Fix confidence scoring flaw

CHANGELOG.md

@@ -1,6 +1,11 @@
 Changelog
 =========
 
+Version 1.10
+------------
+
+- Fixing how confidence works
+
 Version 1.9
 -----------
 

puremagic/main.py

@@ -19,7 +19,7 @@
 from collections import namedtuple
 
 __author__ = "Chris Griffith"
-__version__ = "1.9"
+__version__ = "1.10"
 __all__ = [
     "magic_file",
     "magic_string",
@@ -79,8 +79,8 @@ def _confidence(matches, ext=None):
     """ Rough confidence based on string length and file extension"""
     results = []
     for match in matches:
-        con = 0.8 if len(match.extension) > 9 else float("0.{0}".format(len(match.extension)))
-        if ext == match.extension:
+        con = 0.8 if len(match.byte_match) > 9 else float("0.{0}".format(len(match.byte_match)))
+        if ext and ext == match.extension:
             con = 0.9
         results.append(PureMagicWithConfidence(confidence=con, **match._asdict()))
     return sorted(results, key=lambda x: x.confidence, reverse=True)

test/test_common_extensions.py

@@ -68,18 +68,6 @@ def test_string(self):
         ext = puremagic.from_string(bytes(self.mp4magic))
         self.assertEqual(self.expect_ext, ext)
 
-    def test_string_with_filename_hint(self):
-        """String identification with filename hint """
-        filename = os.path.join(OFFICE_DIR, "test.xlsx")
-        with open(filename, "rb") as f:
-            data = f.read()
-        ext = puremagic.from_string(data)
-        # .docx and .xlsx have same signature
-        self.assertEqual(".docx", ext)
-        # with the hint from_string() shoud find the correct extension
-        ext = puremagic.from_string(data, filename=filename)
-        self.assertEqual(".xlsx", ext)
-
     def test_string_with_confidence(self):
         """String identification: magic_string """
         ext = puremagic.magic_string(bytes(self.mp4magic))