Replace imghdr by filetype for python 3.11

keep imghdr for python 2.7 to reduce dependency
imghdr is deprecated from python 3.11
https://docs.python.org/3/library/imghdr.html

Use alternative filetype
Adding test

Author: mathben

flanker/mime/message/part.py

@@ -1,6 +1,6 @@
 import base64
-import imghdr
 import logging
+import sys
 import mimetypes
 import quopri
 from contextlib import closing
@@ -9,6 +9,11 @@
 import six
 from six.moves import StringIO
 
+if sys.version_info[0] >= 3:
+    import filetype as _filetype
+else:
+    import imghdr as _imghdr
+
 from flanker import metrics, _email
 from flanker.mime import bounce
 from flanker.mime.message import headers, charsets
@@ -113,9 +118,16 @@ def adjust_content_type(content_type, body=None, filename=None):
         if six.PY3 and isinstance(body, six.text_type):
             image_preamble = image_preamble.encode('utf-8', 'ignore')
 
-        sub = imghdr.what(None, image_preamble)
-        if sub:
-            content_type = ContentType('image', sub)
+        if sys.version_info[0] >= 3:
+            kind = _filetype.guess(image_preamble)
+            if kind and kind.mime.startswith('image/'):
+                sub = kind.extension
+                sub = {'jpg': 'jpeg', 'tif': 'tiff'}.get(sub, sub)
+                content_type = ContentType('image', sub)
+        else:
+            sub = _imghdr.what(None, image_preamble)
+            if sub:
+                content_type = ContentType('image', sub)
 
     elif content_type.main == 'audio' and body:
         sub = _email.detect_audio_type(body)

setup.py

@@ -41,7 +41,8 @@
           'regex>=0.1.20110315',
           'six',
           'tld',
-          'WebOb>=0.9.8'],
+          'WebOb>=0.9.8',
+          'filetype; python_version >= "3"'],
       extras_require={
           'validator': [
               'dnsq>=1.1.6',

tests/mime/message/test_adjust_content_type.py

@@ -0,0 +1,111 @@
+# coding:utf-8
+"""
+Tests for adjust_content_type() in flanker/mime/message/part.py.
+
+These tests specifically cover the filetype-based image detection that
+replaced the deprecated imghdr module.
+"""
+import pytest
+
+from flanker.mime.message.part import adjust_content_type
+from flanker.mime.message.headers import ContentType
+from tests import MAILGUN_PNG
+
+
+# Magic bytes for common image formats
+JPEG_MAGIC = b'\xff\xd8\xff\xe0' + b'\x00' * 28
+PNG_MAGIC = b'\x89PNG\r\n\x1a\n' + b'\x00' * 24
+GIF_MAGIC = b'GIF89a' + b'\x00' * 26
+BMP_MAGIC = b'BM' + b'\x00' * 30
+# WebP requires: RIFF + size(4) + WEBP + VP8 chunk type(4)
+WEBP_MAGIC = b'RIFF\x28\x00\x00\x00WEBPVP8 \x1c\x00\x00\x00' + b'\x00' * 16
+# TIFF little-endian (II) and big-endian (MM)
+TIFF_LE_MAGIC = b'II\x2a\x00' + b'\x00' * 28
+TIFF_BE_MAGIC = b'MM\x00\x2a' + b'\x00' * 28
+
+
+class TestAdjustContentTypeImageDetection:
+    """Tests for filetype-based image subtype detection from body bytes."""
+
+    def _image_ct(self, sub='octet-stream'):
+        return ContentType('image', sub)
+
+    def test_jpeg_detected_and_mapped(self):
+        """filetype returns 'jpg', must be mapped to 'jpeg' for MIME."""
+        ct = adjust_content_type(self._image_ct(), body=JPEG_MAGIC)
+        assert str(ct) == 'image/jpeg'
+
+    def test_png_detected(self):
+        ct = adjust_content_type(self._image_ct(), body=PNG_MAGIC)
+        assert str(ct) == 'image/png'
+
+    def test_gif_detected(self):
+        ct = adjust_content_type(self._image_ct(), body=GIF_MAGIC)
+        assert str(ct) == 'image/gif'
+
+    def test_bmp_detected(self):
+        ct = adjust_content_type(self._image_ct(), body=BMP_MAGIC)
+        assert str(ct) == 'image/bmp'
+
+    def test_webp_detected(self):
+        ct = adjust_content_type(self._image_ct(), body=WEBP_MAGIC)
+        assert str(ct) == 'image/webp'
+
+    def test_tiff_le_detected_and_mapped(self):
+        """filetype returns 'tif', must be mapped to 'tiff' for MIME."""
+        ct = adjust_content_type(self._image_ct(), body=TIFF_LE_MAGIC)
+        assert str(ct) == 'image/tiff'
+
+    def test_tiff_be_detected_and_mapped(self):
+        ct = adjust_content_type(self._image_ct(), body=TIFF_BE_MAGIC)
+        assert str(ct) == 'image/tiff'
+
+    def test_unknown_body_keeps_original_subtype(self):
+        """Non-image bytes: content type should not change."""
+        ct = adjust_content_type(self._image_ct('jpeg'), body=b'notanimage' * 4)
+        assert str(ct) == 'image/jpeg'
+
+    def test_no_body_keeps_original_subtype(self):
+        ct = adjust_content_type(self._image_ct('jpeg'), body=None)
+        assert str(ct) == 'image/jpeg'
+
+    def test_real_png_fixture(self):
+        """Detection works on a real PNG binary (mailgun.png fixture)."""
+        ct = adjust_content_type(self._image_ct('octet-stream'), body=MAILGUN_PNG)
+        assert str(ct) == 'image/png'
+
+    def test_only_preamble_used(self):
+        """Detection is based on first 32 bytes; trailing garbage is ignored."""
+        png_with_garbage = PNG_MAGIC + b'\xff' * 10000
+        ct = adjust_content_type(self._image_ct(), body=png_with_garbage)
+        assert str(ct) == 'image/png'
+
+    def test_non_image_content_type_not_affected(self):
+        """adjust_content_type should not touch non-image/non-audio content types."""
+        ct = adjust_content_type(ContentType('text', 'plain'), body=PNG_MAGIC)
+        assert str(ct) == 'text/plain'
+
+
+class TestAdjustContentTypeFilename:
+    """Tests for filename-based content type guessing (pre-existing behavior)."""
+
+    def test_bz2_filename(self):
+        ct = adjust_content_type(
+            ContentType('application', 'octet-stream'), filename='archive.bz2')
+        assert str(ct) == 'application/x-bzip2'
+
+    def test_gz_filename(self):
+        ct = adjust_content_type(
+            ContentType('application', 'octet-stream'), filename='archive.gz')
+        assert str(ct) == 'application/x-gzip'
+
+    def test_png_filename(self):
+        ct = adjust_content_type(
+            ContentType('application', 'octet-stream'), filename='photo.png')
+        assert str(ct) == 'image/png'
+
+    def test_filename_ignored_when_not_octet_stream(self):
+        """Filename guessing only triggers for application/octet-stream."""
+        ct = adjust_content_type(
+            ContentType('image', 'jpeg'), filename='photo.png')
+        assert str(ct) == 'image/jpeg'