}
)
print(data_grid)
weight = data_grid.loc[:, "Weight percent"]
weight
import matplotlib.pyplot as plt
fig1, ax1 = plt.subplots()
ax1.pie(weight, shadow = True)
ax1.axis('equal')
plt.show()
NESN = eikon.get_timeseries(["NESN.S"], start_date="2014-01-06",
end_date="2018-12-20", interval="daily")
plt.plot(NESN.loc[:, "CLOSE"])
print('Will be back {}'.format((curr_time + datetime.timedelta(seconds = number)).strftime('%Y-%m-%d %H:%M:%S')))
event.cancel()
time.sleep(number)
else:
print(ric, ' failed - ', e.message)
else:
print(ric, ' failed - ', e)
for ric in rics:
for date in date_generated:
with eventlet.Timeout(10,False) as event:
try:
news = ek.get_news_headlines(ric+' AND Language:LEN',
date_from=date.strftime("%Y-%m-%d"),
date_to=(date + datetime.timedelta(days=1)).strftime("%Y-%m-%d"),
count=100
)
except Exception as e:
_handle_time(e, event)
continue
for i in range(len(news)):
time.sleep(0.2)
storyId = news.iat[i,2]
storyid = re.sub(':', '_', storyId)
path = "{}\\{}\\{}\\{}\\{}.txt".format(date.year, date.month, date.day, ric, storyid)
os.makedirs(os.path.dirname(path), exist_ok=True)
# HTML story
with eventlet.Timeout(10,False) as event:
try:
def get_news_and_sentiments(key_id, ticker_and_params, ticker, start_date,
end_date):
ek.set_app_key(key_id)
Date_range = pd.date_range(start=start_date, end=end_date, freq='D')
date_range = []
for i in range(len(Date_range) - 1):
start_date = pd.to_datetime(Date_range)[i].strftime(
"%Y-%m-%dT%H:%M:%S")
end_date = pd.to_datetime(Date_range)[i +
1].strftime("%Y-%m-%dT%H:%M:%S")
date_range.append([])
for j in range(1):
date_range[i].append(start_date)
date_range[i].append(end_date)
example_data = pd.DataFrame()
for i in range(len(date_range)):
example_data = example_data.append(ek.get_news_headlines(
f'{ticker_and_params}',
count=10,
date_from=date_range[i][0],
date_to=date_range[i][1]),
ignore_index=False)
news = example_data
stories = []
for i, storyId in enumerate(news['storyId']):
try:
html = ek.get_news_story(storyId)
story = BeautifulSoup(html, 'html5lib').get_text()
stories.append(story)
except:
stories.append('')
news['story'] = stories
## Sentiment Over Time
sentiment = pd.DataFrame()
sid = SentimentIntensityAnalyzer()
for storyId in news['storyId']:
row = news[news['storyId'] == storyId]
scores = sid.polarity_scores(row['story'][0])
sentiment = sentiment.append(
pd.DataFrame(scores, index=[row['versionCreated'][0]]))
sentiment.index = pd.DatetimeIndex(sentiment.index)
sentiment.sort_index(inplace=True)
sentiment_list = list(sentiment['compound'])
news['sentiment'] = sentiment_list
#group by day
dates_normal = []
for i in range(len(news['versionCreated'])):
dates_normal.append(news['versionCreated'][i].strftime("%Y-%m-%d"))
dates_normal_2 = pd.to_datetime(dates_normal)
news['dates_normal'] = dates_normal_2
#save to csv
news.to_csv(f"{ticker}_news.csv")
daily_sentiments_listed = news.groupby(
pd.Grouper(key="dates_normal",
freq="D"))['sentiment'].apply(list).reset_index()
#Daily mood index
DMI = []
for i in range(len(daily_sentiments_listed['sentiment'])):
pos = 0
neg = 0
for j in range(len(daily_sentiments_listed['sentiment'][i])):
try:
if daily_sentiments_listed['sentiment'][i][j] > 0:
pos += 1
elif daily_sentiments_listed['sentiment'][i][j] < 0:
neg += 1
except:
pass
DMI.append(np.log((1 + pos) / (1 + neg)))
daily_sentiments_listed['DMI'] = DMI
#Average Sentiment
Average_S = []
for i in range(len(daily_sentiments_listed['sentiment'])):
Average_S.append(np.mean(daily_sentiments_listed['sentiment'][i]))
daily_sentiments_listed['Average_Sentiment'] = pd.DataFrame(
Average_S).fillna(0)
#save sentiments to csv
daily_sentiments_listed.to_csv(f"{ticker}_sentiment.csv", index=False)
return daily_sentiments_listed, news
import eikon as ek
ek.set_app_key('ba4e52456ba64a87be3b82001782535f159b564e')
print(
ek.get_news_headlines('R:IBM.N',
date_from='2019-09-27T09:00:00',
date_to='2019-09-27T10:00:00'))
# RIC: reuters instrument code, Data item browser에서 검색하여 찾는다.
rics = [
'GE', # general electronic stock
'AAPL.O', # apple stock
'.SPX', # S&P 500 index
'.VIX', # VIX volatility
'EUR=', # EUR/USD exchange rate
'XAU=' # Gold
'DE10YT=RR' # 10yr Germany Bond
]
# get time series data
ts = ek.get_timeseries('AAPL.O', start_date='2019-01-01')
# get news headline, date_to를 기준으로 하루치만 가능
nh = ek.get_news_headlines('R:AAPL.O', date_from='2018-01-01', date_to='2018-01-31')
# convert RICs into ISINs(international securities identification numbers) and ticker
isins_tickers = ek.get_symbology(rics, from_symbol_type='RIC', to_symbol_type=['ISIN', 'ticker'])
# other symbol types(ex. SEDOL: stock exchange daily official list) can also be converted to RICs or ISINs
sedols = ['B1YW440', '0673123', 'B02J639', 'B1XZS82', '0045614', '0053673', '0989529', '0216238', '0969703', '0263494']
transed_rics = ek.get_symbology(sedols, from_symbol_type='SEDOL', to_symbol_type=['RIC', 'ISIN'])
# historical data will be retrieved for the following stocks
symbols = ['US0378331005', 'US0231351067', 'US30303M1027', 'US4581401001']
rics = ek.get_symbology(symbols, from_symbol_type='ISIN', to_symbol_type='RIC')
rics = list(rics.RIC.values)
data = ek.get_timeseries(rics, # the RICs
fields='CLOSE', # close field
start_date='2017-10-01', # start date
import eikon as ek
import numpy
import pandas
ek.set_app_key('cd027055730e46e08ec6eeeaf0986d86824ec3ff')
ford_news = ek.get_news_headlines('FORD', date_from='2019-03-14T09:00:00', date_to='2019-03-15T18:00:00')
print(ford_news[1:])
print(len(ford_news))
ford_news.to_csv('ford1.csv', sep='\t', encoding='utf-8')
headlines = ek.get_news_headlines('EU AND POL',1)
story_id = headlines.iat[0,2]
print(ek.get_news_story(story_id))
# In[81]:
Amazon_stock.iplot(kind="lines")
# In[52]:
Amazon_stock["CLOSE"].iplot()
# In[55]:
ek.get_news_headlines("R:AMZN.O", date_from = "2020-07-08T13:00:00", date_to="2020-07-08T22:00:00", count = 5)
# In[65]:
Amazon_stock_1 = ek.get_timeseries("AMZN.O", fields="CLOSE", start_date="2019-01-01", end_date="2020-07-08")
# In[66]:
Amazon_stock_1
# In[95]:
import eikon as ek
from datetime import date
import re
import lxml.html
import pandas as pd
import win32com.client
outlook = win32com.client.Dispatch("Outlook.Application")
mail = outlook.CreateItem(0)
ek.set_app_id('<Eikon_token_id>')
#start_date, end_date = date(2017, 10, 22), date.today()
start_date, end_date = date(2017, 10, 1), date(2017, 12, 14)
q = "Product:IFRC AND Topic:ISU AND (\"PRICED\" OR \"DEAL\")"
headlines = ek.get_news_headlines(query=q, date_from=start_date, date_to=end_date, count=100)
headlines = headlines[headlines['storyId'].str.contains('nIFR')]
headlines.to_csv('test.csv')
def termsheet_to_dict(storyId):
x = ek.get_news_story(storyId)
story = lxml.html.document_fromstring(x).text_content()
matches = dict(re.findall(pattern=r"\[(.*?)\]:\s?([A-Z,a-z,0-9,\-,\(,\),\+,/,\n,\r,\.,%,\&,>, ]+)", string=story))
clean_matches = {key.strip(): item.strip() for key, item in matches.items()}
return clean_matches
result = []
index = pd.DataFrame(headlines, columns=['storyId']).values.tolist()
for i, storyId in enumerate(index):
x = termsheet_to_dict(storyId[0])
