def sampleColumn(self, numerical):
default = None
if Environment.hasSpark:
from pyspark.sql import functions as F
for field in self.entity.schema.fields:
# Ignore unique ids
if field.name.lower() != 'id' and (
not numerical
or dataFrameMisc.isNumericType(field.dataType)):
# Find a good column to display in pie ChartDisplay
default = default or field.name.decode(
"utf-8") if PY2 else field.name
count = self.entity.count()
sample = self.entity.sample(
False,
(float(200) / count)) if count > 200 else self.entity
orderedSample = sample.groupBy(field.name).agg(
F.count(field.name).alias("agg")).orderBy(
F.desc("agg")).select("agg")
if orderedSample.take(1)[0]["agg"] > 10:
return [
field.name.decode("utf-8") if PY2 else field.name
]
# Otherwise, return first non-id column
return [default]
def canRenderChart(self, handlerId, aggregation, fieldNames):
if (aggregation == "COUNT"):
return (True, None)
else:
for field in self.entity.schema.fields:
if dataFrameMisc.isNumericType(field.dataType):
return (True, None)
return (False, "At least one numerical column required.")
def getDefaultValueFields(self, handlerId, aggregation):
fieldNames = []
for field in self.entity.schema.fields:
if dataFrameMisc.isNumericType(field.dataType):
fieldNames.append(field.name)
if len(fieldNames) == self.getPreferredDefaultValueFieldCount(handlerId):
break
return fieldNames
def getDefaultKeyFields(self, handlerId, aggregation):
if self.supportsKeyFields(handlerId) == False:
return []
defaultFields = []
for field in self.entity.schema.fields:
if (dataFrameMisc.isNumericType(field.dataType) == False and field.name.lower() != "id"):
defaultFields.append(field.name)
if len(defaultFields) == self.getPreferredDefaultKeyFieldCount(handlerId):
break
if len(defaultFields) == 0:
defaultFields.append(self.entity.schema.fields[0].name)
return defaultFields
def sampleColumn(self, numerical):
default=None
if Environment.hasSpark:
from pyspark.sql import functions as F
for field in self.entity.schema.fields:
# Ignore unique ids
if field.name.lower() != 'id' and ( not numerical or dataFrameMisc.isNumericType(field.dataType) ):
# Find a good column to display in pie ChartDisplay
default = default or field.name.decode("utf-8") if PY2 else field.name
count = self.entity.count()
sample = self.entity.sample(False, (float(200) / count)) if count > 200 else self.entity
orderedSample = sample.groupBy(field.name).agg(F.count(field.name).alias("agg")).orderBy(F.desc("agg")).select("agg")
if orderedSample.take(1)[0]["agg"] > 10:
return [field.name.decode("utf-8") if PY2 else field.name]
# Otherwise, return first non-id column
return [default]
def isNumericType(self, field):
return dataFrameMisc.isNumericType(field.dataType)
def isNumericType(self, field):
return dataFrameMisc.isNumericType(field.dataType)
