Question

I need to process a zip archive which itself contains zip archives which all follow the same structure:

main.zip/
  foo-1.txt
  foo-1.zip/
    ... 
    tests/
    ...
  bar-1.txt
  bar-1.zip/
    ... 
    tests/
    ...
  bar-2.txt
  bar-2.zip/
    ... 
    tests/
    ...
  foobar-1.txt
  foobar-1.zip/
    ... 
    tests/
    ...

I want to extract the txt files and the tests directories into a structure like this:

 foo/
   1/
     submission.txt
     tests/
 bar/
   1/
     submission.txt
     tests/
   2/
     submission.txt
     tests/
 foobar/
   1/
     submission.txt
     tests/

I managed to get the directory structure and the text files into the right place:

<target name="extract submissions">
  <copy todir="${basedir}">
    <zipfileset refid="submissions.zip"/>
<mapper type="regexp" from="(.*)-(.*)\.txt" to="\1/\2/\submission.txt"/>
  </copy>
</target>

But now I am stuck how to access the tests directory inside the inner zips. I tried an intermediate step where I copied the inner zips into the place where the tests directories should end up. But how can I access the content of a zip archive without knowing its exact path?

Answer 1

I finally found a pure Ant solution:

<target name="extract submissions">
  <copy todir="${basedir}">
    <zipfileset refid="submissions.zip"/>
    <mapper type="regexp" from="(.*)-(.*)\.txt" to="\1/\2/\submission.txt"/>
  </copy>

  <subant genericantfile="${basedir}/build.xml" target="extract submission">
    <dirset dir="${basedir}" includes="*/*"/>
  </subant>
 </target>

<target name="extract submission">
  <copy todir="tests">
    <zipfileset refid="submission.zip"/>
    <mapper type="regexp" from="tests/(.*)" to="\1"/>
  </copy>
</target>

The first target extracts the outer zip file and calls the second target on each extracted directory. This is achieved by a subant. The second target then unzips an inner zip file. Since it is applied in any subdirectory, all the inner zips are extracted.