add matching of string and better handling of binary encoding

lib/mimemagic.rb

 require 'mimemagic/tables'
 require 'mimemagic/version'
-require 'stringio'
 
 # Mime type detection
 class MimeMagic
@@ -77,11 +76,15 @@ class MimeMagic
   # Lookup mime type by magic content analysis.
   # This is a slow operation.
   def self.by_magic(io)
-    if !(io.respond_to?(:seek) && io.respond_to?(:read))
-      str = io.respond_to?(:read) ? io.read : io.to_s
-      io = StringIO.new(str, 'rb:binary')
-    end
-    mime = MAGIC.find {|type, matches| magic_match(io, matches) }
+    mime =
+      unless io.respond_to?(:seek) && io.respond_to?(:read)
+        str = io.respond_to?(:read) ? io.read : io.to_s
+        str = str.b if str.respond_to? :b
+        MAGIC.find {|type, matches| magic_match_str(str, matches) }
+      else
+        io.binmode
+        MAGIC.find {|type, matches| magic_match_io(io, matches) }
+      end
     mime && new(mime[0])
   end
 
@@ -101,31 +104,39 @@ class MimeMagic
 
   alias == eql?
 
-  private
-
   def self.child?(child, parent)
     child == parent || TYPES.key?(child) && TYPES[child][1].any? {|p| child?(p, parent) }
   end
 
-  def self.magic_match(io, matches)
+  def self.magic_match_io(io, matches)
+    matches.any? do |offset, value, children|
+      match =
+        if Range === offset
+          io.seek(offset.begin)
+          io.read(offset.end - offset.begin + value.length).include?(value)
+        else
+          io.seek(offset)
+          io.read(value.length) == value
+        end
+      match && (!children || magic_match_io(io, children))
+    end
+  rescue
+    false
+  end
+
+  def self.magic_match_str(str, matches)
     matches.any? do |offset, value, children|
-      value = value.b if value.respond_to?(:b)
-      match = if Range === offset
-                io.seek(offset.begin)
-                io_val = io.read(offset.end - offset.begin + value.length)
-                io_val = io_val.b if io_val.respond_to?(:b)
-                io_val.include?(value)
-              else
-                io.seek(offset)
-                io_val = io.read(value.length)
-                io_val = io_val.b if io_val.respond_to?(:b)
-                value == io_val
-              end
-      match && (!children || magic_match(io, children))
+      match =
+        if Range === offset
+          str[offset.begin, offset.end - offset.begin + value.length].include?(value)
+        else
+          str[offset, value.length] == value
+        end
+      match && (!children || magic_match_str(str, children))
     end
   rescue
     false
   end
 
-  private_class_method :magic_match
+  private_class_method :magic_match_io, :magic_match_str
 end

lib/mimemagic/tables.rb

+# -*- coding: binary -*-
 # Generated from freedesktop.org.xml
 class MimeMagic
-  private
+  # @private
+  # :nodoc:
   EXTENSIONS = {
     '123' => 'application/vnd.lotus-1-2-3',
     '3ds' => 'image/x-3ds',
@@ -808,6 +810,8 @@ class MimeMagic
     'zip' => 'application/zip',
     'zoo' => 'application/x-zoo',
   }
+  # @private
+  # :nodoc:
   TYPES = {
     'application/andrew-inset' => [%w(ez), %w(), 'ATK inset'],
     'application/annodex' => [%w(anx), %w(), 'Annodex exchange format'],
@@ -1386,6 +1390,8 @@ class MimeMagic
     'video/x-theora+ogg' => [%w(ogg ogv), %w(video/ogg), 'Ogg Theora video'],
     'x-epoc/x-sisx-app' => [%w(sisx), %w(), 'SISX package'],
   }
+  # @private
+  # :nodoc:
   MAGIC = [
     ['application/annodex', [[0, 'OggS', [[28, "fishead\000", [[56..512, "CMML\000\000\000\000"]]]]]]],
     ['application/atom+xml', [[0..256, '<feed ']]],

script/generate-mime.rb

@@ -90,14 +90,18 @@ end
 
 magics = magics.sort {|a,b| b[0] <=> a[0] }.map {|x| [x[1], x[2]] }
 
+puts "# -*- coding: binary -*-"
 puts "# Generated from #{FILE}"
 puts "class MimeMagic"
-puts "  private"
+puts "  # @private"
+puts "  # :nodoc:"
 puts "  EXTENSIONS = {"
 extensions.keys.sort.each do |key|
   puts "    '#{key}' => '#{extensions[key]}',"
 end
 puts "  }"
+puts "  # @private"
+puts "  # :nodoc:"
 puts "  TYPES = {"
 types.keys.sort.each do |key|
   exts = types[key][0].sort.join(' ')
@@ -106,6 +110,8 @@ types.keys.sort.each do |key|
   puts "    '#{key}' => [%w(#{exts}), %w(#{parents}), #{comment}],"
 end
 puts "  }"
+puts "  # @private"
+puts "  # :nodoc:"
 puts "  MAGIC = ["
 magics.sort.each do |type, matches|
   puts "    ['#{type}', #{matches.inspect}],"

test/mimemagic_test.rb

 require 'bacon'
 require 'mimemagic'
+require 'stringio'
 
 describe 'MimeMagic' do
   it 'should have type, mediatype and subtype' do
@@ -73,14 +74,30 @@ describe 'MimeMagic' do
     MimeMagic.add('application/mimemagic-test',
                   :magic => [[0, 'MAGICTEST'], # MAGICTEST at position 0
                              [1, 'MAGICTEST'], # MAGICTEST at position 1
+                             [9..12, 'MAGICTEST'], # MAGICTEST starting at position 9 to 12
                              [2, 'MAGICTEST', [[0, 'X'], [0, 'Y']]]]) # MAGICTEST at position 2 and (X at 0 or Y at 0)
 
     MimeMagic.by_magic('MAGICTEST').should.equal 'application/mimemagic-test'
     MimeMagic.by_magic('XMAGICTEST').should.equal 'application/mimemagic-test'
     MimeMagic.by_magic(' MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic('123456789MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic('123456789ABMAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic('123456789ABCMAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic('123456789ABCDMAGICTEST').should.equal nil
     MimeMagic.by_magic('X MAGICTEST').should.equal 'application/mimemagic-test'
     MimeMagic.by_magic('Y MAGICTEST').should.equal 'application/mimemagic-test'
     MimeMagic.by_magic('Z MAGICTEST').should.equal nil
+
+    MimeMagic.by_magic(StringIO.new 'MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new 'XMAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new ' MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new '123456789MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new '123456789ABMAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new '123456789ABCMAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new '123456789ABCDMAGICTEST').should.equal nil
+    MimeMagic.by_magic(StringIO.new 'X MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new 'Y MAGICTEST').should.equal 'application/mimemagic-test'
+    MimeMagic.by_magic(StringIO.new 'Z MAGICTEST').should.equal nil
   end
 
   it 'should handle different file objects' do