def test_open_file_handle():
xbrl_parser = XBRLParser()
file_to_parse = "tests/sam-20130629.xml"
try:
xbrl_parser.parse(file(file_to_parse))
except NameError:
pass
def test_parse_Custom10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
custom_obj = xbrl_parser.parseCustom(xbrl)
if six.PY3:
result = len(custom_obj())
assert result == 13
if six.PY2 and not __pypy__:
result = custom_obj()
assert result[0] == ('conversionofclassbcommonstocktoclassacommonstockshares', '100000')
assert result[1] == ('percentageofproductionvolumes', '0.90')
assert result[2] == ('sharebasedcompensationarrangementbysharebasedpaymentawardoptionstovestineachtranche', '5000')
assert result[3] == ('weightedaveragenumberofsharesoutstandingbasicincludingnonvestedparticipatingsecurities', '12866000')
assert result[4] == ('incrementalcommonsharesattributabletoconversionofcommonstock', '4007000')
assert result[5] == ('sharebasedcompensationarrangementbysharebasedpaymentawardinvestmentsharesweightedaveragegrantdat', '59.62')
assert result[6] == ('incomeallocatedtoequityinstrumentsotherthanoptionnonvested', '7000')
assert result[7] == ('netproceedsfromsaleofinvestmentshares', '531000')
assert result[8] == ('weightedaveragenumberofbasicsharesoutstandingequityinstrumentsotherthanoptionnonvested', '94000')
assert result[9] == ('sharebasedcompensationarrangementbysharebasedpaymentawardemployeeinvestmentsharespurchase', '12894')
assert result[10] == ('provisionforreductionofdoubtfulaccounts', '-28000')
assert result[11] == ('receiptofgovernmentgrantsforfacilitiesexpansion', '770000')
assert result[12] == ('netincomelossallocatedtoequityinstrumentsotherthanoptionnonvested', '-143000')
def test_parse_GAAP10Q_Ez_XBRL():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/ggho-20140930.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"10-Q", "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 34273.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 127.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == -1078.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == -7593.0
assert result.data['liabilities_and_equity'] == 79451.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 13026.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 30543.0
assert result.data['net_income_loss'] == -8642.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 0.0
assert result.data['equity_attributable_interest'] == 309.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 0.0
assert result.data['income_before_equity_investments'] == -8531.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 1078.0
assert result.data['gross_profit'] == 5433.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 0.0
assert result.data['interest_and_debt_expense'] == 0.0
def test_open_file_handle():
xbrl_parser = XBRLParser()
file_to_parse = "tests/sam-20130629.xml"
try:
xbrl_parser.parse(file(file_to_parse))
except NameError:
pass
def get_page(self, url):
page = requests.get(url)
xbrl_parser = XBRLParser()
# print(dir(xbrl_parser))
b = io.StringIO(page.text)
obj = xbrl_parser.parse(b)
# print(obj)
return obj
def test_parse_Custom10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
custom_obj = xbrl_parser.parseCustom(xbrl)
result = len(custom_obj())
assert result == 13
def test_parse_DEI10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
dei_obj = xbrl_parser.parseDEI(xbrl)
serializer = DEISerializer()
result = serializer.dump(dei_obj)
assert result.data['trading_symbol'] == "SAM"
assert result.data['company_name'] == "BOSTON BEER CO INC"
assert result.data['shares_outstanding'] == 4007355.0
assert result.data['public_float'] == 0.0
def test_parse_DEI10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
dei_obj = xbrl_parser.parseDEI(xbrl)
serializer = DEISerializer()
result = serializer.dump(dei_obj)
assert result.data['trading_symbol'] == "SAM"
assert result.data['company_name'] == "BOSTON BEER CO INC"
assert result.data['shares_outstanding'] == 4007355.0
assert result.data['public_float'] == 0.0
def test_parse_Custom10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
custom_obj = xbrl_parser.parseCustom(xbrl)
if six.PY3:
result = len(custom_obj())
assert result == 13
if six.PY2 and not __pypy__:
result = custom_obj()
assert result[0] == (
'conversionofclassbcommonstocktoclassacommonstockshares', '100000')
assert result[1] == ('percentageofproductionvolumes', '0.90')
assert result[2] == (
'sharebasedcompensationarrangementbysharebasedpaymentawardoptionstovestineachtranche',
'5000')
assert result[3] == (
'weightedaveragenumberofsharesoutstandingbasicincludingnonvestedparticipatingsecurities',
'12866000')
assert result[4] == (
'incrementalcommonsharesattributabletoconversionofcommonstock',
'4007000')
assert result[5] == (
'sharebasedcompensationarrangementbysharebasedpaymentawardinvestmentsharesweightedaveragegrantdat',
'59.62')
assert result[6] == (
'incomeallocatedtoequityinstrumentsotherthanoptionnonvested',
'7000')
assert result[7] == ('netproceedsfromsaleofinvestmentshares', '531000')
assert result[8] == (
'weightedaveragenumberofbasicsharesoutstandingequityinstrumentsotherthanoptionnonvested',
'94000')
assert result[9] == (
'sharebasedcompensationarrangementbysharebasedpaymentawardemployeeinvestmentsharespurchase',
'12894')
assert result[10] == ('provisionforreductionofdoubtfulaccounts',
'-28000')
assert result[11] == (
'receiptofgovernmentgrantsforfacilitiesexpansion', '770000')
assert result[12] == (
'netincomelossallocatedtoequityinstrumentsotherthanoptionnonvested',
'-143000')
def get_facts_info(self):
"""
return(element_id, amount, context_ref, unit_ref, decimals)
"""
# parse xbrl file
xbrl = XBRLParser.parse(file(self.xbrl_filename)) # beautiful soup type object
facts_dict = defaultdict(list)
#print xbrl
name_space = 'jp*'
for node in xbrl.find_all(name=re.compile(name_space+':*')):
if 'xsi:nil' in node.attrs:
if node.attrs['xsi:nil']=='true':
continue
facts_dict['element_id'].append( node.name.replace(':','_') )
facts_dict['amount'].append( node.string )
facts_dict['context_ref'].append(
self.get_attrib_value(node, 'contextref') )
facts_dict['unit_ref'].append(
self.get_attrib_value(node, 'unitref') )
facts_dict['decimals'].append(
self.get_attrib_value(node, 'decimals') )
return pd.DataFrame( facts_dict )
def parse_xbrl(self):
# parse xbrl file
with open(self.xbrl_filepath, 'r', encoding='utf-8') as of:
xbrl = XBRLParser.parse(of) # beautiful soup type object
result_dicts = defaultdict(list)
_idx = 0
#print xbrl
name_space = 'jp*'
for node in xbrl.find_all(name=re.compile(name_space + ':*')):
if self.ignore_pattern(node):
continue
row_dict = defaultdict(list)
#default tag
row_dict['file_nm'] = self.xbrl_filepath.rsplit('\\' or '/', 1)[1]
row_dict['element_id'] = node.name
row_dict['amount'] = node.string
#cutom tag
for tag in custom_tag:
row_dict[tag] = self.get_attrib_value(node, tag)
result_dicts[_idx] = row_dict
_idx += 1
return result_dicts
def parse_file(cls, xbrl_file_path):
data_container = EdinetXbrlObject()
with codecs.open(xbrl_file_path, 'r', 'utf-8') as of:
parser = XBRLParser.parse(of)
for node in parser.find_all():
cls.put_node(data_container, node)
return data_container
def parse_xbrl(self):
# parse xbrl file
with open(self.xbrl_filepath,'r',encoding='utf-8') as of:
xbrl = XBRLParser.parse(of) # beautiful soup type object
result_dicts = defaultdict(list)
_idx = 0
#print xbrl
name_space = 'jp*'
for node in xbrl.find_all(name=re.compile(name_space+':*')):
if self.ignore_pattern(node):
continue
row_dict = defaultdict(list)
#default tag
row_dict['file_nm'] = self.xbrl_filepath.rsplit('\\' or '/', 1)[1]
row_dict['element_id'] = node.name
row_dict['amount'] = node.string
#cutom tag
for tag in custom_tag:
row_dict[tag] = self.get_attrib_value(node,tag)
result_dicts[_idx] = row_dict
_idx += 1
return result_dicts
def get_facts_info(self):
"""
return(element_id, amount, context_ref, unit_ref, decimals)
"""
# parse xbrl file
xbrl = XBRLParser.parse(open(
self.xbrl_filename)) # beautiful soup type object
facts_dict = defaultdict(list)
#print xbrl
name_space = 'jp*'
for node in xbrl.find_all(name=re.compile(name_space + ':*')):
if 'xsi:nil' in node.attrs:
if node.attrs['xsi:nil'] == 'true':
continue
facts_dict['element_id'].append(node.name.replace(':', '_'))
facts_dict['amount'].append(node.string)
facts_dict['context_ref'].append(
self.get_attrib_value(node, 'contextref'))
facts_dict['unit_ref'].append(
self.get_attrib_value(node, 'unitref'))
facts_dict['decimals'].append(
self.get_attrib_value(node, 'decimals'))
return pd.DataFrame(facts_dict)
def parse_xbrl(self):
with open(self.xbrl_file, 'r', encoding=settings.encode_type) as of:
xbrl = XBRLParser.parse(of)
result_dicts = {}
_idx = 0
name_space = 'jp*'
for node in xbrl.find_all(name=re.compile(name_space + ':*')):
if self.ignore_pattern(node):
continue
row_dict = {}
row_dict['file_name'] = self.xbrl_file.rsplit(os.sep, 1)[1]
row_dict['element_id'] = node.name
row_dict['amount'] = node.string
for tag in custom_tag:
row_dict[tag] = self.get_attrib_value(node, tag)
result_dicts[_idx] = row_dict
_idx += 1
return result_dicts
import tensorflow as tf
import os
from xbrl import XBRLParser, GAAP, GAAPSerializer
num_steps = 500
batch_size = 128
data_x = []
data_y = []
test_x = []
test_y = []
for subdir, dirs, files in os.walk("data/"):
print(os.path.split(subdir)[-1])
for f in files:
xbrl_parser = XBRLParser()
print(subdir + "/" + f)
try:
xbrl = xbrl_parser.parse(subdir + "/" + f)
except:
continue
gaap_obj = xbrl_parser.parseGAAP(xbrl,
context="current",
ignore_errors=0)
print("We here 2")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
if "-13-" in f:
test_x.append([])
for i, k in result.data:
test_x[-1].append(k)
else:
def test_parse_GAAP10K_ThomsonReuters():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/aaoi-20131231.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 39057.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 0.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 177.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == -297.0
assert result.data['liabilities_and_equity'] == 111057.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 6973.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 63077.0
assert result.data['net_income_loss'] == -520.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 0.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 0.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 111057.0
assert result.data['gross_profit'] == 6676.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 77936.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0
assert result.data['common_shares_authorized'] == 0.0
#! /usr/bin/env python
# encoding: utf-8
from xbrl import XBRLParser, GAAP, GAAPSerializer
xbrl_parser = XBRLParser(precision=0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(file("../tests/sam-20130629.xml"))
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20130629",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
# Print out the serialized GAAP data
print result
# Parse just the DEI data from the xbrl object
dei_obj = xbrl_parser.parseDEI(xbrl)
# Serialize the DEI data
serializer = DEISerializer()
result = serializer.dump(dei_obj)
# Print out the serialized DEI data
#! /usr/bin/env python
# encoding: utf-8
from __future__ import print_function
from xbrl import XBRLParser, GAAP, GAAPSerializer, DEISerializer
xbrl_parser = XBRLParser(precision=0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse("../tests/sam-20130629.xml")
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20130629",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
# Print out the serialized GAAP data
print(result)
# Parse just the DEI data from the xbrl object
dei_obj = xbrl_parser.parseDEI(xbrl)
# Serialize the DEI data
serializer = DEISerializer()
result = serializer.dump(dei_obj)
def test_parse_GAAP10K_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20131228.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"10-K", "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 104377.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 0.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 9556.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 29120.0
assert result.data['liabilities_and_equity'] == 69900.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 0.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 302085.0
assert result.data['net_income_loss'] == 18079.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 0.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 0.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 1050.0
assert result.data['gross_profit'] == 104628.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 0.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
def test_parse_empty_file():
xbrl_parser = XBRLParser()
file_to_parse = "tests/nothing.xml"
with pytest.raises(XBRLParserException):
xbrl_parser.parse(file_to_parse)
def test_parse_empty_file():
xbrl_parser = XBRLParser()
file_to_parse = "tests/nothing.xml"
with pytest.raises(XBRLParserException):
xbrl_parser.parse(file_to_parse)
# -*- coding: utf-8 -*-
"""
Created on Thu May 7 12:51:22 2015
This file was built to test python-xbrl package.
@author: ptorre
"""
from xbrl import XBRLParser, GAAP, GAAPSerializer
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(file("rht-20150228.xml"))
#%%
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_type="10-Q",
context="current",
ignore_errors=0)
serialized = GAAPSerializer(gaap_obj)
print serialized.data
#! /usr/bin/env python
# encoding: utf-8
from __future__ import print_function
from xbrl import XBRLParser, GAAP, GAAPSerializer
xbrl_parser = XBRLParser(0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(file("tests/sam-20131228.xml"))
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20131228",
doc_type="10-K",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
# Print out the serialized GAAP data
print(result.data)
#! /usr/bin/env python
# encoding: utf-8
from __future__ import print_function
from xbrl import XBRLParser, GAAP, GAAPSerializer, DEISerializer
xbrl_parser = XBRLParser(0)
# Parse an incoming XBRL file
file_to_parse = "../tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20131228",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
# Print out the serialized GAAP data
print(result.data)
# Parse just the DEI data from the xbrl object
dei_obj = xbrl_parser.parseDEI(xbrl)
from xbrl import XBRLParser, GAAP, GAAPSerializer
import requests
def download_file():
global dump
url = "http://randomsite.com/file.gz"
file = requests.get(url, stream=True)
dump = file.raw
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(file("sam-20131228.xml"))
#! /usr/bin/env python
# encoding: utf-8
import os
import re
import pandas as pd
from xbrl import XBRLParser, GAAP, GAAPSerializer, DEISerializer
Extracted_fields=['liabilities', 'temporary_equity', 'other_comprehensive_income', 'assets', 'common_shares_issued', 'other_operating_income', 'commitments_and_contingencies', 'income_loss', 'current_assets', 'operating_expenses', 'interest_and_debt_expense', 'common_shares_authorized', 'comprehensive_income_parent', 'net_income_loss', 'common_shares_outstanding', 'net_income_loss_noncontrolling', 'net_cash_flows_investing_discontinued', 'net_cash_flows_discontinued', 'nonoperating_income_loss', 'net_cash_flows_financing_continuing', 'equity_attributable_parent', 'noncurrent_liabilities', 'income_tax_expense_benefit', 'revenue', 'net_cash_flows_investing_continuing', 'cost_of_revenue', 'comprehensive_income', 'net_cash_flows_operating_discontinued', 'stockholders_equity', 'income_from_equity_investments', 'comprehensive_income_interest', 'net_income_shareholders', 'redeemable_noncontrolling_interest', 'net_cash_flows_financing', 'equity_attributable_interest', 'extraordary_items_gain_loss', 'preferred_stock_dividends', 'income_before_equity_investments', 'net_cash_flows_operating', 'equity', 'non_current_assets', 'net_income_parent', 'net_cash_flows_operating_continuing', 'gross_profit', 'costs_and_expenses', 'liabilities_and_equity', 'operating_income_loss', 'net_cash_flows_investing', 'current_liabilities']
xbrl_parser = XBRLParser(precision=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
df = pd.DataFrame(columns=Extracted_fields)
df.index.names = ['Company-date']
xml_path="~/apps/repo/python-xbrl/tests"
for root, dirs, files in os.walk("."):
path = root.split(os.sep)
print((len(path) -1) * '---', os.path.basename(root))
for file in files:
print (file)
if re.match(r"[a-z]{3,5}\-[0-9]{8}\.xml",file):
#if re.match(r"aaww-20131231.xml",file):
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(file)
print("Process {}".format(os.path.join(root,file)))
date_of_doc = re.search(r"[0-9]{8}",file).group()
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date=date_of_doc,
context="current",
ignore_errors=0)
def test_parse_GAAP10Q_QXInteractive():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/aaoi-20140630.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"10-Q", "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 5606.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 85.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 978.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 1730.0
assert result.data['liabilities_and_equity'] == 153524.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 9458.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 111781.0
assert result.data['net_income_loss'] == 1919.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 2058.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 2058.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 106114.0
assert result.data['gross_profit'] == 11188.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 46431.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
def data_processing(self,
elements,
xbrl,
ignore_errors,
logger,
context_ids=[],
**kwargs):
"""
Process a XBRL tag object and extract the correct value as
stated by the context.
"""
options = kwargs.get('options', {
'type': 'Number',
'no_context': False
})
if options['type'] == 'String':
if len(elements) > 0:
return elements[0].text
if options['no_context'] == True:
if len(elements) > 0 and XBRLParser().is_number(elements[0].text):
return elements[0].text
try:
# Extract the correct values by context
correct_elements = []
for element in elements:
contextrefs = element.attrs['contextref'].split('_')
if any(
[context_id in contextrefs for context_id in context_ids]):
correct_elements.append(element)
elements = correct_elements
if len(elements) > 0 and XBRLParser().is_number(elements[0].text):
return float(elements[0].text)
else:
return 0
# if len(elements) > 0 and XBRLParser().is_number(elements[0].text):
# decimals = elements[0].attrs['decimals']
# if decimals is not None:
# attr_precision = decimals
# if xbrl.precision != 0 and xbrl.precison != attr_precision:
# xbrl.precision = attr_precision
# if elements:
# return XBRLParser().trim_decimals(elements[0].text, int(xbrl.precision))
# else:
# return 0
# else:
# return 0
except Exception as e:
if ignore_errors == 0:
raise XBRLParserException('value extraction error')
elif ignore_errors == 1:
return 0
elif ignore_errors == 2:
logger.error(str(e) + " error at " + ''.join(elements[0].text))
def test_parse_GAAP10Q_Ez_XBRL():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/ggho-20140930.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 34273.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 127.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 58615.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == -7593.0
assert result.data['liabilities_and_equity'] == 79451.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 13026.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 30543.0
assert result.data['net_income_loss'] == -8642.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 0.0
assert result.data['equity_attributable_interest'] == 309.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 0.0
assert result.data['income_before_equity_investments'] == -8531.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 79451.0
assert result.data['gross_profit'] == 5433.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 20836.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0
assert result.data['common_shares_authorized'] == 0.0
def test_parse_GAAP10Q_QXInteractive():
xbrl_parser = XBRLParser()
file_to_parse = "tests/aaoi-20140630.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 5606.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 85.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 978.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 1730.0
assert result.data['liabilities_and_equity'] == 153524.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 9458.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 111781.0
assert result.data['net_income_loss'] == 1919.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 2058.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 2058.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 153524.0
assert result.data['gross_profit'] == 11188.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 106114.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 14798000.0
assert result.data['common_shares_issued'] == 14798000.0
assert result.data['common_shares_authorized'] == 45000000.0
from xbrl import XBRLParser, GAAP, GAAPSerializer
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(
open(
"/Users/f1v-13/Downloads/0001558370-20-001080-xbrl/adc-20191231x10k4f0b3a.htm"
))
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20131228",
context="current",
ignore_errors=0)
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
print(result.data)
def test_parse_GAAP10Q_Rivet():
xbrl_parser = XBRLParser()
file_to_parse = "tests/c289-20140503.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 12535.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 14.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 708.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == -983.0
assert result.data['liabilities_and_equity'] == 13261.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 3497.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 726.0
assert result.data['net_income_loss'] == -983.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == -977.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == -977.0
assert result.data['income_before_equity_investments'] == -969.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 13261.0
assert result.data['gross_profit'] == 2687.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 10747.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 146033000.0
assert result.data['common_shares_authorized'] == 650000000.0
#! /usr/bin/env python
# encoding: utf-8
import pandas as pd
from xbrl import XBRLParser, GAAP, GAAPSerializer, DEISerializer
xbrl_parser = XBRLParser(precision=0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(open("../tests/aaww-20131231.xml"))
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20131231",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
attribute = [x.encode('ascii') for x in result[0].keys()]
values = result[0].values()
df = pd.DataFrame(columns=attribute, index=['aaww'])
df.loc['aaww'] = values
print(df)
## Parse just the DEI data from the xbrl object
#dei_obj = xbrl_parser.parseDEI(xbrl)
#
## Serialize the DEI data
#serializer = DEISerializer()
def test_parse_GAAP10K_Fujitsu():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/aaww-20131231.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 194292.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 0.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 3124306.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 0.0
assert result.data['liabilities_and_equity'] == 3718259.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 4870.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 0.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 1317773.0
assert result.data['net_income_loss'] == 0.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 0.0
assert result.data['equity_attributable_interest'] == 4352.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 0.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 3718259.0
assert result.data['gross_profit'] == 0.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 593953.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0
assert result.data['common_shares_authorized'] == 0.0
#! /usr/bin/env python
# encoding: utf-8
from __future__ import print_function
from xbrl import XBRLParser, GAAP, GAAPSerializer, DEISerializer
xbrl_parser = XBRLParser(precision=0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(
"../../essembly-cli/tests/import_files/Journal_Entry_704171.xml")
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20130629",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
# Print out the serialized GAAP data
print(result)
# Parse just the DEI data from the xbrl object
dei_obj = xbrl_parser.parseDEI(xbrl)
# Serialize the DEI data
serializer = DEISerializer()
#filename = sys.argv[1]
filename = "/Users/thw/test.xml"
#xbrl_import = xml.dom.minidom.parse( filename)
#for record in xbrl_import.getElementsByTagName('ko'):
# name = record.getAttribute('unitRef')
# print(record)
#print(str(xbrl_import))
from xbrl import XBRLParser, GAAP, GAAPSerializer
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(filename)
#Revenue = xbrl.find_all(name=re.compile("(fsa:GrossProfitLoss$)", re.IGNORECASE | re.MULTILINE))
Gross_Profit = xbrl.find_all(name=re.compile("(fsa:GrossProfitLoss$)", re.IGNORECASE | re.MULTILINE))
'''
Selling_and_General_Expenses = xbrl.find_all(name=re.compile("(us-gaap:SellingGeneralAndAdministrativeExpense$)", re.IGNORECASE | re.MULTILINE))
Depreciation = xbrl.find_all(name=re.compile("(us-gaap:DepreciationDepletionAndAmortization$)", re.IGNORECASE | re.MULTILINE))
Operating_Income = xbrl.find_all(name=re.compile("(us-gaap:OperatingIncomeLoss$)", re.IGNORECASE | re.MULTILINE))
Interest_Expense = xbrl.find_all(name=re.compile("(us-gaap:InterestExpense$)", re.IGNORECASE | re.MULTILINE))
from xbrl import XBRLParser, GAAP, GAAPSerializer
# https://pypi.org/project/python-xbrl/
homedir = '../data/'
addondir = 'xbrl/'
specificfile = 'rf-20190331.xml'
fileLoc = homedir + addondir + specificfile
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(fileLoc)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20190401",
context="current",
ignore_errors=0)
# Serialize the GAAP data
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
pass
#! /usr/bin/env python
# encoding: utf-8
from xbrl import XBRLParser
from xbrl_middleware import ExcelDump, OpenPyxlAdapter
xbrl_parser = XBRLParser(precision=0)
# Parse an incoming XBRL file
xbrl = xbrl_parser.parse(file("sam-20130629.xml"))
# Parse just the GAAP data from the xbrl object
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20130629",
doc_type="10-Q",
context="current",
ignore_errors=0)
# Dump the GAAP data to XLSX (Excel)
xbrl_dump = ExcelDump(gaap_obj)
dumper = xbrl_middleware.OpenPyxlAdapter(xbrl_dump)
dumper.write('demo.xlsx')
dumper.dump()
def test_parse_GAAP10Q_Fujitsu():
xbrl_parser = XBRLParser()
file_to_parse = "tests/aaww-20140630.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 233079.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == -23815.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 414512.0
assert result.data['non_current_assets'] == 3568474.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 26657.0
assert result.data['liabilities_and_equity'] == 4100064.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 4870.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 0.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 1355551.0
assert result.data['net_income_loss'] == 0.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == -515.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == -515.0
assert result.data['income_before_equity_investments'] == 0.0
assert result.data['comprehensive_income_interest'] == -515.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 4100064.0
assert result.data['gross_profit'] == 0.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 531590.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == -612.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 25256124.0
assert result.data['common_shares_issued'] == 28547138.0
assert result.data['common_shares_authorized'] == 50000000.0
def test_parse_GAAP10Q_Webfilings():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/goog-20140630.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 3683.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 913.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 11697.0
assert result.data['non_current_assets'] == 43703.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 3490.0
assert result.data['liabilities_and_equity'] == 121608.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 0.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 95749.0
assert result.data['net_income_loss'] == 3422.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 3579.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 3579.0
assert result.data['income_before_equity_investments'] == 4403.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 121608.0
assert result.data['gross_profit'] == 0.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 77905.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0
assert result.data['common_shares_authorized'] == 0.0
def test_parse_GAAP10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(
xbrl,
str(
file_to_parse.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1].split(".")[0][4:6] +
file_to_parse.split("-")[1].split(".")[0][6:8]), "current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 98032.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 12107.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 5417.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 19715.0
assert result.data['liabilities_and_equity'] == 60263.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 65084.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 253536.0
assert result.data['net_income_loss'] == 19715.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 19715.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 19715.0
assert result.data['income_before_equity_investments'] == 31822.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 376766.0
assert result.data['gross_profit'] == 97132.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 138996.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0
from xbrl import XBRLParser, GAAPSerializer, DEISerializer
xbrl_parser = XBRLParser()
xbrl = xbrl_parser.parse(open("sample_report/apple-2013-qtr1-10Q-xbrl.xml"))
gaap_obj = xbrl_parser.parseGAAP(xbrl,
doc_date="20121229",
context="current",
ignore_errors=0)
print("Cash generated by operating activities: " +
str(gaap_obj.net_cash_flows_operating))
print("Cash generated by/(used in) investing activities: " +
str(gaap_obj.net_cash_flows_investing))
print("Cash used in financing activities: " +
str(gaap_obj.net_cash_flows_financing))
print("Total shareholders’ equity: " + str(gaap_obj.stockholders_equity))
print("Total assets: " + str(gaap_obj.assets))
print("Operating income: " + str(gaap_obj.operating_income_loss))
print("Net income: " + str(gaap_obj.net_income_loss))
print("Liabilities: " + str(gaap_obj.assets - gaap_obj.stockholders_equity))
print("Total net sales: " + str(gaap_obj.revenues))
dei_obj = xbrl_parser.parseDEI(xbrl)
# Serialize the DEI data
serializer = DEISerializer()
result = serializer.dump(dei_obj)
# serializer = GAAPSerializer()
# result = serializer.dump(gaap_obj)
#
# print(result.data)
def test_parse_GAAP10Q_RRDonnelley():
xbrl_parser = XBRLParser(0)
file_to_parse = "tests/sam-20130629.xml"
xbrl = xbrl_parser.parse(file_to_parse)
gaap_obj = xbrl_parser.parseGAAP(xbrl,
str(file_to_parse
.split("-")[1].split(".")[0][:4] +
file_to_parse.split("-")[1]
.split(".")[0][4:6] +
file_to_parse.split("-")[1]
.split(".")[0][6:8]),
"current")
serializer = GAAPSerializer()
result = serializer.dump(gaap_obj)
assert result.data['liabilities'] == 98032.0
assert result.data['net_cash_flows_financing_continuing'] == 0.0
assert result.data['revenue'] == 0.0
assert result.data['income_tax_expense_benefit'] == 12107.0
assert result.data['income_from_equity_investments'] == 0.0
assert result.data['preferred_stock_dividends'] == 0.0
assert result.data['redeemable_noncontrolling_interest'] == 0.0
assert result.data['extraordary_items_gain_loss'] == 0.0
assert result.data['temporary_equity'] == 0.0
assert result.data['costs_and_expenses'] == 0.0
assert result.data['non_current_assets'] == 5417.0
assert result.data['net_cash_flows_discontinued'] == 0.0
assert result.data['income_loss'] == 19715.0
assert result.data['liabilities_and_equity'] == 60263.0
assert result.data['other_operating_income'] == 0.0
assert result.data['operating_income_loss'] == 0.0
assert result.data['net_income_parent'] == 0.0
assert result.data['equity'] == 0.0
assert result.data['net_cash_flows_operating_discontinued'] == 0.0
assert result.data['cost_of_revenue'] == 0.0
assert result.data['operating_expenses'] == 65084.0
assert result.data['noncurrent_liabilities'] == 0.0
assert result.data['current_liabilities'] == 0.0
assert result.data['net_cash_flows_investing'] == 0.0
assert result.data['stockholders_equity'] == 253536.0
assert result.data['net_income_loss'] == 19715.0
assert result.data['net_cash_flows_investing_continuing'] == 0.0
assert result.data['nonoperating_income_loss'] == 0.0
assert result.data['net_cash_flows_financing'] == 0.0
assert result.data['net_income_shareholders'] == 0.0
assert result.data['comprehensive_income'] == 19715.0
assert result.data['equity_attributable_interest'] == 0.0
assert result.data['commitments_and_contingencies'] == 0.0
assert result.data['comprehensive_income_parent'] == 19715.0
assert result.data['income_before_equity_investments'] == 31822.0
assert result.data['comprehensive_income_interest'] == 0.0
assert result.data['other_comprehensive_income'] == 0.0
assert result.data['equity_attributable_parent'] == 0.0
assert result.data['assets'] == 376766.0
assert result.data['gross_profit'] == 97132.0
assert result.data['net_cash_flows_operating_continuing'] == 0.0
assert result.data['current_assets'] == 138996.0
assert result.data['interest_and_debt_expense'] == 0.0
assert result.data['net_income_loss_noncontrolling'] == 0.0
assert result.data['net_cash_flows_operating'] == 0.0
assert result.data['common_shares_outstanding'] == 0.0
assert result.data['common_shares_issued'] == 0.0