Merge pull request #166 from rstudio/pins-req

Author: isabelizimm

vetiver/__init__.py

@@ -7,7 +7,7 @@
 from .server import VetiverAPI, vetiver_endpoint, predict  # noqa
 from .mock import get_mock_data, get_mock_model  # noqa
 from .pin_read_write import vetiver_pin_write  # noqa
-from .attach_pkgs import load_pkgs  # noqa
+from .attach_pkgs import load_pkgs, get_board_pkgs  # noqa
 from .meta import VetiverMeta  # noqa
 from .write_docker import write_docker, prepare_docker  # noqa
 from .write_fastapi import write_app, vetiver_write_app  # noqa

vetiver/attach_pkgs.py

@@ -1,31 +1,33 @@
 import tempfile
 import os
+import warnings
+from typing import List
 from .vetiver_model import VetiverModel
 from .meta import VetiverMeta
 
 
 def load_pkgs(model: VetiverModel = None, packages: list = None, path=""):
     """Load packages necessary for predictions
 
-    Args
-    ----
-        model: VetiverModel
-            VetiverModel to extract packages from
-        packages: list
-            List of extra packages to include
-        path: str
-            Where to save output file
+    Parameters
+    ----------
+    model: VetiverModel
+        VetiverModel to extract packages from
+    packages: list
+        List of extra packages to include
+    path: str
+        Where to save output file
     """
 
     required_pkgs = ["vetiver"]
     if packages:
-        required_pkgs = list(set(required_pkgs + packages))
+        required_pkgs += packages
 
     if isinstance(model.metadata, dict):
         model.metadata = VetiverMeta.from_dict(model.metadata)
 
     if model.metadata.required_pkgs:
-        required_pkgs = list(set(required_pkgs + model.metadata.required_pkgs))
+        required_pkgs += model.metadata.required_pkgs
 
     tmp = tempfile.NamedTemporaryFile(suffix=".in", delete=False)
     tmp.close()
@@ -36,3 +38,37 @@ def load_pkgs(model: VetiverModel = None, packages: list = None, path=""):
 
     os.system(f"pip-compile {tmp.name} --output-file={path}requirements.txt")
     os.remove(tmp.name)
+
+
+def get_board_pkgs(board) -> List[str]:
+    """
+    Extract packages required for pin board authorization
+
+    Parameters
+    ----------
+    board:
+        A pin board, created by `pins.board_folder()` or another `board_` function.
+
+    Returns
+    --------
+    list[str]
+    """
+    prot = board.fs.protocol
+
+    if prot == "rsc":
+        return ["rsconnect-python"]
+    elif prot == "file":
+        return []
+    elif prot == ["s3", "s3a"]:
+        return ["s3fs"]
+    elif prot == "abfs":
+        return ["adlfs"]
+    elif prot == ("gcs", "gs"):
+        return ["gcsfs"]
+    else:
+        warnings.warn(
+            f"required packages unknown for board protocol: {prot}, "
+            "add to model's metadata to export",
+            UserWarning,
+        )
+        return []

vetiver/pin_read_write.py

@@ -43,7 +43,10 @@ def vetiver_pin_write(board, model: VetiverModel, versioned: bool = True):
     >>> vetiver.vetiver_pin_write(model_board, v)
     """
     if not board.allow_pickle_read:
-        raise NotImplementedError  # must be pickle-able
+        raise ValueError(
+            "board does not allow pickled models. Set "
+            "allow_pickle_read to True on board creation."
+        )
 
     inform(
         _log,

vetiver/tests/test_pin_write.py

@@ -16,7 +16,7 @@ def test_board_pin_write_error():
         model=model, prototype_data=X_df, model_name="model", versioned=None
     )
     board = pins.board_temp()
-    with pytest.raises(NotImplementedError):
+    with pytest.raises(ValueError):
         vetiver_pin_write(board=board, model=v)
 
 

vetiver/tests/test_prepare_docker.py

@@ -1,14 +1,19 @@
 import pytest
 import vetiver
+import pins
+from pathlib import Path
+from tempfile import TemporaryDirectory
 import pandas as pd
 import numpy as np
 
 DOCKER_URL = "http://0.0.0.0:8080/predict"
 
-pytestmark = pytest.mark.docker  # noqa
+# uses GitHub Actions to deploy model into Docker
+# see vetiver-python/script/setup-docker for files
 
 
-def test_predict_sklearn_df_check_ptype():
+@pytest.mark.docker
+def test_deployed_dockerfile():
     np.random.seed(500)
 
     X, y = vetiver.mock.get_mock_data()
@@ -17,3 +22,42 @@ def test_predict_sklearn_df_check_ptype():
     assert isinstance(response, pd.DataFrame), response
     assert response.iloc[0, 0] == 44.47
     assert len(response) == 100
+
+
+@pytest.fixture()
+def create_vetiver_model():
+    X, y = vetiver.get_mock_data()
+    model = vetiver.get_mock_model()
+
+    return vetiver.VetiverModel(model.fit(X, y), "model", prototype_data=X)
+
+
+@pytest.fixture(scope="module")
+def test_warning_if_no_protocol(create_vetiver_model):
+    with pytest.warns(UserWarning):
+        board = pins.board_temp(allow_pickle_read=True)
+        board.fs.protocol = "abc"
+
+        vetiver.get_board_pkgs(board)
+
+
+@pytest.mark.parametrize(
+    "prot,output",
+    [
+        (["s3", "s3a"], "s3fs"),
+        ("abfs", "adlfs"),
+        (("gcs", "gs"), "gcsfs"),
+    ],
+)
+@pytest.fixture(scope="module")
+def test_get_board_pkgs(prot, output, create_vetiver_model):
+    board = pins.board_temp(allow_pickle_read=True)
+    board.fs.protocol = prot
+
+    vetiver.vetiver_pin_write(board, create_vetiver_model)
+
+    with TemporaryDirectory() as tempdir:
+        vetiver.prepare_docker(board, "model", path=tempdir)
+        file = Path(tempdir, "vetiver_requirements.txt")
+        contents = open(file).read()
+        assert f"{output}==" in contents

vetiver/tests/test_write_app.py

@@ -1,38 +1,39 @@
-import os
 import pins
 import vetiver
+import pytest
+from tempfile import TemporaryDirectory
+from pathlib import Path
 
-# Load data, model
-X_df, y = vetiver.get_mock_data()
-model = vetiver.get_mock_model().fit(X_df, y)
 
+@pytest.fixture
+def vetiver_model_creation():
+    X_df, y = vetiver.get_mock_data()
+    model = vetiver.get_mock_model().fit(X_df, y)
+    return vetiver.VetiverModel(model, "model")
 
-def test_write_app():
-    file = "app.py"
-    v = vetiver.VetiverModel(
-        model=model, prototype_data=X_df, model_name="model", versioned=None
-    )
-    model_board = pins.board_folder(path=".", versioned=True, allow_pickle_read=True)
-    vetiver.vetiver_pin_write(board=model_board, model=v)
-    vetiver.write_app(model_board, "model", file="app.py")
-    contents = open(file).read()
-    os.remove(file)
-    version = model_board.pin_versions("model").sort_values(
-        by="created", ascending=False
-    )
-    version = version.version[0]
-    assert (
-        contents
-        == f"""from vetiver import VetiverModel
+
+def test_write_app(vetiver_model_creation):
+    with TemporaryDirectory() as tempdir:
+        file = Path(tempdir, "app.py")
+        model_board = pins.board_folder(path=tempdir, allow_pickle_read=True)
+        vetiver.vetiver_pin_write(model_board, vetiver_model_creation)
+        vetiver.write_app(model_board, "model", file=file)
+        contents = open(file).read()
+        version = model_board.pin_versions("model").sort_values(
+            by="created", ascending=False
+        )
+        version = version.version[0]
+        assert (
+            contents
+            == f"""from vetiver import VetiverModel
 import vetiver
 import pins
 
 
-b = pins.board_folder('.', allow_pickle_read=True)
-v = VetiverModel.from_pin(b, 'model', version = '{version}')
+b = pins.board_folder({repr(tempdir)}, allow_pickle_read=True)
+v = VetiverModel.from_pin(b, 'model', version = {repr(version)})
 
 vetiver_api = vetiver.VetiverAPI(v)
 api = vetiver_api.app
 """
-    )
-    model_board.pin_delete("model")
+        )

vetiver/vetiver_model.py

@@ -52,8 +52,6 @@ class VetiverModel:
     Parameter `ptype_data` was changed to `prototype_data`. Handling of `ptype_data`
     will be removed in a future version.
 
-
-
     Examples
     -------
     >>> from vetiver import mock, VetiverModel
@@ -104,15 +102,9 @@ def from_pin(cls, board, name: str, version: str = None):
             required_pkgs = meta.user.get("vetiver_meta").get("required_pkgs", None)
             python_version = meta.user.get("vetiver_meta").get("python_version", None)
             meta.user.pop("vetiver_meta")
+        # old pin type
         else:
-            # ptype = meta.user.get("ptype", None)
-
             get_prototype = meta.user.get("ptype")
-            # elif meta.user.get("prototype"):
-            #     get_prototype = meta.user.get("prototype")
-            # else:
-            #     get_prototype = None
-
             required_pkgs = meta.user.get("required_pkgs")
             python_version = meta.user.get("python_version")
 

vetiver/write_docker.py

@@ -3,7 +3,7 @@
 from pathlib import Path
 
 from .write_fastapi import write_app
-from .attach_pkgs import load_pkgs
+from .attach_pkgs import load_pkgs, get_board_pkgs
 from .vetiver_model import VetiverModel
 
 
@@ -137,11 +137,20 @@ def prepare_docker(
     >>> v = vetiver.VetiverModel(model, "my_model", prototype_data = X)
     >>> vetiver.vetiver_pin_write(board, v)
     >>> vetiver.prepare_docker(board = board, pin_name = "my_model", path = tmp.name)
+
+    Notes
+    ------
+    This function uses `vetiver.get_board_pkgs(board)` for generating requirements.
+    For more complex use cases, call `write_docker()`, `load_pkgs()`, and
+    `write_app()` individually.
     """
 
     v = VetiverModel.from_pin(board=board, name=pin_name, version=version)
+
     write_app(
         board=board, pin_name=pin_name, version=version, file=Path(path, "app.py")
     )
-    load_pkgs(v, path=Path(path, "vetiver_"))
+
+    load_pkgs(v, path=Path(path, "vetiver_"), packages=get_board_pkgs(board))
+
     write_docker(path=path, rspm_env=rspm_env, host=host, port=port)