Support for specifying custom date format for date and timestamp types.

[HyukjinKwon]

https://github.com/databricks/spark-csv/issues/279
https://github.com/databricks/spark-csv/issues/262
https://github.com/databricks/spark-csv/issues/266

This PR adds the support to specify custom date format for DateType and TimestampType.

For TimestampType, this uses the given format to infer schema and also to convert the values
For DateType, this uses the given format to convert the values.
If the dateFormat is not given, then it works with Timestamp.valueOf() and Date.valueOf() for backwords compatibility.
When it's given, then it uses SimpleDateFormat for parsing data.

In addition, IntegerType, DoubleType and LongType have a higher priority than TimestampType in type inference. This means even if the given format is yyyy or yyyy.MM, it will be inferred as IntegerType or DoubleType. Since it is type inference, I think it is okay to give such precedences.

[HyukjinKwon]

@falaki Just to let you know, the original functions, tryParse...() were not mostly modified but just moved into inferField() as inner functions at src/main/scala/com/databricks/spark/csv/util/InferSchema.scala.

[codecov-io]

Current coverage is 86.66%

Merging #280 into master will increase coverage by +0.19% as of 89d6e92

@@            master    #280   diff @@
======================================
  Files           12      12       
  Stmts          525     540    +15
  Branches       155     160     +5
  Methods          0       0       
======================================
+ Hit            454     468    +14
  Partial          0       0       
- Missed          71      72     +1

Review entire Coverage Diff as of 89d6e92

Powered by Codecov. Updated on successful CI builds.

[falaki]

This line needs to be removed.

[falaki]

Indent is off for this entire block

[HyukjinKwon]

Um.. Do you mean the indentation correction as below?

• from

  private[csv] def inferField(typeSoFar: DataType,
    field: String,
    nullValue: String = "",
    dateFormatter: SimpleDateFormat = null): DataType = {
    def tryParseInteger(field: String): DataType = if ((allCatch opt field.toInt).isDefined) {
      IntegerType
    } else {
      tryParseLong(field)
    }
...

• to

  private[csv] def inferField(typeSoFar: DataType,
    field: String,
    nullValue: String = "",
    dateFormatter: SimpleDateFormat = null): DataType = {
      def tryParseInteger(field: String): DataType = if ((allCatch opt field.toInt).isDefined) {
        IntegerType
      } else {
        tryParseLong(field)
      }
...

[falaki]

Oh I see. The problem is for lines above:

  private[csv] def inferField(typeSoFar: DataType,
      field: String,
      nullValue: String = "",
      dateFormatter: SimpleDateFormat = null): DataType = {
    def tryParseInteger(field: String): DataType = if ((allCatch opt field.toInt).isDefined) {
      IntegerType
    } else {
      tryParseLong(field)
    }

[HyukjinKwon]

I see. Thanks!

[falaki]

@HyukjinKwon left one more comment. Otherwise looks good. I can merge this before I cut the branch tonight.

[HyukjinKwon]

@falaki (although this is not the right thread to say this), what do you think about uncompressed and none options for compression codecs to set explicitly no compression?

This was merged in Spark, apache/spark#11464 and I was working on this for this library as well.

However, I just realised that this might not be some kind of what we must do identically for this, like compression option is supported for codec at Spark CSV data source but compression is not acceptable at this library.

If you think it is good to support uncompressed and none, then I will submit a PR about this within today.

[falaki]

Let's open an issue for it.

[HyukjinKwon]

Filed in https://github.com/databricks/spark-csv/issues/281.

[barrybecker4]

Thank you for adding this! I will pull and build a local snapshot until 1.4.0 officially releases.
(Minor: there is a type in the doc. "specificy" should be "specify")

[HyukjinKwon]

@barrybecker4 Maybe would you create a PR for that typo (also for other typos if you know)?

[5ean]

I still face error in python. Maybe i did not use it correctly. Please kindly advise.
df = sqlContext.read.format('com.databricks.spark.csv').options(header='true',dateFormat='dd/MM/yyyy hh:mm:ss').load('test.csv',schema = schema)

schema has timestamp type.

And the string in csv file '"25/02/2014 00:00:00"

exception:

Caused by: java.lang.IllegalArgumentException: Timestamp format must be yyyy-mm-dd hh:mm:ss[.fffffffff]

[raam86]

The docs are not clear on how to do this so hopefully this can help:

https://stackoverflow.com/questions/43259485/how-to-load-csvs-with-timestamps-in-custom-format

spark.read
      .option("header", true)
      .option("inferSchema", true)
      .option("timestampFormat", "MM/dd/yyyy h:mm:ss a")
      .csv("PATH_TO_CSV")

[shatestest]

this is not working as expected. https://stackoverflow.com/questions/55965978/how-to-set-jdbc-partitioncolumn-type-to-date-in-spark-2-4-1/55966481#55966481