Add BlobHelper#include_in_language_stats?

This just extracts some logic from Repository#compute_stats and makes it
testable.

lib/linguist/blob_helper.rb

@@ -332,5 +332,15 @@ module Linguist
     def tm_scope
       language && language.tm_scope
     end
+
+    DETECTABLE_TYPES = [:programming, :markup].freeze
+
+    # Internal: Should this blob be included in repository language statistics?
+    def include_in_language_stats?
+      !vendored? &&
+      !documentation? &&
+      !generated? &&
+      language && DETECTABLE_TYPES.include?(language.type)
+    end
   end
 end

lib/linguist/repository.rb

@@ -8,8 +8,6 @@ module Linguist
   # Its primary purpose is for gathering language statistics across
   # the entire project.
   class Repository
-    DETECTABLE_TYPES = [:programming, :markup].freeze
-
     attr_reader :repository
 
     # Public: Create a new Repository based on the stats of
@@ -158,14 +156,10 @@ module Linguist
 
           blob = Linguist::LazyBlob.new(repository, delta.new_file[:oid], new, mode.to_s(8))
 
-          # Skip vendored or generated blobs
+          next unless blob.include_in_language_stats?
-          next if blob.vendored? || blob.documentation? || blob.generated? || blob.language.nil?
-
-          if DETECTABLE_TYPES.include?(blob.language.type)
           file_map[new] = [blob.language.group.name, blob.size]
         end
       end
-      end
 
       file_map
     end

test/test_blob.rb

@@ -517,4 +517,29 @@ class TestBlob < Minitest::Test
     refute blob.new(" ").empty?
     refute blob.new("nope").empty?
   end
+
+  def test_include_in_language_stats
+    vendored = sample_blob("bower_components/custom/custom.js")
+    assert_predicate vendored, :vendored?
+    refute_predicate vendored, :include_in_language_stats?
+
+    documentation = fixture_blob("README")
+    assert_predicate documentation, :documentation?
+    refute_predicate documentation, :include_in_language_stats?
+
+    generated = sample_blob("CSS/bootstrap.min.css")
+    assert_predicate generated, :generated?
+    refute_predicate generated, :include_in_language_stats?
+
+    data = sample_blob("Ant Build System/filenames/ant.xml")
+    assert_equal :data, data.language.type
+    refute_predicate data, :include_in_language_stats?
+
+    prose = sample_blob("Markdown/tender.md")
+    assert_equal :prose, prose.language.type
+    refute_predicate prose, :include_in_language_stats?
+
+    included = sample_blob("HTML/pages.html")
+    assert_predicate included, :include_in_language_stats?
+  end
 end