search_by_key.rb

# [単語\t絶対パス\t行番号]のファイルから単語検索する
require "digest/md5"

wanna_search = ARGV[1]
ext = ARGV[0]
context = (ARGV[2] || 1).to_i
max_output = (ARGV[3] || 10).to_i

unless wanna_search && ext
  STDERR.puts "args error"
end

$verbose = false
def get_key_path_num(str)
  if str =~ /^([0-9a-f]+)\t([^\t]*)\t([0-9]+)$/
    target_key = $1.to_i(16)
    target_path = $2
    target_num = $3.to_i
    #STDERR.puts "#{target_key},#{target_path},#{target_num}"
  end
  return target_key, target_path, target_num
end

# range: 整数, 前後どのくらいの行数を表示するか
def search_by_key(key,lines)
  raise "search_by_key: keyは整数で指定する" unless key.kind_of?(Integer)
  STDERR.puts "search_by_key(#{key},#{lines.size})" if $verbose
  half_point = (lines.size/2).floor
  #STDERR.puts "lines[half_point]:#{lines[half_point]}"
  if lines[half_point] =~ /^([0-9a-f]+)\t([^\t]*)\t([0-9]+)$/
    target_key = $1.to_i(16)
    target_path = $2
    target_num = $3.to_i
    #STDERR.puts "#{target_key},#{target_path},#{target_num}"
  end
  STDERR.puts "target_key:#{target_key} <-> key:#{key}" if $verbose
  STDERR.puts "half_point: #{half_point}" if $verbose
  if target_key == key
    STDERR.puts "==" if $verbose
    surround_results = [lines[half_point]]
    i=0
    # 前に同じ結果が並んでいればそれも得る
    while (lines[half_point-i]||"") =~ /^[0-9a-f]+/ && $~[0].to_i(16)==key
      surround_results << lines[half_point-i]
      i+=1
    end
    i=0
    # 後ろに同じ結果が並んでいればそれも得る
    while (lines[half_point+i]||"") =~ /^[0-9a-f]+/ && $~[0].to_i(16)==key
      surround_results << lines[half_point+i]
      i+=1
    end
    #return surround_results.map{|ln| ln=~/^([0-9a-f]+)\t([^\t]*)\t([0-9]+)$/ && [$2,$3] }.compact
    return surround_results
  elsif target_key < key && half_point < lines.size-1
    STDERR.puts "<" if $verbose
    return search_by_key(key,lines[half_point+1..-1])
  elsif target_key > key && half_point > 0
    STDERR.puts ">" if $verbose
    return search_by_key(key,lines[0..half_point-1])
  else
    return nil
  end
end

path = "C:/sample_codes"
results = search_by_key(Digest::MD5.hexdigest(wanna_search).to_i(16),File.read("#{path}/indexed_of_#{ext}.txt").split("\n").map{|s| s.strip})
results[0..max_output].compact.map{|e| e.split("\t")}.each do |key,path,num|
  puts "="*30
  ((num.to_i - context)..(num.to_i+context)).to_a.each do |i|
    puts File.read(path).split("\n")[i]
  end
end