Better language detection

Dmitriy Zaporozhets
1 parent b73a33ed
Showing 1 changed file with 9 additions and 29 deletions Show diff stats
lib/utils.rb
@@ -22,40 +22,20 @@ module Utils
     end
     def system_colorize(data, file_name)
+      options = { :encoding => 'utf-8', :linenos => 'True' }
+
+      # Try detect language with pygments
+      Pygments.highlight data, :filename => file_name, :options => options
+    rescue 
+      # if it fails use manual detection
       ft = handle_file_type(file_name)
-      Pygments.highlight(data, :lexer => ft, :options => { :encoding => 'utf-8', :linenos => 'True' })
+      Pygments.highlight(data, :lexer => ft, :options => options)
     end
-    def handle_file_type(file_name, mime_type = nil)
+    def handle_file_type(file_name)
       case file_name
-      when /(\.pl|\.scala|\.java|\.haml|\.jade|\.scaml|\.html|\.sass|\.scss|\.php|\.erb)$/
-        $1[1..-1].to_sym
-      when /(\.c|\.h|\.idc)$/
-        :c
-      when /(\.cpp|\.hpp|\.c++|\.h++|\.cc|\.hh|\.cxx|\.hxx)$/
-        :cpp
-      when /(\.d|\.di)$/
-        :d
-      when /(\.hs|\.lhs)$/
-        :haskell
-      when /(\.rb|\.ru|\.rake|Rakefile|\.gemspec|\.rbx|Gemfile)$/
+      when /(\.ru|Gemfile)$/
         :ruby
-      when /(\.py|\.pyw|\.sc|SConstruct|SConscript|\.tac)$/
-        :python
-      when /(\.js|\.json)$/
-        :javascript
-      when /(\.xml|\.xsl|\.rss|\.xslt|\.xsd|\.wsdl)$/
-        :xml
-      when /(\.vm|\.fhtml)$/
-        :velocity
-      when /\.sh$/
-        :bash
-      when /\.coffee$/
-        :coffeescript
-      when /(\.yml|\.yaml)$/
-        :yaml
-      when /\.md$/
-        :minid
       else
         :text
       end
	@@ -22,40 +22,20 @@ module Utils		@@ -22,40 +22,20 @@ module Utils
22	end	22	end
23		23
24	def system_colorize(data, file_name)	24	def system_colorize(data, file_name)
		25	+ options = { :encoding => 'utf-8', :linenos => 'True' }
		26	+
		27	+ # Try detect language with pygments
		28	+ Pygments.highlight data, :filename => file_name, :options => options
		29	+ rescue
		30	+ # if it fails use manual detection
25	ft = handle_file_type(file_name)	31	ft = handle_file_type(file_name)
26	- Pygments.highlight(data, :lexer => ft, :options => { :encoding => 'utf-8', :linenos => 'True' })	32	+ Pygments.highlight(data, :lexer => ft, :options => options)
27	end	33	end
28		34
29	- def handle_file_type(file_name, mime_type = nil)	35	+ def handle_file_type(file_name)
30	case file_name	36	case file_name
31	- when /(\.pl\|\.scala\|\.java\|\.haml\|\.jade\|\.scaml\|\.html\|\.sass\|\.scss\|\.php\|\.erb)$/
32	- $1[1..-1].to_sym
33	- when /(\.c\|\.h\|\.idc)$/
34	- :c
35	- when /(\.cpp\|\.hpp\|\.c++\|\.h++\|\.cc\|\.hh\|\.cxx\|\.hxx)$/
36	- :cpp
37	- when /(\.d\|\.di)$/
38	- :d
39	- when /(\.hs\|\.lhs)$/
40	- :haskell
41	- when /(\.rb\|\.ru\|\.rake\|Rakefile\|\.gemspec\|\.rbx\|Gemfile)$/	37	+ when /(\.ru\|Gemfile)$/
42	:ruby	38	:ruby
43	- when /(\.py\|\.pyw\|\.sc\|SConstruct\|SConscript\|\.tac)$/
44	- :python
45	- when /(\.js\|\.json)$/
46	- :javascript
47	- when /(\.xml\|\.xsl\|\.rss\|\.xslt\|\.xsd\|\.wsdl)$/
48	- :xml
49	- when /(\.vm\|\.fhtml)$/
50	- :velocity
51	- when /\.sh$/
52	- :bash
53	- when /\.coffee$/
54	- :coffeescript
55	- when /(\.yml\|\.yaml)$/
56	- :yaml
57	- when /\.md$/
58	- :minid
59	else	39	else
60	:text	40	:text
61	end	41	end