custom-markup.KQNEWEJ6.wok

#!/usr/bin/wok -f

/** custom-markup.KQNEWEJ6.wok
  * 
  * This script modifies the search result for a word in Kenkyusha Shin EiWa Cyujiten (converted
  * to Epwing by dessed). The following changes will be applied to the data:
  *
  * 1. Fix too many ebkw problem.
  * 2. Add a class `pos` to lines start with the image file of zB125.
  * 2. Split `sense1` and `sense1` when those are connected.
  * 4. Norimalize indicators, "◆" and "◇", inserted before example sentences, to "・".
  * 
  * Usage:
  *   java -jar wok-0.1.0.jar -f custom-markup.KQNEWEJ6.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 = """<span class="ebin"><span class="ebkw"></span>(<span class="ebul">((?!<br>).)*)""".r
val p2 = """(━((?!<br>).)*)""".r
val p3 = """<br><span class="ebkw"></span><span class="ebul">([0-9]{1,2})a</span>""".r
val p4 = """<br><span class="ebkw"></span>(<span class="ebul">[0-9]{1,2}</span>((?!<br>).)*)""".r
val p5 = """<br><span class="ebkw"></span>(<span class="ebul">[a-z]</span>((?!<br>).)*)""".r
val p6 = """<br><span class="ebkw"></span>(\(<span class="ebul">[0-9]{1,2}</span>\)((?!<br>).)*)""".r
val p7 = """<span class="ebkw"></span>""".r

In { _ 
  .filter (_.size > 0)
  .map (_.padTo(idx+1, ""))
  .map { row =>
    var str = row(idx)
    str = p1.replaceAllIn(str, """<span class="ebin"><span class="ebkw">$1</span>""")
    str = p2.replaceAllIn(str, """<span class="pos">$1</span>""")
    str = p3.replaceAllIn(str, """<br><span class="ebkw"></span><span class="ebul">$1</span><br><span class="ebkw"></span><span class="ebul">a</span>""")
    str = p4.replaceAllIn(str, """<br><span class="sense1">$1</span>""")
    str = p5.replaceAllIn(str, """<br><span class="sense2">$1</span>""")
    str = p6.replaceAllIn(str, """<br><span class="sense3">$1</span>""")
    str = p7.replaceAllIn(str, "")
    row.updated(idx, str)
  }
  .foreach (row => println(row: _*))
}