archive Google Code wiki to HTML, replace old wiki with links to archive

gistfile1.scala

import org.jsoup._
import java.io.{File, FileWriter}

/*
libraryDependencies += "org.jsoup" % "jsoup" % "1.7.1"
scalaVersion := "2.10.0-M7"
*/
object Main {
	val WikiSuffix = ".wiki"
	val HtmlSuffix = ".html"
	val Href = "href"
	val WikiAuthorClass = "#wikiauthor"
	val WikiLinkPrefix = "/p/simple-build-tool/wiki/"
	val SbtMoved = "sbt_has_moved"
	val RobotsMeta = """<meta name="robots" content="noindex, nofollow"></meta>"""

	def main(args: Array[String]) {
		val inDir = new File(args(0))
		val outDir = new File(args(1))

		for(page <- inDir.listFiles().take(1); name <- wikiPageName(page)) {
			write(generateHtml(name), new File(outDir, name + HtmlSuffix), name)
			write(newWikiContent(name), page, name)
		}
	}

	def write(content: String, file: File, name: String)
	{
		println("Writing " + file.getAbsolutePath + "...")
		file.getParentFile.mkdirs()
		val out = new FileWriter(file)
		try { out.write(content) }
		finally { out.close() }
		println("Transformed " + name)
	}
	def generateHtml(pagename: String): String =
	{
		val url = wikiURL(pagename)
		println("Retrieving " + url)
		val doc = Jsoup.connect(url).get();
		trim(doc)
		transformInternalLinks(doc)
		addRobotsMeta(doc)
		doc.outerHtml()
	
	}
	def addRobotsMeta(doc: nodes.Document)
	{
		doc.select("head").append(RobotsMeta)
	}
	def trim(doc: nodes.Document)
	{
		doc.select(WikiAuthorClass).remove()
		doc.select(s"a.label[$Href*=label:Featured]").remove()
		removeSbtMoved(doc)
	}
	def removeSbtMoved(doc: nodes.Document)
	{
		for(moved <- select(doc, s"h1 > a[name=$SbtMoved]")) {
			val h1 = moved.parent
			val next = h1.nextElementSibling
			next.nextElementSibling.remove()
			next.remove()
			h1.remove()
		}
		for(a <- select(doc, s"a[href=#$SbtMoved]"))
			a.parent.remove()
	}
	def transformInternalLinks(doc: nodes.Document)
	{
		for(a <- select(doc, s"a[$Href^=$WikiLinkPrefix]"))
			a.attr(Href, a.attr(Href).stripPrefix(WikiLinkPrefix))
	}
	def select(doc: nodes.Document, s: String) =
	{
			import collection.JavaConverters._
		doc.select(s).iterator.asScala
	}

	def wikiPageName(file: File): Option[String] =
		if(file.getName.endsWith(WikiSuffix))
			Some(file.getName.stripSuffix(WikiSuffix))
		else
			None
		
	def wikiURL(name: String) = s"http://code.google.com/p/simple-build-tool/wiki/$name?show=content"


	def newWikiContent(name: String): String = s"""
= sbt has moved =

*sbt is now hosted on !GitHub*: https://github.com/harrah/xsbt.

This page previously documented sbt 0.7.7, which is *no longer maintained*.

The original content is archived at http://www.scala-sbt.org/0.7.7/docs/$name.html
"""

}