gistfile1.txt

photo_organizer.rb

require 'find'
require 'exifr'
require "fileutils"

class PhotoOrganizer
  SOURCE_DIRS = ["/media/truecrypt1/nick_picks"]
  EXCLUDES = ["CVS"]
  DEST_DIR = "/media/truecrypt1/dow_jones"
  DEST_DIR_OTHER = "/media/truecrypt1/dow_jones/other"
  DUPLICATES = "duplicates"
  IMAGE_FILE_EXTS = [".jpg", ".jpeg"]

  STATS_LOG_FILE = "stats.log"
  DUPLICATES_LOG_FILE = "duplicate_files.log"

  def run
    FileUtils.mkdir_p "#{DEST_DIR}#{File::SEPARATOR}#{DUPLICATES}"
    FileUtils.mkdir_p DEST_DIR_OTHER

    @files_with_valid_exif_data = []
    @files_with_invalid_exif_data = []
    @files_with_invalid_ext = []
    @duplicate_files = []
    @files_with_same_timestamp = []
    @unique_files = []

    walk_the_source_dir
    write_log_file
  end

  def write_log_file
    log = <<EOS
Files with valid EXIF data = #{@files_with_valid_exif_data.size}
Files with invalid EXIF data = #{@files_with_invalid_exif_data.size}
Files with invalid_ext = #{@files_with_invalid_ext.size}
Duplicate files = #{@duplicate_files.size}
Files with same timestamp = #{@files_with_same_timestamp.size}
Unique files = #{@unique_files.size}
EOS
    File.open("#{DEST_DIR}#{File::SEPARATOR}#{STATS_LOG_FILE}", 'w') {|f| f.write(log) }
    
    File.open("#{DEST_DIR}#{File::SEPARATOR}#{DUPLICATES_LOG_FILE}", 'w') {|f| 
      @duplicate_files.each { |entry|
        f.puts "duplicate of #{entry[:original]} found in #{entry[:duplicate]}"
      }
    }
  end

  def walk_the_source_dir
    SOURCE_DIRS.each do |dir|
      Find.find(dir) do |path|
        if FileTest.directory?(path)
          if EXCLUDES.include?(File.basename(path))
            Find.prune       # Don't look any further into this directory.
          else
            next
          end
        else
          #p "processing #{path}..."
          process_file path
        end
      end
    end
  end

  def process_file source_path
    file_ext = File.extname(source_path.downcase)
    if IMAGE_FILE_EXTS.include? file_ext
      exif_file = EXIFR::JPEG.new(source_path)
      if  exif_file.exif? && exif_file.date_time
        @files_with_valid_exif_data << source_path
        copy_file(source_path, exif_file.date_time)
      else
        p "#{source_path}: doesn't contain valid EXIF data, hence using ctime"
        @files_with_invalid_exif_data << source_path
        copy_file(source_path, File.new(source_path).ctime)
      end
    else
      p "#{source_path}:doesn't have a valid file ext. Coping to Other"
      @files_with_invalid_ext << source_path
      FileUtils.cp  source_path, "#{DEST_DIR_OTHER}#{File::SEPARATOR}#{File.basename(source_path)}"
    end
  end

  def copy_file source_path, timestamp
    dest_dir = "#{DEST_DIR}#{File::SEPARATOR}#{timestamp.year}#{File::SEPARATOR}#{timestamp.strftime("%b")}#{File::SEPARATOR}#{timestamp.strftime("%d - %a")}"
    dest_file_path = "#{dest_dir}#{File::SEPARATOR}#{timestamp.strftime("%p %I:%M:%S")}"
    FileUtils.mkdir_p dest_dir
    if File.exists?(dest_file_path)
      if FileUtils.compare_file(source_path, dest_file_path)
        @count = 0
        @duplicate_files << {:duplicate => source_path, :original => dest_file_path}
        FileUtils.cp  source_path, "#{DEST_DIR}#{File::SEPARATOR}#{DUPLICATES}#{File::SEPARATOR}#{File.basename(source_path)}-#{rand(51252)}"
      else #two or more uniuqe files with the same timestamp
        @count = @count + 1
        @files_with_same_timestamp << source_path
        FileUtils.cp  source_path,"#{dest_file_path} - #{@count}"
      end
    else
      @unique_files << source_path
      @count = 0
      FileUtils.cp  source_path, dest_file_path
    end

    p "copied #{source_path} to #{dest_file_path}"
  end
end