docbook52pdf

#!/bin/bash
# Author: Jens Getreu

# apt install pandoc docbook5-xml docbook-xsl-ns xsltproc fop xmlto libxml2-utils xmlstarlet

#set -x

render () {
    ### parse args

    #set -x
    InPath="$1"
    InFile="${InPath##*/}"
    InFileExt="${InPath##*.}"
    InBase="${InFile%.*}"
    InDir="${InPath%/*}"
    if [ "$InDir" = "$InPath" ] ; then
        InDir="."
    fi

    OutPath="$2"
    OutFile="${OutPath##*/}"
    OutBase="${OutFile%.*}"
    OutDir="${OutPath%/*}"
    if [ "$OutDir" = "$OutPath" ] ; then
        OutDir="."
    fi


    ### Prepare
    XmlFile="$OutBase.xml"
    XmlPath="$OutDir/$XmlFile"
    FoFile="$OutBase.fo"
    FoPath="$OutDir/$FoFile"
    PdfFile="$OutBase.pdf"
    PdfPath="$OutDir/$PdfFile"
    XslPath="$(realpath $InDir/docbook-xsl/fo-pdf.xsl)"
    mkdir -p "$OutDir"


    ### Generate XML

    # unfortunately the chain does not honor --number-section yet
    if [ $InFileExt == 'xml' ]
    then
        cp "$InPath" "$XmlPath"
    else
        if ! pandoc -s -t docbook5 -o "$XmlPath" "$InPath"
        then
            echo Fatal error: Pandoc failed: \"$InPath\"
            exit 1
        fi
    fi

    # Workaround bug
    # [Docbook5 Writer: produces invalid output when author is given ·
    # Issue #6244 · jgm/pandoc](https://github.com/jgm/pandoc/issues/6244)

    if grep -q '<author>' "$XmlPath" &&  ! grep -q '<personname>' "$XmlPath"
    then
        echo change
        sed -i 's/<author>/<author><personname>/g' "$XmlPath"
        sed -i 's/<\/author>/<\/personname><\/author>/g' "$XmlPath"
    fi


    ### Validate

    if ! xmlstarlet val --err \
        --xsd /usr/share/xml/docbook/schema/xsd/5.0/docbook.xsd \
        "$XmlPath"
        then
            echo Fatal error: docbook file \"$XmlFile\" is not valid.
            exit 1
        fi



    ### Generate .pdf
    # xsltproc also take parameters e.g. --stringparam use.extensions 0\

    cp -r  "$InDir/images/" "$OutDir"

    cd "$OutDir"

    xsltproc --output "$FoFile" \
           "$XslPath"\
           "$XmlFile" && \
        fop -fo "$FoFile" -pdf "$PdfFile" && \
        rm  "$FoFile" && \
	    if [ "$OutDir" != "$OutDir" ] ; then
        	rm  "$XmlFile" && \
        	rm  -r "images"
        fi
}



### Main
# usage:
# render FILE [FILE]
# render report.md ./rendition/report.pdf

if [[ -n "${2/[ ]*\n/}" ]] ; then
        OutPath="$2"
else
        OutPath="${1%.*}.pdf" # $2 is empty
fi
render "$1" "$OutPath"