def test_DateString_Ymd(self):
self.assertEqual(
wd.dateCleanup('20200205').strftime("%Y%m%d%H%M%S"),
'20200205000000')
def test_DateWithColonTimeNoSeperationFormat(self):
self.assertEqual(
wd.dateCleanup('2020:02:2812:30:00').strftime("%Y%m%d%H%M%S"),
'20200228123000')
def test_DayLessThanTweleve(self):
self.assertEqual(
wd.dateCleanup('2020-02-05').strftime("%Y%m%d%H%M%S"),
'20200205000000')
def test_alphaDateShortFormat(self):
self.assertEqual(
wd.dateCleanup('MAR 25 2020').strftime("%Y%m%d%H%M%S"),
'20200325000000')
def test_DateWithTimeFormat(self):
self.assertEqual(
wd.dateCleanup('2020-02-28 12:30:00').strftime("%Y%m%d%H%M%S"),
'20200228123000')
def test_germanDateFormatWithTime(self):
self.assertEqual(
wd.dateCleanup('07.04.2020 12:12:12').strftime("%Y%m%d%H%M%S"),
'20200407121212')
def test_alphaDateFormat(self):
self.assertEqual(
wd.dateCleanup('march 25, 2020').strftime("%Y%m%d%H%M%S"),
'20200325000000')
def test_germanDateFormat(self):
self.assertEqual(
wd.dateCleanup('07.04.2020').strftime("%Y%m%d%H%M%S"),
'20200407000000')
def test_TZFormat(self):
self.assertEqual(
wd.dateCleanup('2015-03-26T10:58:51Z').strftime("%Y%m%d%H%M%S"),
'20150326105851')
def test_epochFormat(self):
self.assertEqual(
wd.dateCleanup(1571824800000, epoch=True).strftime("%Y%m%d%H%M%S"),
'20191023100000')
def dc(val):
result = wd.dateCleanup(val, epoch=True)
return result
def __init__(self, spark):
# ---------------------------------------------------------------------
# set config attributes
# ---------------------------------------------------------------------
if len(sys.argv) > 1:
with open(sys.argv[1]) as dependencyFile:
conf = json.load(dependencyFile)
for x in conf:
setattr(parent, x, conf.get(x, ""))
# ---------------------------------------------------------------------
# set the process state
# ---------------------------------------------------------------------
processStart = wd.dateCleanup(
datetime.utcnow().strftime("%Y%m%d%H%M%S"))
# ---------------------------------------------------------------------
# create udf dateCleanup
# ---------------------------------------------------------------------
def dc(val):
result = wd.dateCleanup(val, epoch=True)
return result
dateCleanup = F.udf(dc, TimestampType())
# ---------------------------------------------------------------------
# set audit
# ---------------------------------------------------------------------
tableId = self.name
columnAudit = "tableid,ins_gmt_ts,process_timestamp"
namespaceAudit = "ea_sc_kif"
tableAudit = "batch_process_times"
cfAudit = "o"
dfAudit = DataFrame(
spark.sparkContext._jvm.com.hpe.hbase.HbaseManager.getDF(
columnAudit, namespaceAudit, tableAudit, cfAudit),
spark).where(F.col("tableid") == tableId)
# ---------------------------------------------------------------------
# get last proccesed time
# ---------------------------------------------------------------------
col = "ins_gmt_ts"
lastProcessedDate = dfAudit.select(col).collect()
if len(lastProcessedDate) > 0:
DTS = str([ele[col] for ele in lastProcessedDate][0])
deltaFilter = (F.col('epoch_ts') > DTS)
else:
DTS = ""
deltaFilter = ""
# ---------------------------------------------------------------------
# get mapper attributes
# ---------------------------------------------------------------------
with open(self.app["mapper-properties"]) as mapperFile:
topMap = json.load(mapperFile)
# ---------------------------------------------------------------------
# set hbase table properties
# ---------------------------------------------------------------------
hbase = topMap["hbase"]
table = hbase.get("table", "none")
cf = hbase.get("cf", "none")
namespace = hbase.get("namespace", "none")
column = hbase.get("column", "none")
latestVersionMapped = hbase.get("latestVersion", "none")
if latestVersionMapped == "True":
latestVersion = True
else:
latestVersion = False
# ---------------------------------------------------------------------
# create data frame
# ---------------------------------------------------------------------
df = spark.sparkContext \
._jvm.com.hpe.hbase \
.HbaseManager.getDF(column,
namespace,
table,
cf,
latestVersion)
pyDF = DataFrame(df, spark)
pyDF = pyDF.withColumn(
"epoch_ts",
F.to_timestamp(dateCleanup(pyDF["ts"]), 'yyyy-MM-dd HH:mm:ss'))
sqlDF = pyDF
if type(deltaFilter) is not str:
sqlDF = sqlDF.where(deltaFilter)
# ---------------------------------------------------------------------
# max date
# ---------------------------------------------------------------------
maxDate = dfColumnToString(
sqlDF.agg(F.max("epoch_ts")).select("max(epoch_ts)"),
"max(epoch_ts)")
# ---------------------------------------------------------------------
# if no delta results available then end else process delta
# ---------------------------------------------------------------------
if maxDate:
printInc = ("Found delta maxdate of '{}'." +
" Starting incremental update on process of " +
"delta records.")
print(printInc.format(maxDate))
# -----------------------------------------------------------------
# set checkpoint dir
# -----------------------------------------------------------------
spark \
.sparkContext \
.setCheckpointDir(self.app["checkpoint"])
sqlDF.persist(StorageLevel.MEMORY_AND_DISK)
sqlDF.take(1)
# -----------------------------------------------------------------
# set recursive table properties
# -----------------------------------------------------------------
recursive = topMap["recursive"]
id = str(recursive.get("id", "none"))
parentid = str(recursive.get("parentid", "none"))
parentLookupCol = str(recursive.get("parentLookupCol", "none"))
lookupCol = str(recursive.get("lookupCol", "none"))
levels = recursive.get("levels", "2")
# -----------------------------------------------------------------
# set output DF columns
# -----------------------------------------------------------------
outCols = stringToList(column)
outCols.insert(0, id)
outCols.append(lookupCol)
# -----------------------------------------------------------------
# run recursiveLookup
# -----------------------------------------------------------------
recursiveDF = recursiveLookup(sqlDF, id, parentid, parentLookupCol,
lookupCol, levels).select(outCols)
recursiveDF = recursiveDF.withColumn(
"parentlevel",
F.col("parentlevel").cast(StringType()))
recursiveDF = recursiveDF.withColumn("key", F.col(id))
# -----------------------------------------------------------------
# store df to hbase
# -----------------------------------------------------------------
outputCols = appendString("key", "parentlevel", lookupCol)
print("writing recursive output to Hbase . . .")
spark.sparkContext \
._jvm.com.hpe.hbase \
.HbaseManager.setDF(recursiveDF._jdf,
outputCols,
"key",
namespace,
table,
cf)
# -----------------------------------------------------------------
# close process state
# -----------------------------------------------------------------
processEnd = wd.dateCleanup(
datetime.utcnow().strftime("%Y%m%d%H%M%S"))
dfAuditWrite = sqlDF.agg(F.max("epoch_ts")
.alias("ins_gmt_ts")) \
.select(F.col("ins_gmt_ts").cast(StringType())) \
.withColumn("tableid", F.lit(tableId)) \
.withColumn("process_start_ts",
F.lit(processStart).cast(StringType())) \
.withColumn("process_end_ts", F.lit(processEnd
).cast(StringType()))
auditWriteCols = "tableid,ins_gmt_ts,process_start_ts" \
+ ",process_end_ts"
# -----------------------------------------------------------------
# write to the audit log
# -----------------------------------------------------------------
print("updating audit log . . .")
spark.sparkContext \
._jvm.com.hpe.hbase \
.HbaseManager \
.setDF(dfAuditWrite._jdf,
auditWriteCols,
"tableid",
namespaceAudit,
tableAudit,
cfAudit)
# -----------------------------------------------------------------
# add finial df to class (if run outside of __main__)
# -----------------------------------------------------------------
self.df = recursiveDF
else:
# -----------------------------------------------------------------
# No delta results
# -----------------------------------------------------------------
print("No delta available.")
