feat: Detect duplicated backgrounds and icons (#66)

pyproject.toml

@@ -14,6 +14,7 @@ dependencies = [
     "pyyaml>=6.0.2",
     "requests>=2.32.3",
     "typer>=0.15.1",
+    "xxhash>=3.6.0",
 ]
 
 [build-system]

toolbox/commands/verify.py

@@ -8,6 +8,7 @@
 from jsonschema import ValidationError, validate
 
 from toolbox.utils.config import EventConfig, LabelsConfig, RegistrationConfig
+from toolbox.utils.hash import hash_file
 from toolbox.utils.tasks import find_tasks
 
 app = typer.Typer()
@@ -89,6 +90,8 @@ def tasks(context: typer.Context):
 
     valid_count = 0
     invalid_count = 0
+    tasks_icons = {}
+    tasks_backgrounds = {}
 
     rich.print("[dim]Validating tasks...")
     for subdir_path in find_tasks(tasks_directory):
@@ -119,7 +122,7 @@ def tasks(context: typer.Context):
             invalid_count += 1
             continue
 
-        if not verify_pictures(subdir_path / "pictures"):
+        if not verify_pictures(subdir_path / "pictures", task_id, tasks_icons, tasks_backgrounds):
             invalid_count += 1
             continue
 
@@ -136,6 +139,13 @@ def tasks(context: typer.Context):
     rich.print(f"\nFinished validating all tasks: {total_tasks} tasks processed.")
     rich.print(f"[green]{valid_count} tasks are valid.")
 
+    for task_list in tasks_icons.values():
+        if len(task_list) > 1:
+            rich.print("[yellow]Following tasks have the same icons: " + ", ".join(task_list))
+    for task_list in tasks_backgrounds.values():
+        if len(task_list) > 1:
+            rich.print("[yellow]Following tasks have the same backgrounds: " + ", ".join(task_list))
+
     if invalid_count > 0:
         rich.print(f"[red]{invalid_count} tasks are invalid.")
         raise Exit(code=1)
@@ -167,12 +177,21 @@ def verify_assets(yaml_data: dict, assets_path: Path, subdir_path: Path) -> bool
     return True
 
 
-def verify_pictures(subdir_path: Path) -> bool:
+def verify_pictures(
+    subdir_path: Path, task_id: str, tasks_icons: dict[str, list[str]], tasks_backgrounds: dict[str, list[str]]
+) -> bool:
     required_pictures = ["background.png", "icon.png"]
     for picture in required_pictures:
         picture_path = subdir_path.joinpath(picture)
         if not picture_path.is_file():
             rich.print(f"[red]Missing file pictures/{picture} for {subdir_path}")
             return False
+        picture_hash = hash_file(picture_path)
+        if picture == "icon.png":
+            tasks_icons[picture_hash] = tasks_icons.get(picture_hash, []) + [task_id]
+        elif picture == "background.png":
+            tasks_backgrounds[picture_hash] = tasks_backgrounds.get(picture_hash, []) + [task_id]
+        else:
+            rich.print("[red]Unsupported file type for duplicate detection: " + picture)
 
     return True

toolbox/tests/test_verify.py

@@ -247,6 +247,43 @@ def test_verify_invalid_difficulty(
         tasks(mock_context)
 
 
+@patch("rich.print")
+@patch("toolbox.commands.verify.hash_file")
+@patch.object(Path, "iterdir")
+@patch.object(Path, "is_dir")
+@patch.object(Path, "is_file")
+@patch.object(Path, "read_text")
+def test_verify_duplicated_pictures(
+    mock_read_text,
+    mock_is_file,
+    mock_is_dir,
+    mock_iterdir,
+    mock_hash_file,
+    mock_rich_print,
+    mock_context,
+    valid_schema,
+    valid_task_config,
+    valid_labels_config,
+):
+    second_task = valid_task_config.copy()
+    second_task["id"] = "second_task"
+    mock_iterdir.side_effect = [[Path("valid_task"), Path("second_task")], [], []]
+    mock_is_dir.return_value = True
+    mock_is_file.return_value = True
+    mock_read_text.side_effect = [
+        yaml.dump(valid_labels_config),
+        json.dumps(valid_schema),
+        yaml.dump(valid_task_config),
+        yaml.dump(second_task),
+    ]
+    mock_hash_file.return_value = "duplicate_hash"
+
+    tasks(mock_context)
+
+    mock_rich_print.assert_any_call("[yellow]Following tasks have the same icons: valid_task, second_task")
+    mock_rich_print.assert_any_call("[yellow]Following tasks have the same backgrounds: valid_task, second_task")
+
+
 @patch.object(Path, "iterdir")
 @patch.object(Path, "is_file")
 @patch.object(Path, "is_dir")
@@ -360,15 +397,29 @@ def test_labels_missing_icons(mock_iterdir, mock_read_text, mock_context, valid_
         labels(mock_context)
 
 
+@patch("toolbox.commands.verify.hash_file")
+@patch.object(Path, "is_file")
+def test_valid_verify_pictures(mock_is_file, _mock_hash_file):
+    mock_is_file.return_value = True
+
+    assert verify_pictures(Path("assets"), "", {}, {}) is True
+
+
+@patch("toolbox.commands.verify.hash_file")
 @patch.object(Path, "is_file")
-def test_valid_verify_pictures(mock_is_file):
+def test_valid_verify_pictures_duplicated(mock_is_file, mock_hash_file):
     mock_is_file.return_value = True
+    mock_hash_file.return_value = "duplicate_hash"
+    tasks_icons = {"duplicate_hash": ["41"]}
+    tasks_backgrounds = {"duplicate_hash": ["UwU"]}
 
-    assert verify_pictures(Path("assets")) is True
+    assert verify_pictures(Path("assets"), "67", tasks_icons, tasks_backgrounds) is True
+    assert tasks_icons["duplicate_hash"] == ["41", "67"]
+    assert tasks_backgrounds["duplicate_hash"] == ["UwU", "67"]
 
 
 @patch.object(Path, "is_file")
 def test_missing_verify_pictures(mock_is_file):
     mock_is_file.return_value = False
 
-    assert verify_pictures(Path("assets")) is False
+    assert verify_pictures(Path("assets"), "", {}, {}) is False

toolbox/utils/hash.py

@@ -0,0 +1,13 @@
+from pathlib import Path
+
+from xxhash import xxh64
+
+
+def hash_file(path: Path, chunk_size=8192):
+    buffer = xxh64()
+
+    with path.open("rb") as file:
+        while chunk := file.read(chunk_size):
+            buffer.update(chunk)
+
+    return buffer.hexdigest()