Detect language from shebang script

lib/linguist/blob.rb

@@ -1,3 +1,4 @@
+require 'linguist/language'
 require 'linguist/mime'
 require 'linguist/pathname'
 
@@ -77,5 +78,65 @@ module Linguist
     def viewable?
       !file? && !large?
     end
+
+    def language
+      if text?
+        shebang_language || name.language
+      else
+        Language['Text']
+      end
+    end
+
+    def lexer
+      language.lexer
+    end
+
+    def shebang_script
+      return if !text? || large?
+
+      if (match = data.match(/(.+)\n?/)) && (bang = match[0]) =~ /^#!/
+        bang.sub!(/^#! /, '#!')
+        tokens = bang.split(' ')
+        pieces = tokens.first.split('/')
+        if pieces.size > 1
+          script = pieces.last
+        else
+          script = pieces.first.sub('#!', '')
+        end
+
+        script = script == 'env' ? tokens[1] : script
+
+        # python2.4 => python
+        if script =~ /((?:\d+\.?)+)/
+          script.sub! $1, ''
+        end
+
+        script
+      end
+    end
+
+    def shebang_language
+      if script = shebang_script
+        case script
+        when 'bash'
+          Language['Shell']
+        when 'groovy'
+          Language['Java']
+        when 'macruby'
+          Language['Ruby']
+        when 'node'
+          Language['JavaScript']
+        when 'rake'
+          Language['Ruby']
+        when 'sh'
+          Language['Shell']
+        when 'zsh'
+          Language['Shell']
+        else
+          lang = Language.find_by_lexer(shebang_script)
+          lang != Language['Text'] ? lang : nil
+        end
+      end
+    end
   end
 end

test/fixtures/blob/Capfile

@@ -0,0 +1,3 @@
+load 'deploy'
+Dir['vendor/plugins/*/recipes/*.rb'].each { |plugin| load(plugin) }
+load 'config/deploy'

test/fixtures/blob/README

@@ -0,0 +1,2 @@
+README
+======

test/fixtures/blob/Rakefile

@@ -0,0 +1,3 @@
+task :default do
+  puts "Rake"
+end

test/fixtures/blob/dude-thing-okay--001.patch

@@ -0,0 +1,4 @@
+diff --git a/lib/linguist.rb b/lib/linguist.rb
+index d472341..8ad9ffb 100644
+--- a/lib/linguist.rb
++++ b/lib/linguist.rb

test/fixtures/blob/dude.el

@@ -0,0 +1 @@
+(print "Dude!")

test/fixtures/blob/dude.js

@@ -0,0 +1 @@
+alert("dude!")

test/fixtures/blob/grit.rb

@@ -0,0 +1,2 @@
+module Grit
+end

test/fixtures/blob/script.bash

@@ -0,0 +1,2 @@
+#!/bin/bash
+echo "bash"

test/fixtures/blob/script.foo

@@ -0,0 +1,2 @@
+#!/bin/foo
+???

test/fixtures/blob/script.groovy

@@ -0,0 +1,2 @@
+#!/usr/bin/env groovy
+println "Groovy!"

test/fixtures/blob/script.js

@@ -0,0 +1,2 @@
+#!/usr/bin/env node
+console.log("Node")

test/fixtures/blob/script.mrb

@@ -0,0 +1,2 @@
+#!/usr/bin/env macruby
+puts "MacRuby"

test/fixtures/blob/script.pl

@@ -0,0 +1,2 @@
+#!/usr/local/bin/perl
+print "Perl\n"

test/fixtures/blob/script.py

@@ -0,0 +1,2 @@
+#!/usr/bin/env python2.4
+print "Python"

test/fixtures/blob/script.rake

@@ -0,0 +1,4 @@
+#!/usr/bin/env rake
+task :default do
+  puts "Rake"
+end

test/fixtures/blob/script.rb

@@ -0,0 +1,2 @@
+#!/usr/bin/env ruby
+puts "Ruby"

test/fixtures/blob/script.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+echo "sh"

test/fixtures/blob/script.zsh

@@ -0,0 +1,2 @@
+#!/bin/zsh
+echo "zsh"

test/fixtures/blob/script2.rb

@@ -0,0 +1,2 @@
+#! /usr/bin/env ruby -w -Ilib:test
+echo "Ruby"

test/fixtures/blob/subdir/Rakefile

@@ -0,0 +1,3 @@
+task :default do
+  puts "Rake (subdir)"
+end

test/test_blob.rb

@@ -92,4 +92,53 @@ class TestBlob < Test::Unit::TestCase
     assert blob("octocat.gif").image?
     assert !blob("octocat.psd").image?
   end
+
+  def test_language
+    assert_equal Language['Ruby'], blob("foo.rb").language
+    assert_equal Language['Ruby'], blob("script.rb").language
+    assert_equal Language['Text'], blob("octocat.png").language
+  end
+
+  def test_lexer
+    assert_equal 'ruby', blob("grit.rb").lexer
+    assert_equal 'text', blob("README").lexer
+    assert_equal 'diff', blob("dude-thing-okay--001.patch").lexer
+    assert_equal 'scheme', blob("dude.el").lexer
+    assert_equal 'javascript', blob("dude.js").lexer
+    assert_equal 'ruby', blob("Capfile").lexer
+
+    assert_equal 'ruby', blob("Rakefile").lexer
+    assert_equal 'ruby', blob("subdir/Rakefile").lexer
+  end
+
+  def test_shebang_script
+    assert_equal 'sh', blob("script.sh").shebang_script
+    assert_equal 'bash', blob("script.bash").shebang_script
+    assert_equal 'zsh', blob("script.zsh").shebang_script
+    assert_equal 'perl', blob("script.pl").shebang_script
+    assert_equal 'ruby', blob("script.rb").shebang_script
+    assert_equal 'ruby', blob("script2.rb").shebang_script
+    assert_equal 'python', blob("script.py").shebang_script
+    assert_equal 'node', blob("script.js").shebang_script
+    assert_equal 'groovy', blob("script.groovy").shebang_script
+    assert_equal 'macruby', blob("script.mrb").shebang_script
+    assert_equal 'rake', blob("script.rake").shebang_script
+    assert_equal 'foo', blob("script.foo").shebang_script
+    assert_equal nil, blob("foo.rb").shebang_script
+  end
+
+  def test_shebang_language
+    assert_equal Language['Shell'], blob("script.sh").shebang_language
+    assert_equal Language['Shell'], blob("script.bash").shebang_language
+    assert_equal Language['Shell'], blob("script.zsh").shebang_language
+    assert_equal Language['Perl'], blob("script.pl").shebang_language
+    assert_equal Language['Ruby'], blob("script.rb").shebang_language
+    assert_equal Language['Python'], blob("script.py").shebang_language
+    assert_equal Language['JavaScript'], blob("script.js").shebang_language
+    assert_equal Language['Java'], blob("script.groovy").shebang_language
+    assert_equal Language['Ruby'], blob("script.mrb").shebang_language
+    assert_equal Language['Ruby'], blob("script.rake").shebang_language
+    assert_equal nil, blob("script.foo").shebang_language
+    assert_equal nil, blob("foo.rb").shebang_language
+  end
 end