From 49da9728a69ae8a63af8a4630fccc55c10e66392 Mon Sep 17 00:00:00 2001
From: Thomas Scholtes <geigerzaehler@axiom.fm>
Date: Fri, 8 Mar 2024 12:17:43 +0100
Subject: [PATCH 1/3] Replace deprecated `imghdr` package with `filetype`

The `imghdr` package from the Python standard library is [deprecated and
will be removed in Python 3.13][1]. The deprecated PEP suggests packages
filetype, puremagic, python-magic as replacements. I chose filetype
because it has a [larger user-base][2] than puremagic and does not have
a C dependency, like python-magic.

[1]: https://peps.python.org/pep-0594/#deprecated-modules
[2]: https://libraries.io/pypi/filetype
---
 mediafile.py                   |  45 +++------------------------------
 pyproject.toml                 |   1 +
 test/rsrc/only-magic-bytes.jpg | Bin 622 -> 0 bytes
 test/test_mediafile_edge.py    |  11 --------
 4 files changed, 4 insertions(+), 53 deletions(-)
 delete mode 100644 test/rsrc/only-magic-bytes.jpg

diff --git a/mediafile.py b/mediafile.py
index 5f488d7..5600c49 100644
--- a/mediafile.py
+++ b/mediafile.py
@@ -46,8 +46,8 @@
 import codecs
 import datetime
 import enum
+import filetype
 import functools
-import imghdr
 import logging
 import math
 import os
@@ -78,8 +78,6 @@
     'wav':  'WAVE',
 }
 
-PREFERRED_IMAGE_EXTENSIONS = {'jpeg': 'jpg'}
-
 
 # Exceptions.
 
@@ -346,52 +344,15 @@ def _sc_encode(gain, peak):
 
 
 # Cover art and other images.
-def _imghdr_what_wrapper(data):
-    """A wrapper around imghdr.what to account for jpeg files that can only be
-    identified as such using their magic bytes
-    See #1545
-    See https://github.com/file/file/blob/master/magic/Magdir/jpeg#L12
-    """
-    # imghdr.what returns none for jpegs with only the magic bytes, so
-    # _wider_test_jpeg is run in that case. It still returns None if it didn't
-    # match such a jpeg file.
-    return imghdr.what(None, h=data) or _wider_test_jpeg(data)
-
-
-def _wider_test_jpeg(data):
-    """Test for a jpeg file following the UNIX file implementation which
-    uses the magic bytes rather than just looking for the bytes that
-    represent 'JFIF' or 'EXIF' at a fixed position.
-    """
-    if data[:2] == b'\xff\xd8':
-        return 'jpeg'
-
 
 def image_mime_type(data):
     """Return the MIME type of the image data (a bytestring).
     """
-    # This checks for a jpeg file with only the magic bytes (unrecognized by
-    # imghdr.what). imghdr.what returns none for that type of file, so
-    # _wider_test_jpeg is run in that case. It still returns None if it didn't
-    # match such a jpeg file.
-    kind = _imghdr_what_wrapper(data)
-    if kind in ['gif', 'jpeg', 'png', 'tiff', 'bmp']:
-        return 'image/{0}'.format(kind)
-    elif kind == 'pgm':
-        return 'image/x-portable-graymap'
-    elif kind == 'pbm':
-        return 'image/x-portable-bitmap'
-    elif kind == 'ppm':
-        return 'image/x-portable-pixmap'
-    elif kind == 'xbm':
-        return 'image/x-xbitmap'
-    else:
-        return 'image/x-{0}'.format(kind)
+    return filetype.guess_mime(data)
 
 
 def image_extension(data):
-    ext = _imghdr_what_wrapper(data)
-    return PREFERRED_IMAGE_EXTENSIONS.get(ext, ext)
+    return filetype.guess_extension(data)
 
 
 class ImageType(enum.Enum):
diff --git a/pyproject.toml b/pyproject.toml
index f558885..1f08159 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -10,6 +10,7 @@ home-page = "https://github.com/beetbox/mediafile"
 description-file = "README.rst"
 requires = [
     "mutagen>=1.46",
+    "filetype>=1.2.0",
 ]
 requires-python = ">=3.7"
 classifiers = [
diff --git a/test/rsrc/only-magic-bytes.jpg b/test/rsrc/only-magic-bytes.jpg
deleted file mode 100644
index ca0cf220fd40788729bd33b867ff33ae84706a3f..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 622
zcmex=<NpH&Mn*;k4<Katf1AOX;XkDSXb2;vX8b?EAjrYM%)rFVD9FGh$jB_n`2Prl
zJOcwGE6}wd2SWiP6Eh1d8#@Ol7dKGBRsjYkMrLLv7G_pf78aoDTA(}wiy*6zqM;+3
za9|?4QlW@Z<HUs=%1#>(iUwW$pkka<)WpdpCN3c<rK+Z`p{ZqJYG!U>Y31zV>gMj@
z=@lFj8WtWA8I_!pnwFlCnN?g;T2@|BS=HRq+ScCD*)?hMl&RCE&zL!D(c&dbmn~nh
za@D5ITefc7zGLUELx+zXJ$C%W$y1juU%7hi`i+~n9zJ^f<mt2LFJ8X-`04YPuiw7^
z`1uRuFGdDth_~QAL-Q9f5Ez+QSeRMZLH=T7DhFa_K^9g;LpCADK=#B!VI`wR4iTq`
z3pXC*R5lL!AevNkkxNX)<e}<Ekk`OIBhF(@Wcdv4F@(QvG4L=m0wa%Ekini|QiJ{f
Fn*cYfIUN81

diff --git a/test/test_mediafile_edge.py b/test/test_mediafile_edge.py
index 337c610..4796525 100644
--- a/test/test_mediafile_edge.py
+++ b/test/test_mediafile_edge.py
@@ -78,17 +78,6 @@ def test_old_ape_version_bitrate(self):
         f = mediafile.MediaFile(media_file)
         self.assertEqual(f.bitrate, 0)
 
-    def test_only_magic_bytes_jpeg(self):
-        # Some jpeg files can only be recognized by their magic bytes and as
-        # such aren't recognized by imghdr. Ensure that this still works thanks
-        # to our own follow up mimetype detection based on
-        # https://github.com/file/file/blob/master/magic/Magdir/jpeg#L12
-        magic_bytes_file = os.path.join(_common.RSRC, b'only-magic-bytes.jpg')
-        with open(magic_bytes_file, 'rb') as f:
-            jpg_data = f.read()
-        self.assertEqual(
-            mediafile._imghdr_what_wrapper(jpg_data), 'jpeg')
-
     def test_soundcheck_non_ascii(self):
         # Make sure we don't crash when the iTunes SoundCheck field contains
         # non-ASCII binary data.

From 52b5e23d4f910182396048ed8280b77058c2f688 Mon Sep 17 00:00:00 2001
From: Jack Wilsdon <jack@wilsdon.me>
Date: Sun, 3 Nov 2024 17:50:18 +0000
Subject: [PATCH 2/3] Use "tiff" extension for TIFF files to match imghdr

---
 mediafile.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/mediafile.py b/mediafile.py
index 5600c49..8a8e835 100644
--- a/mediafile.py
+++ b/mediafile.py
@@ -352,7 +352,9 @@ def image_mime_type(data):
 
 
 def image_extension(data):
-    return filetype.guess_extension(data)
+    ext = filetype.guess_extension(data)
+    # imghdr returned "tiff", so we should keep returning it with filetype.
+    return ext if ext != 'tif' else 'tiff'
 
 
 class ImageType(enum.Enum):

From 182f3dd9f28dda8be5499c82badf8cc3bbc4fc64 Mon Sep 17 00:00:00 2001
From: Jack Wilsdon <jack@wilsdon.me>
Date: Sun, 3 Nov 2024 17:52:19 +0000
Subject: [PATCH 3/3] Add changelog entry

---
 docs/index.rst | 1 +
 1 file changed, 1 insertion(+)

diff --git a/docs/index.rst b/docs/index.rst
index f9b892d..c31655b 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -141,6 +141,7 @@ v0.13.0
 - Add a mapping compatible with Plex and ffmpeg for the "original date"
   fields.
 - Remove an unnecessary dependency on `six`.
+- Replace `imghdr` with `filetype` to support Python 3.13.
 
 v0.12.0
 '''''''