Handle XMP data from an UNDEFINED TIFF tag (#8997)

Author: hugovk

Tests/test_file_tiff.py

@@ -14,6 +14,7 @@
     ImageFile,
     JpegImagePlugin,
     TiffImagePlugin,
+    TiffTags,
     UnidentifiedImageError,
 )
 from PIL.TiffImagePlugin import RESOLUTION_UNIT, X_RESOLUTION, Y_RESOLUTION
@@ -900,6 +901,29 @@ def test_getxmp(self) -> None:
                 assert description[0]["format"] == "image/tiff"
                 assert description[3]["BitsPerSample"]["Seq"]["li"] == ["8", "8", "8"]
 
+    def test_getxmp_undefined(self, tmp_path: Path) -> None:
+        tmpfile = tmp_path / "temp.tif"
+        im = Image.new("L", (1, 1))
+        ifd = TiffImagePlugin.ImageFileDirectory_v2()
+        ifd.tagtype[700] = TiffTags.UNDEFINED
+        with Image.open("Tests/images/lab.tif") as im_xmp:
+            ifd[700] = im_xmp.info["xmp"]
+        im.save(tmpfile, tiffinfo=ifd)
+
+        with Image.open(tmpfile) as im_reloaded:
+            if ElementTree is None:
+                with pytest.warns(
+                    UserWarning,
+                    match="XMP data cannot be read without defusedxml dependency",
+                ):
+                    assert im_reloaded.getxmp() == {}
+            else:
+                assert "xmp" in im_reloaded.info
+                xmp = im_reloaded.getxmp()
+
+                description = xmp["xmpmeta"]["RDF"]["Description"]
+                assert description[0]["format"] == "image/tiff"
+
     def test_get_photoshop_blocks(self) -> None:
         with Image.open("Tests/images/lab.tif") as im:
             assert isinstance(im, TiffImagePlugin.TiffImageFile)

Tests/test_image.py

@@ -995,7 +995,7 @@ def test_getxmp_padded(self) -> None:
         im = Image.new("RGB", (1, 1))
         im.info["xmp"] = (
             b'<?xpacket begin="\xef\xbb\xbf" id="W5M0MpCehiHzreSzNTczkc9d"?>\n'
-            b'<x:xmpmeta xmlns:x="adobe:ns:meta/" />\n<?xpacket end="w"?>\x00\x00'
+            b'<x:xmpmeta xmlns:x="adobe:ns:meta/" />\n<?xpacket end="w"?>\x00\x00 '
         )
         if ElementTree is None:
             with pytest.warns(

src/PIL/Image.py

@@ -1511,7 +1511,7 @@ def get_value(element: Element) -> str | dict[str, Any] | None:
             return {}
         if "xmp" not in self.info:
             return {}
-        root = ElementTree.fromstring(self.info["xmp"].rstrip(b"\x00"))
+        root = ElementTree.fromstring(self.info["xmp"].rstrip(b"\x00 "))
         return {get_name(root.tag): get_value(root)}
 
     def getexif(self) -> Exif:

src/PIL/TiffImagePlugin.py

@@ -1260,7 +1260,10 @@ def _seek(self, frame: int) -> None:
         self.fp.seek(self._frame_pos[frame])
         self.tag_v2.load(self.fp)
         if XMP in self.tag_v2:
-            self.info["xmp"] = self.tag_v2[XMP]
+            xmp = self.tag_v2[XMP]
+            if isinstance(xmp, tuple) and len(xmp) == 1:
+                xmp = xmp[0]
+            self.info["xmp"] = xmp
         elif "xmp" in self.info:
             del self.info["xmp"]
         self._reload_exif()