Sakai disk usage reports by binary properties

binary-resource-report.rb

#!/usr/bin/env ruby
#
# binary-resource-report.rb
#   Reports disk usage broken down by properties in the binary_entity field.
#
# WARNING: takes a long time to run
# (~5 hours for 10,000,000 records)
#
# Written by David Adams ([email protected])
#
# This software is licensed to the Public Domain; No Rights Reserved
#
# Requirements
#   You must install the sakai-info gem as well as a database driver.
#   Tested drivers are:
#     For Oracle: ruby-oci8
#     For MySQL: mysql2
#
#   MySQL access may also work with the mysqlplus or mysql drivers.
#

######################################################################
#
# Variables
#
# connect to database using Sequel connection strings
#  (http://sequel.jeremyevans.net/rdoc/files/doc/opening_databases_rdoc.html)
#

CONNECTION_STRING = "oracle://username:password@dbsid"

#
# Report filenames
#

CREATORS_REPORT = "creators.csv"
CREATEDATES_REPORT = "creation-dates.csv"
MIMETYPES_REPORT = "mimetypes.csv"
EXTENSIONS_REPORT = "file-extensions.csv"

#
######################################################################

STDOUT.sync = true

require 'sakai-info'

SakaiInfo::DB.configure({ "db" => CONNECTION_STRING })

creators = {}
createdates = {}
mimetypes = {}
extensions = {}

counter = 0
creator_missing = 0
created_errors = 0
mimetype_errors = 0
no_display_name = 0

print "Collecting data..."
SakaiInfo::DB.connect[:content_resource].select(:resource_id, :binary_entity).all.each do |dbrow|
  begin
    meta = SakaiInfo::ContentBinaryEntity.new(dbrow[:binary_entity])

    # creators
    creator = meta["CHEF:creator"]
    if creator.nil?
      creator_missing += 1
    else
      if creators[creator].nil?
        creators[creator] = { :count => 1, :size => meta["content_length"] }
      else
        creators[creator][:count] += 1
        creators[creator][:size] += meta["content_length"]
      end
    end

    # created dates
    created = meta["DAV:creationdate"]
    if not created.nil?
      created = created.slice(0,8)

      if createdates[created].nil?
        createdates[created] = { :count => 1, :size => meta["content_length"] }
      else
        createdates[created][:count] += 1
        createdates[created][:size] += meta["content_length"]
      end
    else
      created_errors += 1
    end

    # mimetypes
    mimetype = meta["content_type"]
    if not mimetype.nil?
      if mimetypes[mimetype].nil?
        mimetypes[mimetype] = { :count => 1, :size => meta["content_length"] }
      else
        mimetypes[mimetype][:count] += 1
        mimetypes[mimetype][:size] += meta["content_length"]
      end
    else
      mimetype_errors += 1
    end

    # file extensions
    extension = ""
    if not meta["DAV:displayname"].nil?
      extension = meta["DAV:displayname"].split(".").last
      extension = "" if extension == meta["DAV:displayname"]
    else
      no_display_name += 1
    end
    if extensions[extension].nil?
      extensions[extension] = { :count => 1, :size => meta["content_length"] }
    else
      extensions[extension][:count] += 1
      extensions[extension][:size] += meta["content_length"]
    end

    # UI
    counter += 1
    if counter > 500
      print "."
      counter = 0
    end
  rescue => e
    STDERR.puts "ERROR: #{dbrow[:resource_id]}"
    STDERR.puts "    #{e}"
  end
end
puts " OK"

puts
puts "Quick summary:"
puts "  Unique file creators: #{creators.keys.count}"
puts "  Unique creation dates: #{createdates.keys.count}"
puts "  Unique mime types: #{mimetypes.keys.count}"
puts "  Unique file extensions: #{extensions.keys.count}"
puts
puts "  Files with no creator: #{creator_missing}"
puts "  Created date errors: #{created_errors}"
puts "  Mime type errors: #{mimetype_errors}"
puts "  Missing display name: #{no_display_name}"
puts


print "Writing out #{CREATORS_REPORT}... "
File.open(CREATORS_REPORT, "w") do |f|
  f.puts "user_id,file count,total size"
  creators.keys.each do |user_id|
    f.puts "#{user_id},#{creators[user_id][:count]},#{creators[user_id][:size]}"
  end
end
puts "OK"

print "Writing out #{CREATEDATES_REPORT}... "
File.open(CREATEDATES_REPORT, "w") do |f|
  f.puts "creation date,file count,total size"
  createdates.keys.sort.each do |createdate|
    f.puts "#{createdate},#{createdates[createdate][:count]},#{createdates[createdate][:size]}"
  end
end
puts "OK"

print "Writing out #{MIMETYPES_REPORT}... "
File.open(MIMETYPES_REPORT, "w") do |f|
  f.puts "mimetype,file count,total size"
  mimetypes.keys.each do |mimetype|
    f.puts "#{mimetype},#{mimetypes[mimetype][:count]},#{mimetypes[mimetype][:size]}"
  end
end
puts "OK"

print "Writing out #{EXTENSIONS_REPORT}... "
File.open(EXTENSIONS_REPORT, "w") do |f|
  f.puts "extension,file count,total size"
  extensions.keys.each do |extension|
    f.puts "#{extension},#{extensions[extension][:count]},#{extensions[extension][:size]}"
  end
end
puts "OK"