def updateJSON_prices(sym):
dailyDat = rqstStockTSDataDaily(sym)
dat = read_articles()
for article in dat:
# round down to nearest hour
t = (datetime.datetime.strptime(
article["date_published"], "%Y-%m-%dT%H:%M:%SZ") - datetime.timedelta(hours=5)).replace(second=0, minute=0)
# print(dailyDat)
result = getTSDataDailyForceSuccessFuture(dailyDat, t, 1000)
if not result:
article["delta"] = float(0)
else:
article["delta"] = result[1]["close"] - result[1]["open"]
if t.date() == result[0].date():
article["const"] = True
else:
article["const"] = False
write_articles(dat)
def json_to_df(nkeywords):
def get_datetime(article):
return datetime.datetime.strptime(article['date_published'], "%Y-%m-%dT%H:%M:%SZ")
def get_dtnearest_hr(article):
return (get_datetime(article) - datetime.timedelta(hours=5)).replace(second=0, minute=0)
def get_timeframe(data):
dt_start = get_dtnearest_hr(data[0])
dt_end = get_dtnearest_hr(data[-1])
diff = dt_end - dt_start
tot_hours = diff.days * 24 + diff.seconds / 3600 + 1
timeframe = pd.date_range(start=str(dt_start), end=str(dt_end), periods=tot_hours)
return timeframe
data = read_articles()
keywords = read_keywords()[0:nkeywords]
hourlyStock = read_hourlyStock()
data = sorted(data, key=lambda entry: get_datetime(entry))
timeframe = get_timeframe(data)
feat_mat = np.zeros((len(timeframe), len(keywords)), int)
labels = np.array(list(hourlyStock.values()))
j = 0
for i in range(len(timeframe)):
if j < len(data) and timeframe[i] == get_dtnearest_hr(data[j]):
# labels[i] = data[j]['delta']
while j < len(data) and timeframe[i] == get_dtnearest_hr(data[j]):
feat_row = [1 if keyword in data[j]['keywords'] else 0 for keyword in keywords]
feat_mat[i] += feat_row
j += 1
else:
feat_mat[i] = feat_mat[i - 1]
X = pd.DataFrame(feat_mat, timeframe, keywords)
y = pd.DataFrame(labels, timeframe, ['stock_change'])
return X, y
from load_articles import read_articles
import json
def write_valuation(valuation_dictionary):
with open('valuation.json', 'w') as fp:
json.dump(valuation_dictionary, fp)
def read_valuation():
with open('valuation.json') as f:
return json.load(f)
if __name__ == '__main__':
articles = read_articles()
valuation_dictionary = {}
for article in articles:
valuation = article['delta']
for keyword in article['keywords']:
if keyword not in valuation_dictionary:
valuation_dictionary[keyword] = 0
valuation_dictionary[keyword] += valuation
write_valuation(valuation_dictionary)
def updateJSON_prices(sym):
APIKey = "442ONKXSVHA79170" # redact in submissions
APIbase = "https://www.alphavantage.co/query"
def rqstStockTSDataDaily(sym):
r = None
tries = 0
maxTries = 10
while not r and tries < maxTries:
r = requests.get(APIbase,
params={
"function": "TIME_SERIES_DAILY",
"symbol": sym,
"outputsize": "full",
"apikey": APIKey
})
tries += 1
if not r:
raise ValueError("Something unexpected happened.")
return r.json()["Time Series (Daily)"]
def getTSDataDaily(stockDat, t):
try:
e = stockDat[t.strftime("%Y-%m-%d")]
return {
"open": round(float(e["1. open"]), 2),
"close": round(float(e["4. close"]), 2),
"high": round(float(e["2. high"]), 2),
"low": round(float(e["3. low"]), 2),
"volume": int(e["5. volume"])
}
except KeyError:
return False
def getTSDataDailyForceSuccessFuture(stockDat, t, maxFail):
y = getTSDataDaily(stockDat, t)
i = 0
while not y:
t += datetime.timedelta(days=1)
y = getTSDataDaily(stockDat, t)
i += 1
if i == maxFail:
return False
return (t, y)
def get_datetime(article):
return datetime.datetime.strptime(article['date_published'],
"%Y-%m-%dT%H:%M:%SZ")
def get_dtnearest_hr(article):
return (get_datetime(article) - datetime.timedelta(hours=5)).replace(
second=0, minute=0)
def get_timeframe(data):
dt_start = get_dtnearest_hr(data[0])
dt_end = get_dtnearest_hr(data[-1])
diff = dt_end - dt_start
tot_hours = diff.days * 24 + diff.seconds / 3600 + 1
timeframe = pd.date_range(start=str(dt_start),
end=str(dt_end),
periods=tot_hours)
return timeframe
# Updates hourlyStock
dailyDat = rqstStockTSDataDaily(sym)
dat = read_articles()
dat = sorted(dat, key=lambda entry: get_datetime(entry))
timeframe = get_timeframe(dat)
hourlyStock = dict.fromkeys(timeframe)
for t in timeframe:
result = getTSDataDailyForceSuccessFuture(dailyDat, t, 1000)
hourlyStock[t] = result[1]["close"] - result[1]["open"]
hourlyStock = {str(k): v for k, v in hourlyStock.items()}
write_hourlyStock(hourlyStock)
# Updates the articles with delta values
for article in dat:
# round down to nearest hour
t = (datetime.datetime.strptime(article["date_published"],
"%Y-%m-%dT%H:%M:%SZ") -
datetime.timedelta(hours=5)).replace(second=0, minute=0)
# print(dailyDat)
result = getTSDataDailyForceSuccessFuture(dailyDat, t, 1000)
if not result:
article["delta"] = float(0)
else:
article["delta"] = result[1]["close"] - result[1]["open"]
if t.date() == result[0].date():
article["const"] = True
else:
article["const"] = False
write_articles(dat)