Datatyping Details
By default, AutomaticCSV checks every single cell of your CSV file for its datatype to ensure robustness when selecting a datatype for the corresponding field. If you have well formatted data, or a large dataset, you can use the .setHeuristicDatatyping(true) method to make assumptions of datatype based on the first 100 rows of the file to speed up the initial run. Once a class file has been generated for your CSV, this setting is irrelevant to parsing speed.
Overview
No primitive datatypes are used. All primitives can be represented using a wrapper class for enhanced functionality, so that is what we have adopted.
String: “Arkansas Razorbacks”
Double: 12.37, NaN
Integer: 13, NaN
Enhanced Types
The ability to parse the following datatypes has been enhanced beyond what is present by default.
Boolean
Columns with any combination of 0,1,True,False,true,false,Yes,No,yes,no are read in as Boolean.
Dates and DateTimes
Parsed as LocalDateTime and LocalDate objects with any of the following formats.
| Format | Example |
|---|---|
| yyyy-mm-dd | 2020-12-03 |
| M/d/yyyy | 3/12/2020 or 03/12/2020 |
| M/d/yy | 3/12/20 or 03/12/20 |
| M-d-yyyy | 3-12-2020 or 03-12-2020 |
| M-dd-yy | 3-12-20 or 03-12-20 |
| yyyy/M/d | 2020-3-12 or 2020-03-12 |
| yyyy-mm-ddThh:mm:ss | 2020-03-12T17:19:23 |
| M/d/y H:m | 03/12/22 17:19 or 3/12/2022 17:19 or 03/12/2022 17:19 |
| M/d/y H: m:s | 03/12/22 17:19:01 or 3/12/2022 17:19:01 or 03/12/2022 17:19:01 |
These dataypes were selected to support the vast majority of tabular data with as few datatypes as possible.