WIP: Extract EULAs from a .dmg file

lib/cask/container.rb

@@ -6,6 +6,7 @@ class Cask::Container; end
 require 'cask/container/cab'
 require 'cask/container/criteria'
 require 'cask/container/dmg'
+require 'cask/container/dmg_eula'
 require 'cask/container/generic_unar'
 require 'cask/container/gzip'
 require 'cask/container/naked'

lib/cask/container/base.rb

@@ -4,4 +4,12 @@ def initialize(cask, path, command)
     @path = path
     @command = command
   end
+
+  def eula?
+    !eulas.empty?
+  end
+
+  def eulas
+    []
+  end
 end

lib/cask/container/dmg.rb

@@ -9,6 +9,10 @@ def initialize(*args)
     @mounts = []
   end
 
+  def eulas
+    @eulas ||= Cask::Container::DmgEula.all(realpath)
+  end
+
   def extract
     mount!
     assert_mounts_found
@@ -34,8 +38,7 @@ def mount!
     plist = @command.run('/usr/bin/hdiutil',
       # :startup may not be the minimum necessary privileges
       :bsexec => :startup,
-      # realpath is a failsafe against unusual filenames
-      :args => %w[mount -plist -nobrowse -readonly -noidme -mountrandom /tmp] + [Pathname.new(@path).realpath],
+      :args => %w[mount -plist -nobrowse -readonly -noidme -mountrandom /tmp] + [realpath],
       :input => %w[y]
     ).plist
     @mounts = mounts_from_plist(plist)
@@ -75,4 +78,11 @@ def eject!
       raise CaskError.new "Failed to eject #{mountpath}"
     end
   end
+
+  private
+
+  def realpath
+    # realpath is a failsafe against unusual filenames
+    Pathname.new(@path).realpath
+  end
 end

lib/cask/container/dmg_eula.rb

@@ -0,0 +1,99 @@
+class Cask::Container::DmgEula
+  attr_reader :dmg_filename, :format, :language_id, :language_name
+
+  LANGUAGE_RANK = lambda do |eula|
+    region_code = eula.mac_system_region_code
+    locale = Cask::Utils::Locale.locale
+    locale_match = locale.match_system_region?(region_code)
+
+    [
+      (locale_match ? 0 : 1),   # Prefer exact matches
+      eula.language_id,         # If inexact, prefer lower IDs (0 is English)
+      (eula.rich_text? ? 0 : 1) # Also, prefer rich text over plain text
+    ]
+  end
+
+  def self.all(dmg_filename)
+    EULASpecParser.new(dmg_filename)
+      .parse
+      .sort_by(&LANGUAGE_RANK)
+  end
+
+  def show!(title)
+    less_prompt = [
+      %Q[License agreement for "#{ title }"],
+      'Page %dm ?Bof %D.',
+      '[H]elp  [Q]uit'
+    ].join('    ')
+
+    system("#{ content_command } | fmt | less -P '#{ less_prompt }'")
+  end
+
+  def content
+    @content ||= `#{ content_command }`
+  end
+
+  def mac_system_region_code
+    language_id - Cask::Utils::Locale.resource_fork_region_base_id
+  end
+
+  def plain_text?
+    format === 'TEXT'
+  end
+
+  def rich_text?
+    format === 'RTF '
+  end
+
+  def to_s
+    "#{ language_name } language EULA (id: #{ language_id }, format: #{ file_extension })"
+  end
+
+  def initialize(dmg_filename, format, language_id, language_name)
+    @dmg_filename = dmg_filename
+    @format = format
+    @language_id = language_id
+    @language_name = language_name
+  end
+
+  private
+
+  def content_command
+    %Q<hdiutil udifderez -rez "#{ dmg_filename }" | awk -F\\" '/^data .#{ format }. \\(#{ language_id }[,)]/{current=1} /^\\t\\$/ {if(current) {print $2}} /};/{current=0}' | xxd -r -p | iconv -f MAC -t UTF-8 | tr '\\r' '\\n' | textutil -stdin -format #{ file_extension } -convert txt -stdout>
+  end
+
+  def file_extension
+    case
+      when plain_text? then 'txt'
+      when rich_text?  then 'rtf'
+    end
+  end
+
+  class EULASpecParser
+    attr_reader :dmg_filename
+
+    def initialize(dmg_filename)
+      @dmg_filename = dmg_filename
+    end
+
+    def parse
+      eula_spec = YAML::load(eula_spec_yaml) || []
+      eula_spec.map do |raw_spec|
+        odebug "EULA spec parsed from DMG: #{ raw_spec }"
+        language_id, language_name = raw_spec[:language]
+        Cask::Container::DmgEula.new(dmg_filename,
+          raw_spec[:format], language_id, language_name)
+      end
+    end
+
+    private
+
+    def eula_spec_yaml
+      `#{ eula_spec_yaml_command }`
+    end
+
+    def eula_spec_yaml_command
+      %Q<hdiutil udifderez -rez "#{ dmg_filename }" | iconv -f MAC -t UTF-8 | sed -n -E "s/^data ('TEXT'|'RTF ') \\((.*)\\) {/- :format: \\1\\\\\n  :language: [ \\2 ]/p" | awk 'BEGIN { print "---" } 1'>
+    end
+  end
+end

lib/cask/installer.rb

@@ -57,6 +57,7 @@ def install(force=false)
     begin
       satisfy_dependencies
       download
+      display_eula
       extract_primary_container
       install_artifacts
       save_caskfile force
@@ -86,8 +87,29 @@ def download
     @downloaded_path
   end
 
+  def display_eula
+    unless primary_container.eula?
+      odebug "No EULA found"
+      return
+    end
+    eula = primary_container.eulas.first
+    odebug "Selecting #{ eula }"
+    odebug "Mac system region code: #{ eula.mac_system_region_code }"
+    odebug "EULA source format: #{ eula.format }"
+
+    eula.show!(@cask.token)
+  end
+
   def extract_primary_container
     odebug "Extracting primary container"
+    primary_container.extract
+  end
+
+  def primary_container
+    @primary_container ||= load_primary_container
+  end
+
+  def load_primary_container
     FileUtils.mkdir_p @cask.staged_path
     container = if @cask.container and @cask.container.type
        Cask::Container.from_type(@cask.container.type)
@@ -98,7 +120,7 @@ def extract_primary_container
       raise CaskError.new "Uh oh, could not identify primary container for '#{@downloaded_path}'"
     end
     odebug "Using container class #{container} for #{@downloaded_path}"
-    container.new(@cask, @downloaded_path, @command).extract
+    container.new(@cask, @downloaded_path, @command)
   end
 
   def install_artifacts

lib/cask/utils.rb

@@ -89,6 +89,8 @@ def odebug title, *sput
 end
 
 module Cask::Utils
+  require 'cask/utils/locale'
+
   def dumpcask
     if Cask.respond_to?(:debug) and Cask.debug
       odebug "Cask instance dumps in YAML:"

lib/cask/utils/locale.rb

@@ -0,0 +1,145 @@
+require 'yaml'
+
+module Cask::Utils::Locale
+  GLOBAL_PREFERENCES_DOMAIN = '.GlobalPreferences'
+  APPLE_LOCALE_KEY = "AppleLocale"
+
+  RESOURCE_FORK_REGION_BASE_ID = 5000
+
+  # Sources:
+  #
+  # - /System/Library/Frameworks/CoreServices.framework/Frameworks/ \
+  #   CarbonCore.framework/Headers/Script.h
+  #
+  # - http://whitefiles.org/b1_s/1_free_guides/fg3mo/pgs/t02.htm
+  #
+  # - http://www.filibeto.org/unix/macos/lib/dev/documentation/mac/ \
+  #     pdf/Text/Script_Manager.pdf
+
+  OS_LOCALES = YAML::load(<<-EOF.undent).freeze
+    ---
+    0:   'en_US'      # United States
+    1:   'fr_FR'      # France
+    2:   'en_GB'      # Britain
+    3:   'de_DE'      # Germany
+    4:   'it_IT'      # Italy
+    5:   'nl_NL'      # Netherlands
+    6:   'nl_BE'      # Flemish (Dutch) for Belgium
+    7:   'sv_SE'      # Sweden
+    8:   'es_ES'      # Spanish for Spain
+    9:   'da_DK'      # Denmark
+    10:  'pt_PT'      # Portuguese for Portugal
+    11:  'fr_CA'      # French for Canada
+    12:  'nb_NO'      # Bokmål
+    13:  'he_IL'      # Hebrew
+    14:  'ja_JP'      # Japan
+    15:  'en_AU'      # English for Australia
+    16:  'ar'         # Arabic
+    17:  'fi_FI'      # Finland
+    18:  'fr_CH'      # French Swiss
+    19:  'de_CH'      # Swiss German
+    20:  'el_GR'      # German Swiss
+    21:  'is_IS'      # Iceland
+    22:  'mt_MT'      # Malta
+    23:  'el_CY'      # Cyprus
+    24:  'tr_TR'      # Turkey
+    25:  'hr_HR'      # Yugo/Croatian
+    33:  'hi_IN'      # India/Hindi
+    34:  'ur_PK'      # Pakistan/Urdu
+    35:  'tr_TR'      # Turkish (Modified)
+    36:  'it_CH'      # Italian Swiss
+    37:  'en-ascii'   # English for international use
+    39:  'ro_RO'      # Romania
+    40:  'grc'        # Ancient Greek
+    41:  'lt_LT'      # Lithuania
+    42:  'pl_PL'      # Poland
+    43:  'hu_HU'      # Hungary
+    44:  'et_EE'      # Estonia
+    45:  'lv_LV'      # Latvia
+    46:  'se'         # Lapland
+    47:  'fo_FO'      # Faroe Islands
+    48:  'fa_IR'      # Iran
+    49:  'ru_RU'      # Russia
+    50:  'ga_IE'      # Ireland
+    51:  'ko_KR'      # Korea
+    52:  'zh_CN'      # China
+    53:  'zh_TW'      # Taiwan
+    54:  'th_TH'      # Thailand
+    56:  'cs_CZ'      # Czech
+    57:  'sk_SK'      # Slovak
+    60:  'bn'         # Bangladesh or India
+    61:  'be_BY'      # Belarus
+    62:  'uk_UA'      # Ukraine
+    65:  'sr_CS'      # Serbian
+    66:  'sl_SI'      # Slovenian
+    67:  'mk_MK'      # Macedonian
+    68:  'hr_HR'      # Croatia
+    70:  'de-1996'    # German (reformed)
+    71:  'pt_BR'      # Portuguese for Brazil
+    72:  'bg_BG'      # Bulgaria
+    73:  'ca_ES'      # Catalonia
+    75:  'gd'         # Scottish Gaelic
+    76:  'gv'         # Isle of Man
+    77:  'br'         # Breton
+    78:  'iu_CA'      # Inuktitut for Canada
+    79:  'cy'         # Welsh
+    81:  'ga-Latg_IE' # Irish Gaelic for Ireland
+    82:  'en_CA'      # English for Canada
+    83:  'dz_BT'      # Dzongkha for Bhutan
+    84:  'hy_AM'      # Armenian
+    85:  'ka_GE'      # Georgian
+    86:  'es_XL'      # Spanish for Latin America
+    88:  'to_TO'      # Tonga
+    91:  'fr'         # French generic
+    92:  'de_AT'      # German for Austria
+    94:  'gu_IN'      # Gujarati
+    95:  'pa'         # Pakistan or India
+    96:  'ur_IN'      # Urdu for India
+    97:  'vi_VN'      # Vietnam
+    98:  'fr_BE'      # French for Belgium
+    99:  'uz_UZ'      # Uzbek
+    100: 'en_SG'      # Singapore
+    101: 'nn_NO'      # Norwegian Nynorsk
+    102: 'af_ZA'      # Afrikaans
+    103: 'eo'         # Esperanto
+    104: 'mr_IN'      # Marathi
+    105: 'bo'         # Tibetan
+    106: 'ne_NP'      # Nepal
+    107: 'kl'         # Greenland
+    108: 'en_IE'      # English for Ireland, with Euro for currency
+  EOF
+
+  Locale = Struct.new(:locale_string) do
+    def match_system_region?(region_code)
+      os_locale_string = OS_LOCALES[region_code]
+
+      if os_locale_string
+        os_locale_string === locale_string.to_s
+      else
+        false
+      end
+    end
+
+    def to_s
+      locale_string
+    end
+  end
+
+  def self.locale
+    @@locale ||= load_locale
+  end
+
+  def self.resource_fork_region_base_id
+    RESOURCE_FORK_REGION_BASE_ID
+  end
+
+  def self.load_locale
+    options = {
+      :args => ['read', GLOBAL_PREFERENCES_DOMAIN, APPLE_LOCALE_KEY]
+    }
+    locale_string = Cask::SystemCommand.run('/usr/bin/defaults', options)
+    locale = Locale.new(locale_string.to_s.chomp)
+    odebug "Current OS locale is: #{ locale }"
+    locale
+  end
+end