Do not extract text from JPEG and Gif #74

Signed-off-by: Philippe Ombredanne <pombredanne@nexb.com>
aboutcode-org · May 2, 2018 · c57b5b6 · c57b5b6
1 parent 5b5cca0
commit c57b5b6
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 5 deletions.
diff --git a/src/typecode/contenttype.py b/src/typecode/contenttype.py
@@ -454,9 +454,11 @@ def is_media_with_meta(self):
         """
         Return True if the file is a media file that may contain text metadata.
         """
-        # For now we only exclude PNGs, though there are likely several other
-        # mp(1,2,3,4), jpeg, gif all have support for metadata
-        if self.is_media and 'png image' in self.filetype_file.lower():
+        # For now we only exclude PNGs, JEPG and Gifs, though there are likely
+        # several other
+        # mp(1,2,3,4), jpeg, gif all have support for metadata but we exclude some
+        if (self.is_media and self.filetype_file.lower().startswith(
+                ('gif image', 'png image','jpeg image'))):
             return False
         else:
             return True

diff --git a/...textcode/data/media_with_text/sample1.gif → ...tcode/data/media_without_text/sample1.gif b/...textcode/data/media_with_text/sample1.gif → ...tcode/data/media_without_text/sample1.gif
diff --git a/...textcode/data/media_with_text/sample1.jpg → ...tcode/data/media_without_text/sample1.jpg b/...textcode/data/media_with_text/sample1.jpg → ...tcode/data/media_without_text/sample1.jpg
diff --git a/tests/typecode/test_contenttype.py b/tests/typecode/test_contenttype.py
@@ -811,6 +811,7 @@ def test_media_image_gif(self):
         test_file = self.get_test_loc('contenttype/media/Image1.gif')
         assert is_media(test_file)
         assert is_binary(test_file)
+        assert not contains_text(test_file)
 
     def test_media_image_ico(self):
         test_file = self.get_test_loc('contenttype/media/Image1.ico')
@@ -840,13 +841,13 @@ def test_media_image_jpeg(self):
         test_file = self.get_test_loc('contenttype/media/Image1.jpeg')
         assert is_media(test_file)
         assert is_binary(test_file)
-        assert contains_text(test_file)
+        assert not contains_text(test_file)
 
     def test_media_image_jpg(self):
         test_file = self.get_test_loc('contenttype/media/Image1.jpg')
         assert is_media(test_file)
         assert is_binary(test_file)
-        assert contains_text(test_file)
+        assert not contains_text(test_file)
 
     def test_media_image_pbm(self):
         test_file = self.get_test_loc('contenttype/media/Image1.pbm')