custom-markup.SAITOWAEI_EP.wok

#!/usr/bin/wok -f

/** custom-markup.SAITOWAEI_EP.wok
  * 
  * This script modifies the search result for a word in NEW Saito Waei Daijiten (converted
  * to Epwing by dessed). The following changes will be applied to the data:
  *
  * 1. Split pos and sense by `<br>` when those are connected.
  * 2. Add a class `pos` to lines start with "〈".
  * 2. Add a class `sense1` to lines start with "NUMBER. " (e.g, "1. ").
  * 4. Norimalize indicators, "◆" and "◇", inserted before example sentences, to "・".
  * 
  * Usage:
  *   java -jar wok-0.1.0.jar -f custom-markup.SAITOWAEI_EP.wok -v idx=0 input > output
  *
  * Option:
  *   -v idx:  The index number of the target column.
  *
  */

FS = '\t'
FQ = Quote.Min
OFS = '\t'
OFQ = Quote.All

val p1 = """<br>(〈[^〉]+〉)(<span class="ebbo">[0-9]{1,2}</span>\.)""".r
val p2 = """<br>(〈((?!<br>).)*)""".r
val p3 = """<br>( ?<span class="ebbo">[0-9]{1,2}</span>\.((?!<br>).)*)""".r
val p4 = """<br>(◆|◇)""".r

In { _ 
  .filter (_.size > 0)
  .map (_.padTo(idx+1, ""))
  .map { row =>
    var str = row(idx)
    str = p1.replaceAllIn(str, """<br>$1<br>$2""")
    str = p2.replaceAllIn(str, """<br><span class="pos">$1</span>""")
    str = p3.replaceAllIn(str, """<br><span class="sense1">$1</span>""")
    str = p4.replaceAllIn(str, """<br>・""")
    row.updated(idx, str)
  }
  .foreach (row => println(row: _*))
}