def setUpClass(cls):
#----------------------
# set up db
#----------------------
cls.db = MySQLdb.connect(
ks_db_settings.setting('host'),
ks_db_settings.setting('user'),
ks_db_settings.setting('password'),
ks_db_settings.setting('database'))
cls.ks_fh = filehandler(cls.db)
cls.ks_fh.reset()
cls.company_id = 1
register_raw_files("./tests/data2/Sales.csv", cls.company_id, cls.db)
register_raw_files("./tests/data2/Currencyv2.csv", cls.company_id, cls.db)
register_raw_files("./tests/data2/CountryRegion.csv", cls.company_id, cls.db)
register_raw_files("./tests/data2/ComissionTax.csv",cls.company_id, cls.db)
ks_precompute = precompute(cls.db)
ks_precompute.reset()
precompute
ks_merge = merge(cls.db)
load_precompute_normalize_URL(cls.company_id, cls.db)
id = ks_precompute.getMaxBigTableIdForCompany(cls.company_id)
ks_merge = merge(cls.db)
mergeBigTable = ks_merge.getTables()
metaData = ks_merge.getMetaDataFromTable(mergeBigTable[0])
cls.ks_analytics = analytics(cls.db)
newBigTable = "BigTable"+ str(ks_precompute.getMaxBigTableIdForCompany(cls.company_id))
cls.ks_analytics.reset()
cls.ks_analytics.addBigTable(mergeBigTable[0], newBigTable, metaData)
#clean up
sql ="update %s set TaxRate = TaxRate/100;"%("analytics."+newBigTable)
cls.db.cursor().execute(sql)
# ProductType changed from D to M see documentation of test case
sql ="update %s set ProductType = 'M' where VendorId='0268_20140114_SOFA_ENGLIS' and DownloadDate='6/1/14';"%("analytics."+newBigTable)
cls.db.cursor().execute(sql)
cls.ks_analytics.addFactUsingBinaryOpAPI("NET_REVENUE", "Units", "RoyaltyPrice", "*", newBigTable)
cls.ks_analytics.addFactUsingBinaryOpAPI("TAXES", "NET_REVENUE","TaxRate","*", newBigTable)
cls.ks_analytics.addFactUsingBinaryOpAPI("REVENUE_AFTER_TAX", "NET_REVENUE","TAXES","-", newBigTable)
cls.ks_fh.registerFormula("", "Plus", "Plus", "Units+RoyaltyPrice", "sum")
cls.ks_fh.registerFormula("", "Mult", "Mult", "Units*RoyaltyPrice", "sum")
cls.ks_fh.registerFormula("", "Individual_Tax", "Individual_Tax", "RoyaltyPrice*TaxRate", "sum")
cls.ks_fh.registerFormula("", "NET_REVENUE", "NET_REVENUE", "Units*RoyaltyPrice", "sum")
cls.ks_fh.registerFormula("", "SumPlus", "SumPlus", "Sum(Units)+Sum(RoyaltyPrice)", "sum")
cls.ks_fh.registerFormula("", "SumMult", "SumMult", "Sum(Units)*Sum(RoyaltyPrice)", "sum")
cls.ks_fh.registerFormula("", "Individual_TaxSum", "Individual_TaxSum", "Sum(RoyaltyPrice)*Sum(TaxRate)", "sum")
cls.ks_fh.registerFormula("", "NonsenseSum", "Individual_Tax", "Sum(RoyaltyPrice)+Sum(TaxRate)", "sum")
cls.ks_fh.registerFormula("", "Nonsense", "Individual_Tax", "RoyaltyPrice+TaxRate", "sum")
cls.ks_fh.registerFormula("", "REVENUE_AFTER_TAX", "REVENUE_AFTER_TAX", "", "sum")
def test_load_files(self):
ks_fh = filehandler(self.db)
ks_fh.reset()
company_id = 1
register_raw_files("./tests/data2/Sales.csv", company_id, self.db)
register_raw_files("./tests/data2/Currencyv2.csv", company_id, self.db)
register_raw_files("./tests/data2/CountryRegion.csv", company_id, self.db)
register_raw_files("./tests/data2/ComissionTax.csv",company_id, self.db)
ks_precompute = precompute(self.db)
ks_precompute.reset()
precompute
ks_merge = merge(self.db)
load_precompute_normalize(company_id, self.db)
ks_fh.registerFormula("", "Plus", "Plus", "Units+RoyaltyPrice", "sum")
ks_fh.registerFormula("", "Mult", "Mult", "Units*RoyaltyPrice", "sum")
plus_id = ks_fh.getMeasureID("Plus")
mult_id = ks_fh.getMeasureID("Mult")
units_id = ks_fh.getMeasureID("Units")
royality_id = ks_fh.getMeasureID("RoyaltyPrice")
# MEASURE DATA DEMO raw_facts + measures with formulas
print(measure_data(self.db, company_id, [plus_id,mult_id,units_id,royality_id],"day","2014-06-01","2014-06-01"))
# MEASURE DATA DEMO raw_facts group by
print(measure_data(self.db, company_id, [units_id, royality_id],"day","2014-06-01","2014-06-01","Region"))
self.db.commit()
def load_precompute_normalize_URL(company_name, db):
ks_fh = filehandler(db)
rows = ks_fh.getLatestTablesByCompany(company_name)
ks_merge = merge(db)
ks_merge.reset()
for row in rows:
table_name =row[1]
file_path = row[2]
ks_merge.addTableURL(file_path, table_name)
ks_merge.automaticMerge()
mergeBigTable = ks_merge.getTables()
ks_precompute = precompute(db)
meta_data = ks_merge.getMetaDataFromTable(mergeBigTable[0])
ks_precompute.reset()
ks_precompute.addBigTable(meta_data,mergeBigTable[0],company_name)
id = ks_precompute.getMaxBigTableIdForCompany(company_name)
metaData = ks_merge.getMetaDataFromTable(mergeBigTable[0])
ks_analytics = analytics(db)
newBigTable = "BigTable"+ str(ks_precompute.getMaxBigTableIdForCompany(company_name))
ks_analytics.reset()
ks_analytics.addBigTable(mergeBigTable[0], newBigTable, metaData)
def test_AddBigTable(self):
# compute BigTable
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
third_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
fourth_table = "./ks_filehandler/ks_filehandler/data/ComissionTax.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table, "Sales")
ks_merge.addTable(second_table, "Currencyv2")
ks_merge.addTable(third_table, "CountryRegion")
ks_merge.addTable(fourth_table, "ComissionTax")
# ks_merge.automaticMerge()
ks_precompute = precompute(self.db)
meta_data = {
"VendorId": "dim",
"ProductType": "dim",
"Units": "fact",
"RoyaltyPrice": "fact",
"DownloadDate": "date",
"CustomerCurrency": "dim",
"CountryCode": "dim",
"Region": "dim",
"ExchangeRate": "fact",
"TaxRate": "fact",
"RightsHolder": "dim",
"ComissionRate": "fact",
"id": "sys",
}
ks_precompute.reset()
ks_precompute.addBigTable(meta_data, "Sales", 1)
def test_Two_Tables_One_Link2(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"Currencyv2")
self.assertEqual(False, ks_merge.isUniqueOneLink(first_table, second_table))
def test_Two_Tables_One_Link1(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"CountryRegion")
self.assertEqual(True, ks_merge.isUniqueOneLink("Sales", "CountryRegion"))
def test_JOIN_TWO_TABLES_TWO_UNIQUE_LINK(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"Currencyv2")
print (ks_merge.isUniqueTwoLinks("Sales", "Currencyv2"))
ks_merge.joinUniqueTwoLinks("Sales", "Currencyv2")
def test_Two_Tables_One_Link_Case3(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"CountryRegion")
general_links = generalLinksDB(["Sales","CountryRegion"], ks_merge)
self.assertEqual(False, general_links.isEdge('Sales.csv:Cou ntryCode', 'CountryRegion.csv:CountryCode'))
self.db.commit()
def test_JOIN_TWO_TABLES_ONE_UNIQUE_LINK(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"CountryRegion")
print (ks_merge.getLinks())
ks_merge.joinUniqueOneLink("Sales", "CountryRegion")
print (ks_merge.getLinks())
def test_AutomaticFourTableMerge(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
third_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
fourth_table = "./ks_filehandler/ks_filehandler/data/ComissionTax.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"Currencyv2")
ks_merge.addTable(third_table,"CountryRegion")
ks_merge.addTable(fourth_table,"ComissionTax")
ks_merge.automaticMerge()
def test_JOIN_TABLEWISE(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"Currencyv2")
ks_merge.joinUniqueTwoLinks("Sales", "Currencyv2")
third_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
ks_merge.addTable(third_table,"CountryRegion")
ks_merge.joinUniqueOneLink("Sales", "CountryRegion")
fourth_table = "./ks_filehandler/ks_filehandler/data/ComissionTax.csv"
ks_merge.addTable(fourth_table,"ComissionTax")
ks_merge.joinUniqueTwoLinks("Sales", "ComissionTax")
def test_Four_Tables_Five_Links(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
second_table = "./ks_filehandler/ks_filehandler/data/graph/CountryRegion.csv"
third_table = "./ks_filehandler/ks_filehandler/data/ComissionTax.csv"
fourth_table = "./ks_filehandler/ks_filehandler/data/graph/Currencyv2.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
ks_merge.addTable(second_table,"CountryRegion")
ks_merge.addTable(third_table,"ComissionTax")
ks_merge.addTable(fourth_table,"Currencyv2")
general_links = generalLinksDB(["Sales","Currencyv2","ComissionTax","CountryRegion"], ks_merge)
print ("Links:")
print(general_links.getLinks())
self.assertEqual(True, general_links.isEdge('ComissionTax:Region', 'CountryRegion:Region'))
self.assertEqual(True, general_links.isEdge('ComissionTax:VendorId', 'Sales:VendorId'))
self.assertEqual(True, general_links.isEdge('Sales:CountryCode', 'CountryRegion:CountryCode'))
self.assertEqual(True, general_links.isEdge('Sales:DownloadDate', 'Currencyv2:DownloadDate'))
self.assertEqual(True, general_links.isEdge('Currencyv2:CustomerCurrency', 'Sales:CustomerCurrency'))
self.db.commit()
def test_IS_UNIQUE_COL_CASE2(self):
first_table = "./ks_filehandler/ks_filehandler/data/Sales.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"Sales")
self.assertEqual(False, ks_merge.isUniqueCol("Sales","CountryCode"))
def test_IS_UNIQUE_COL_CASE1(self):
first_table = "./ks_filehandler/ks_filehandler/data/CountryRegion.csv"
ks_merge = merge(self.db)
ks_merge.reset()
ks_merge.addTable(first_table,"CountryRegion")
self.assertEqual(True, ks_merge.isUniqueCol("CountryRegion","CountryCode"))
def setUpClass(cls):
#----------------------
# set up db
#----------------------
cls.db = MySQLdb.connect(
ks_db_settings.setting('host'),
ks_db_settings.setting('user'),
ks_db_settings.setting('password'),
ks_db_settings.setting('database'))
#----------------------
# filehandler
#----------------------
file_handler = filehandler(cls.db)
file_handler.reset()
file_handler.addTable("Sales", "1","Sales.csv")
file_handler.addTable("Sales", "2","SalesCustomerTwo.csv")
file_handler.addTable("ComissionTax", "1","ComissionTax.csv")
file_handler.addTable("ComissionTax","2","ComissionTaxCustomerTwo.csv")
file_handler.addTable("CountryRegion", "1","CountryRegion.csv")
file_handler.addTable("CountryRegion","2","CountryRegionCustomerTwo.csv")
file_handler.addTable("Currency2","1","Currencyv2.csv")
time.sleep(1)
file_handler.addTable("Sales", "1","SalesNewVersion.csv")
print(file_handler.getLatestTable("Sales", "2"))
#----------------------
# merge
#----------------------
ks_merge = merge(cls.db)
ks_merge.reset()
ks_merge.addTable("./ks_filehandler/ks_filehandler/data/Sales.csv","Sales")
ks_merge.addTable("./ks_filehandler/ks_filehandler/data/CountryRegion.csv","CountryRegion")
ks_merge.addTable("./ks_filehandler/ks_filehandler/data/ComissionTax.csv","ComissionTax")
sql_BigTable = "CREATE TABLE BigTable(id INT PRIMARY KEY AUTO_INCREMENT, \
VendorId VARCHAR(25), \
ProductType VARCHAR(25), \
Units FLOAT, \
RoyaltyPrice FLOAT, \
DownloadDate VARCHAR(25), \
CustomerCurrency VARCHAR(25), \
CountryCode VARCHAR(25), \
Region VARCHAR(25), \
RightsHolder VARCHAR(25), \
ComissionRate VARCHAR(25), \
TaxRate VARCHAR(25))"
sql_join = "insert into BigTable select S.id,S.VendorId,S.ProductType, "\
"S.Units, S.RoyaltyPrice, S.DownloadDate, S.CustomerCurrency, "\
"S.CountryCode, C.Region, T.RightsHolder, T.ComissionRate, "\
"T.TaxRate from Sales S Inner Join CountryRegion C on "\
"S.CountryCode=C.CountryCode Inner join ComissionTax T on " \
"S.VendorId = T.VendorId and C.Region = T.Region;"
ks_merge.join(sql_join, sql_BigTable)
#----------------------
# clean up
#----------------------
cursor = cls.db.cursor()
sql = "use merge;"
cursor.execute(sql)
sql = "ALTER TABLE BigTable change ComissionRate ComissionRate FLOAT;"
cursor.execute(sql)
sql = "ALTER TABLE BigTable change TaxRate TaxRate FLOAT;"
cursor.execute(sql)
sql ="update BigTable set TaxRate = TaxRate/100;"
cursor.execute(sql)
#----------------------
# analytics
#----------------------
cls.ks_analytics = analytics(cls.db)
cls.ks_analytics.addFactUsingBinaryOp("NET_REVENUE", "Units", "RoyaltyPrice", "*")
cls.ks_analytics.addFactUsingBinaryOp("TAXES", "NET_REVENUE","TaxRate","*")
cls.ks_analytics.addFactUsingBinaryOp("REVENUE_AFTER_TAX", "NET_REVENUE","TAXES","-")
