suggestions from review

Author: isabelizimm

docs/source/advancedusage/custom_handler.md

@@ -1,41 +1,45 @@
 # Custom Handlers
 
-There are two different ways that vetiver supports flexible handling for models that do not work automatically with the vetiver framework. The first way is with [new model types,](#new-model-type) where there is no current implementation for the type of model you would like to deploy. The second way is when you would like to implement a current handler, but [in a different way](#different-model-implementation). In either case, you *must* make a custom handler from the base `VetiverHandler`. A minimal custom handler could look like the following:
+There are two different ways that vetiver supports flexible handling for models that do not work automatically with the vetiver framework. The first way is with new model types where there is no current implementation for the type of model you would like to deploy. The second way is when you would like to implement a current handler, but in a different way. In either case, you should create a custom handler from vetiver's `BaseHandler()`. At a minimum, you must give the type of your model via `model_type` how predictions should be made, via the method `handler_predict()`. Then, initialize your handler with your model, and pass the object into `VetiverModel`.
+
+This example shows a custom handler of `newmodeltype` type.
 
 ```python
-from vetiver.handlers.base import VetiverHandler
+from vetiver.handlers.base import BaseHandler
 
-class SampleCustomHandler(VetiverHandler):
+class CustomHandler(BaseHandler):
     def __init__(model, ptype_data):
         super().__init__(model, ptype_data)
 
-    def handler_predict(self, input_data, check_ptype):
+    model_type = staticmethod(lambda: newmodeltype)
+
+    def handler_predict(self, input_data, check_ptype: bool):
         """
-        handler_predict defines how to make predictions from your model
+        Generates method for /predict endpoint in VetiverAPI
+
+        The `handler_predict` function executes at each API call. Use this
+        function for calling `predict()` and any other tasks that must be executed at each API call.
+
+        Parameters
+        ----------
+        input_data:
+            Test data
+        check_ptype: bool
+            Whether the ptype should be enforced
+
+        Returns
+        -------
+        prediction
+            Prediction from model
         """
         # your code here
-```
+        prediction = model.fancy_new_predict(input_data)
 
-## New model type
-If your model type is not supported by vetiver, you should create and then register the handler using [single dispatch](https://docs.python.org/3/library/functools.html#functools.singledispatch). Once the new type is registered, you are able to use `VetiverModel()` as normal. Here is a template for such a function:
-
-```python
-from vetiver.handlers._interface import create_handler
+        return prediction
 
-@create_handler.register
-def _(model: {_model_type}, ptype_data):
-    return SampleCustomHandler(model, ptype_data)
+new_model = CustomHandler(model, ptype_data)
 
-VetiverModel(your_model, "your_model")
+VetiverModel(new_model, "integer_model")
 ```
 
 If your model is a common type, please consider [submitting a pull request](https://github.com/rstudio/vetiver-python/pulls).
-
-## Different model implementation
-If your model's prediction function is different than vetiver's, you should create a custom handler with a `handler_predict` method to make predictions. Then, initialize your handler with your model, and pass the object into `VetiverModel`.
-
-```python
-new_model = SampleCustomHandler(your_model, your_ptype_data)
-
-VetiverModel(new_model, "your_model")
-```

vetiver/__init__.py

@@ -14,7 +14,7 @@
 from .meta import *  # noqa
 from .write_docker import write_docker  # noqa
 from .write_fastapi import write_app  # noqa
-from .handlers.base import VetiverHandler, create_handler, InvalidModelError  # noqa
+from .handlers.base import BaseHandler, create_handler, InvalidModelError  # noqa
 from .handlers.sklearn import SKLearnHandler  # noqa
 from .handlers.torch import TorchHandler  # noqa
 from .rsconnect import deploy_rsconnect  # noqa

vetiver/handlers/base.py

@@ -53,7 +53,10 @@ def create_handler(model, ptype_data):
     )
 
 
-class VetiverHandler:
+# BaseHandler uses create_handler to register subclasses based on model_class
+
+
+class BaseHandler:
     """Base handler class for creating VetiverModel of different type.
 
     Parameters
@@ -68,11 +71,7 @@ class VetiverHandler:
     def __init_subclass__(cls, **kwargs):
         super().__init_subclass__(**kwargs)
         with suppress(AttributeError, NameError):
-            create_handler.register(cls.base_class(), cls)
-
-    # def __new__(cls, value=None):
-    #     implementation_cls = create_handler.registry[type(value)]
-    #     return super().__new__(implementation_cls)
+            create_handler.register(cls.model_class(), cls)
 
     def __init__(self, model, ptype_data):
         self.model = model
@@ -140,8 +139,12 @@ def handler_predict(self, input_data, check_ptype):
         ...
 
 
+# BaseHandler for subclassing, Handler for new model types
+Handler = BaseHandler
+
+
 @create_handler.register
-def _(model: base.VetiverHandler, ptype_data):
+def _(model: base.BaseHandler, ptype_data):
     if model.ptype_data is None and ptype_data is not None:
         model.ptype_data = ptype_data
 

vetiver/handlers/sklearn.py

@@ -2,10 +2,10 @@
 import sklearn
 
 from ..meta import _model_meta
-from .base import VetiverHandler
+from .base import BaseHandler
 
 
-class SKLearnHandler(VetiverHandler):
+class SKLearnHandler(BaseHandler):
     """Handler class for creating VetiverModels with sklearn.
 
     Parameters
@@ -14,7 +14,7 @@ class SKLearnHandler(VetiverHandler):
         a trained sklearn model
     """
 
-    base_class = staticmethod(lambda: sklearn.base.BaseEstimator)
+    model_class = staticmethod(lambda: sklearn.base.BaseEstimator)
 
     def __init__(self, model, ptype_data):
         super().__init__(model, ptype_data)

vetiver/handlers/torch.py

@@ -1,7 +1,7 @@
 import numpy as np
 
 from ..meta import _model_meta
-from .base import VetiverHandler
+from .base import BaseHandler
 
 torch_exists = True
 try:
@@ -10,7 +10,7 @@
     torch_exists = False
 
 
-class TorchHandler(VetiverHandler):
+class TorchHandler(BaseHandler):
     """Handler class for creating VetiverModels with torch.
 
     Parameters
@@ -19,7 +19,7 @@ class TorchHandler(VetiverHandler):
         a trained torch model
     """
 
-    base_class = staticmethod(lambda: torch.nn.Module)
+    model_class = staticmethod(lambda: torch.nn.Module)
 
     def __init__(self, model, ptype_data):
         super().__init__(model, ptype_data)

vetiver/tests/test_custom_handler.py

@@ -2,13 +2,15 @@
 import pydantic
 import pandas as pd
 
-from vetiver import mock, VetiverModel, VetiverHandler
+from vetiver import mock, VetiverModel, BaseHandler
 
 
-class CustomHandler(VetiverHandler):
+class CustomHandler(BaseHandler):
     def __init__(self, model, ptype_data):
         super().__init__(model, ptype_data)
 
+    model_type = staticmethod(lambda: sklearn.dummy.DummyRegressor)
+
     def handler_predict(self, input_data, check_ptype):
         if check_ptype is True:
             if isinstance(input_data, pd.DataFrame):

vetiver/tests/test_pytorch.py

@@ -1,12 +1,12 @@
-from vetiver.vetiver_model import VetiverModel
-from vetiver import VetiverAPI
-from fastapi.testclient import TestClient
 import pytest
-import numpy as np
 
-torch = pytest.importorskip("torch")
+torch = pytest.importorskip("torch", reason="torch library not installed")
 
-import torch  # noqa
+import numpy as np  # noqa
+from fastapi.testclient import TestClient  # noqa
+
+from vetiver.vetiver_model import VetiverModel  # noqa
+from vetiver import VetiverAPI  # noqa
 
 
 def _build_torch_v():