<tool id="unzip" name="Unzip" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>Unzip a file</description>
    <macros>
        <import>creators.xml</import>
        <token name="@TOOL_VERSION@">6.0</token>
        <token name="@VERSION_SUFFIX@">2</token>
        <token name="@PROFILE@">22.01</token>
        <xml name="regex_sanitizer">
            <sanitizer>
                <valid>
                    <add preset="string.printable"/>
                    <remove value="\"/>
                    <remove value="'"/>
                </valid>
                <mapping initial="none">
                    <add source="\" target="__backslash__"/>
                    <add source="'" target="__sq__"/>
                </mapping>
            </sanitizer>
        </xml>
    </macros>
    <creator>
        <expand macro="creators/bmcv"/>
        <expand macro="creators/tuncK"/>
    </creator>
    <requirements>
        <!--
            BioContainer for: unzip=6.0,tar=1.34,findutils=4.10.0
            Enforced containerization to minimize potential security issues due to regex hacking.
        -->
        <container type="docker">quay.io/biocontainers/mulled-v2-6a2891161dcf5f35b38c6a49fff923163de7a66d:159a5b483a078f27b97a401b0626abac09e5f2d1-0</container>
    </requirements>
    <command> 
    <![CDATA[
        mkdir ./out &&
        #if $input_file.is_of_type("zip"):
            unzip
            #if $extract_options['target'] == 'single'
                -p '$input_file' '$extract_options["pathtofile"]' > out.data
            #else
                '$input_file' -d out/
            #end if
        #end if
        #if $input_file.is_of_type("tar"):
            tar
            #if $extract_options['target'] == 'single'
              xfO '$input_file' '$extract_options["pathtofile"]' > out.data
            #else
              -xf '$input_file' -C ./out/
            #end if
        #end if
        &&

        ## Filter out files that do not match the regex
        #if $extract_options['target'] != 'single'
            pushd out &&
            find . -regextype posix-extended -type f -not -regex '$extract_options["keep_regex"]' -delete &&
            popd &&
        #end if

        ## Flatten any potential directory structure, e.g.
        ## out/sub1/subsub2/fileA.ext -> out/sub1_subsub2_fileA.ext
        find out -type f | cut -d '/' -f 2- | grep '/' | xargs -I % sh -c "mv out/% out/\\$(echo % | tr / _)"
    ]]>
    </command>
    <inputs>
        <param label="Input file" name="input_file" type="data" format="zip,tar"/>
        <conditional name="extract_options">
            <param label="What to extract" name="target" type="select">
                <option selected="True" value="all">All files</option>
                <option value="all_nondot">All files except dotfiles</option>
                <option value="all_regex">All files matching a regex</option>
                <option value="single">Single file</option>
            </param>
            <when value="all">
                <param name="keep_regex" type="hidden" value=".*"/>
            </when>
            <when value="all_nondot">
                <param name="keep_regex" type="hidden" value=".*/[^.][^/]*">
                    <expand macro="regex_sanitizer"/>
                </param>
            </when>
            <when value="all_regex">
                <param name="keep_regex" type="text" value=".*" optional="False" label="Regular expression" help="POSIX regular expression, that will be used for filtering the extracted files. Only files that match this expression will be retained. Example: To only retain files from a sub-directory &lt;tt&gt;subdir&lt;/tt&gt;, use &lt;tt&gt;./subdir/.*&lt;/tt&gt; as the regular expression.">
                    <expand macro="regex_sanitizer"/>
                </param>
            </when>
            <when value="single">
                <param name="pathtofile" type="text" value="" label="File path">
                    <validator type="expression" message="No two dots (..) allowed">'..' not in value</validator>
                    <validator type="expression" message="No comma (,) allowed">',' not in value</validator>
                    <validator type="expression" message="No asterisk (*) allowed">'*' not in value</validator>
                    <validator type="expression" message="No leading slash (/) allowed">str(value)[0] != '/'</validator>
                </param>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data auto_format="true" name="unzipped_single" from_work_dir="out.data" label="${tool.name} on ${on_string}">
            <filter>extract_options['target'] == 'single'</filter>
        </data>
        <collection name="unzipped" type="list" label="${tool.name} on ${on_string} all files as collection">
            <discover_datasets directory="out" pattern="(?P&lt;designation&gt;\.*[^\.]+)\.?(?P&lt;ext&gt;.*)" visible="false" format="auto"/>
            <filter>extract_options['target'] != 'single'</filter>
        </collection>
    </outputs>
    <tests>
        <test expect_num_outputs="1">
            <param name="input_file" value="input.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="single"/>
                <param name="pathtofile" value="input.png"/>
            </conditional>
            <output name="unzipped_single" file="input.png" ftype="png" compare="sim_size"/>
        </test>
        <test expect_num_outputs="1">
            <param name="input_file" value="input.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="2">
                <element name="input" file="input.png" ftype="png" compare="sim_size"/>
                <element name="res" file="res.tiff" ftype="tiff" compare="sim_size"/>
            </output_collection>
        </test>
        <!-- Binary files without file extension -->
        <test expect_num_outputs="1">
            <param name="input_file" value="input_binary.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="2">
                <element name="yelp" file="yelp" compare="sim_size"/>
                <element name="yes" file="yes" compare="sim_size"/>
            </output_collection>
        </test>
        <!-- Compressed object with subfolders, each containing files -->
        <test expect_num_outputs="1">
            <param name="input_file" value="subfolders.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="4">
                <element name="binaries_yelp" file="yelp" compare="sim_size"/>
                <element name="binaries_yes" file="yes" compare="sim_size"/>
                <element name="images_input" file="input.png" ftype="png" compare="sim_size"/>
                <element name="images_res" file="res.tiff" ftype="tiff" compare="sim_size"/>
            </output_collection>
        </test>
        <test expect_num_outputs="1">
            <param name="input_file" value="input.tar" ftype="tar"/>
            <conditional name="extract_options">
                <param name="target" value="single"/>
                <param name="pathtofile" value="input.png"/>
            </conditional>
            <output name="unzipped_single" file="input.png" ftype="png" compare="sim_size"/>
        </test>
        <test expect_num_outputs="1">
            <param name="input_file" value="input.tar" ftype="tar"/>
            <conditional name="extract_options">
                <param name="target" value="all"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="2">
                <element name="input" file="input.png" ftype="png" compare="sim_size"/>
                <element name="res" file="res.tiff" ftype="tiff" compare="sim_size"/>
            </output_collection>
        </test>
        <!-- ZIP containing dotfiles and non-dotfiles -->
        <test expect_num_outputs="1">
            <param name="input_file" value="input_with_dotfiles.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="4">
                <element name=".dotfile">
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
                <element name="nondotfile">
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
                <element name="subdir_"><!-- The suffix `.dotfile` is recognized as the extension -->
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
                <element name="subdir_nondotfile">
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
            </output_collection>
        </test>
        <test expect_num_outputs="1">
            <param name="input_file" value="input_with_dotfiles.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all_nondot"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="2">
                <element name="nondotfile">
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
                <element name="subdir_nondotfile">
                    <assert_contents>
                        <has_size value="0"/>
                    </assert_contents>
                </element>
            </output_collection>
        </test>
        <!-- Custom Regex -->
        <test expect_num_outputs="1">
            <param name="input_file" value="subfolders.zip" ftype="zip"/>
            <conditional name="extract_options">
                <param name="target" value="all_regex"/>
                <param name="keep_regex" value="./images/.*"/>
            </conditional>
            <output_collection name="unzipped" type="list" count="2">
                <element name="images_input" file="input.png" ftype="png" compare="sim_size"/>
                <element name="images_res" file="res.tiff" ftype="tiff" compare="sim_size"/>
            </output_collection>
        </test>
    </tests>
    <help>
        **What it does**
          
        Unzip a folder containing file(s) of various types. If multiple files are asked to be retained, the output will be a collection containing all files within the zip or tar archive.
    </help>
    <citations>
        <citation type="doi">10.1016/j.jbiotec.2017.07.019</citation>
    </citations>
</tool>