From: kjana@... (YANAGAWA Kazuhisa)
Date: 1998-12-21T23:10:01+09:00
Subject: [ruby-list:11397] fold.rb

ちょっとしたスクリプトを書いてみました．割と使えると思うんで，公表して
みようかな，と．

    * テキストの折り畳みをします．

    * <- こういう箇条書やプリフィクスがついた文章もそれなりに扱えるつ
      もりです．

      # filladapt.el 並に便利に使えるといいな，と．

    * はなはだいい加減な禁則処理もついてます．

      # ちなみに禁則用の正規表現は JISX-0208 で定義された文字を EUC か
      # SJIS で表現し，正しく e/s を振ったものでなければなりません．

    * 日本語 EUC か ShiftJIS が入力になると安易に仮定しているため違う
      ものが入ってくるとわやになります．

    * パラグラフの区切りは空行か箇条書のプリフィクスがついている行のみ
      で，インデントの変化なんかは別に見ていません．

      # 簡単に直せると思うけど．

    * プリフィクスつきの箇条書はうまく扱えません．

      # * こんな感じの
      #   文章．

    * あんまり効率が良くないと思います．

    * コメントの英語に似た文字列が変です (^^;

なにか御意見頂ければ幸い．

# unexpand もつけようかと思っていたけど簡単にできなかったのでちょっと
# くじけてしまった．まあぼちぼちと，ね．

===========================================================================
  柳川和久 @ 東大阪市 . 大阪府
  kjana@os.xaxon.ne.jp                                  December 21, 1998
「....生き伸びられたら労災長者になれるわね」
「やっぱりあんた何にも考えてないな!」


#!/usr/local/bin/ruby

class NilClass
  def ascii?
    false
  end
end

class Fixnum
  def ascii?
    (0..127).include? self
  end
end

class String
  def normalize
    strip.gsub(/\s+/, " ")
  end

  def expand(tabs = 8)
    if tabs.is_a? Integer
      tw = tabs
      tabs = []
      for n in 0..(120/tw)
        tabs << tw*(n+1)
      end
    end

    nc = 0
    ts = 0
    col = 0
    ostr = ""
    while nc < size
      c = self[nc]
      case c
      when ?\t
        while not tabs[ts] > col
          ts += 1
        end
        ostr << " "*(tabs[ts]-col)
        col = tabs[ts]
      when ?\n
        ostr << "\n"
        col = 0
        ts = 0
      else
        ostr << c.chr
        col += 1
      end
      nc += 1
    end
    ostr
  end

  def join
    self.dup.join!
  end

  def join!
    n = 0
    while n < size
      n = index("\n", n)
      break if n.nil?
      if self[n-1].ascii? or self[n+1].ascii?
        self[n] = " "
        n += 1
      else
        self[n, 1] = ""
      end
    end
    self
  end

  def can_break?(n)
    if self[n, 1] =~ /\s/
      true
    elsif self[n].ascii?
      false
    else
      i = 0
      while i < n
        if not self[i].ascii?
          i += 2
        else
          i += 1
        end
      end
      if i == n
        true
      else
        false
      end
    end
  end

  # Regular expressions below should be given in same coding system of
  # expected input texts.

  HeadInhibitCharsHang = /^[、。，．！？：；…‥）｝］】〕」』＞》’”°]/e
  HeadInhibitChars = /^[ー〜ヽヾゝゞ〃仝々んっゃゅょぁぃぅぇぉンッャュョァィゥェォ]/e
  TailInhibitChars = /[（｛［【〔「『＜《‘“]$/e

  def fold(len)
    istr = normalize
    ostr = ""
    if size < len
      ostr << istr
    else
      while istr.size > len
        n = len
        while n > 0 and not istr.can_break? n
          n -= 1
        end
        if n == 0
          n = len
          while n < istr.size and not istr.can_break? n
            n += 1
          end
        end

        # not proper KINSOKU processing :-P
        n -= 2 while n > 1 and istr[0..n-1] =~ TailInhibitChars
        n -= 2 while n > 0 and istr[n..-1] =~ HeadInhibitChars
        n += 2 while n < istr.size and istr[n..-1] =~ HeadInhibitCharsHang

        ostr << istr[0..n-1]+"\n"
        n += 1 if istr[n] == ? # space
        istr = istr[n..-1]
      end
      ostr << istr
    end
    ostr
  end
end

if $0 == __FILE__

# Usage:
#   fold.rb [-wwidth] [file]
#
# Description:
#   Format a text file `file' in given width `width.' If `file' is omitted,
#   text from standarad input is formatted. If `width' is omitted, default
#   (70) is used.
#
#   Output texts are pretty good than those of fold(1). Prefixed texts,
#   such as items in a list or comments, are properly treated, I hope :-P
#
# Bugs:
#   * Japanese-EUC/SJIS texts are expected for inputs. So ISO-2022 or
#     ISO-8859 texts are not properly handled.
#   * Nested prefixed form (like this section) is not supported.
#   * More efficient execution is desired.
#   * ....somthing more :-<

  DotPrefix = /^(\s*(?:\d+(?:\.|\))|\*|\+|\-)\s*)/
  DupPrefix = /^(\s*(?:#)?\s*)/

  w = 70
  dotp = ""
  pref = ""
  pline = ""

  if ARGV[0] =~ /-w(\d+)/
    w = $1.to_i
    ARGV.shift
  end

  while not eof?
    line = pline
    while not eof?
      gets
      if $_ =~ DotPrefix or $_ =~ /^\s*$/
        pline = $_
        break
      end
      line << $_
    end

    case line
    when DotPrefix
      dotp = $1
      pref = " "*$1.expand.size
      line = $'
    when DupPrefix
      dotp = ""
      pref = $1
      line.gsub! DupPrefix, ""
    else
      dotp = ""
      pref = ""
    end

    line.join.fold(w-pref.size).each do
      |ln|
      if dotp == ""
        puts pref+ln
      else
        puts dotp+ln
        dotp = ""
      end
    end

    if pline =~ /^\s*$/
      puts
      pline = ""
    end
  end
end