Export AutoModerator from main package for cleaner imports

README.md

@@ -49,7 +49,7 @@ Run a model in a few lines.
 
 Python API:
 ```python
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 # Load from the Hugging Face Hub (e.g., NSFW image classifier)
 moderator = AutoModerator.from_pretrained("viddexa/nsfw-mini")
@@ -101,7 +101,7 @@ JSON shape (CLI output):
 Tip (Python):
 ```python
 from dataclasses import asdict
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 moderator = AutoModerator.from_pretrained("viddexa/nsfw-mini")
 result = moderator("/path/to/image.jpg")

examples/README.md

@@ -18,10 +18,10 @@ uv add moderators
 
 ## Quickstart (Image, Python)
 ```python
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 # NSFW image classification model from the Hub
-moderator = AutoModerator.from_pretrained("viddexa/nsfw-mini")
+moderator = AutoModerator.from_pretrained("viddexa/nsfw-detector-mini")
 
 # Run on a local image path
 result = moderator("/path/to/image.jpg")
@@ -30,7 +30,7 @@ print(result)
 
 ## Quickstart (Image, CLI)
 ```bash
-moderators viddexa/nsfw-mini /path/to/image.jpg
+moderators viddexa/nsfw-detector-mini /path/to/image.jpg
 ```
 
 Tip: Add `--local-files-only` to force offline usage if the files are already cached.
@@ -40,10 +40,10 @@ Process a directory of images and print the top result per file.
 
 ```python
 from pathlib import Path
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 images_dir = Path("/path/to/images")
-model = AutoModerator.from_pretrained("viddexa/nsfw-mini")
+model = AutoModerator.from_pretrained("viddexa/nsfw-detector-mini")
 
 for img_path in images_dir.glob("**/*"):
     if img_path.suffix.lower() in {".jpg", ".jpeg", ".png", ".webp", ".bmp", ".gif", ".avif"}:
@@ -56,7 +56,7 @@ You can also load a text classifier.
 
 Python:
 ```python
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 text_model = AutoModerator.from_pretrained("distilbert/distilbert-base-uncased-finetuned-sst-2-english")
 print(text_model("I love this!"))
@@ -78,13 +78,13 @@ python examples/benchmarks.py <model_id> <image_path> [--warmup N] [--repeats N]
 Examples:
 ```bash
 # Default backend (auto-detected)
-python examples/benchmarks.py viddexa/nsfw-mini /path/to/image.jpg --warmup 3 --repeats 20
+python examples/benchmarks.py viddexa/nsfw-detector-mini /path/to/image.jpg --warmup 3 --repeats 20
 
 ```
 
 Expected output (sample):
 ```
-Model: viddexa/nsfw-mini
+Model: viddexa/nsfw-detector-mini
 Backend: auto
 Runs: 20, avg: 12.34 ms, p50: 11.80 ms, p90: 14.10 ms
 ```

examples/benchmarks.py

@@ -4,7 +4,7 @@
 import statistics
 import time
 
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 
 def summarize(times: list[float]) -> dict[str, float]:

pyproject.toml

@@ -43,10 +43,10 @@ packages = ["src/moderators"]
 
 [tool.ruff]
 line-length = 120
-target-version = "py310"
+target-version = "py39"
 
 [tool.ruff.lint]
-extend-select = ["I", "D", "UP"]
+extend-select = ["F", "I", "D", "UP", "RUF", "FA"]
 ignore = [
     "D100",  # Missing docstring in public module
     "D104",  # Missing docstring in public package
@@ -58,6 +58,9 @@ ignore = [
     "D406",  # Section name should end with a newline
     "D407",  # Missing dashed underline after section
     "D413",  # Missing blank line after last section
+    "RUF001", # Ambiguous Unicode characters in strings
+    "RUF002", # Ambiguous Unicode characters in docstrings
+    "RUF012", # Mutable default values in class attributes
 ]
 
 [tool.ruff.lint.per-file-ignores]

src/moderators/__init__.py

@@ -1 +1,5 @@
-__version__ = "0.1.0"
+__version__ = "0.1.1"
+
+from .auto_model import AutoModerator
+
+__all__ = ["AutoModerator"]

src/moderators/cli.py

@@ -3,7 +3,7 @@
 import json
 from dataclasses import asdict, is_dataclass
 
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 
 def _to_jsonable(obj):

src/moderators/utils/__init__.py

@@ -4,8 +4,8 @@
 
 __all__ = [
     "auto_install",
-    "ensure_transformers",
     "ensure_dl_framework",
     "ensure_pillow_for_task",
+    "ensure_transformers",
     "preprocess_image_input",
 ]

src/moderators/utils/deps.py

@@ -29,7 +29,7 @@ def auto_install(packages: list[str]) -> bool:
 def ensure_transformers(install_fn: Callable[[list[str]], bool]):
     """Ensure 'transformers' is importable; optionally auto-install and retry."""
     try:
-        import transformers as _transformers  # noqa: F401
+        import transformers as _transformers
 
         return _transformers
     except Exception:
@@ -48,7 +48,7 @@ def ensure_dl_framework(install_fn: Callable[[list[str]], bool]) -> str:
     Tries to auto-install torch first.
     """
     try:
-        import torch  # noqa: F401
+        import torch
 
         return "pt"
     except Exception:
@@ -81,7 +81,7 @@ def ensure_pillow_for_task(task: str, install_fn: Callable[[list[str]], bool]) -
     if "image" not in str(task).lower():
         return
     try:
-        import PIL  # noqa: F401
+        import PIL
     except Exception:
         if not install_fn(["Pillow"]):
             raise ImportError("This image task requires Pillow. Install with: uv pip install Pillow")

src/moderators/utils/image.py

@@ -44,8 +44,8 @@ def _process(obj: Any):
                 w, h = img.size
                 if w < min_side or h < min_side:
                     scale = max(min_side / w, min_side / h)
-                    new_w = int(round(w * scale))
-                    new_h = int(round(h * scale))
+                    new_w = round(w * scale)
+                    new_h = round(h * scale)
                     resample = getattr(getattr(Image, "Resampling", Image), "BILINEAR")
                     img = img.resize((new_w, new_h), resample)
             except Exception:

tests/test_auto_model.py

@@ -4,7 +4,7 @@
 
 import pytest
 
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 from moderators.integrations.base import PredictionResult
 from moderators.integrations.transformers_moderator import TransformersModerator
 

tests/test_push_to_hub.py

@@ -1,7 +1,7 @@
 import json
 from pathlib import Path
 
-from moderators.auto_model import AutoModerator
+from moderators import AutoModerator
 
 
 class _FakeTempDir:
@@ -14,7 +14,7 @@ def __enter__(self):
         return self.name
 
     def __exit__(self, exc_type, exc, tb):
-        # Klasörü bilerek silmiyoruz; test sonrasında pytest tmp cleanup zaten yapar.
+        # Intentionally not deleting folder; pytest tmp cleanup handles it after tests.
         return False
 
 
@@ -25,7 +25,7 @@ def write_config(tmp_path: Path, data: dict) -> Path:
 
 
 def test_push_to_hub_offline(tmp_path, monkeypatch, fake_transformers):
-    # Ağ ve ağır bağımlılıkları devre dışı bırak
+    # Disable network and heavy dependencies
     monkeypatch.setenv("MODERATORS_DISABLE_AUTO_INSTALL", "1")
     monkeypatch.setattr(
         "moderators.integrations.transformers_moderator.ensure_dl_framework",
@@ -56,35 +56,35 @@ def upload_folder(self, repo_id, repo_type, folder_path, commit_message=None, to
             }
             return {"ok": True}
 
-    # Doğru namespace: hub_mixin içindeki HfApi sembolünü patch'le
+    # Correct namespace: patch the HfApi symbol in hub_mixin
     monkeypatch.setattr(
         "huggingface_hub.hub_mixin.HfApi",
         lambda *a, **k: FakeApi(),
         raising=True,
     )
 
-    # Geçici klasörün silinmesini engelle: SoftTemporaryDirectory'yi patch'le
+    # Prevent deletion of temp folder: patch SoftTemporaryDirectory
     monkeypatch.setattr(
         "huggingface_hub.hub_mixin.SoftTemporaryDirectory",
         lambda *a, **k: _FakeTempDir(prefix="moderators_push_"),
         raising=True,
     )
 
-    # Yerel bir TransformersModerator konfigürasyonu ile yükle
+    # Load with a local TransformersModerator configuration
     model_dir = write_config(tmp_path, {"architecture": "TransformersModerator", "task": "text-classification"})
     mod = AutoModerator.from_pretrained(str(model_dir))
 
-    # push_to_hub çağrısı (ağ yok, FakeApi çalışacak)
+    # push_to_hub call (no network, FakeApi will run)
     repo_id = "user/repo-for-tests"
     mod.push_to_hub(repo_id, commit_message="test commit", token="fake-token")
 
-    # upload_folder çağrıldı mı?
+    # Was upload_folder called?
     assert "upload_folder" in calls
     up = calls["upload_folder"]
     folder_path = Path(up["folder_path"])
     assert folder_path.exists()
 
-    # Kaydedilen config.json doğrula
+    # Verify saved config.json
     cfg = json.loads((folder_path / "config.json").read_text(encoding="utf-8"))
     assert cfg.get("architecture") == "TransformersModerator"
     assert cfg.get("task") == "text-classification"