use most accurate language from modeline lexer and filename lexer Pygments guessing functions

wakatime/stats.py

@@ -20,7 +20,7 @@ if sys.version_info[0] == 2:
     sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'packages', 'pygments_py2'))
 else:
     sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'packages', 'pygments_py3'))
-from pygments.lexers import guess_lexer_for_filename
+from pygments.lexers import guess_lexer, guess_lexer_for_filename
 
 
 log = logging.getLogger('WakaTime')
@@ -47,20 +47,67 @@ TRANSLATIONS = {
 
 
 def guess_language(file_name):
-    language, lexer = None, None
-    try:
-        with open(file_name, 'r', encoding='utf-8') as fh:
-            lexer = guess_lexer_for_filename(file_name, fh.read(512000))
-    except:
-        pass
+    """Guess lexer and language for a file.
+
+    Returns (language, lexer) tuple where language is a unicode string.
+    """
+
+    lexer = smart_guess_lexer(file_name)
+
+    language = None
+
+    # guess language from file extension
     if file_name:
         language = guess_language_from_extension(file_name.rsplit('.', 1)[-1])
-    if lexer and language is None:
+
+    # get language from lexer if we didn't have a hard-coded extension rule
+    if language is None and lexer:
         language = translate_language(u(lexer.name))
+
     return language, lexer
 
 
+def smart_guess_lexer(file_name):
+    """Guess Pygments lexer for a file.
+
+    Looks for a vim modeline in file contents, then compares the accuracy
+    of that lexer with a second guess. The second guess looks up all lexers
+    matching the file name, then runs a text analysis for the best choice.
+    """
+    lexer = None
+
+    text = get_file_contents(file_name)
+
+    try:
+        guess_1 = guess_lexer(text)
+    except:
+        guess_1 = None
+    try:
+        guess_2 = guess_lexer_for_filename(file_name, text)
+    except:
+        guess_2 = None
+    try:
+        accuracy_1 = guess_1.analyse_text(text)
+    except:
+        accuracy_1 = None
+    try:
+        accuracy_2 = guess_2.analyse_text(text)
+    except:
+        accuracy_2 = None
+
+    if accuracy_1:
+        lexer = guess_1
+    if (accuracy_2 and
+        (not accuracy_1 or accuracy_2 > accuracy_1)):
+        lexer = guess_2
+
+    return lexer
+
+
 def guess_language_from_extension(extension):
+    """Checks hard-coded extension map for a matching language.
+    """
+
     if extension:
         if extension in EXTENSIONS:
             return EXTENSIONS[extension]
@@ -70,6 +117,9 @@ def guess_language_from_extension(extension):
 
 
 def translate_language(language):
+    """Turns Pygments lexer class name string into human-readable language.
+    """
+
     if language in TRANSLATIONS:
         language = TRANSLATIONS[language]
     return language
@@ -107,3 +157,16 @@ def get_file_stats(file_name, notfile=False, lineno=None, cursorpos=None):
             'cursorpos': cursorpos,
         }
     return stats
+
+
+def get_file_contents(file_name):
+    """Returns the first 512000 bytes of the file's contents.
+    """
+
+    text = None
+    try:
+        with open(file_name, 'r', encoding='utf-8') as fh:
+            text = fh.read(512000)
+    except:
+        pass
+    return text