A Groovy script to convert pragprog PML (a DocBook-like language) to Asciidoc

pml2adoc

def replacements = [
 "<bookname[^>]*>([^<]+?)</bookname>":" '\$1'"
,"<acronym>([^<]+?)</acronym>":" _\$1_"
,"<emph>([^<]+?)</emph>":" *\$1*"
,"<url>([^<]+?)</url>":"link:\$1[]"
,"<keystroke>([^<]+?)</keystroke>":"kbd:[\$1]"
,"<variablename>([^<]*?)</variablename>":"`\$1`"
,"<classname>([^<]+?)</classname>":"`\$1`"
,"<dirname>([^<]+?)</dirname>":"`\$1`"
,"<filename>([^<]+?)</filename>":"`\$1`"
,"<cf>([^<]+?)</cf>":"`\$1`"
,"<inlinecode>([^<]+?)</inlinecode>":"`\$1`"
,"<stringinfile>([^<]+?)</stringinfile>":"`\$1`"
,"<firstuse>([^<]+?)</firstuse>":"*_\$1_*"
,"<ldots.*?/>":"&#8230;"
,"``([^']*)''":"\"\$1\""
,"<pref[^l]+linkend=\"(.*?)\".*?/>":"<<\$1>>"
,"<ii.*?>.+?</ii>":""
,"<i.*?>[^<]+?</i>":""
,"<layout[^>]*>(.+?)</layout>":""
,"<chapter[^>]*>([^<]*)<title>([^<]+?)</title>":"= \$2\n\n"
,"</chapter>":""
,"<sect1[^>]*?>([^<]*)<title>([^<]+?)</title>":"== \$2\n\n"
,"</sect1>":""
,"<sect2[^>]*?>([^<]*)<title>([^<]+?)</title>":"=== \$2\n\n"
,"</sect2>":""
,"<sect3[^>]*?>([^<]*)<title>([^<]+?)</title>":"==== \$2\n\n"
,"</sect3>":""
,"<sect4[^>]*?>([^<]*)<title>([^<]+?)</title>":"===== \$2\n\n"
,"</sect4>":""
,"<li>\n[\\s]*([\\S]*)":"* \$1"
,"</li>":""
,"<ul.*?>":""
,"</ul>":""
,"<ol>":""
,"</ol>":""
,"<p>":""
,"</p>":""
,"<footnote>([^<]*)</footnote>":"\n\n[TIP]\n====\n\$1\n====\n\n"
,"<figure.*?>([^<]*)<title>([^<]*)</title>":"== \n\n.\$2"
,"</figure>":""
,"<imagedata.+?fileref=\"(.+?)\"[^/]*?/>":"image::\$1[]"
,"<code.+?file=\"(.+?)\"[^>]*?>.*?</code>":"[source,java]\n----\ninclude::\$1\n----"
,"<joeasks>([^<]*)<title>([^<]*)</title>":"== \n\n.\$2\n****"
,"</joeasks>":"****"
,"<ddd>([^<]*)<title>([^<]*)</title>([^<]*)</ddd>":"== \n\n.\$2\n****\$3\n****\n\n"
,"<!--[^-]+?-->":""
,"<!--.+?-->":""
]

def stripLeadingSpaces(str) {
  lines=str.split("\n")
  buf=""
  lines.each{line -> buf+= line.trim() +"\n" }
  return buf
}

def currentDir = new File(".");
currentDir.eachFileRecurse { file ->

  if (file.path.endsWith(".pml")) {
    adocFile = new File(file.path.replaceAll("[.]pml",".adoc"))
    println("${file.path} ---> ${adocFile.path}" )

    def fileText = file.text;
    replacements.each { from, to -> 
      //println(  from + " ---> " + to)

      p = java.util.regex.Pattern.compile(from)
      m = p.matcher(fileText);
      newFileText = m.replaceAll(to)
      fileText = newFileText;
    }    

    // strip off leading spaces
    newFileText = stripLeadingSpaces(newFileText)
    adocFile.write(newFileText);
  }
}