AssembleArchiveOrgBorrowedBook.scala

val dirIn = userHome / "Downloads" / "bla"
val pIn = dirIn.children(_.name.endsWith(".html")).sorted(File.NameOrdering)

def getPageFiles(htmlF: File): Seq[File] = {
  val dir = htmlF.parent
  val in  = new java.io.FileInputStream(htmlF)
  val arr = new Array[Byte](in.available)
  in.read(arr)
  in.close()
  val html = new String(arr, "UTF-8")
  var res = Seq.empty[File]
  
  def getPage(c: Char): Unit = {
    val iL = html.indexOf(s"""data-side="$c"""")
    if (iL > 0) {
      val jL = html.indexOf("""src="""", iL) + 5
      val kL = html.indexOf("\"", jL)
      val name = html.substring(jL, kL)
      val f = dir / name
      if (!f.isFile) println(s"! Warning. File $name does not exist")
      res :+= dir / name
    }
  }
  getPage('L')
  getPage('R')
  res
}


val imagesIn = pIn.flatMap(getPageFiles)
imagesIn.size

// BookReaderImages.jpeg
// BookReaderImages_002.jpeg
// BookReaderPreview_002.jpeg
// BookReaderPreview.jpeg

// val dirImgOut = dirIn / "pages"
// dirImgOut.mkdir()
// imagesIn.zipWithIndex.foreach { case (f, fi) =>
//   import sys.process._
//   val ext = f.extL
//   val fOut = dirImgOut / s"page-${fi + 1}.$ext"
//   Seq("cp", f.path, fOut.path).!
// }

val cmd = Seq("convert", "-density", "150") ++ imagesIn.map(_.path) ++ Seq((dirIn / "output.pdf").path)

// NOTE: if running out of memory (""),
// then edit /etc/ImageMagick-6/policy.xml
// and change <policy domain="resource" name="disk" value="1GiB"/> to <policy domain="resource" name="disk" value="8GiB"/>

{ import sys.process._; cmd.! }