initial commit

Manifest.txt

+lib/mimemagic.rb
+lib/mimemagic_tables.rb
+test/test_mimemagic.rb
+script/freedesktop.org.xml
+script/generate-mime.rb
+Rakefile
+README

README

+MimeMagic is a library to detect the mime type of a file by extension or by content.
+
+Usage
+=====
+
+   require 'mimemagic'
+   MimeMagic.by_extension('html').text?
+   MimeMagic.by_extension('.html').child_of? 'text/plain'
+   MimeMagic.by_magic(File.open('test.html'))
+   etc...
+
+Authors
+=======
+
+Daniel Mendler
\ No newline at end of file

Rakefile

+require 'hoe'
+
+$:.unshift 'lib'
+require 'mimemagic'
+
+Hoe.new 'mimemagic', MimeMagic::VERSION do |mimemagic|
+  mimemagic.developer 'Daniel Mendler', 'mail@daniel-mendler.de'
+  mimemagic.summary = 'Mime detection by extension or content'
+end
+

lib/mimemagic.rb

+require 'mimemagic_tables'
+require 'stringio'
+
+# Mime type detection
+class MimeMagic
+  VERSION = '0.1'
+
+  attr_reader :type, :mediatype, :subtype
+
+  # Mime type by type string
+  def initialize(type)
+    @type      = type
+    @mediatype = @type.split('/')[0]
+    @subtype   = @type.split('/')[1]
+  end
+
+  # Add custom mime type. You have to
+  # specify the type, a string list of file extensions,
+  # a string list of parent mime types and an optional
+  # detector block for magic detection.
+  def self.add(type, extensions, parents, *magics)
+    TYPES[type] = [extensions, parents, block_given? ? proc(&block) : nil]
+    extensions.each do |ext|
+      EXTENSIONS[ext] = type
+    end
+    MAGIC.unshift [type, magics] if magics
+  end
+
+  # Returns true if type is a text format
+  def text?
+    child_of? 'text/plain'
+  end
+
+  # Returns true if type is child of parent type
+  def child_of?(parent)
+    child?(type, parent)
+  end
+
+  # Get string list of file extensions
+  def extensions
+    TYPES.key?(type) ? TYPES[type][0] : []
+  end
+
+  # Lookup mime type by file extension
+  def self.by_extension(ext)
+    ext = ext.downcase
+    mime = EXTENSIONS[ext] || (ext[0..0] == '.' && EXTENSIONS[ext[1..-1]])
+    mime ? new(mime) : nil
+  end
+
+  # Lookup mime type by magic content analysis
+  # That could be slow
+  def self.by_magic(content)
+    io = content.respond_to?(:seek) ? content : StringIO.new(content.to_s, 'rb')
+    mime = MAGIC.find {|type, matches| magic_match(io, matches) }
+    mime ? new(mime[0]) : nil
+  end
+
+  # Return type as string
+  def to_s
+    type
+  end
+
+  # Allow comparison with string
+  def ==(x)
+    type == x.to_s
+  end
+
+  private
+
+  def child?(child, parent)
+    return true if child == parent
+    TYPES.key?(child) ? TYPES[child][1].any? {|p| child?(p, parent) } : false
+  end
+
+  def self.magic_match(io, matches)
+    matches.any? do |offset, value, children|
+      if Range === offset
+        io.seek(offset.begin)
+        match = io.read(offset.end - offset.begin + value.length).include?(value)
+      else
+        io.seek(offset)
+        match = value == io.read(value.length)
+      end
+      match && (!children || magic_match(io, children))
+    end
+  rescue
+    false
+  end
+end

mimemagic.gemspec

+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{mimemagic}
+  s.version = "0.1"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Daniel Mendler"]
+  s.date = %q{2009-05-09}
+  s.email = ["mail@daniel-mendler.de"]
+  s.files = ["lib/mimemagic.rb", "lib/mimemagic_tables.rb", "test/test_mimemagic.rb", "script/freedesktop.org.xml", "script/generate-mime.rb", "Rakefile"]
+  s.has_rdoc = true
+  s.rdoc_options = ["--main", "README.txt"]
+  s.require_paths = ["lib"]
+  s.rubyforge_project = %q{mimemagic}
+  s.rubygems_version = %q{1.3.1}
+  s.summary = %q{Mime detection by extension or content}
+  s.test_files = ["test/test_mimemagic.rb"]
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 2
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+      s.add_development_dependency(%q<hoe>, [">= 1.8.3"])
+    else
+      s.add_dependency(%q<hoe>, [">= 1.8.3"])
+    end
+  else
+    s.add_dependency(%q<hoe>, [">= 1.8.3"])
+  end
+end
+

script/generate-mime.rb

+#!/usr/bin/ruby
+
+require 'rexml/document'
+
+def str2int(s)
+  return s.to_i(16) if s[0..1].downcase == '0x'
+  return s.to_i(8) if s[0..0].downcase == '0'
+  s.to_i(10)
+end
+
+def get_matches(parent)
+  parent.get_elements('match').map {|match|
+    if match.attributes['mask']
+      nil
+    else
+      type = match.attributes['type']
+      value = match.attributes['value']
+      offset = match.attributes['offset'].split(':').map {|x| x.to_i }
+      offset = offset.size == 2 ? offset[0]..offset[1] : offset[0]
+      case type
+      when 'string'
+        value.gsub!(/\\(x[\dA-Fa-f]{1,2}|0\d{1,3}|\d{1,3}|.)/) { eval("\"\\#{$1}\"") }
+      when 'big16'
+        value = str2int(value)
+        value = ((value >> 8).chr + (value & 0xFF).chr)
+      when 'big32'
+        value = str2int(value)
+        value = (((value >> 24) & 0xFF).chr + ((value >> 16) & 0xFF).chr + ((value >> 8) & 0xFF).chr + (value & 0xFF).chr)
+      when 'little16'
+        value = str2int(value)
+        value = ((value & 0xFF).chr + (value >> 8).chr)
+      when 'little32'
+        value = str2int(value)
+        value = ((value & 0xFF).chr + ((value >> 8) & 0xFF).chr + ((value >> 16) & 0xFF).chr + ((value >> 24) & 0xFF).chr)
+      when 'host16' # use little endian
+        value = str2int(value)
+        value = ((value & 0xFF).chr + (value >> 8).chr)
+      when 'host32' # use little endian
+        value = str2int(value)
+        value = ((value & 0xFF).chr + ((value >> 8) & 0xFF).chr + ((value >> 16) & 0xFF).chr + ((value >> 24) & 0xFF).chr)
+      when 'byte'
+        value = str2int(value)
+        value = value.chr
+      end
+      children = get_matches(match)
+      children.empty? ? [offset, value] : [offset, value, children]
+    end
+  }.compact
+end
+
+FILE = ARGV[0] || '/usr/share/mime/packages/freedesktop.org.xml'
+file = File.new(FILE)
+doc = REXML::Document.new(file)
+extensions = {}
+types = {}
+magics = []
+doc.each_element('mime-info/mime-type') do |mime|
+  type = mime.attributes['type']
+  subclass = mime.get_elements('sub-class-of').map{|x| x.attributes['type']}
+  exts = mime.get_elements('glob').map{|x| x.attributes['pattern'] =~ /^\*\.([^\[\]]+)$/ ? $1.downcase : nil }.compact
+  mime.get_elements('magic').each do |magic|
+    priority = magic.attributes['priority'].to_i
+    matches = get_matches(magic)
+    magics << [priority, type, matches]
+  end
+  if !exts.empty?
+    exts.each{|x|
+      extensions[x] = type if !extensions.include?(x)
+    }
+    types[type] = [exts,subclass]
+  end
+end
+
+magics = magics.sort {|a,b| b[0] <=> a[0] }.map {|x| [x[1], x[2]] }
+
+puts "# Generated from #{FILE}"
+puts "class MimeMagic"
+puts "  private"
+puts "  EXTENSIONS = {"
+extensions.keys.sort.each do |key|
+  puts "    '#{key}' => '#{extensions[key]}',"
+end
+puts "  }"
+puts "  TYPES = {"
+types.keys.sort.each do |key|
+  exts = types[key][0].sort.inspect
+  parents = types[key][1].sort.inspect
+  puts "    '#{key}' => [#{exts}, #{parents}],"
+end
+puts "  }"
+puts "  MAGIC = ["
+magics.each do |type, matches|
+  puts "    ['#{type}', #{matches.inspect}],"
+end
+puts "  ]"
+puts "end"
+

test/test_mimemagic.rb

+require 'mimemagic'
+
+class TC_MimeMagic < Test::Unit::TestCase
+  def test_text?
+    assert MimeMagic.new('text/plain').text?
+    assert MimeMagic.new('text/html').text?
+    assert !MimeMagic.new('application/octet-stream').text?
+    assert !MimeMagic.new('image/png').text?
+  end
+
+  def test_child_of?
+    assert MimeMagic.new('text/html').child_of?('text/plain')
+    assert MimeMagic.new('text/x-java').child_of?('text/plain')
+  end
+
+  def test_extensions
+    assert_equal %w(htm html), MimeMagic.new('text/html').extensions
+  end
+
+  def test_by_extension
+    assert_equal 'text/html', MimeMagic.by_extension('html').to_s
+    assert_equal 'application/x-ruby', MimeMagic.by_extension('rb').to_s
+    assert_nil MimeMagic.by_extension('crazy')
+    assert_nil MimeMagic.by_extension('')
+  end
+
+  def test_by_magic
+    assert_equal 'application/x-executable', MimeMagic.by_magic(File.open('/bin/ls')).to_s
+    assert_equal 'application/x-sharedlib', MimeMagic.by_magic(File.open('/lib/libc.so.6')).to_s
+  end
+end