Adding test to check that languages.yml includes all extensions represented in samples folder

lib/linguist/language.rb

@@ -532,6 +532,7 @@ module Linguist
     if extnames = extensions[name]
       extnames.each do |extname|
         if !options['extensions'].include?(extname)
+          warn "#{name} has a sample with extension (#{extname}) that isn't explicitly defined in languages.yml" unless extname == '.script!'
           options['extensions'] << extname
         end
       end

lib/linguist/languages.yml

@@ -260,6 +260,7 @@ C:
   extensions:
   - .c
   - .cats
+  - .h
   - .w
 
 C#:
@@ -288,6 +289,7 @@ C++:
   - .cc
   - .cxx
   - .H
+  - .h
   - .h++
   - .hh
   - .hpp
@@ -444,6 +446,7 @@ Coq:
   type: programming
   extensions:
   - .coq
+  - .v
 
 Cpp-ObjDump:
   type: data
@@ -539,6 +542,7 @@ Dart:
 Diff:
   extensions:
   - .diff
+  - .patch
 
 Dogescript:
   type: programming
@@ -623,6 +627,7 @@ Erlang:
   color: "#0faf8d"
   extensions:
   - .erl
+  - .escript
   - .hrl
 
 F#:
@@ -698,6 +703,7 @@ Forth:
   extensions:
   - .fth
   - .4th
+  - .forth
 
 Frege:
   type: programming
@@ -806,6 +812,9 @@ Gosu:
   color: "#82937f"
   extensions:
   - .gs
+  - .gst
+  - .gsx
+  - .vark
 
 Grace:
   type: programming
@@ -841,6 +850,7 @@ Groovy:
   color: "#e69f56"
   extensions:
   - .groovy
+  - .gradle
   - .grt
   - .gtpl
   - .gvy
@@ -945,6 +955,7 @@ IDL:
   color: "#e3592c"
   extensions:
   - .pro
+  - .dlm
 
 INI:
   type: data
@@ -1019,6 +1030,7 @@ JSON:
   searchable: false
   extensions:
   - .json
+  - .lock
   - .sublime-keymap
   - .sublime-mousemap
   - .sublime-project
@@ -1147,6 +1159,9 @@ Lasso:
   color: "#2584c3"
   extensions:
   - .lasso
+  - .las
+  - .lasso9
+  - .ldml
 
 Latte:
   type: markup
@@ -1232,6 +1247,7 @@ Lua:
   extensions:
   - .lua
   - .nse
+  - .pd_lua
   - .rbxs
   interpreters:
   - lua
@@ -1377,6 +1393,7 @@ Myghty:
 NSIS:
   extensions:
   - .nsi
+  - .nsh
 
 Nemerle:
   type: programming
@@ -1441,6 +1458,7 @@ OCaml:
   color: "#3be133"
   extensions:
   - .ml
+  - .eliom
   - .eliomi
   - .ml4
   - .mli
@@ -1461,6 +1479,7 @@ Objective-C:
   - objc
   extensions:
   - .m
+  - .h
 
 Objective-C++:
   type: programming
@@ -1508,6 +1527,7 @@ OpenEdge ABL:
   - abl
   extensions:
   - .p
+  - .cls
 
 Org:
   type: prose
@@ -1546,6 +1566,7 @@ PHP:
   - .php
   - .aw
   - .ctp
+  - .module
   - .php3
   - .php4
   - .php5
@@ -1594,6 +1615,7 @@ Pascal:
   extensions:
   - .pas
   - .dfm
+  - .dpr
   - .lpr
 
 Perl:
@@ -1603,12 +1625,14 @@ Perl:
   extensions:
   - .pl
   - .PL
+  - .fcgi
   - .perl
   - .ph
   - .plx
   - .pm
   - .pod
   - .psgi
+  - .t
   interpreters:
   - perl
 
@@ -1817,6 +1841,7 @@ Racket:
   - .rkt
   - .rktd
   - .rktl
+  - .scrbl
 
 Ragel in Ruby Host:
   type: programming
@@ -1886,7 +1911,10 @@ Ruby:
   - .god
   - .irbrc
   - .mspec
+  - .pluginspec
   - .podspec
+  - .rabl
+  - .rake
   - .rbuild
   - .rbw
   - .rbx
@@ -1960,6 +1988,7 @@ Sass:
   group: CSS
   extensions:
   - .sass
+  - .scss
 
 Scala:
   type: programming
@@ -1967,6 +1996,7 @@ Scala:
   color: "#7dd3b0"
   extensions:
   - .scala
+  - .sbt
   - .sc
 
 Scaml:
@@ -1982,6 +2012,7 @@ Scheme:
   - .scm
   - .sld
   - .sls
+  - .sps
   - .ss
   interpreters:
   - guile
@@ -1993,6 +2024,8 @@ Scilab:
   type: programming
   extensions:
   - .sci
+  - .sce
+  - .tst
 
 Self:
   type: programming
@@ -2012,8 +2045,10 @@ Shell:
   - zsh
   extensions:
   - .sh
+  - .bash
   - .bats
   - .tmux
+  - .zsh
   interpreters:
   - bash
   - sh
@@ -2080,6 +2115,7 @@ Standard ML:
   extensions:
   - .ML
   - .fun
+  - .sig
   - .sml
 
 Stata:
@@ -2280,6 +2316,7 @@ Visual Basic:
   extensions:
   - .vb
   - .bas
+  - .cls
   - .frm
   - .frx
   - .vba
@@ -2308,6 +2345,7 @@ XML:
   - wsdl
   extensions:
   - .xml
+  - .ant
   - .axml
   - .ccxml
   - .clixml
@@ -2321,6 +2359,7 @@ XML:
   - .fsproj
   - .glade
   - .grxml
+  - .ivy
   - .jelly
   - .kml
   - .launch

test/test_samples.rb

@@ -35,15 +35,33 @@ class TestSamples < Test::Unit::TestCase
     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 } }
   end
-  
+
+  # Check that there aren't samples with extensions that aren't explicitly defined in languages.yml
+  def test_parity
+    extensions = Samples::DATA['extnames']
+    languages_yml = File.expand_path("../../lib/linguist/languages.yml", __FILE__)
+    languages = YAML.load_file(languages_yml)
+
+    languages.each do |name, options|
+      options['extensions'] ||= []
+
+      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
+    end
+  end
+
   # If a language extension isn't globally unique then make sure there are samples
   def test_presence
     Linguist::Language.all.each do |language|
       language.all_extensions.each do |extension|
         language_matches = Language.find_by_filename("foo#{extension}")
-        
+
         # If there is more than one language match for a given extension
-        # then check that there are examples for that language with the extension 
+        # then check that there are examples for that language with the extension
         if language_matches.length > 1
           language_matches.each do |language|
             assert File.directory?("samples/#{language.name}"), "#{language.name} is missing a samples directory"