#! /usr/bin/env ruby
#
# mailhist - generate a histogram of the number of mails per day.
#
# Copyright (C) 2003 Satoru Takabayashi <satoru@namazu.org>
#     All rights reserved.
#     This is free software with ABSOLUTELY NO WARRANTY.
#
# You can redistribute it and/or modify it under the terms of 
# the GNU General Public License version 2.
#

require 'find'
require 'time'
require 'mailread'
require 'getoptlong'

class Time
  def yymmdd
    self.strftime("%Y-%m-%d")
  end
end
    
def show_histogram (stat)
  first_day = Time.parse(stat.keys.sort.first)
  last_day = Time.parse(stat.keys.sort.last)

  first_day.to_i.step(last_day.to_i, 86400) {|time|
    date = Time.at(time).yymmdd
    stat[date] = 0 unless stat[date]
  }

  ndays = 0
  total = 0
  hist = []
  stat.each {|key, value|
    hist[value] = 0 unless hist[value]
    hist[value] += 1
    ndays += 1
    total += value
  }

  puts "一日に出したメールの数の棒グラフ"
  printf("%s 〜 %s (全%d日, 全%d通, 平均%.2f通/日)\n", 
         first_day.yymmdd, last_day.yymmdd, ndays, total,
         total.to_f / ndays)

  max_day = stat.values.max
  ratio = 60.0 / hist.find_all{|x| x}.max
  ratio = 1.0 if ratio > 1.0

  (0..max_day).each {|i|
    val = (hist[i] or 0)
    printf("%4d通 %4d日	%s\n", i, val, "*" * (val * ratio))
  }
  puts
end

def parse_options
  options = Hash.new
  parser = GetoptLong.new
  parser.set_options(['--from', '-f', GetoptLong::REQUIRED_ARGUMENT])
  parser.each_option {|name, arg|
    options[name.sub(/^--/, "")] = arg
  }
  if ARGV.empty?
    puts "Usage: mailhist [OPTION] FOLDER..."
    puts("  -f, --from=RE    count mails only matching RE in From:")
    exit(1)
  end
  return options
end

def collect_timestamps (folders, from_re)
  timestamps = []
  folders.each {|folder|
    Find.find(folder) {|file_name|
      next unless File.file?(file_name)
      begin
        mail = Mail.new(file_name)
        if from_re.match(mail['From'])
          timestamp = Time.parse(mail['Date'])
          timestamps.push(timestamp)
        end
      rescue # malformed mail?
      end
    }
  }
  return timestamps.sort
end

def process (timestamps)
  prev_year = timestamps.first.year
  first_year = prev_year

  total_stat = {}
  yearly_stat = {}

  timestamps.each {|timestamp|
    if timestamp.year != prev_year
      show_histogram(yearly_stat)
      yearly_stat = {}
    end
    date = timestamp.yymmdd
    yearly_stat[date] = 0 unless yearly_stat[date]
    yearly_stat[date] += 1
    total_stat[date] = 0 unless total_stat[date]
    total_stat[date] += 1
    prev_year = timestamp.year
  }

  show_histogram(yearly_stat)
  show_histogram(total_stat) if first_year != prev_year
end

def main
  options = parse_options

  folders = ARGV
  from_re = if options['from'] then Regexp.new(options['from']) else // end

  timestamps = collect_timestamps(folders, from_re)
  process(timestamps)
end

main