Splunk Search

How to map values in case statement?

karthi2809
Builder

Hi All,

I am using case statement to map values instead of other values. But i am not getting the values.I am getting UNknown values.
BucketFolder values is like:

inbound/concur

 

|rename  bucketFolder as BucketFolder| eval  InterfaceName=case(BucketFolder="%inbound%epm%","EPM", BucketFolder="%inbound%KPIs%","APEX_File_Upload", BucketFolder="%inbound%concur%","ConcurFile_Upload ",true(),"Unknown")| stats   values(InterfaceName) as InterfaceName   min(timestamp) as Timestamp values(BucketFolder) as BucketFolder   values(Status) as Status by correlationId  
| table Status  InterfaceName Timestamp FileName Bucket BucketFolder correlationId 

 

Labels (2)
0 Karma
1 Solution

yuanliu
SplunkTrust
SplunkTrust

There are multiple issues with this search.  But first, let me clarify your use case.  For events where bucketFolder has exemplified value of "inbound/concur", you want to assign "ConcurFile_Upload" as value of Interface; if bucketFolder is "<blah>inbound<bleh>epm<blih>", you want to assign "EPM" to Interface, and so on.  Is this correct?  In other words, given this dataset

bucketFoldercorrelationId
inbound/concur1
inbound/epm1
inbound/KPIs2
inbound/epm2

you want the output to be

Status
InterfaceName
TimestampFileNameBucket
BucketFolder
correlationId
 
ConcurFile_Upload
EPM
   
inbound/concur
inbound/epm
1
 
APEX_File_Upload
EPM
   
inbound/KPIs
inbound/epm
2

(Take care to clarify such carefully and in plain language and data illustration the next time you ask a question.  Using SPL to represent use case is often self defeating, even more so when you already know the SPL doesn't give you desired results.  You are basically inviting volunteers to read your mind, and mind-reading is not only painful for volunteers, but most often leads to wrong conclusions.)

Assuming that the above is a faithful representation of your requirement, we can discuss SPL problems.  But before that, I want to hammer on use case/requirement even further.  Do you really mean to map ""<blah>inbound<bleh>epm<blih>"" instead of simply "inbound/epm"? In other words, are wildcard uses like   "%inbound%epm%" truly essential to your dataset?  Decisions like this bears a lot of weight on optimal solution, sometimes can change suitable solution, too.  In the following, I will assume that you really, really want wildcards as your original SPL implied. (But I hope that's not the case.)

Now, to SPL diagnosis.  The first problem is, you cannot use wildcard in the righthand side of equal sign (=) outside of search context.  case function is not a search context.  Secondly, percent sign (%) is not a wildcard in the righthand side of equal sign.  As such, @tej57 's suggestion of using LIKE is correct.  You can also use searchmatch as @gcusello suggested, but use it directly as boolean AND use asterisk (*) instead of percent sign as wildcard.

 

| rename bucketFolder as BucketFolder 
| eval InterfaceName=case(searchmatch("BucketFolder = *inbound*epm*"),"EPM", searchmatch("BucketFolder = *inbound*KPIs*"), "APEX_File_Upload", searchmatch("BucketFolder = *inbound*concur*"), "ConcurFile_Upload", true(),"Unknown") 
| stats values(InterfaceName) as InterfaceName min(timestamp) as Timestamp values(BucketFolder) as BucketFolder values(Status) as Status by correlationId 
| table Status InterfaceName Timestamp FileName Bucket BucketFolder correlationId

 

You can also use regex match.

 

| rename bucketFolder as BucketFolder 
| eval InterfaceName=case(match(BucketFolder, "inbound.*epm"),"EPM", match(BucketFolder, "inbound.*KPIs"), "APEX_File_Upload", match(BucketFolder, "inbound.*concur"), "ConcurFile_Upload", true(),"Unknown") 
| stats values(InterfaceName) as InterfaceName min(timestamp) as Timestamp values(BucketFolder) as BucketFolder values(Status) as Status by correlationId 
| table Status InterfaceName Timestamp FileName Bucket BucketFolder correlationId

 

Each of these three can give the above illustrated results table using the following emulation that gives the mock data table illustrated above

 

| makeresults format=csv data="bucketFolder,correlationId
inbound/concur,1
inbound/epm,1
inbound/KPIs,2
inbound/epm,2"
``` data emulation above ```

 

Play with it and compare with real data.

 


View solution in original post

yuanliu
SplunkTrust
SplunkTrust

There are multiple issues with this search.  But first, let me clarify your use case.  For events where bucketFolder has exemplified value of "inbound/concur", you want to assign "ConcurFile_Upload" as value of Interface; if bucketFolder is "<blah>inbound<bleh>epm<blih>", you want to assign "EPM" to Interface, and so on.  Is this correct?  In other words, given this dataset

bucketFoldercorrelationId
inbound/concur1
inbound/epm1
inbound/KPIs2
inbound/epm2

you want the output to be

Status
InterfaceName
TimestampFileNameBucket
BucketFolder
correlationId
 
ConcurFile_Upload
EPM
   
inbound/concur
inbound/epm
1
 
APEX_File_Upload
EPM
   
inbound/KPIs
inbound/epm
2

(Take care to clarify such carefully and in plain language and data illustration the next time you ask a question.  Using SPL to represent use case is often self defeating, even more so when you already know the SPL doesn't give you desired results.  You are basically inviting volunteers to read your mind, and mind-reading is not only painful for volunteers, but most often leads to wrong conclusions.)

Assuming that the above is a faithful representation of your requirement, we can discuss SPL problems.  But before that, I want to hammer on use case/requirement even further.  Do you really mean to map ""<blah>inbound<bleh>epm<blih>"" instead of simply "inbound/epm"? In other words, are wildcard uses like   "%inbound%epm%" truly essential to your dataset?  Decisions like this bears a lot of weight on optimal solution, sometimes can change suitable solution, too.  In the following, I will assume that you really, really want wildcards as your original SPL implied. (But I hope that's not the case.)

Now, to SPL diagnosis.  The first problem is, you cannot use wildcard in the righthand side of equal sign (=) outside of search context.  case function is not a search context.  Secondly, percent sign (%) is not a wildcard in the righthand side of equal sign.  As such, @tej57 's suggestion of using LIKE is correct.  You can also use searchmatch as @gcusello suggested, but use it directly as boolean AND use asterisk (*) instead of percent sign as wildcard.

 

| rename bucketFolder as BucketFolder 
| eval InterfaceName=case(searchmatch("BucketFolder = *inbound*epm*"),"EPM", searchmatch("BucketFolder = *inbound*KPIs*"), "APEX_File_Upload", searchmatch("BucketFolder = *inbound*concur*"), "ConcurFile_Upload", true(),"Unknown") 
| stats values(InterfaceName) as InterfaceName min(timestamp) as Timestamp values(BucketFolder) as BucketFolder values(Status) as Status by correlationId 
| table Status InterfaceName Timestamp FileName Bucket BucketFolder correlationId

 

You can also use regex match.

 

| rename bucketFolder as BucketFolder 
| eval InterfaceName=case(match(BucketFolder, "inbound.*epm"),"EPM", match(BucketFolder, "inbound.*KPIs"), "APEX_File_Upload", match(BucketFolder, "inbound.*concur"), "ConcurFile_Upload", true(),"Unknown") 
| stats values(InterfaceName) as InterfaceName min(timestamp) as Timestamp values(BucketFolder) as BucketFolder values(Status) as Status by correlationId 
| table Status InterfaceName Timestamp FileName Bucket BucketFolder correlationId

 

Each of these three can give the above illustrated results table using the following emulation that gives the mock data table illustrated above

 

| makeresults format=csv data="bucketFolder,correlationId
inbound/concur,1
inbound/epm,1
inbound/KPIs,2
inbound/epm,2"
``` data emulation above ```

 

Play with it and compare with real data.

 


tej57
Communicator

Hello @karthi2809,

I'm not sure if searchmatch function can work with case condition or not. I've read that it must be used inside if condition. Can you try running the following search:

| rename bucketFolder as BucketFolder 
| eval InterfaceName=case(BucketFolder like "%inbound%epm%","EPM", BucketFolder like "%inbound%KPIs%","APEX_File_Upload", BucketFolder like "%inbound%concur%","ConcurFile_Upload",true(),"Unknown") 
| stats values(InterfaceName) as InterfaceName min(timestamp) as Timestamp values(BucketFolder) as BucketFolder values(Status) as Status by correlationId 
| table Status InterfaceName Timestamp FileName Bucket BucketFolder correlationId

 

Thanks,
Tejas.

---

If the above solution helps, an upvote is appreciated.

gcusello
SplunkTrust
SplunkTrust

Hi @karthi2809,

you can use % only using the like function otherwise you have to use *:

| rename  bucketFolder as BucketFolder
| eval  InterfaceName=case(
     BucketFolder=searchmatch("inbound") AND searchmatch("epm"), "EPM",
     BucketFolder=searchmatch("inbound") AND searchmatch("KPIs"), "APEX_File_Upload", 
     BucketFolder=searchmatch("inbound") AND searchmatch("concur"), "ConcurFile_Upload",
     true(), "Unknown")
| stats 
     values(InterfaceName) AS InterfaceName   
     min(timestamp) AS Timestamp 
     values(BucketFolder) AS BucketFolder   
     values(Status) AS Status 
     BY correlationId  
| table Status  InterfaceName Timestamp FileName Bucket BucketFolder correlationId 

Ciao.

Giuseppe

karthi2809
Builder

I am getting below error.

Error in 'EvalCommand': Type checking failed. The '==' operator received different types.

0 Karma
Get Updates on the Splunk Community!

Enter the Splunk Community Dashboard Challenge for Your Chance to Win!

The Splunk Community Dashboard Challenge is underway! This is your chance to showcase your skills in creating ...

.conf24 | Session Scheduler is Live!!

.conf24 is happening June 11 - 14 in Las Vegas, and we are thrilled to announce that the conference catalog ...

Introducing the Splunk Community Dashboard Challenge!

Welcome to Splunk Community Dashboard Challenge! This is your chance to showcase your skills in creating ...