Fix for XMP metadata handling #69

Author: bigcat88

pillow_heif/misc.py

@@ -51,14 +51,27 @@ def set_orientation(info: dict, orientation: int = 1) -> Union[int, None]:
         except Exception:  # noqa # pylint: disable=broad-except
             pass
     if info.get("xmp", None):
-        xmp_data = info["xmp"].decode("utf-8")
-        match = re.search(r'tiff:Orientation(="|>)([0-9])', xmp_data)
-        if match:
-            if original_orientation is None and int(match[2]) != 1:
-                original_orientation = int(match[2])
-            xmp_data = re.sub(r'tiff:Orientation="([0-9])"', "", xmp_data)
-            xmp_data = re.sub(r"<tiff:Orientation>([0-9])</tiff:Orientation>", "", xmp_data)
-            info["xmp"] = xmp_data.encode("utf-8")
+        xmp_data = info["xmp"].rsplit(b"\x00", 1)
+        if xmp_data[0]:
+            decoded_xmp_data = None
+            for encoding in ("utf-8", "latin1"):
+                try:
+                    decoded_xmp_data = xmp_data[0].decode(encoding)
+                    break
+                except Exception:  # noqa # pylint: disable=broad-except
+                    pass
+            if decoded_xmp_data:
+                _original_orientation = 1
+                match = re.search(r'tiff:Orientation(="|>)([0-9])', decoded_xmp_data)
+                if match:
+                    _original_orientation = int(match[2])
+                    if original_orientation is None and _original_orientation != 1:
+                        original_orientation = _original_orientation
+                    decoded_xmp_data = re.sub(r'tiff:Orientation="([0-9])"', "", decoded_xmp_data)
+                    decoded_xmp_data = re.sub(r"<tiff:Orientation>([0-9])</tiff:Orientation>", "", decoded_xmp_data)
+                # should encode in "utf-8" anyway, as `defusedxml` do not work with `latin1` encoding.
+                if encoding != "utf-8" or _original_orientation != 1:
+                    info["xmp"] = b"".join([decoded_xmp_data.encode("utf-8"), b"\x00" if len(xmp_data) > 1 else b""])
     return original_orientation
 
 

pyproject.toml

@@ -31,7 +31,8 @@ profile = "black"
 master.py-version = "3.7"
 master.extension-pkg-allow-list = ["_pillow_heif_cffi"]
 design.max-attributes = 9
-design.max-locals = 16
+design.max-branches = 14
+design.max-locals = 18
 design.max-returns = 8
 basic.good-names = [
     "a", "b", "c", "d", "e", "f", "i", "j", "k", "v",

tests/images/heif_other/xmp_latin1.heic

@@ -52,15 +52,15 @@ def test_pillow_xmp_add_remove():
     im.save(out_heif, format="HEIF", xmp=xmp_data)
     assert "xmp" not in im.info or im.info["xmp"] is None
     im_heif = Image.open(out_heif)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     # filling `info["xmp"]` before save.
     im.info["xmp"] = xmp_data
     im.save(out_heif, format="HEIF")
     im_heif = Image.open(out_heif)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     # setting `xmp` to `None` during save.
     im_heif.save(out_heif_no_xmp, format="HEIF", xmp=None)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     im_heif_no_xmp = Image.open(out_heif_no_xmp)
     assert "xmp" not in im_heif_no_xmp.info or im_heif_no_xmp.info["xmp"] is None
     # filling `info["xmp"]` with `None` before save.
@@ -85,16 +85,16 @@ def test_heif_xmp_add_remove():
     im_heif.save(out_heif, xmp=xmp_data)
     assert "xmp" not in im_heif.info or im_heif.info["xmp"] is None
     im_heif = pillow_heif.open_heif(out_heif)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     # test filling `info["xmp"]` before save.
     im_heif = pillow_heif.from_pillow(Image.new("RGB", (15, 15), 0))
     im_heif.info["xmp"] = xmp_data
     im_heif.save(out_heif)
     im_heif = pillow_heif.open_heif(out_heif)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     # setting `xmp` to `None` during save.
     im_heif.save(out_heif_no_xmp, xmp=None)
-    assert im_heif.info["xmp"]
+    assert im_heif.info["xmp"] == xmp_data
     im_heif_no_xmp = pillow_heif.open_heif(out_heif_no_xmp)
     assert "xmp" not in im_heif_no_xmp.info or im_heif_no_xmp.info["xmp"] is None
     # filling `info["xmp"]` with `None` before save.
@@ -107,3 +107,13 @@ def test_heif_xmp_add_remove():
     im_heif.save(out_heif_no_xmp)
     im_heif_no_xmp = pillow_heif.open_heif(out_heif_no_xmp)
     assert "xmp" not in im_heif_no_xmp.info or im_heif_no_xmp.info["xmp"] is None
+
+
+@pytest.mark.skipif(not helpers.hevc_enc(), reason="Requires HEVC encoder.")
+def test_heif_xmp_latin1_with_zero_byte():
+    im = Image.open("images/heif_other/xmp_latin1.heic")
+    out_heif = BytesIO()
+    im.save(out_heif, format="HEIF")
+    out_im = Image.open(out_heif)
+    assert im.getxmp() == out_im.getxmp()  # noqa
+    assert out_im.info["xmp"][-1] == 0  # check null byte not to get lost