Merge pull request #1750 from github/interpreters-in-samples

Fix for interpreters from samples

lib/linguist/languages.yml

@@ -559,6 +559,8 @@ Crystal:
   - .cr
   ace_mode: ruby
   tm_scope: source.ruby
+  interpreters:
+  - crystal
 
 Cucumber:
   extensions:
@@ -736,6 +738,8 @@ Erlang:
   - .es
   - .escript
   - .hrl
+  interpreters:
+  - escript
 
 F#:
   type: programming
@@ -931,6 +935,8 @@ Gnuplot:
   - .gnuplot
   - .plot
   - .plt
+  interpreters:
+  - gnuplot
 
 Go:
   type: programming
@@ -1196,6 +1202,8 @@ Ioke:
   color: "#078193"
   extensions:
   - .ik
+  interpreters:
+  - ioke
 
 Isabelle:
   type: programming
@@ -1703,6 +1711,8 @@ Nu:
   filenames:
   - Nukefile
   tm_scope: source.scheme
+  interpreters:
+  - nush
 
 NumPy:
   group: Python
@@ -1889,6 +1899,8 @@ Parrot Assembly:
   - pasm
   extensions:
   - .pasm
+  interpreters:
+  - parrot
   tm_scope: none
 
 Parrot Internal Representation:
@@ -1899,6 +1911,8 @@ Parrot Internal Representation:
   - pir
   extensions:
   - .pir
+  interpreters:
+  - parrot
 
 Pascal:
   type: programming
@@ -1941,6 +1955,8 @@ Perl6:
   - .p6m
   - .pl6
   - .pm6
+  interpreters:
+  - perl6
   tm_scope: none
 
 PigLatin:
@@ -2005,6 +2021,8 @@ Prolog:
   - .ecl
   - .pro
   - .prolog
+  interpreters:
+  - swipl
 
 Propeller Spin:
   type: programming
@@ -2068,6 +2086,8 @@ Python:
   - wscript
   interpreters:
   - python
+  - python2
+  - python3
 
 Python traceback:
   type: data
@@ -2088,6 +2108,8 @@ QMake:
   extensions:
   - .pro
   - .pri
+  interpreters:
+  - qmake
 
 R:
   type: programming
@@ -2242,6 +2264,8 @@ Ruby:
   - .watchr
   interpreters:
   - ruby
+  - macruby
+  - rake
   filenames:
   - .pryrc
   - Appraisals
@@ -2328,6 +2352,8 @@ Scala:
   - .scala
   - .sbt
   - .sc
+  interpreters:
+  - scala
 
 Scaml:
   group: HTML

lib/linguist/samples.rb

@@ -52,14 +52,16 @@ module Linguist
               })
             end
           else
+            path = File.join(dirname, filename)
+
             if File.extname(filename) == ""
-              raise "#{File.join(dirname, filename)} is missing an extension, maybe it belongs in filenames/ subdir"
+              raise "#{path} is missing an extension, maybe it belongs in filenames/ subdir"
             end
 
             yield({
-              :path     => File.join(dirname, filename),
+              :path     => path,
               :language => category,
-              :interpreter => File.exist?(filename) ? Linguist.interpreter_from_shebang(File.read(filename)) : nil,
+              :interpreter => Linguist.interpreter_from_shebang(File.read(path)),
               :extname  => File.extname(filename)
             })
           end
@@ -131,18 +133,19 @@ module Linguist
 
       script = script == 'env' ? tokens[1] : script
 
-      # "python2.6" -> "python"
+      # If script has an invalid shebang, we might get here
-      if script =~ /((?:\d+\.?)+)/
+      return unless script
-        script.sub! $1, ''
+
-      end
+      # "python2.6" -> "python2"
+      script.sub! $1, '' if script =~ /(\.\d+)$/
 
       # Check for multiline shebang hacks that call `exec`
       if script == 'sh' &&
         lines[0...5].any? { |l| l.match(/exec (\w+).+\$0.+\$@/) }
         script = $1
       end
-
+      
-      script
+      File.basename(script)
     else
       nil
     end

samples/Ruby/wrong_shebang.rb

@@ -1,2 +0,0 @@
-#!/usr/bin/env python
-puts "Not Python"

test/test_samples.rb

@@ -31,23 +31,29 @@ class TestSamples < Test::Unit::TestCase
     assert_equal data['languages_total'], data['languages'].inject(0) { |n, (_, c)| n += c }
     assert_equal data['tokens_total'], data['language_tokens'].inject(0) { |n, (_, c)| n += c }
     assert_equal data['tokens_total'], data['tokens'].inject(0) { |n, (_, ts)| n += ts.inject(0) { |m, (_, c)| m += c } }
+    assert !data["interpreters"].empty?
   end
 
-  # Check that there aren't samples with extensions that aren't explicitly defined in languages.yml
+  # Check that there aren't samples with extensions or interpreters that
-  def test_parity
+  # aren't explicitly defined in languages.yml
-    extensions = Samples.cache['extnames']
+  languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
-    languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
+  YAML.load_file(languages_yml).each do |name, options|
-    languages = YAML.load_file(languages_yml)
+    define_method "test_samples_have_parity_with_languages_yml_for_#{name}" do
-
-    languages.each do |name, options|
       options['extensions'] ||= []
-
+      if extnames = Samples.cache['extnames'][name]
-      if extnames = extensions[name]
         extnames.each do |extname|
           next if extname == '.script!'
           assert options['extensions'].include?(extname), "#{name} has a sample with extension (#{extname}) that isn't explicitly defined in languages.yml"
         end
       end
+
+      options['interpreters'] ||= []
+      if interpreters = Samples.cache['interpreters'][name]
+        interpreters.each do |interpreter|
+          # next if extname == '.script!'
+          assert options['interpreters'].include?(interpreter), "#{name} has a sample with an interpreter (#{interpreter}) that isn't explicitly defined in languages.yml"
+        end
+      end
     end
   end
 
@@ -76,4 +82,9 @@ class TestSamples < Test::Unit::TestCase
       end
     end
   end
+
+  def test_shebang
+    assert_equal "crystal", Linguist.interpreter_from_shebang("#!/usr/bin/env bin/crystal")
+    assert_equal "python2", Linguist.interpreter_from_shebang("#!/usr/bin/python2.4")
+  end
 end