fix_tsv.scala

// fix a tsv with unescaped newlines

val lines = io.Source.fromFile(argv(0)).getLines

val header = lines.next()
val tabCount = header.count(_ == '\t')

println(header)

while (lines.hasNext) {
  var lineTabCount = 0
  while (lineTabCount < tabCount) {
    val line = lines.next()
    print(line)
    lineTabCount += line.count(_ == '\t')
  }
  println()
}

validate_tsv.scala

// check the number of tabs on each line in a tsv

val lines = io.Source.fromFile(argv(0)).getLines

val header = lines.next()
val tabCount = header.count(_ == '\t')

println("%d columns" format tabCount)

var i = 1

while (lines.hasNext) {
  val lineTabCount = lines.next().count(_ == '\t')
  if (lineTabCount != tabCount) {
    println("line %10d has %d columns" format (i, lineTabCount))
  }
  i += 1
}