octocat
Unified Split

Showing with 36 additions and 11 deletions.

+9 −0 lib/linguist/heuristics.rb

+27 −11 test/test_heuristics.rb
diff --git a/lib/linguist/heuristics.rb b/lib/linguist/heuristics.rb
@@ -23,6 +23,15 @@ def self.call(blob, languages)
       [] # No heuristics matched
     end
 
+    # Public: Do we have a heuristic for these candidate languages?
+    #
+    # languages - Array of Language objects
+    #
+    # Returns true if there are matching heuristics or nil
+    def self.defined_for?(languages)
+      @heuristics.select { |h| h.matches?(languages) }.any?
+    end
+
     # Internal: Define a new heuristic.
     #
     # languages - String names of languages to disambiguate.

diff --git a/test/test_heuristics.rb b/test/test_heuristics.rb
@@ -22,6 +22,17 @@ def test_no_match
     assert_equal [], results
   end
 
+  def assert_heuristics(hash)
+    candidates = hash.keys.map { |l| Language[l] }
+
+    hash.each do |language, blobs|
+      Array(blobs).each do |blob|
+        result = Heuristics.call(file_blob(blob), candidates)
+        assert_equal [Language[language]], result, "Failed for #{blob}"
+      end
+    end
+  end
+
   # Candidate languages = ["C++", "Objective-C"]
   def test_obj_c_by_heuristics
     # Only calling out '.h' filenames as these are the ones causing issues
@@ -148,17 +159,6 @@ def test_cs_by_heuristics
     })
   end
 
-  def assert_heuristics(hash)
-    candidates = hash.keys.map { |l| Language[l] }
-
-    hash.each do |language, blobs|
-      Array(blobs).each do |blob|
-        result = Heuristics.call(file_blob(blob), candidates)
-        assert_equal [Language[language]], result, "Failed for #{blob}"
-      end
-    end
-  end
-
   def test_ls_by_heuristics
     assert_heuristics({
       "LiveScript" => "LiveScript/hello.ls",
@@ -172,4 +172,20 @@ def test_ts_by_heuristics
       "XML" => all_fixtures("XML", "*.ts")
     })
   end
+
+  def output_languages(languages)
+    languages.map { |l| l.name }.join(", ")
+  end
+
+  # If a language extension isn't globally unique then check if we have a heuristic
+  Linguist::Language.all.each do |language|
+    define_method "test_#{language.name}_has_adequate_heuristics" do
+      language.extensions.each do |extension|
+        languages = Language.find_by_extension(extension)
+        next if extension == ".script!" || languages.size == 1
+
+        assert Heuristics.defined_for?(languages), "No heuristic for #{extension} and #{output_languages(languages)}"
+      end
+    end
+  end
 end