Always keep ICC profile and orientation to avoid display issues; keep copyright and artist fields by default (#103)

Author: stefmolin

.github/release.yml

@@ -1,5 +1,5 @@
 changelog:
   exclude:
     authors:
-      - dependabot
-      - pre-commit-ci
+      - dependabot[bot]
+      - pre-commit-ci[bot]

src/exif_stripper/cli.py

@@ -43,17 +43,29 @@ def main(argv: Sequence[str] | None = None) -> int:
     exif_options_group.add_argument(
         '--fields',
         nargs='+',
-        choices=list(FieldGroup),
-        default=FieldGroup.ALL,
+        choices=[group for group in FieldGroup if group != FieldGroup.COPYRIGHT],
+        default=(FieldGroup.ALL,),
         help=(
-            'The fields to remove from the EXIF metadata. '
-            'By default, all EXIF metadata is removed.'
+            'The fields to remove from the EXIF metadata. By default, all EXIF metadata '
+            'that can safely be deleted is removed.'
+        ),
+    )
+    exif_options_group.add_argument(
+        '--remove-copyright',
+        action='store_true',
+        help=(
+            'Whether to remove the image copyright information. '
+            'By default, the artist and copyright tags will be preserved, if present.'
         ),
     )
 
     args = parser.parse_args(argv)
 
-    results = [process_image(filename, args.fields) for filename in args.filenames]
+    fields = args.fields
+    if args.remove_copyright:
+        fields = [*fields, FieldGroup.COPYRIGHT]
+
+    results = [process_image(filename, fields=fields) for filename in args.filenames]
     return int(any(results))
 
 

src/exif_stripper/fields.py

@@ -4,6 +4,18 @@
 
 from PIL import ExifTags
 
+PRESERVE_FIELDS: tuple[IntEnum] = (
+    ExifTags.Base.InterColorProfile,
+    ExifTags.Base.Orientation,
+)
+"""Locations of EXIF information that should be preserved."""
+
+OWNERSHIP_FIELDS: tuple[IntEnum] = (
+    ExifTags.Base.Artist,
+    ExifTags.Base.Copyright,
+)
+"""Locations of EXIF information related to copyright/ownership."""
+
 
 class FieldGroup(StrEnum):
     """Enum for groups of fields to target."""
@@ -14,6 +26,9 @@ class FieldGroup(StrEnum):
     CAMERA = auto()
     """Field group for EXIF tags containing the make and model of the camera."""
 
+    COPYRIGHT = auto()
+    """Field group for EXIF tags related to the creator of the image."""
+
     GPS = auto()
     """Field group for all GPS information in EXIF metadata."""
 
@@ -27,13 +42,14 @@ class FieldGroup(StrEnum):
 # References for EXIF tags:
 # - meanings: https://exiv2.org/tags.html
 # - enums: https://pillow.readthedocs.io/en/stable/_modules/PIL/ExifTags.html
-FIELDS: dict[FieldGroup, tuple[IntEnum]] = {
+EXIF_TAG_MAPPING: dict[FieldGroup, tuple[IntEnum]] = {
     FieldGroup.CAMERA: (
         ExifTags.Base.Make,
         ExifTags.Base.Model,
         ExifTags.Base.MakerNote,
         ExifTags.Base.MakerNoteSafety,
     ),
+    FieldGroup.COPYRIGHT: OWNERSHIP_FIELDS,
     FieldGroup.GPS: (ExifTags.IFD.GPSInfo,),
     FieldGroup.LENS: (ExifTags.Base.LensMake, ExifTags.Base.LensModel),
     FieldGroup.SERIALS: (

src/exif_stripper/processing.py

@@ -2,13 +2,15 @@
 
 from __future__ import annotations
 
+import itertools
 from contextlib import suppress
+from copy import deepcopy
 from typing import TYPE_CHECKING
 
 from PIL import Image, UnidentifiedImageError
 
 from .exceptions import UnknownFieldError
-from .fields import FIELDS, FieldGroup
+from .fields import EXIF_TAG_MAPPING, OWNERSHIP_FIELDS, PRESERVE_FIELDS, FieldGroup
 
 if TYPE_CHECKING:
     import os
@@ -40,22 +42,38 @@ def process_image(
         suppress(FileNotFoundError, UnidentifiedImageError),
         Image.open(filename) as image,
     ):
-        if exif := image.getexif():
-            if FieldGroup.ALL in fields:
-                exif.clear()
-                has_changed = True
-            else:
-                for field in fields:
-                    if locations := FIELDS.get(field):
-                        for location in locations:
-                            with suppress(KeyError):
-                                del exif[location]
-                                has_changed = True
-                    else:
-                        raise UnknownFieldError(field, FieldGroup)
-
-            if has_changed:
-                image.save(filename, exif=exif)
-                print(f'Stripped EXIF metadata from {filename}')
+        exif = image.getexif()
+
+        if FieldGroup.ALL in fields:
+            original_exif = deepcopy(exif)
+
+            fields_to_preserve = {
+                location: value
+                for location in itertools.chain(
+                    PRESERVE_FIELDS,
+                    OWNERSHIP_FIELDS if FieldGroup.COPYRIGHT not in fields else {},
+                )
+                if (value := exif.get(location))
+            }
+
+            exif.clear()
+
+            for field, value in fields_to_preserve.items():
+                exif[field] = value
+
+            has_changed = original_exif != exif
+        else:
+            for field in fields:
+                if locations := EXIF_TAG_MAPPING.get(field):
+                    for location in locations:
+                        with suppress(KeyError):
+                            del exif[location]
+                            has_changed = True
+                else:
+                    raise UnknownFieldError(field, FieldGroup)
+
+        if has_changed:
+            image.save(filename, exif=exif)
+            print(f'Stripped EXIF metadata from {filename}')
 
     return has_changed

tests/__init__.py

@@ -0,0 +1 @@
+"""Test suite for exif_stripper."""

tests/common.py

@@ -0,0 +1,32 @@
+"""Utility functions for testing."""
+
+from PIL import Image
+
+from exif_stripper.fields import (
+    EXIF_TAG_MAPPING,
+    OWNERSHIP_FIELDS,
+    PRESERVE_FIELDS,
+    FieldGroup,
+)
+
+
+def has_expected_metadata(filepath, fields, has_copyright, was_stripped) -> bool:
+    """Utility to check if a file has the expected metadata."""
+    with Image.open(filepath) as im:
+        exif = im.getexif()
+
+        preserved_fields_are_present = all(
+            tag in exif for tag in PRESERVE_FIELDS
+        ) and has_copyright == all(tag in exif for tag in OWNERSHIP_FIELDS)
+
+        if fields is None or FieldGroup.ALL in fields:
+            other_fields_present = all(
+                tag in exif
+                for field in set(FieldGroup).difference({FieldGroup.ALL})
+                for tag in EXIF_TAG_MAPPING[field]
+            )
+            return preserved_fields_are_present and was_stripped != other_fields_present
+
+        return preserved_fields_are_present and was_stripped != all(
+            tag in exif for field in fields for tag in EXIF_TAG_MAPPING[field]
+        )

tests/conftest.py

@@ -0,0 +1,60 @@
+"""Common test fixtures."""
+
+import pytest
+from PIL import ExifTags, Image
+
+
+@pytest.fixture
+def image_without_exif_data(tmp_path):
+    """Fixture for an image without EXIF data."""
+    image_without_exif_data = tmp_path / 'clean.png'
+    image_without_exif_data.touch()
+    return image_without_exif_data
+
+
+@pytest.fixture
+def image_with_exif_data(tmp_path):
+    """Fixture for an image with only the required EXIF data."""
+    image_file = tmp_path / 'test.png'
+    with Image.new(mode='1', size=(2, 2)) as im:
+        exif = im.getexif()
+
+        exif[ExifTags.Base.InterColorProfile] = 'some color profile'
+        exif[ExifTags.Base.Orientation] = 1
+
+        im.save(image_file, exif=exif)
+
+    return image_file
+
+
+@pytest.fixture
+def image_with_full_exif_data(image_with_exif_data):
+    """Fixture for an image with EXIF data."""
+    with Image.open(image_with_exif_data) as im:
+        exif = im.getexif()
+
+        exif[ExifTags.Base.CameraOwnerName] = 'Unknown'
+
+        exif[ExifTags.Base.Artist] = 'Stefanie Molin'
+        exif[ExifTags.Base.Copyright] = 'Copyright (c) Stefanie Molin.'
+
+        exif[ExifTags.IFD.GPSInfo] = {
+            ExifTags.GPS.GPSVersionID: 1,
+            ExifTags.GPS.GPSTrackRef: 1,
+        }
+
+        exif[ExifTags.Base.Make] = 'SomeCameraMake'
+        exif[ExifTags.Base.Model] = 'SomeCameraModel'
+        exif[ExifTags.Base.MakerNote] = 'Some maker notes'
+        exif[ExifTags.Base.MakerNoteSafety] = 1
+
+        exif[ExifTags.Base.LensMake] = 'SomeLensMake'
+        exif[ExifTags.Base.LensModel] = 'SomeLensModel'
+
+        exif[ExifTags.Base.BodySerialNumber] = 'ABC123'
+        exif[ExifTags.Base.CameraSerialNumber] = 'DEF456'
+        exif[ExifTags.Base.LensSerialNumber] = 'GHI789'
+
+        im.save(image_with_exif_data, exif=exif)
+
+    return image_with_exif_data