Update UNTAR to account for file-only or multi-directory tar archives (#2055)

* Update main.nf

* Update meta.yml

* Re-add logos as not staged in a way that works with MultiQC config files

* Fix when only files or multiple directories

* Add test for only-files tar

* Fix bash check

modules/untar/main.nf

@@ -25,12 +25,23 @@ process UNTAR {
     """
     mkdir output
 
-    tar \\
+    ## Ensures --strip-components only applied when top level of tar contents is a directory
-        -C output --strip-components 1 \\
+    ## If just files or multiple directories, place all in output
-        -xzvf \\
+    if [[ \$(tar -tzf ${archive} | grep "/\$" | wc -l) -eq 1 ]]; then
-        $args \\
+        tar \\
-        $archive \\
+            -C output --strip-components 1 \\
-        $args2
+            -xzvf \\
+            $args \\
+            $archive \\
+            $args2
+    else
+        tar \\
+            -C output \\
+            -xzvf \\
+            $args \\
+            $archive \\
+            $args2
+    fi
 
     mv output ${untar}
 

modules/untar/meta.yml

@@ -26,9 +26,9 @@ output:
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
   - untar:
-      type: file
+      type: directory
-      description:
+      description: Directory containing contents of archive
-      pattern: "*.*"
+      pattern: "*/"
   - versions:
       type: file
       description: File containing software versions
@@ -36,3 +36,5 @@ output:
 authors:
   - "@joseespinosa"
   - "@drpatelh"
+  - "@matthdsm"
+  - "@jfy133"

tests/config/test_data.config

@@ -446,6 +446,9 @@ params {
                 ipython_md                      = "${test_data_dir}/generic/notebooks/jupyter/ipython_notebook.md"
                 ipython_ipynb                   = "${test_data_dir}/generic/notebooks/jupyter/ipython_notebook.ipynb"
             }
+            'tar' {
+                tar_gz                           = "${test_data_dir}/generic/tar/hello.tar.gz"
+            }
             'tsv' {
                 test_tsv                        = "${test_data_dir}/generic/tsv/test.tsv"
             }

tests/modules/untar/main.nf

@@ -22,3 +22,13 @@ workflow test_untar_different_output_path {
 
     UNTAR ( input )
 }
+
+
+workflow test_untar_onlyfiles {
+    input = [
+        [],
+        file(params.test_data['generic']['tar']['tar_gz'], checkIfExists: true)
+    ]
+
+    UNTAR ( input )
+}

tests/modules/untar/test.yml

@@ -15,5 +15,13 @@
   tags:
     - untar
   files:
-    - path: output/untar/flowcell/RunInfo.xml
+    - path: output/untar/flowcell/220422_M11111_0222_000000000-K9H97/RunInfo.xml
       md5sum: 03038959f4dd181c86bc97ae71fe270a
+
+- name: untar test_untar_onlyfiles
+  command: nextflow run ./tests/modules/untar -entry test_untar_onlyfiles -c ./tests/config/nextflow.config  -c ./tests/modules/untar/nextflow.config
+  tags:
+    - untar
+  files:
+    - path: output/untar/hello/hello.txt
+      md5sum: e59ff97941044f85df5297e1c302d260