Tests for getAs() for Double, Integer types

[navinrathore]

sample input/output
#355 (comment)

[navinrathore]

• Ref. issue Test null types for double/integers #355

[navinrathore]

Null can be cast to any type without any exception. reference

[sonalgoyal]

what happens in these tests with the wrong values?

[navinrathore]

They are converted to null. at present schema has "Nullable" field set to true
But is converted to 0.0, when this field is set to false.

[sonalgoyal]

ok. so we should move the df creation here and add that test - nullable change. this will be good documentation when we later encounter issues with user data.

[navinrathore]

Note
Nullable 'true' in schema. Unsupported values change to null

+-----------+------------+
|fieldDouble|fieldInteger|
+-----------+------------+
|       0.55|          55|
|      1.234|        1234|
|       34.0|        null|
|      99.56|        9956|
|       null|          56|
|       23.0|        null|
|       null|          45|
|       65.0|        null|
|       null|        null|
|       null|          23|
|       56.0|          56|
|       null|        null|
|       null|          34|
|       78.0|        null|
|       78.0|          87|
+-----------+------------+

Nullable 'false' in schema. Unsupported values change to Zero(0.0/0)

+-----------+------------+
|fieldDouble|fieldInteger|
+-----------+------------+
|       0.55|          55|
|      1.234|        1234|
|       34.0|           0|
|      99.56|        9956|
|        0.0|          56|
|       23.0|           0|
|        0.0|          45|
|       65.0|           0|
|        0.0|           0|
|        0.0|          23|
|       56.0|          56|
|        0.0|           0|
|        0.0|          34|
|       78.0|           0|
|       78.0|          87|
+-----------+------------+

[navinrathore]

Now the testcases includes Nullable - true and false for IntegerType and DoubleType.

[sonalgoyal]

why not use @before ?

[navinrathore]

Changed to @BeforeAll

[sonalgoyal]

Comment for next time, it would be simpler to set up different dfs for different types - doubles/integers rather than having one df. tests should be as atomic as possible.