<?xml version="1.0"?>
<tool id="twobit_mask" name="twoBitMask" version="1.0">
    <description>Apply repeat masking to a twoBit file</description>

    <macros>
        <import>ucsc_macros.xml</import>
    </macros>

    <expand macro="requirements_twobit" />

    <command detect_errors="exit_code">
<![CDATA[
    #if str($maskfile_type.maskfile_type_selector) == "out":
        ## twoBitMask expects the first column of the RepeatMasker header to
        ## contain only three leading spaces. RepeatMasker output with large
        ## SW scores could contain more leading spaces.

        awk '{ sub(/^\s+SW/, "   SW"); print }' "${out_maskfile}" |
            twoBitMask
                ${add_mask}
                -type=.out
                "${twobit_input}" stdin "${twobit_output}"
    #else:
        ## The BED Output from TRF is in bed4+ format, but twoBitMask
        ## only uses the first three fields of a BED file.
        ## The extra columns result in a warning from twoBitMask

        awk 'BEGIN { OFS="\t" } { print $1, $2, $3 }' "${bed_maskfile}" |
            twoBitMask
                ${add_mask}
                -type=.bed
                "${twobit_input}" stdin "${twobit_output}"
    #end if
]]>
    </command>
    <inputs>
        <param name="twobit_input" type="data" format="twobit" label="twoBit input file" />

        <param name="add_mask" type="boolean" checked="true"
            truevalue="-add" falsevalue=""
            label="Keep pre-existing masking"
            help="-add" />

        <conditional name="maskfile_type">
            <param name="maskfile_type_selector" type="select"
                    label="Choose the type of mask file"
                    help="Mask file can be a RepeatMasker .out file or a BED file">
                <option value="out">RepeatMasker .out file</option>
                <option value="bed" selected="true">BED file</option>
            </param>

            <when value="out">
                <param name="out_maskfile" type="data" format="txt"
                        label="Mask file in RepeatMasker .out format" />
            </when>

            <when value="bed">
                <param name="bed_maskfile" type="data" format="bed"
                        label="Mask file in BED format" />
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="twobit_output" format="twobit" />
    </outputs>
    <tests>
        <test>
            <!-- Test standard RepeatMasker .out file -->
            <param name="twobit_input" value="contig1.2bit" ftype="twobit" />
            <param name="maskfile_type_selector" value="out" />
            <param name="out_maskfile" value="contig1.fasta.out" />
            <output name="twobit_output" file="contig1.out.2bit" />
        </test>
        <test>
            <!-- Test RepeatMasker .out file with extra space in header -->
            <param name="twobit_input" value="contig1.2bit" ftype="twobit" />
            <param name="maskfile_type_selector" value="out" />
            <param name="out_maskfile" value="contig1.fasta.extraspace.out" />
            <output name="twobit_output" file="contig1.out.2bit" />
        </test>
        <test>
            <!-- Test TRF .bed output -->
            <param name="twobit_input" value="contig1.2bit" ftype="twobit" />
            <param name="maskfile_type_selector" value="bed" />
            <param name="bed_maskfile" value="contig1.fasta.bed" ftype="bed" />
            <output name="twobit_output" file="contig1.bed.2bit" />
        </test>
        <test>
            <!-- Test keep pre-existing masking -->
            <param name="twobit_input" value="contig1.bed.2bit" ftype="twobit" />
            <param name="maskfile_type_selector" value="out" />
            <param name="out_maskfile" value="contig1.fasta.out" />
            <output name="twobit_output" file="contig1.both.2bit" />
        </test>
        <test>
            <!-- Test ignore pre-existing masking -->
            <param name="twobit_input" value="contig1.bed.2bit" ftype="twobit" />
            <param name="maskfile_type_selector" value="out" />
            <param name="out_maskfile" value="contig1.fasta.out" />
            <param name="add_mask" value="" />
            <output name="twobit_output" file="contig1.out.2bit" />
        </test>
    </tests>
    <help>
<![CDATA[
**What it does**

twoBitMask applies a mask to a twoBit file based on repeat information
from a `RepeatMasker <http://www.repeatmasker.org/>`_ .out file or
a BED file (e.g., from
`Tandem Repeats Finder <https://tandem.bu.edu/trf/trf.html>`_).

    ]]></help>

    <expand macro="citations" />
</tool>