Move specific filename samples into their own dir

lib/linguist/language.rb

@@ -442,11 +442,13 @@ module Linguist
   end
 
   extensions = Sample.extensions
+  filenames = Sample.filenames
   popular = YAML.load_file(File.expand_path("../popular.yml", __FILE__))
 
   YAML.load_file(File.expand_path("../languages.yml", __FILE__)).each do |name, options|
     aliases = [name.downcase.gsub(/\s/, '-') ] + (options[:aliases] || [])
     options['extensions'] ||= []
+    options['filenames'] ||= []
     aliases.each do |name|
       if extnames = extensions[name]
         extnames.each do |extname|
@@ -457,6 +459,16 @@ module Linguist
           end
         end
       end
+
+      if fns = filenames[name]
+        fns.each do |filename|
+          if !options['filenames'].include?(filename)
+            options['filenames'] << filename
+          else
+            warn "#{name} #{filename.inspect} is already defined in samples/. Remove from languages.yml."
+          end
+        end
+      end
     end
 
     lang = Language.create(

lib/linguist/languages.yml

@@ -1064,11 +1064,9 @@ Ruby:
   - .thor
   - .watchr
   filenames:
-  - Capfile
   - Gemfile
   - Guardfile
   - Podfile
-  - Rakefile
   - Thorfile
   - Vagrantfile
 
@@ -1151,7 +1149,6 @@ Shell:
   - .zshrc
   - bashrc
   - zshrc
-  - PKGBUILD
 
 Smalltalk:
   type: programming

lib/linguist/sample.rb

@@ -22,7 +22,24 @@ module Linguist
         dirname = File.join(PATH, category)
         Dir.entries(dirname).each do |filename|
           next if filename == '.' || filename == '..'
-          yield({ :path => File.join(dirname, filename), :language => category })
+
+          if filename == 'filenames'
+            Dir.entries(File.join(dirname, filename)).each do |subfilename|
+              next if subfilename == '.' || subfilename == '..'
+
+              yield({
+                :path    => File.join(dirname, filename, subfilename),
+                :language => category,
+                :filename => subfilename
+              })
+            end
+          else
+            yield({
+              :path     => File.join(dirname, filename),
+              :language => category,
+              :extname  => File.extname(filename)
+            })
+          end
         end
       end
 
@@ -36,15 +53,29 @@ module Linguist
     def self.extensions
       extensions = {}
       each do |sample|
-        extname = File.extname(sample[:path])
         # TODO: For now skip empty extnames
-        next if extname == ""
+        next if sample[:extname].nil? || sample[:extname] == ""
         extensions[sample[:language]] ||= Set.new
-        extensions[sample[:language]] << extname
+        extensions[sample[:language]] << sample[:extname]
       end
       extensions
     end
 
+    # Get all filenames listed in samples/
+    #
+    # Returns Hash of sample language keys with a Set of filename
+    # Strings.
+    def self.filenames
+      filenames = {}
+      each do |sample|
+        # TODO: For now skip empty extnames
+        next if sample[:filename].nil?
+        filenames[sample[:language]] ||= Set.new
+        filenames[sample[:language]] << sample[:filename]
+      end
+      filenames
+    end
+
     # Public: Build Classifier from all samples.
     #
     # Returns trained Classifier.