はてなダイアリーはてなブックマークランキング

bookmark_ranking.rb

# -*- coding: utf-8 -*-
require 'open-uri'
require 'nokogiri'
require 'cgi'
require 'digest/sha1'

class Entry
  attr_accessor :url, :title
  def initialize(url, title)
    @url = url
    @title = title
  end

  def bookmark_count
    return @bookmark_count if @bookmark_count
    warn self.url

    # キャッシュ+sleepしながら
    begin
      @bookmark_count = open(Digest::SHA1.hexdigest(url)).read.to_i
    rescue
      @bookmark_count = open("http://api.b.st-hatena.com/entry.count?url=#{CGI.escape(url)}").read.to_i
      open(Digest::SHA1.hexdigest(url), 'w'){ |f|
        f.write @bookmark_count
      }
      sleep 0.5
    end

    @bookmark_count
  end

  def date
    date_text = self.url.match(/http:\/\/d\.hatena\.ne\.jp\/.+?\/(\d+)/)[1]
    Time.local(date_text[0..3], date_text[4..5], date_text[6..7])
  end

  def as_hatena_syntax
    "[#{self.url}:title=#{self.title}:bookmark]"
  end
end

def fetch(hatena_id, offset = 0)
  url = "http://d.hatena.ne.jp/#{hatena_id}/archive?of=#{offset}"
  page = Nokogiri::HTML open(url)
  page.search('.archive-section a')
end

hatena_id = ARGV.first
raise 'usage: bookmark_ranking.rb HATENA_ID' unless hatena_id

# エントリが多いとこれではたりないことがありそう
entries = (0..3).map{|page|
  fetch(ARGV.first, page * 50)
}.flatten

entries.map{|a|
  Entry.new(a['href'], a.content.strip)
}.select{ |entry|
  entry.url.match(/http/) && entry.date.year == 2011
}.sort_by(&:bookmark_count).
  reverse.
  each_with_index{|entry, index|
  puts "|*#{index+1}位|#{entry.as_hatena_syntax}|"
}