"Unable to use @Grab in Groovy script to add maven dependencies to classpath"

polluxpollux Member Posts: 3 Contributor I
edited May 2019 in Help

I want to use the "Execute Script" RM operator to run a Groovy script using the HTTPBuilder module.

The Groovy script uses @Grab to add the required dependencies and runs perfectly in the GroovyConsole.

But when I run it from the RapidMiner UI in the "Execute Script" operator, I get the following error :

Process failed:
The scripting engine Groovy reported an error in the script: org.codehaus.groovy.control.
MultipleCompilationErrorsException: startup failed, Script1.groovy: 8: unexpected token: import @ line 8, column 1.
1 error

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<process version="5.1.001">
  <operator activated="true" class="process" compatibility="5.1.001" expanded="true" name="Process">
    <process expanded="true" height="684" width="884">
      <operator activated="true" class="execute_script" compatibility="5.1.001" expanded="true" height="60" name="Execute Script" width="90" x="255" y="165">
        <parameter key="script" value="@Grab(group='org.codehaus.groovy.modules.http-builder', module='http-builder', version='0.5.0-RC2' )&#10;import groovyx.net.http.*&#10;import static groovyx.net.http.ContentType.*&#10;import static groovyx.net.http.Method.*&#10;&#10;def http = new HTTPBuilder( 'http://ajax.googleapis.com' )&#10;&#10;// perform a GET request, expecting JSON response data&#10;http.request( GET, JSON ) {&#10;  uri.path = '/ajax/services/search/web'&#10;  uri.query = [ v:'1.0', q: 'Calvin and Hobbes' ]&#10;&#10;  headers.'User-Agent' = 'Mozilla/5.0 Ubuntu/8.10 Firefox/3.0.4'&#10;&#10;  // response handler for a success response code:&#10;  response.success = { resp, json -&gt;&#10;    println resp.statusLine&#10;&#10;    // parse the JSON response object:&#10;    json.responseData.results.each {&#10;      println &quot;  ${it.titleNoFormatting} : ${it.visibleUrl}&quot;&#10;    }&#10;  }&#10;&#10;  // handler for any failure status code:&#10;  response.failure = { resp -&gt;&#10;    println &quot;Unexpected error: ${resp.statusLine.statusCode} : ${resp.statusLine.reasonPhrase}&quot;&#10;  }&#10;}"/>
      <portSpacing port="source_input 1" spacing="0"/>
      <portSpacing port="sink_result 1" spacing="0"/>
Any fix or workaround available ?

NB: I also tried to use the modified execute script plugin proposed by Steffen in another post but it didn't work either.



  • Options
    steffensteffen Member Posts: 347 Maven
    Hi pollux,

    Back then carol aka chaney ignored by begging to give me feedback on the modified groovy operator, so I did not bother anymore, too. But now that you are here ... Frankly, I was a little lazy and didn't test it with the groovy http module itself.

    So I will definitely have a look at it, but not before sunday (so little time, so much to code). If you are still working on this issue, please post everything you find out ;).


  • Options
    polluxpollux Member Posts: 3 Contributor I
    Hi Steffen,

    Thanks for the quick update.

    To be more precise about my testing of your GroovyScripting operator, I have installed it as a plugin in a fresh RM 5.1 installation, and tried it with the sample process you've included in your post (the one using the commons-math-2.1 library).

    This sample mathlab process doesn't compile, Groovy inside RM says it is unable to resolve class org.apache.commons.math.stat.descriptive.DescriptiveStatistics

    Unfortunately I've neither the time nore the immediate knowledge to investigate deeper by myself the class loading problem :'(, but if you manage to solve it I'll be happy to test and give feedback.

    best regards

  • Options
    steffensteffen Member Posts: 347 Maven
    Hi pollux

    I am verrry sorry for the late reply ... the last two weeks I was really grateful that there is some time left where I do not have to stare at a monitor (aka my free time) ;).

    1. My plugins works within rm5.1, the process with apache commons math, too. Are you sure that you have pointed the parameter to the correct location of the library ?
    2. The grab-functionality is around since groovy 1.6.0, so this should not be an issue. However, notice that rm5.1 uses groovy 1.6.3 (< current version).

    So ... this seems to be a bug happening when the classloader magic of groovy,maven and rm collide. Since I do not even have a maven installation on my home computer, I will need some time to investigate.

    Sorry again for the late reply and for not-providing a fix


  • Options
    steffensteffen Member Posts: 347 Maven
    Ok ...

    The reason seems to be the groovy version. Although groovy grape has been introduced around 1.6.2-BETA, the syntax (among other things) have changed since then. Since I really did not want to download a bunch of outdated software to find a fix , here is my advice for the latest groovy version.
    • Download the latest groovy zip file and replace the groovy-all.jar under rapidminer/lib/ with the groovy-all-1.7.7,jar
    • Additionally download apache ivy (http://ant.apache.org/ivy/download.cgi) and place ivy-2.2.0.jar in the same folder. Reason: An bug/"undesired behavior" of grape (see here)
    • now start rapidminer using the startup-script, i.e. either rapidminer/scripts/RapidMinerGUI.bat (windows) or rapidminer/scripts/RapidminerGUI (linux). Now the execute-script operator together with your groovy-httpbuilder-script should work (at least it worked for me)
    All in all the grab-functionality is really cool. It is much better than my proposed modification (the one drawback is that you need maven installation).

    on a sidenote: I added a feature request for this, so hopefully grape will work in 5.2+



  • Options
    polluxpollux Member Posts: 3 Contributor I
    Hi Steffen,

    Thanks for the answer.

    I gave it a try and made some progress (Grape classes are loading), but now I come into the Ivy "bug" you mentioned, despite having ivy-2.2.0.jar in the RM classpath when starting with RapidMinerGUI.bat.

    Anyway I managed to workaround my "real world" process problem without using Groovy, so I can live without it until 5.2+...

    Thanks for adding a feature request for 5.2+,  it'll unleash the full power of the RM/Groovy combination.


Sign In or Register to comment.