def get_data_from_one_year(company_name, year):
stock = Stock(company_name)
data = []
for i in range(1, 5):
filing = stock.get_filing(period='quarterly', year=year, quarter=i)
try:
income_statement = filing.get_income_statements().reports[0].map
#balance_sheet = filing.get_balance_sheets().reports[0].map
#cash_flow = filing.get_cash_flows().reports[0].map
except:
continue
one_quarter_data = income_statement
if filing.get_balance_sheets():
balance_sheet = filing.get_balance_sheets().reports[0].map
one_quarter_data.update(balance_sheet)
if filing.get_cash_flows():
cash_flow = filing.get_cash_flows().reports[0].map
one_quarter_data.update(cash_flow)
#one_quarter_data.update(balance_sheet)
#one_quarter_data.update(cash_flow)
# trick to maintain the same structure of the data
my_date_obj = namedtuple('MyStruct', 'value')
date = my_date_obj(value=f"{year}_{i}")
one_quarter_data["date"] = date
data.append(one_quarter_data)
return data
def test_get_cash_flows():
stock = Stock(symbol='SPWR')
filing = stock.get_filing(period='quarterly', year=2018, quarter=4)
result = filing.get_cash_flows()
print(FinancialReportEncoder().encode(result)) # for easy QA using JSON
# ensure certain data points are correct
profit_loss = result.reports[0].map['us-gaap_ProfitLoss'].value
assert profit_loss == -745351000.0
def test_get_statement_of_financial_position():
# synonymous with balance sheet
stock = Stock(symbol='IBM')
filing = stock.get_filing(period='annual', year=2018, quarter=1)
result = filing.get_balance_sheets()
print(FinancialReportEncoder().encode(result)) # for easy QA using JSON
# ensure certain data points are correct
assets = result.reports[0].map['us-gaap_Assets'].value
assert assets == 125356000000.0
def test_get_balance_sheets():
stock = Stock(symbol='SPWR')
filing = stock.get_filing(period='quarterly', year=2018, quarter=4)
result = filing.get_balance_sheets()
print(FinancialReportEncoder().encode(result)) # for easy QA using JSON
# ensure certain data points are correct
assets = result.reports[0].map['us-gaap_Assets'].value
assert assets == 3126117000.0
def test_get_statement_of_earnings():
# synonymous with income statement
stock = Stock(symbol='IBM')
filing = stock.get_filing(period='annual', year=2018, quarter=1)
result = filing.get_income_statements()
print(FinancialReportEncoder().encode(result)) # for easy QA using JSON
# ensure certain data points are correct
revenue = result.reports[0].map['us-gaap_Revenues'].value
assert revenue == 79139000000.0
def test_get_income_statements():
stock = Stock(symbol='AAPL')
filing = stock.get_filing(period='quarterly', year=2016, quarter=1)
result = filing.get_income_statements()
print(FinancialReportEncoder().encode(result)) # for easy QA using JSON
# ensure certain data points are correct
revenue = result.reports[0].map['us-gaap_SalesRevenueNet'].value
assert revenue == 75872000000.0
def test_get_filing_no_filing_found_exception():
stock = Stock(symbol='FB')
try:
# IPO was in 2012
filing = stock.get_filing(period='quarterly', year=2011, quarter=1)
assert False
except NoFilingInfoException:
assert True
def get_data_from_one_year(company_name, year):
stock = Stock(company_name)
one_year_income_statements = []
one_year_balance_sheets = []
one_year_cash_flows = []
for i in range(1, 5):
filing = stock.get_filing(period='quarterly', year=year, quarter=i)
income_statements = filing.get_income_statements().reports
balance_sheets = filing.get_balance_sheets().reports
cash_flows = filing.get_cash_flows().reports
income_statement = None
for report in income_statements:
# choose the report from the correct year, covering only 3 months (not cumulative)
report_year = int(report.date.year)
report_period = int(report.months)
# this condition can be relaxed later
if report_year == year and report_period == 3:
income_statement = report.map
income_statement['exact_date'] = report.date
income_statement['period'] = 3
break
one_year_income_statements.append(income_statement)
balance_sheet = None
for report in balance_sheets:
report_year = int(report.date.year)
if report_year == year:
balance_sheet = report.map
balance_sheet['exact_date'] = report.date
balance_sheet['period'] = None
break
one_year_balance_sheets.append(balance_sheet)
cash_flow = None
for report in cash_flows:
report_year = int(report.date.year)
report_period = int(report.months)
if report_year == year:
cash_flow = report.map
cash_flow['exact_date'] = report.date
cash_flow['period'] = report_period
break
one_year_cash_flows.append(cash_flow)
return {
"income_statements": one_year_income_statements,
"balance_sheets": one_year_balance_sheets,
"cash_flows": one_year_cash_flows
}
def getData(ticket, ty, ignored_cmp, period='annual', year=2018, quarter=0):
# period = 'annual' # or 'quarterly', which is the default
# year = 2018 # can use default of 0 to get the latest
# quarter = 1 # 1, 2, 3, 4, or default value of 0 to get the latest
try:
stock = Stock(ticket)
filing = stock.get_filing(period, year, quarter)
# financial reports (contain data for multiple years)
if ty == 'income_statements':
statements = filing.get_income_statements()
elif ty == "balance_sheets":
statements = filing.get_balance_sheets()
elif ty == "cash_flows":
statements = filing.get_cash_flows()
jsonstr = FinancialReportEncoder().encode(statements)
data = json.loads(jsonstr)
# print(data.keys()) # dict_keys(['company', 'date_filed', 'reports'])
listreports = data['reports']
columns = []
data = []
map_keys = []
for report in listreports:
# print(report.keys()) # dict_keys(['date', 'months', 'map']): string, int, dict
for report_key in report.keys():
if report_key == 'map':
for map_key in report['map'].keys():
map_keys.append(map_key)
# print(map_key, report['map'][map_key].keys()) # dict_keys(['label', 'value'])
for key in report['map'][map_key].keys():
# print(key, report['map'][map_key][key])
if key == 'label':
columns.append(report['map'][map_key][key])
else:
data.append(report['map'][map_key][key])
sec_data = pd.DataFrame([data], columns=columns)
sec_data = sec_data.transpose()
sec_data.reset_index(inplace=True)
sec_data['ind'] = np.arange(len(sec_data))
sec_data.set_index("ind", inplace=True)
map_data = pd.Series(map_keys)
return sec_data, True, ignored_cmp
except:
print('Invalid company input', cmp, ty, period, year, quarter)
ignored_cmp.append(cmp)
return None, False, ignored_cmp
#
#
# # In[18]:
#
#
cik_not_found = []
year_not_found = []
income_statements_error = []
balance_sheets_error = []
cash_flows_error = []
for y in range(2017, 2018):
for k in stock_list:
try:
stock = Stock(k)
except:
cik_not_found.append(k)
continue
period = 'quarterly'
try:
filing = stock.get_filing('annual', y, 1)
except:
year_not_found.append([k, y])
try:
income_statements = filing.get_income_statements()
newdict_income_statements = income_statements.reports[0]
except:
income_statements_error.append(k)
def filing(cikList):
# This is inspired from examples
# This uses the full version of the library, not the lite version, therefore not using it
from edgar.stock import Stock
import datetime
tickers = []
'''with open('../data/spy500 - test.csv') as csv_file:
csv_reader = csv.reader(csv_file, delimiter=',')
for row in csv_reader:
tickers.append(row[0])'''
tickers = ['XYL']
print(tickers)
for tuple in cikList:
print('********************************')
print('CIK: %s' % tuple[0])
print('URL: %s' % tuple[1])
try:
# stock = Stock(ticker)
stock = Stock(cik=tuple[0].lstrip('0'))
except IndexError as e:
print(e)
continue
period = 'quarterly' # or 'annual', which is the default
year = 0 # can use default of 0 to get the latest
quarter = 0 # 1, 2, 3, 4, or default value of 0 to get the latest
# using defaults to get the latest annual, can simplify to stock.get_filing()
# Get CIK from .csv, get 10-K / 10-Q / file link from master.idx lookup, process .txt
try:
filing = stock.get_filing(period,
year,
quarter,
filing_url=tuple[1])
except Exception as e:
print(e)
continue
# financial reports (contain data for multiple years)
try:
income_statements = filing.get_income_statements()
except Exception as e:
print(e)
if str(e) == 'list index out of range':
continue
# balance_sheets = filing.get_balance_sheets()
# cash_flows = filing.get_cash_flows()
labelDict = {
'revenue': [
'us-gaap_RevenueFromContractWithCustomerExcludingAssessedTax',
'us-gaap_RevenueFromContractWithCustomerIncludingAssessedTax',
'us-gaap_Revenues'
],
'netIncome': [
'us-gaap_NetIncomeLoss',
'us-gaap_NetIncomeLossAvailableToCommonStockholdersBasicAbstract',
'us-gaap_NetIncomeLossAvailableToCommonStockholdersBasic', ''
],
'epsBasic': [
'us-gaap_EarningsPerShareBasic',
'us-gaap_EarningsPerShareBasicAndDiluted',
'us-gaap_IncomeLossFromContinuingOperationsPerBasicShare'
],
'epsDiluted': [
'us-gaap_EarningsPerShareDiluted',
'us-gaap_EarningsPerShareBasicAndDiluted',
'us-gaap_IncomeLossFromContinuingOperationsPerDilutedShare'
]
}
for key in labelDict:
for label in labelDict[key][:]:
if label not in income_statements.reports[0].map and ('defref_' + label) not in \
income_statements.reports[
0].map:
labelDict[key].remove(label)
for report in income_statements.reports:
print(report.date)
print("Last %d months" % report.months)
for key in labelDict:
print(key + ':')
for label in labelDict[key]:
print(label)
try:
print(report.map[label])
except KeyError as e:
print(e)
continue
def test_init_unknown_symbol():
try:
Stock(symbol='ZZZZZZZZZZZZZZZ')
assert False
except IndexError:
assert True
def test_get_filing():
stock = Stock(symbol='AAPL')
filing = stock.get_filing(period='quarterly', year=2016, quarter=1)
assert True
def test_init():
stock = Stock(symbol='AAPL')
assert stock.symbol == 'AAPL'
assert stock.cik == '320193'
from edgar.stock import Stock
stockAlexion = Stock('ALXN')
stockRegeneron = Stock('REN')
stockVertex = Stock('VRTX')
stockBiomarin = Stock('BMRN')
stockIncyte = Stock('INCY')
period = 'quarterly'
filingAlexion2018Q4 = stockAlexion.get_filing(period, 2018, 4)
filingAlexion2019Q1 = stockAlexion.get_filing(period, 2019, 1)
filingAlexion2019Q2 = stockAlexion.get_filing(period, 2019, 2)
filingAlexion2019Q3 = stockAlexion.get_filing(period, 2019, 3)
filingAlexion2019Q4 = stockAlexion.get_filing(period, 2019, 4)
filingAlexion2020Q1 = stockAlexion.get_filing(period, 2020, 1)
bSheetAlexion2018Q4 = filingAlexion2018Q4.get_balance_sheets()
bSheetAlexion2019Q1 = filingAlexion2019Q1.get_balance_sheets()
bSheetAlexion2019Q2 = filingAlexion2019Q2.get_balance_sheets()
bSheetAlexion2019Q3 = filingAlexion2019Q3.get_balance_sheets()
bSheetAlexion2019Q4 = filingAlexion2019Q4.get_balance_sheets()
bSheetAlexion2020Q1 = filingAlexion2020Q1.get_balance_sheets()
filingRegeneron2018Q4 = stockRegeneron.get_filing(period, 2018, 4)
filingRegeneron2019Q1 = stockRegeneron.get_filing(period, 2019, 1)
filingRegeneron2019Q2 = stockRegeneron.get_filing(period, 2019, 2)
filingRegeneron2019Q3 = stockRegeneron.get_filing(period, 2019, 3)
filingRegeneron2019Q4 = stockRegeneron.get_filing(period, 2019, 4)
filingRegeneron2020Q1 = stockRegeneron.get_filing(period, 2020, 1)