「ことばのおもみ」自動測定スクリプト http://doratex.hatenablog.jp/entry/20150331/1427755389

wordweight

#!/usr/bin/ruby -W0
# coding: utf-8
#
# This software is derived from bmp.rb provided by Yoichi Yokogawa.
# https://github.com/yoyoichi/bmp-ruby
#
# This software is released under the MIT License.
# http://opensource.org/licenses/mit-license.php

### BitMap class
class BitMap
  def initialize(width, height, dpi = 96)
    @width = width
    @height = height
    @line_size = width * 3 + (4 - (width * 3) % 4) % 4
    @buf_size = @line_size * height
    @buf = ("\000" * @buf_size).encode('BINARY')
    @bit_count = 24
    @compression = 0
    @size_image = 0
    @x_pix_per_meter = (39.375 * dpi).round
    @y_pix_per_meter = (39.375 * dpi).round
    @clr_used = 0
    @cir_important = 0
    @black_count = -1
    @black_ratio = -1
  end

  attr_writer :buf
  attr_reader :width, :height

  def BitMap.read(filename)
    buf = nil
    open(filename, "rb") do |f|
      buf = f.read
    end

    if buf[0] != ?B or buf[1] != ?M
      raise '[Error] read: Invalid Header'
    end
    real_buf_size = buf.size
    buf_size = (buf[2, 4].unpack("l*"))[0]
    if buf_size > real_buf_size
      raise '[Error] read: Invalid Buffer Size'
    end
    data_offset = (buf[10, 4].unpack("l*"))[0]
    if data_offset != 54
      raise '[Error] read: Invalid Data Offset'
    end

    width = (buf[18, 4].unpack("l*"))[0]
    height = (buf[22, 4].unpack("l*"))[0]

    bit_count = (buf[28, 2].unpack("s*"))[0]
    if bit_count != 24
      raise '[Error] read: Unsupported Color Depth'
    end

    compression = (buf[30, 4].unpack("l*"))[0]
    if compression != 0
      raise '[Error] read: Compression Not Supported'
    end

    pix_per_meter = (buf[38, 4].unpack("l*"))[0]
    dpi = pix_per_meter / 39.375

    image_buf = buf[54, buf_size]

    image = BitMap.new(width, height, dpi)
    image.buf = image_buf

    return image
  end

  def pget(x, y)
    x = 0  if x < 0
    x = @width - 1  if x >= @width
    y = 0  if y < 0
    y = @height - 1  if y >= @height

    addr = (@height - 1 - y) * @line_size + x * 3
    b = @buf[addr    ].ord
    g = @buf[addr + 1].ord
    r = @buf[addr + 2].ord

    return [r, g, b]
  end
  
  def blackPixelCount
    return @black_count if @black_count >= 0
    
    count = 0
    (0...@width).each{|x|
      (0...@height).each{|y|
        count +=1 if pget(x, y) == [0, 0, 0]
      }
    }
    @black_count = count
  end
  
  def blackPixelRatio
    return @black_ratio if @black_ratio >= 0
    
    @black_ratio = blackPixelCount.to_f / (@width*@height)
  end
end

### main

require 'fileutils'
require 'optparse'

SCRIPT = $0
TEMPNAME = "wordweight"
$texfile = "#{TEMPNAME}.tex"
$bmpfile = "#{TEMPNAME}.bmp"

def usage
  puts "Usage : #{File.basename(SCRIPT)} [\e[1m-s\e[0m \e[4msize\e[0m] [\e[1m-r\e[0m \e[4mresolution\e[0m] [\e[1m-d\e[0m] \e[4mfontname\e[0m \e[4mwords...\e[0m "
  puts "Options:"
  puts "  \e[1m-s\e[0m \e[4msize\e[0m       : set font size (pt) (default: 100)"
  puts "  \e[1m-r\e[0m \e[4mresolution\e[0m : set resoluion for font rendering (default: 600)"
  puts "  \e[1m-d\e[0m            : do not delete temporary files for debug"
  exit
end

def clean
  FileUtils.rm_rf([$texfile, "#{TEMPNAME}.aux", "#{TEMPNAME}.log", "#{TEMPNAME}.pdf", $bmpfile])
end

def measure(word)
  clean
  zw = $fontsize.to_f
  ht = 0.88*zw
  dp = 0.12*zw

  open($texfile, "w") {|f| f.write <<EOS
\\documentclass{article}
\\usepackage{fontspec}
\\newfontfamily\\theFont{#{$fontname}}
\\pagestyle{empty}
\\usepackage[tightpage,active]{preview}
\\PreviewEnvironment{minipage}
\\parindent=0pt
\\begin{document}
\\setbox0=\\hbox{\\theFont\\fontsize{#{$fontsize}pt}{0pt}\\selectfont #{word}}%
\\ht0=#{ht}pt
\\dp0=#{dp}pt
\\begin{minipage}{\\wd0}\\box0\\end{minipage}
\\end{document}
EOS
}

  system("xelatex -no-shell-escape -interaction=batchmode #{TEMPNAME} > /dev/null")
  system("gs -sDEVICE=bmp16m -dNOPAUSE -dBATCH -sOutputFile=#{$bmpfile} -r#{$resolution} #{TEMPNAME}.pdf > /dev/null") if $? == 0

  if $? == 0 && File.exist?($bmpfile)
    bmp = BitMap.read($bmpfile)
    puts "Font name    : #{$fontname}"
    puts "Font size    : #{$fontsize}pt"
    puts "Word         : #{word}"
    puts "Resolution   : #{$resolution}dpi"
    puts "Total pixels : #{bmp.width * bmp.height}"
    puts "Black pixels : #{bmp.blackPixelCount}"
    puts "Ratio        : #{sprintf("%.2f%%", 100*bmp.blackPixelRatio)}"
  else
    puts "ERROR"
  end

  clean unless $debug
end


opt = OptionParser.new

$fontsize = 100
$resolution = 600
$debug = false

begin
  opt.on('-s VAL') {|v|
    $fontsize =  v.to_i
    usage unless $fontsize > 0
  }
  opt.on('-r VAL') {|v|
    $resolution = v.to_i
    usage unless $resolution > 0
  }
  opt.on('-d') { $debug = true }
  opt.parse!(ARGV)
  usage unless ARGV.length >= 2
  $fontname = ARGV.shift
  words = ARGV
rescue
  usage
end

words.each{|w| measure(w)}