def test_mainly_null_inferred_type_not_null():
counts = {Type.INTEGRAL: 1, Type.NULL: 30}
tracker = SchemaTracker()
multiple_track(tracker, counts)
inferred_type = tracker.infer_type()
assert inferred_type.type == Type.INTEGRAL
counts = {Type.INTEGRAL: 1, Type.STRING: 2, Type.NULL: 30}
tracker = SchemaTracker()
multiple_track(tracker, counts)
inferred_type = tracker.infer_type()
assert inferred_type.type != Type.NULL
def test_all_null_inferred_type_is_null():
counts = {
Type.NULL: 1,
}
tracker = SchemaTracker()
multiple_track(tracker, counts)
inferred_type = tracker.infer_type()
assert inferred_type.type == Type.NULL
def test_majority_int():
tracker = SchemaTracker()
type_counts = {
Type.INTEGRAL: 50,
Type.STRING: 30,
Type.UNKNOWN: 20,
}
multiple_track(tracker, type_counts)
assert tracker.infer_type().type == Type.INTEGRAL
def test_all_types_equal_coerced_to_string():
tracker = SchemaTracker()
multiple_track(tracker,
counts={
Type.INTEGRAL: 20,
Type.FRACTIONAL: 29,
Type.STRING: 50,
})
assert tracker.infer_type().type == Type.STRING
def test_float_and_int():
tracker = SchemaTracker()
multiple_track(tracker,
counts={
Type.INTEGRAL: 50,
Type.FRACTIONAL: 50,
Type.STRING: 10,
})
assert tracker.infer_type().type == Type.FRACTIONAL
def test_summary():
type_counts = {
Type.INTEGRAL: 3,
Type.STRING: 4,
Type.FRACTIONAL: 5,
Type.BOOLEAN: 6,
Type.UNKNOWN: 1,
}
tracker = SchemaTracker()
multiple_track(tracker, type_counts)
summary = tracker.to_summary()
c = summary.type_counts
assert c["INTEGRAL"] == type_counts[Type.INTEGRAL]
assert c["STRING"] == type_counts[Type.STRING]
assert c["FRACTIONAL"] == type_counts[Type.FRACTIONAL]
assert c["BOOLEAN"] == type_counts[Type.BOOLEAN]
assert c["UNKNOWN"] == type_counts[Type.UNKNOWN]
assert summary.inferred_type.type == tracker.infer_type().type
def test_70percent_string():
tracker = SchemaTracker()
type_counts = {Type.INTEGRAL: 29, Type.STRING: 71}
multiple_track(tracker, type_counts)
assert tracker.infer_type().type == Type.STRING
def test_track_nothing_should_return_unknown():
tracker = SchemaTracker()
inferred_type = tracker.infer_type()
assert inferred_type.type == Type.UNKNOWN
assert inferred_type.ratio == 0.0
