analysis.rb

# -*- coding: utf-8 -*-
require 'active_support/time'
require 'json'
require 'scanf'
require 'net/http'
require 'nkf'
require_relative './csv.rb'
require 'pp'

# "147908914338480128","111217 142007","府中二郎の小ラーメン全部（豚・温泉玉子・チーズ）入りの食券 http://t.co/RMh2BDwP"
year = 2011
path = ARGV[0]
$count = 0
$last_count_date = Date.new(2010)
$day_count = 0
$eat_count = 0
$url_regxp = Regexp.new("(http:\/\/[-_.!~*\'()a-zA-Z0-9;\/?:\@&=+\$,%#]+)")
$jiro_regxp = Regexp.new("二郎|マシ|小豚")

data = []
KCSV.foreach(path) do |row|
  date = Date.parse(row[1])
  if date.year == year
    data << [date, row[2]]
  else
    break
  end
end
puts "read done."

# sub routine
def parse(tweet)
  if $url_regxp =~ tweet[1]
    uri = $1
    if /(http:\/\/t\.co.*)/ =~ tweet[1] 
      uri = Net::HTTP.get_response(URI.parse(uri))["Location"]
      begin 
        tweet[1].sub!(/(http:\/\/[-_.!~*\'()a-zA-Z0-9;\/?:\@&=+\$,%#]+)/, uri) 
      rescue
        pp "Error:" + uri.to_s
      end
    end

    if $jiro_regxp =~ tweet[1] 
      $eat_count += 1        
      $count += 1
      if $last_count_date != tweet[0]
        $day_count += 1
        $last_count_date = tweet[0]
      end
      puts tweet[0].to_s + ":" + tweet[1].to_s
    end
  end
end

def compare_date(a, b)
  return a.year == b.year && a.month == b.month && a.day == b.day
end

date = Date.new(2011)
day = 0
data.each do |tweet|
  parse(tweet)
  date = date.since(day.days).to_date
end

puts "合計：#{data.length}"
puts "「二郎」：#{$count}, #{$day_count}, #{$eat_count}回, #{($count*100/data.length)}%"