Merge branch 'avif'

Merges PR #399

14 files changed, 197 insertions, 2 deletions

diff --git a/server/szurubooru/func/image_hash.py b/server/szurubooru/func/image_hash.py
index fc7d141..8d1cd2a 100644
--- a/server/szurubooru/func/image_hash.py
+++ b/server/szurubooru/func/image_hash.py
@@ -6,6 +6,10 @@ from typing import Any, Callable, List, Optional, Set, Tuple
 
 import numpy as np
 from PIL import Image
+import pillow_avif
+import pyheif
+from pyheif_pillow_opener import register_heif_opener
+register_heif_opener()
 
 from szurubooru import config, errors
 
diff --git a/server/szurubooru/func/images.py b/server/szurubooru/func/images.py
index 6413ac8..101bba8 100644
--- a/server/szurubooru/func/images.py
+++ b/server/szurubooru/func/images.py
@@ -4,7 +4,9 @@ import math
 import re
 import shlex
 import subprocess
+from io import BytesIO
 from typing import List
+from PIL import Image as PILImage
 
 from szurubooru import errors
 from szurubooru.func import mime, util
@@ -12,6 +14,13 @@ from szurubooru.func import mime, util
 logger = logging.getLogger(__name__)
 
 
+def convert_heif_to_png(content: bytes) -> bytes:
+    img = PILImage.open(BytesIO(content))
+    img_byte_arr = BytesIO()
+    img.save(img_byte_arr, format='PNG')
+    return img_byte_arr.getvalue()
+
+
 class Image:
     def __init__(self, content: bytes) -> None:
         self.content = content
@@ -252,7 +261,12 @@ class Image:
         ignore_error_if_data: bool = False,
         get_logs: bool = False,
     ) -> bytes:
-        extension = mime.get_extension(mime.get_mime_type(self.content))
+        mime_type = mime.get_mime_type(self.content)
+        if mime.is_heif(mime_type):
+            # FFmpeg does not support HEIF.
+            # https://trac.ffmpeg.org/ticket/6521
+            self.content = convert_heif_to_png(self.content)
+        extension = mime.get_extension(mime_type)
         assert extension
         with util.create_temp_file(suffix="." + extension) as handle:
             handle.write(self.content)
diff --git a/server/szurubooru/func/mime.py b/server/szurubooru/func/mime.py
index e521c25..93c096b 100644
--- a/server/szurubooru/func/mime.py
+++ b/server/szurubooru/func/mime.py
@@ -24,6 +24,15 @@ def get_mime_type(content: bytes) -> str:
     if content[0:2] == b"BM":
         return "image/bmp"
 
+    if content[4:12] in (b"ftypavif", b"ftypavis"):
+        return "image/avif"
+
+    if content[4:12] == b"ftypmif1":
+        return "image/heif"
+
+    if content[4:12] in (b"ftypheic", b"ftypheix"):
+        return "image/heic"
+
     if content[0:4] == b"\x1A\x45\xDF\xA3":
         return "video/webm"
 
@@ -41,6 +50,9 @@ def get_extension(mime_type: str) -> Optional[str]:
         "image/png": "png",
         "image/webp": "webp",
         "image/bmp": "bmp",
+        "image/avif": "avif",
+        "image/heif": "heif",
+        "image/heic": "heic",
         "video/mp4": "mp4",
         "video/webm": "webm",
         "application/octet-stream": "dat",
@@ -63,6 +75,9 @@ def is_image(mime_type: str) -> bool:
         "image/gif",
         "image/webp",
         "image/bmp",
+        "image/avif",
+        "image/heif",
+        "image/heic",
     )
 
 
@@ -72,3 +87,10 @@ def is_animated_gif(content: bytes) -> bool:
         get_mime_type(content) == "image/gif"
         and len(re.findall(pattern, content)) > 1
     )
+
+def is_heif(mime_type: str) -> bool:
+    return mime_type.lower() in (
+        "image/heif",
+        "image/heic",
+        "image/avif",
+    )
diff --git a/server/szurubooru/func/posts.py b/server/szurubooru/func/posts.py
index 0493681..be2259c 100644
--- a/server/szurubooru/func/posts.py
+++ b/server/szurubooru/func/posts.py
@@ -650,7 +650,8 @@ def update_post_content(post: model.Post, content: Optional[bytes]) -> None:
         image = images.Image(content)
         post.canvas_width = image.width
         post.canvas_height = image.height
-    except errors.ProcessingError:
+    except errors.ProcessingError as ex:
+        logger.exception(ex)
         if not config.config["allow_broken_uploads"]:
             raise InvalidPostContentError("Unable to process image metadata")
         else:
diff --git a/server/szurubooru/tests/assets/avif-avis.avif b/server/szurubooru/tests/assets/avif-avis.avif
new file mode 100644
index 0000000..3b27f29
--- /dev/null
+++ b/server/szurubooru/tests/assets/avif-avis.avif
diff --git a/server/szurubooru/tests/assets/avif-similar.avif b/server/szurubooru/tests/assets/avif-similar.avif
new file mode 100644
index 0000000..476f3f5
--- /dev/null
+++ b/server/szurubooru/tests/assets/avif-similar.avif
diff --git a/server/szurubooru/tests/assets/avif.avif b/server/szurubooru/tests/assets/avif.avif
new file mode 100644
index 0000000..edf4960
--- /dev/null
+++ b/server/szurubooru/tests/assets/avif.avif
diff --git a/server/szurubooru/tests/assets/heic-heix.heic b/server/szurubooru/tests/assets/heic-heix.heic
new file mode 100644
index 0000000..27d1268
--- /dev/null
+++ b/server/szurubooru/tests/assets/heic-heix.heic
diff --git a/server/szurubooru/tests/assets/heic.heic b/server/szurubooru/tests/assets/heic.heic
new file mode 100644
index 0000000..f1d2ed0
--- /dev/null
+++ b/server/szurubooru/tests/assets/heic.heic
diff --git a/server/szurubooru/tests/assets/heif-similar.heif b/server/szurubooru/tests/assets/heif-similar.heif
new file mode 100644
index 0000000..54385f2
--- /dev/null
+++ b/server/szurubooru/tests/assets/heif-similar.heif
diff --git a/server/szurubooru/tests/assets/heif.heif b/server/szurubooru/tests/assets/heif.heif
new file mode 100644
index 0000000..7fec72b
--- /dev/null
+++ b/server/szurubooru/tests/assets/heif.heif
diff --git a/server/szurubooru/tests/func/test_image_hash.py b/server/szurubooru/tests/func/test_image_hash.py
index e7028b6..5a5dc71 100644
--- a/server/szurubooru/tests/func/test_image_hash.py
+++ b/server/szurubooru/tests/func/test_image_hash.py
@@ -27,3 +27,53 @@ def test_signature_functions(read_asset, config_injector):
     words2 = image_hash.generate_words(sig2)
     words_match = sum(word1 == word2 for word1, word2 in zip(words1, words2))
     assert words_match == 18
+
+
+def test_signature_heif(read_asset, config_injector):
+    sig1 = image_hash.generate_signature(read_asset("heif.heif"))
+    sig2 = image_hash.generate_signature(read_asset("heif-similar.heif"))
+
+    sig1_repacked = image_hash.unpack_signature(
+        image_hash.pack_signature(sig1)
+    )
+    sig2_repacked = image_hash.unpack_signature(
+        image_hash.pack_signature(sig2)
+    )
+    assert array_equal(sig1, sig1_repacked)
+    assert array_equal(sig2, sig2_repacked)
+
+    dist1 = image_hash.normalized_distance([sig1], sig2)
+    assert abs(dist1[0] - 0.136777724290135) < 1e-8
+
+    dist2 = image_hash.normalized_distance([sig2], sig2)
+    assert abs(dist2[0]) < 1e-8
+
+    words1 = image_hash.generate_words(sig1)
+    words2 = image_hash.generate_words(sig2)
+    words_match = sum(word1 == word2 for word1, word2 in zip(words1, words2))
+    assert words_match == 43
+
+
+def test_signature_avif(read_asset, config_injector):
+    sig1 = image_hash.generate_signature(read_asset("avif.avif"))
+    sig2 = image_hash.generate_signature(read_asset("avif-similar.avif"))
+
+    sig1_repacked = image_hash.unpack_signature(
+        image_hash.pack_signature(sig1)
+    )
+    sig2_repacked = image_hash.unpack_signature(
+        image_hash.pack_signature(sig2)
+    )
+    assert array_equal(sig1, sig1_repacked)
+    assert array_equal(sig2, sig2_repacked)
+
+    dist1 = image_hash.normalized_distance([sig1], sig2)
+    assert abs(dist1[0] - 0.22628712858355998) < 1e-8
+
+    dist2 = image_hash.normalized_distance([sig2], sig2)
+    assert abs(dist2[0]) < 1e-8
+
+    words1 = image_hash.generate_words(sig1)
+    words2 = image_hash.generate_words(sig2)
+    words_match = sum(word1 == word2 for word1, word2 in zip(words1, words2))
+    assert words_match == 12
diff --git a/server/szurubooru/tests/func/test_mime.py b/server/szurubooru/tests/func/test_mime.py
index 62fd4a7..b33746b 100644
--- a/server/szurubooru/tests/func/test_mime.py
+++ b/server/szurubooru/tests/func/test_mime.py
@@ -14,6 +14,11 @@ from szurubooru.func import mime
         ("gif.gif", "image/gif"),
         ("webp.webp", "image/webp"),
         ("bmp.bmp", "image/bmp"),
+        ("avif.avif", "image/avif"),
+        ("avif-avis.avif", "image/avif"),
+        ("heif.heif", "image/heif"),
+        ("heic.heic", "image/heic"),
+        ("heic-heix.heic", "image/heic"),
         ("text.txt", "application/octet-stream"),
     ],
 )
@@ -36,6 +41,9 @@ def test_get_mime_type_for_empty_file():
         ("image/gif", "gif"),
         ("image/webp", "webp"),
         ("image/bmp", "bmp"),
+        ("image/avif", "avif"),
+        ("image/heif", "heif"),
+        ("image/heic", "heic"),
         ("application/octet-stream", "dat"),
     ],
 )
@@ -78,10 +86,16 @@ def test_is_video(input_mime_type, expected_state):
         ("image/png", True),
         ("image/jpeg", True),
         ("image/bmp", True),
+        ("image/avif", True),
+        ("image/heic", True),
+        ("image/heif", True),
         ("IMAGE/GIF", True),
         ("IMAGE/PNG", True),
         ("IMAGE/JPEG", True),
         ("IMAGE/BMP", True),
+        ("IMAGE/AVIF", True),
+        ("IMAGE/HEIC", True),
+        ("IMAGE/HEIF", True),
         ("image/anything_else", False),
         ("not an image", False),
     ],
@@ -99,3 +113,26 @@ def test_is_image(input_mime_type, expected_state):
 )
 def test_is_animated_gif(read_asset, input_path, expected_state):
     assert mime.is_animated_gif(read_asset(input_path)) == expected_state
+
+
+@pytest.mark.parametrize(
+    "input_mime_type,expected_state",
+    [
+        ("image/gif", False),
+        ("image/png", False),
+        ("image/jpeg", False),
+        ("image/avif", True),
+        ("image/heic", True),
+        ("image/heif", True),
+        ("IMAGE/GIF", False),
+        ("IMAGE/PNG", False),
+        ("IMAGE/JPEG", False),
+        ("IMAGE/AVIF", True),
+        ("IMAGE/HEIC", True),
+        ("IMAGE/HEIF", True),
+        ("image/anything_else", False),
+        ("not an image", False),
+    ],
+)
+def test_is_heif(input_mime_type, expected_state):
+    assert mime.is_heif(input_mime_type) == expected_state
diff --git a/server/szurubooru/tests/func/test_posts.py b/server/szurubooru/tests/func/test_posts.py
index 0693741..fa1b3bb 100644
--- a/server/szurubooru/tests/func/test_posts.py
+++ b/server/szurubooru/tests/func/test_posts.py
@@ -401,6 +401,41 @@ def test_update_post_source_with_too_long_string():
         ),
         (
             False,
+            "avif.avif",
+            "image/avif",
+            model.Post.TYPE_IMAGE,
+            "1_244c8840887984c4.avif",
+        ),
+        (
+            False,
+            "avif-avis.avif",
+            "image/avif",
+            model.Post.TYPE_IMAGE,
+            "1_244c8840887984c4.avif",
+        ),
+        (
+            False,
+            "heic.heic",
+            "image/heic",
+            model.Post.TYPE_IMAGE,
+            "1_244c8840887984c4.heic",
+        ),
+        (
+            False,
+            "heic-heix.heic",
+            "image/heic",
+            model.Post.TYPE_IMAGE,
+            "1_244c8840887984c4.heic",
+        ),
+        (
+            False,
+            "heif.heif",
+            "image/heif",
+            model.Post.TYPE_IMAGE,
+            "1_244c8840887984c4.heif",
+        ),
+        (
+            False,
             "gif-animated.gif",
             "image/gif",
             model.Post.TYPE_ANIMATION,
@@ -706,6 +741,38 @@ def test_update_post_content_leaving_custom_thumbnail(
     assert os.path.exists(generated_path)
 
 
+@pytest.mark.parametrize("filename", ("avif.avif", "heic.heic", "heif.heif"))
+def test_update_post_content_convert_heif_to_png_when_processing(
+    tmpdir, config_injector, read_asset, post_factory, filename
+):
+    config_injector(
+        {
+            "data_dir": str(tmpdir.mkdir("data")),
+            "thumbnails": {
+                "post_width": 300,
+                "post_height": 300,
+            },
+            "secret": "test",
+            "allow_broken_uploads": False,
+        }
+    )
+    post = post_factory(id=1)
+    db.session.add(post)
+    posts.update_post_content(post, read_asset(filename))
+    posts.update_post_thumbnail(post, read_asset(filename))
+    db.session.flush()
+    generated_path = (
+        "{}/data/generated-thumbnails/".format(tmpdir)
+        + "1_244c8840887984c4.jpg"
+    )
+    source_path = (
+        "{}/data/posts/custom-thumbnails/".format(tmpdir)
+        + "1_244c8840887984c4.dat"
+    )
+    assert os.path.exists(source_path)
+    assert os.path.exists(generated_path)
+
+
 def test_update_post_tags(tag_factory):
     post = model.Post()
     with patch("szurubooru.func.tags.get_or_create_tags_by_names"):