detect non-C languages in folders with C/C++ files

Improves on 2da75aa119.

tests/test_dependencies.py

@@ -42,7 +42,7 @@ class DependenciesTestCase(utils.TestCase):
 
     def test_token_parser(self):
         with self.assertRaises(NotYetImplemented):
-            source_file = 'tests/samples/codefiles/see.h'
+            source_file = 'tests/samples/codefiles/c_only/non_empty.h'
             parser = TokenParser(source_file)
             parser.parse()
 
@@ -458,7 +458,7 @@ class DependenciesTestCase(utils.TestCase):
         self.patched['wakatime.packages.requests.adapters.HTTPAdapter.send'].return_value = response
 
         with utils.TemporaryDirectory() as tempdir:
-            entity = 'tests/samples/codefiles/see.c'
+            entity = 'tests/samples/codefiles/c_only/non_empty.c'
             shutil.copy(entity, os.path.join(tempdir, 'see.c'))
             entity = os.path.realpath(os.path.join(tempdir, 'see.c'))
 
@@ -510,9 +510,9 @@ class DependenciesTestCase(utils.TestCase):
         self.patched['wakatime.packages.requests.adapters.HTTPAdapter.send'].return_value = response
 
         with utils.TemporaryDirectory() as tempdir:
-            entity = 'tests/samples/codefiles/seeplusplus.cpp'
-            shutil.copy(entity, os.path.join(tempdir, 'seeplusplus.cpp'))
-            entity = os.path.realpath(os.path.join(tempdir, 'seeplusplus.cpp'))
+            entity = 'tests/samples/codefiles/c_and_cpp/non_empty.cpp'
+            shutil.copy(entity, os.path.join(tempdir, 'non_empty.cpp'))
+            entity = os.path.realpath(os.path.join(tempdir, 'non_empty.cpp'))
 
             now = u(int(time.time()))
             config = 'tests/samples/configs/good_config.cfg'
@@ -562,7 +562,7 @@ class DependenciesTestCase(utils.TestCase):
         self.patched['wakatime.packages.requests.adapters.HTTPAdapter.send'].return_value = response
 
         with utils.TemporaryDirectory() as tempdir:
-            entity = 'tests/samples/codefiles/seesharp.cs'
+            entity = 'tests/samples/codefiles/csharp/seesharp.cs'
             shutil.copy(entity, os.path.join(tempdir, 'seesharp.cs'))
             entity = os.path.realpath(os.path.join(tempdir, 'seesharp.cs'))
 

tests/test_languages.py

@@ -23,31 +23,6 @@ class LanguagesTestCase(utils.TestCase):
         ['wakatime.session_cache.SessionCache.connect', None],
     ]
 
-    def test_language_detected_for_header_file(self):
-        response = Response()
-        response.status_code = 500
-        self.patched['wakatime.packages.requests.adapters.HTTPAdapter.send'].return_value = response
-
-        now = u(int(time.time()))
-        config = 'tests/samples/configs/good_config.cfg'
-        entity = 'tests/samples/codefiles/see.h'
-        args = ['--file', entity, '--config', config, '--time', now]
-
-        retval = execute(args)
-        self.assertEquals(retval, 102)
-
-        language = u('C')
-        self.assertEqual(self.patched['wakatime.offlinequeue.Queue.push'].call_args[0][0].get('language'), language)
-
-        entity = 'tests/samples/codefiles/seeplusplus.h'
-        args[1] = entity
-
-        retval = execute(args)
-        self.assertEquals(retval, 102)
-
-        language = u('C++')
-        self.assertEqual(self.patched['wakatime.offlinequeue.Queue.push'].call_args[0][0].get('language'), language)
-
     def test_c_language_detected_for_header_with_c_files_in_folder(self):
         response = Response()
         response.status_code = 500
@@ -80,6 +55,22 @@ class LanguagesTestCase(utils.TestCase):
         language = u('C++')
         self.assertEqual(self.patched['wakatime.offlinequeue.Queue.push'].call_args[0][0].get('language'), language)
 
+    def test_c_not_detected_for_non_header_with_c_files_in_folder(self):
+        response = Response()
+        response.status_code = 500
+        self.patched['wakatime.packages.requests.adapters.HTTPAdapter.send'].return_value = response
+
+        now = u(int(time.time()))
+        config = 'tests/samples/configs/good_config.cfg'
+        entity = 'tests/samples/codefiles/c_and_python/see.py'
+        args = ['--file', entity, '--config', config, '--time', now]
+
+        retval = execute(args)
+        self.assertEquals(retval, 102)
+
+        language = u('Python')
+        self.assertEqual(self.patched['wakatime.offlinequeue.Queue.push'].call_args[0][0].get('language'), language)
+
     def test_guess_language(self):
         with utils.mock.patch('wakatime.stats.smart_guess_lexer') as mock_guess_lexer:
             mock_guess_lexer.return_value = None

wakatime/stats.py

@@ -11,6 +11,7 @@
 
 import logging
 import os
+import re
 import sys
 
 from .compat import u, open
@@ -64,13 +65,18 @@ def get_file_stats(file_name, entity_type='file', lineno=None, cursorpos=None,
 def guess_language(file_name):
     """Guess lexer and language for a file.
 
-    Returns (language, lexer) tuple where language is a unicode string.
+    Returns a tuple of (language_str, lexer_obj).
     """
 
+    lexer = None
+
     language = get_language_from_extension(file_name)
-    lexer = smart_guess_lexer(file_name)
-    if language is None and lexer is not None:
-        language = u(lexer.name)
+    if language:
+        lexer = get_lexer(language)
+    else:
+        lexer = smart_guess_lexer(file_name)
+        if lexer:
+            language = u(lexer.name)
 
     return language, lexer
 
@@ -151,15 +157,17 @@ def guess_lexer_using_modeline(text):
 
 def get_language_from_extension(file_name):
     """Returns a matching language for the given file extension.
+
+    When guessed_language is 'C', does not restrict to known file extensions.
     """
 
     filepart, extension = os.path.splitext(file_name)
 
-    if os.path.exists(u('{0}{1}').format(u(filepart), u('.c'))) or os.path.exists(u('{0}{1}').format(u(filepart), u('.C'))):
-        return 'C'
+    if re.match(r'\.h.*', extension, re.IGNORECASE) or re.match(r'\.c.*', extension, re.IGNORECASE):
+
+        if os.path.exists(u('{0}{1}').format(u(filepart), u('.c'))) or os.path.exists(u('{0}{1}').format(u(filepart), u('.C'))):
+            return 'C'
 
-    extension = extension.lower()
-    if extension == '.h':
         directory = os.path.dirname(file_name)
         available_files = os.listdir(directory)
         available_extensions = list(zip(*map(os.path.splitext, available_files)))[1]
@@ -191,7 +199,7 @@ def number_lines_in_file(file_name):
 def standardize_language(language, plugin):
     """Maps a string to the equivalent Pygments language.
 
-    Returns a tuple of (language_name, lexer_object).
+    Returns a tuple of (language_str, lexer_obj).
     """
 
     if not language: