Hi, One of solutions to use `spark-csv` (See: https://github.com/databricks/spark-csv#features). To load NULL, you can use `nullValue` there.
// maropu On Mon, Jul 11, 2016 at 1:14 AM, Radha krishna <grkmc...@gmail.com> wrote: > I want to apply null comparison to a column in sqlcontext.sql, is there > any way to achieve this? > On Jul 10, 2016 8:55 PM, "Radha krishna" <grkmc...@gmail.com> wrote: > >> Ok thank you, how to achieve the requirement. >> >> On Sun, Jul 10, 2016 at 8:44 PM, Sean Owen <so...@cloudera.com> wrote: >> >>> It doesn't look like you have a NULL field, You have a string-value >>> field with an empty string. >>> >>> On Sun, Jul 10, 2016 at 3:19 PM, Radha krishna <grkmc...@gmail.com> >>> wrote: >>> > Hi All,IS NOT NULL is not working in programmatic sql. check below for >>> input >>> > output and code. >>> > >>> > Input >>> > ==== >>> > 10,IN >>> > 11,PK >>> > 12,US >>> > 13,UK >>> > 14,US >>> > 15,IN >>> > 16, >>> > 17,AS >>> > 18,AS >>> > 19,IR >>> > 20,As >>> > >>> > val cntdat = sc.textFile("/user/poc_hortonworks/radha/gsd/sample.txt"); >>> > case class CNT (id:Int , code : String) >>> > val cntdf = cntdat.map((f) => { val ff=f.split(",");new >>> > CNT(ff(0).toInt,ff(1))}).toDF >>> > cntdf.registerTempTable("cntids"); >>> > sqlContext.sql("SELECT MAX(id),code FROM cntids WHERE code is not null >>> GROUP >>> > BY code").show() >>> > >>> > Output >>> > ===== >>> > +---+----+ >>> > |_c0|code| >>> > +---+----+ >>> > | 18| AS| >>> > | 16| | >>> > | 13| UK| >>> > | 14| US| >>> > | 20| As| >>> > | 15| IN| >>> > | 19| IR| >>> > | 11| PK| >>> > +---+----+ >>> > >>> > i am expecting the below one any idea, how to apply IS NOT NULL ? >>> > >>> > +---+----+ >>> > |_c0|code| >>> > +---+----+ >>> > | 18| AS| >>> > | 13| UK| >>> > | 14| US| >>> > | 20| As| >>> > | 15| IN| >>> > | 19| IR| >>> > | 11| PK| >>> > +---+----+ >>> > >>> > >>> > >>> > Thanks & Regards >>> > Radha krishna >>> > >>> > >>> >> >> >> >> -- >> >> >> >> >> >> >> >> >> Thanks & Regards >> Radha krishna >> >> >> -- --- Takeshi Yamamuro