System Config (#99)

* Add coverage level gate
* Add config parser
* Add config processing utility
* Refactor Encoder --> CategoricalEncoder
* Refactor SmartText --> TextEncoder
* Condense config files into setup.cfg until pyproject.toml is supported
* Remove .travis.yml
* Remove wily from deps
* Remove coveralls and patch pre-commit installation steps
* Address pytest deprecation warning
* Fix config parser docstring and newsfragment
* Patch failing tests and add reset config cache functionality

.pre-commit-config.yaml

@@ -1,6 +1,6 @@
 repos:
 -   repo: https://github.com/asottile/seed-isort-config
-    rev: v1.9.1
+    rev: v1.9.2
     hooks:
     -   id: seed-isort-config
 -   repo: https://github.com/pre-commit/mirrors-isort
@@ -17,11 +17,3 @@ repos:
     rev: v2.2.3
     hooks:
     - id: flake8
--   repo: local
-    hooks:
-    -   id: wily
-        name: wily
-        entry: printf %s "y\ny\n" | wily diff
-        verbose: true
-        language: python
-        additional_dependencies: [wily]

azure-pipelines.yml

@@ -24,7 +24,7 @@ stages:
 
   # Build
   # Linux
-  # TODO: Turn off Linux Testing until https://github.com/sdispater/poetry/issues/1197 is resolved
+  # # scs install is broken on Linux
   # - job: test_linux
   #   dependsOn: lint
   #   pool:
@@ -76,8 +76,8 @@ stages:
 #         brew install gcc@5
 #       displayName: Install Brew packages
 #     - template: ci/install.yml
-#     - script: $HOME/.poetry/bin/poetry build
-#         $HOME/.poetry/bin/poetry publish -u $(pypi_username) -p $(pypi_password)
+#     - script: poetry build
+#         poetry publish -u $(pypi_username) -p $(pypi_password)
 #       displayName: Release to pypi
 #     - task: GitHubRelease@0
 #       inputs:

ci/install.yml

@@ -3,17 +3,23 @@ steps:
     displayName: 'Use Python $(python.version)'
     inputs:
       versionSpec: '$(python.version)'
+    name: python3
   - script: |
       pip install --upgrade pip
       curl -sSL https://raw.githubusercontent.com/sdispater/poetry/master/get-poetry.py | python
     displayName: Install Poetry
-  - script: $HOME/.poetry/bin/poetry install -vvv -E dev
-    condition: eq( variables['Agent.OS'], 'Linux' )
-    displayName: Install all packages (Linux)
-  # TODO: Fix manual patch for pyrfr after https://github.com/automl/auto-sklearn/issues/688 is closed
+  # Unix
+  - bash: |
+      echo "##vso[task.prependpath]$HOME/.poetry/bin"
+    condition: or(eq( variables['Agent.OS'], 'Linux' ), eq( variables['Agent.OS'], 'Darwin' ))
+    displayName: Add Poetry to path (Unix)
   - script: |
-      $HOME/.poetry/bin/poetry install -vvv
-      export CC=gcc-5; export CXX=g++-5; $HOME/.poetry/bin/poetry install -vvv -E dev
-      export CC=gcc-5; export CXX=g++-5; pip install --upgrade pyrfr
-    condition: eq( variables['Agent.OS'], 'Darwin' )
-    displayName: Install all packages (MacOS)
+      poetry --version
+      pip freeze --all
+      pip --version
+      poetry config settings.virtualenvs.create false
+      poetry install -vvv
+      poetry install -vvv
+    displayName: Install all packages
+  # TODO: Fix manual patch for pyrfr after https://github.com/automl/auto-sklearn/issues/688 is closed
+  # TODO: Temporary patch to make CI pass, install poetry twice to mask setuptools error https://github.com/sdispater/poetry/issues/1197 is resolved

ci/lint.yml

@@ -4,6 +4,6 @@ steps:
     versionSpec: '3.6'
 - script: |
     pip install pre-commit
-    pre-commit install-hooks
+    pre-commit install -f --install-hooks
     pre-commit run --all-files
   displayName: Running pre-commit on all files

ci/tests.yml

@@ -1,14 +1,8 @@
 steps: 
-  - script: $HOME/.poetry/bin/poetry run pytest --junitxml=junit/test-results.xml --cov-report=xml
+  - script: poetry run pytest --junitxml=junit/test-results.xml --cov-report=xml
     env:
       FORESHADOW_TESTS: ALL
     displayName: Run pytest
-  # TODO: Switch back to 'Linux' when https://github.com/sdispater/poetry/issues/1197 is resolved
-  - script: $HOME/.poetry/bin/poetry run coveralls
-    condition: and(succeeded(), eq( variables['Agent.OS'], 'Darwin' ))
-    env:
-      COVERALLS_REPO_TOKEN: $(coveralls_token)
-    displayName: Submit coveralls report
   - task: PublishTestResults@2
     inputs:
       testResultsFiles: 'junit/test-results.xml'
@@ -18,4 +12,11 @@ steps:
     inputs:
       codeCoverageTool: Cobertura
       summaryFileLocation: '$(System.DefaultWorkingDirectory)/**/coverage.xml'
-      reportDirectory: '$(System.DefaultWorkingDirectory)/**/htmlcov'
+      reportDirectory: '$(System.DefaultWorkingDirectory)/**/htmlcov'
+  - task: BuildQualityChecks@6
+    inputs:
+      checkCoverage: true
+      coverageFailOption: 'fixed'
+      coverageType: 'branches'
+      coverageThreshold: '90'
+    displayName: 'Check build quality'

foreshadow/config.py

@@ -0,0 +1,109 @@
+"""Foreshadow system config resolver."""
+
+import os
+
+import yaml
+
+from foreshadow.utils import get_config_path, get_transformer
+
+
+CONFIG_FILE_NAME = "config.yml"
+
+DEFAULT_CONFIG = {
+    "cleaner": [],
+    "engineerer": {},
+    "preprocessor": {
+        "numerical": ["Imputer", "Scaler"],
+        "categorical": ["CategoricalEncoder"],
+        "text": ["TextEncoder"],
+    },
+    "reducer": {},
+}
+
+_cfg = {}
+
+
+def get_config(base):
+    """Try to load configuration data from specific folder path.
+
+    Args:
+        base (str): A base path that has a file called `config.yml`
+
+    Returns:
+        dict: If the file does not exist an empty dictionary is returned.
+
+    """
+    data_file_path = os.path.join(base, CONFIG_FILE_NAME)
+    check_file = os.path.exists(data_file_path) and os.path.isfile(
+        data_file_path
+    )
+
+    if not check_file:
+        return {}
+    else:
+        with open(data_file_path) as fopen:
+            data = yaml.safe_load(fopen)
+            if data is None:
+                return {}
+            else:
+                return data
+
+
+def reset_config():
+    """Reset internal configuration
+
+    Note:
+        This is useful in an IDLE setting when the configuration file might
+        have been modified but you don't want to reload the system.
+
+    """
+    global _cfg
+    _cfg = {}
+
+
+def resolve_config():
+    """Resolve the configuration to actual classes.
+
+    Note:
+        The order is resolution is as follows in increasing precedence order:
+        framework, user, local.
+
+    Returns:
+        A dictionary with the same keys as `foreshadow.config.DEFAULT_CONFIG`
+        with the correct overrides.
+
+    """
+    default = DEFAULT_CONFIG
+    user = get_config(get_config_path())
+    local_path = os.path.abspath("")
+    local = get_config(local_path)
+
+    # import pdb; pdb.set_trace()
+
+    global _cfg
+    if local_path in _cfg:
+        return _cfg.get(local_path)
+
+    # Expand the dictionaries in order of precedence
+    _resolved = {**default, **user, **local}
+
+    resolved = {}
+    for step, data in _resolved.items():
+        if not len(data):
+            resolved[step] = data
+        elif isinstance(data, list):
+            resolved[step] = [
+                get_transformer(transformer) for transformer in data
+            ]
+        elif isinstance(data, dict):
+            resolved[step] = {
+                intent: [
+                    get_transformer(transformer)
+                    for transformer in transformer_list
+                ]
+                for intent, transformer_list in data.items()
+            }
+
+    _cfg[local_path] = resolved
+
+    return resolved

foreshadow/console.py

@@ -14,7 +14,7 @@
 from foreshadow.estimators.auto import determine_problem_type
 
 
-def generate_model(args):
+def generate_model(args):  # noqa: C901
     """Process command line args and generate a Foreshadow model to fit.
 
     Args:

foreshadow/core/serializers.py

@@ -87,13 +87,14 @@ def _make_deserializable(data):
         return data
 
 
-def _pickle_cache_path(self, cache_path=None):
+def _pickle_cache_path(cls_name, cache_path=None):
     """Get the pickle cache path of a transformer.
 
     Uses a generated UUID and the class name to come up with a unique
     filename.
 
     Args:
+        cls_name (str): Name of the class to be pickled
         cache_path (str, optional): override the default cache_path which
             is in the root of the user's directory.
 
@@ -105,7 +106,7 @@ def _pickle_cache_path(self, cache_path=None):
     if cache_path is None:
         cache_path = get_cache_path()
 
-    fname = self.__class__.__name__ + uuid.uuid4().hex
+    fname = cls_name + uuid.uuid4().hex
     fpath = "{}.pkl".format(fname)
     path = os.path.join(cache_path, fpath)
 
@@ -325,7 +326,7 @@ def disk_serialize(self, cache_path=None):
             str: The path the data was saved to.
 
         """
-        fpath = _pickle_cache_path(cache_path)
+        fpath = _pickle_cache_path(self.__class__.__name__, cache_path)
         with open(fpath, "wb+") as fopen:
             pickle.dump(self, fopen, protocol=pickle.HIGHEST_PROTOCOL)
 
@@ -434,7 +435,7 @@ def dict_serialize(self, deep=False):
 
         Note:
             This recursively serializes the individual steps to facilitate a
-            human readabel form.
+            human readable form.
 
         Args:
             deep (bool): If True, will return the parameters for this estimator

foreshadow/intents/general.py

@@ -8,11 +8,11 @@
 from foreshadow.intents.base import BaseIntent, PipelineTemplateEntry
 from foreshadow.transformers.concrete import DropFeature
 from foreshadow.transformers.smart import (
-    Encoder,
+    CategoricalEncoder,
     MultiImputer,
     Scaler,
     SimpleImputer,
-    SmartText,
+    TextEncoder,
 )
 
 
@@ -198,7 +198,7 @@ class CategoricalIntent(GenericIntent):
 
     single_pipeline_template = [
         PipelineTemplateEntry("dropper", DropFeature, False),
-        PipelineTemplateEntry("impute_encode", Encoder, True),
+        PipelineTemplateEntry("impute_encode", CategoricalEncoder, True),
     ]
     """Encode the column automatically"""
 
@@ -248,7 +248,7 @@ class TextIntent(GenericIntent):
     """No children"""
 
     single_pipeline_template = [
-        PipelineTemplateEntry("text", SmartText, False)
+        PipelineTemplateEntry("text", TextEncoder, False)
     ]
     """Encodes the column automatically"""