cleanup CodeRay.scan* methods

6 files changed, 45 insertions, 29 deletions

diff --git a/etc/CodeRay.tmproj b/etc/CodeRay.tmproj
index 1886802..285255b 100644
--- a/etc/CodeRay.tmproj
+++ b/etc/CodeRay.tmproj
@@ -5,6 +5,8 @@
 	<key>documents</key>
 	<array>
 		<dict>
+			<key>expanded</key>
+			<true/>
 			<key>name</key>
 			<string>lib</string>
 			<key>regexFolderFilter</key>
@@ -24,27 +26,25 @@
 			<key>filename</key>
 			<string>../coderay.gemspec</string>
 			<key>lastUsed</key>
-			<date>2011-08-19T02:13:23Z</date>
+			<date>2011-08-19T02:43:54Z</date>
 		</dict>
 		<dict>
 			<key>filename</key>
 			<string>../Changes-1.0.textile</string>
 			<key>lastUsed</key>
-			<date>2011-08-19T01:12:40Z</date>
+			<date>2011-08-19T03:20:53Z</date>
 		</dict>
 		<dict>
 			<key>filename</key>
 			<string>../README_INDEX.rdoc</string>
 			<key>lastUsed</key>
-			<date>2011-08-19T02:16:06Z</date>
+			<date>2011-08-19T02:40:24Z</date>
 		</dict>
 		<dict>
 			<key>filename</key>
 			<string>../README.textile</string>
 			<key>lastUsed</key>
-			<date>2011-08-19T02:29:46Z</date>
-			<key>selected</key>
-			<true/>
+			<date>2011-08-19T02:38:26Z</date>
 		</dict>
 		<dict>
 			<key>filename</key>
@@ -145,7 +145,7 @@
 					<key>filename</key>
 					<string>../test/scanners/coderay_suite.rb</string>
 					<key>lastUsed</key>
-					<date>2011-08-19T00:50:30Z</date>
+					<date>2011-08-19T03:16:08Z</date>
 				</dict>
 				<dict>
 					<key>filename</key>
@@ -179,6 +179,6 @@
 	<key>showFileHierarchyDrawer</key>
 	<true/>
 	<key>windowFrame</key>
-	<string>{{214, 4}, {1066, 774}}</string>
+	<string>{{0, 4}, {1066, 774}}</string>
 </dict>
 </plist>
diff --git a/lib/coderay.rb b/lib/coderay.rb
index 9779ff5..2ae58d8 100644
--- a/lib/coderay.rb
+++ b/lib/coderay.rb
@@ -159,8 +159,7 @@ module CodeRay
     # See also demo/demo_simple.
     def scan code, lang, options = {}, &block
       # FIXME: return a proxy for direct-stream encoding
-      scanner = Scanners[lang].new code, options, &block
-      scanner.tokenize
+      scanner(lang, options, &block).tokenize code
     end
     
     # Scans +filename+ (a path to a code file) with the Scanner for +lang+.
@@ -175,11 +174,9 @@ module CodeRay
     #  require 'coderay'
     #  page = CodeRay.scan_file('some_c_code.c').html
     def scan_file filename, lang = :auto, options = {}, &block
-      file = IO.read filename
-      if lang == :auto
-        lang = FileType.fetch filename, :text, true
-      end
-      scan file, lang, options = {}, &block
+      lang = FileType.fetch filename, :text, true if lang == :auto
+      code = File.read filename
+      scan code, lang, options = {}, &block
     end
     
     # Encode a string.
@@ -261,8 +258,8 @@ module CodeRay
     # +options+ to it.
     #
     # See Scanner.new.
-    def scanner lang, options = {}
-      Scanners[lang].new '', options
+    def scanner lang, options = {}, &block
+      Scanners[lang].new '', options, &block
     end
     
     # Extract the options for the scanner from the +options+ hash.
diff --git a/lib/coderay/encoders/_map.rb b/lib/coderay/encoders/_map.rb
index e5cbdeb..4cca196 100644
--- a/lib/coderay/encoders/_map.rb
+++ b/lib/coderay/encoders/_map.rb
@@ -8,7 +8,8 @@ module Encoders
     :remove_comments => :comment_filter,
     :stats           => :statistic,
     :term            => :terminal,
-    :tty             => :terminal
+    :tty             => :terminal,
+    :yml             => :yaml
   
   # No default because Tokens#nonsense should raise NoMethodError.
   
diff --git a/lib/coderay/scanner.rb b/lib/coderay/scanner.rb
index ec89b87..27b6153 100644
--- a/lib/coderay/scanner.rb
+++ b/lib/coderay/scanner.rb
@@ -69,6 +69,8 @@ module CodeRay
         def normalize code
           # original = code
           code = code.to_s unless code.is_a? ::String
+          return code if code.empty?
+          
           if code.respond_to? :encoding
             code = encode_with_encoding code, self.encoding
           else
@@ -183,14 +185,12 @@ module CodeRay
         @tokens = options[:tokens] || @tokens || Tokens.new
         @tokens.scanner = self if @tokens.respond_to? :scanner=
         case source
-        when String
-          self.string = source
         when Array
-          self.string = source.join
+          self.string = self.class.normalize(source.join)
         when nil
           reset
         else
-          raise ArgumentError, 'expected String, Array, or nil'
+          self.string = self.class.normalize(source)
         end
         
         begin
diff --git a/test/functional/basic.rb b/test/functional/basic.rb
index 1702e4e..2654359 100755
--- a/test/functional/basic.rb
+++ b/test/functional/basic.rb
@@ -213,9 +213,7 @@ more code  # and another comment, in-line.
   def test_scanner_tokenize
     assert_equal ['foo', :plain], CodeRay::Scanners::Plain.new.tokenize('foo')
     assert_equal [['foo', :plain], ['bar', :plain]], CodeRay::Scanners::Plain.new.tokenize(['foo', 'bar'])
-    assert_raise ArgumentError do
-      CodeRay::Scanners::Plain.new.tokenize 42
-    end
+    CodeRay::Scanners::Plain.new.tokenize 42
   end
   
   def test_scanner_tokens
diff --git a/test/unit/duo.rb b/test/unit/duo.rb
index e7f3f91..62bd57b 100644
--- a/test/unit/duo.rb
+++ b/test/unit/duo.rb
@@ -16,10 +16,30 @@ class DuoTest < Test::Unit::TestCase
   end
   
   def test_call
-    duo = CodeRay::Duo[:python => :xml]
-    assert_equal <<-'XML'.chomp, duo.call('def test: "pass"')
-<?xml version='1.0'?><coderay-tokens><keyword>def</keyword> <method>test</method><operator>:</operator> <string><delimiter>&quot;</delimiter><content>pass</content><delimiter>&quot;</delimiter></string></coderay-tokens>
-    XML
+    duo = CodeRay::Duo[:python => :yml]
+    assert_equal <<-'YAML', duo.call('def test: "pass"')
+--- 
+- - def
+  - :keyword
+- - " "
+  - :space
+- - test
+  - :method
+- - ":"
+  - :operator
+- - " "
+  - :space
+- - :begin_group
+  - :string
+- - "\""
+  - :delimiter
+- - pass
+  - :content
+- - "\""
+  - :delimiter
+- - :end_group
+  - :string
+    YAML
   end
   
 end