Override for binary mime type based on languages.yml

If the extension already exists in languages.yml, it's probably not a
binary, but code.

lib/linguist/blob_helper.rb

@@ -58,6 +58,15 @@ module Linguist
       _mime_type ? _mime_type.binary? : false
     end
 
+    # Internal: Is the blob binary according to its mime type,
+    # overriding it if we have better data from the languages.yml
+    # database.
+    #
+    # Return true or false
+    def likely_binary?
+       binary_mime_type? and not Language.find_by_filename(name)
+    end
+
     # Public: Get the Content-Type header value
     #
     # This value is used when serving raw blobs.

lib/linguist/repository.rb

@@ -67,8 +67,8 @@ module Linguist
       return if @computed_stats
 
       @enum.each do |blob|
-        # Skip binary file extensions
+        # Skip files that are likely binary
-        next if blob.binary_mime_type?
+        next if blob.likely_binary?
 
         # Skip vendored or generated blobs
         next if blob.vendored? || blob.generated? || blob.language.nil?

test/test_repository.rb

@@ -24,4 +24,8 @@ class TestRepository < Test::Unit::TestCase
   def test_linguist_size
     assert linguist_repo.size > 30_000
   end
+
+  def test_binary_override
+    assert_equal repo(File.expand_path("../../samples/Nimrod", __FILE__)).language, Language["Nimrod"]
+  end
 end