def test_merge_total_counts_match():
x1 = SchemaTracker()
multiple_track(x1, {
Type.INTEGRAL: 10,
Type.FRACTIONAL: 10,
Type.BOOLEAN: 10,
Type.UNKNOWN: 10
})
x2 = SchemaTracker()
multiple_track(x2, {
Type.INTEGRAL: 20,
Type.FRACTIONAL: 20,
Type.BOOLEAN: 20,
Type.UNKNOWN: 20
})
merged = x1.merge(x2)
assert merged.get_count(Type.INTEGRAL) == 30
assert merged.get_count(Type.FRACTIONAL) == 30
assert merged.get_count(Type.BOOLEAN) == 30
assert merged.get_count(Type.UNKNOWN) == 30
# Make sure we can serialize round trip
SchemaTracker.from_protobuf(merged.to_protobuf())
def test_serialization_roundtrip():
tracker = SchemaTracker()
type_count = {
Type.INTEGRAL: 10,
Type.STRING: 100,
}
multiple_track(tracker, type_count)
roundtrip = SchemaTracker.from_protobuf(tracker.to_protobuf())
assert tracker.to_protobuf() == roundtrip.to_protobuf()
assert roundtrip.get_count(Type.INTEGRAL) == 10
assert roundtrip.get_count(Type.STRING) == 100
def from_protobuf(message):
"""
Load from a protobuf message
Returns
-------
column_profile : ColumnProfile
"""
return ColumnProfile(
message.name,
counters=CountersTracker.from_protobuf(message.counters),
schema_tracker=SchemaTracker.from_protobuf(message.schema),
number_tracker=NumberTracker.from_protobuf(message.numbers),
string_tracker=StringTracker.from_protobuf(message.strings),
frequent_items=FrequentItemsSketch.from_protobuf(
message.frequent_items),
cardinality_tracker=HllSketch.from_protobuf(
message.cardinality_tracker),
)