timecrawler

#! /bin/sh
exec ruby -S -x "$0" "$@"

#! ruby
require 'time'
require 'open-uri'
require 'optparse'

## Example of execution
# timecrawler http://example.com/images/%Y%m%d/%H%M%S.png \
#   -d /path/to/dir -since "2016-06-20 00:00:00" --until "2016-06-21 00:00:00" --interval $((60 * 5))

## strftime
# %Y 4-digit years
# %m 2-digit months
# %d 2-digit days
# %H 2-digit hours(24h)
# %M 2-digit minutes
# %S 2-digit seconds

# parse options
conf = {}.tap do |h|
  ARGV << '-h' if ARGV.empty?
  OptionParser.new("Usage: ruby timecrawler.rb [opions] [baseuri(formatted with strftime)]") do |opt|
    opt.on('-y', '確認しない') { |v| h[:continue] = true }
    opt.on('-q', '出力しない') { |v| h[:silent] = true }
    opt.on('-d [VALUE]', '--dest [VALUE]', '保存先パス') { |v| h[:dest] = v }
    opt.on('--interval [VALUE]', '間隔(秒) default:60') { |v| h[:interval] = v.to_i }
    opt.on('--since VALUE', '開始時刻') { |v| h[:since] = v }
    opt.on('--until [VALUE]', '終了時刻') { |v| h[:until] = v }
    opt.on('--duration [VALUE]', '期間(秒)') { |v| h[:duration] = v.to_i }

    opt.parse!(ARGV)
  end
end

# validate parameters
conf[:interval] ||= 60
conf[:dest] ||= "."
baseuri = ARGV.pop
raise ArgumentError.new("Need to specify a uri to crawl") unless baseuri
raise ArgumentError.new("Need to specify either --until or --duration") unless conf[:until].nil? ^ conf[:duration].nil?
raise ArgumentError.new("--interval value must be greater than 0") unless conf[:interval] > 0

current = Time.parse(conf[:since])
finish = conf[:duration].nil? ? Time.parse(conf[:until]) : current + conf[:duration]

puts "The given range: #{current} -> #{finish}"
unless conf[:continue]
  print "continue? (y/n)> "
  exit 1 unless gets.chomp =~ /^y$/i
end

while (current <= finish)
  uri = current.strftime(baseuri)
  puts "Downloading #{uri} ..." unless conf[:silent]
  begin
    open(uri) do |s|
      open(File.join(conf[:dest], File.basename(uri)), 'w') do
        |f| f.write(s.read)
      end
    end
  rescue => e
    warn "Failed to download #{uri}. (#{e.inspect})"
  end
  current += conf[:interval]
end