correct EXIF handling + tests #92

Signed-off-by: bigcat88 <[email protected]>

Author: bigcat88

pillow_heif/misc.py

@@ -63,7 +63,9 @@ def set_orientation(info: dict) -> Optional[int]:
     if info.get("exif", None):
         try:
             tif_tag = info["exif"]
+            skipped_exif00 = False
             if tif_tag.startswith(b"Exif\x00\x00"):
+                skipped_exif00 = True
                 tif_tag = tif_tag[6:]
             endian_mark = "<" if tif_tag[0:2] == b"\x49\x49" else ">"
             pointer = unpack(endian_mark + "L", tif_tag[4:8])[0]
@@ -77,7 +79,9 @@ def set_orientation(info: dict) -> Optional[int]:
                 _original_orientation = unpack(endian_mark + "H", value[0:2])[0]
                 if _original_orientation != 1:
                     original_orientation = _original_orientation
-                    p_value = 6 + pointer + 8
+                    p_value = pointer + 8
+                    if skipped_exif00:
+                        p_value += 6
                     new_orientation = pack(endian_mark + "H", 1)
                     info["exif"] = info["exif"][:p_value] + new_orientation + info["exif"][p_value + 2 :]
                     break
@@ -154,10 +158,11 @@ def _retrieve_exif(metadata: List[dict]) -> Optional[bytes]:
     for i, md_block in enumerate(metadata):
         if md_block["type"] == "Exif":
             _purge.append(i)
-            if md_block["data"][:4] == b"\x00\x00\x00\n":  # Xiaomi EXIF start
-                _data = md_block["data"][8:]  # skip `\0\0\0\n` +  TIFF header -> total 8 bytes
-            else:
-                _data = md_block["data"][4:]  # skip TIFF header, first 4 bytes
+            skip_size = int.from_bytes(md_block["data"][:4], byteorder="big", signed=False)
+            skip_size += 4  # size of skip offset itself
+            if len(md_block["data"]) - skip_size <= 4:  # bad EXIF data, skip first 4 bytes
+                skip_size = 4
+            _data = md_block["data"][skip_size:]
             if not _result and _data:
                 _result = _data
     for i in reversed(_purge):

pyproject.toml

@@ -30,8 +30,8 @@ profile = "black"
 master.py-version = "3.7"
 master.extension-pkg-allow-list = ["_pillow_heif"]
 design.max-attributes = 8
-design.max-branches = 14
-design.max-locals = 16
+design.max-branches = 16
+design.max-locals = 18
 design.max-returns = 8
 basic.good-names = [
     "a", "b", "c", "d", "e", "f", "i", "j", "k", "v",

tests/metadata_exif_test.py

@@ -196,3 +196,27 @@ def test_xiaomi_exif():
     im = Image.open("images/heif_special/xiaomi.heic")
     im.getexif()
     im.load()
+
+
+@pytest.mark.skipif(not helpers.hevc_enc(), reason="Requires HEVC encoder.")
+def test_data_before_exif():
+    exif = Image.Exif()
+    exif_bytes = exif.tobytes()
+    exif_full_data = b"hidden data " + exif_bytes
+    out_im = BytesIO()
+    helpers.gradient_rgb().save(out_im, format="HEIF", exif=exif_full_data)
+    im = Image.open(out_im)
+    out_im.seek(0)
+    assert out_im.read().find(b"hidden data ") != -1  # checking that this was saved
+    assert im.getexif() == exif
+    assert im.info["exif"] == exif_bytes[6:]  # skipping b`Exif\x00\x00` that `exif.tobytes()` returns.
+
+
+@pytest.mark.skipif(not helpers.hevc_enc(), reason="Requires HEVC encoder.")
+def test_empty_exif():
+    exif = Image.Exif()
+    out_im = BytesIO()
+    helpers.gradient_rgb().save(out_im, format="HEIF", exif=exif.tobytes())
+    im = Image.open(out_im)
+    assert im.getexif() == exif
+    assert im.info["exif"] == exif.tobytes()[6:]