nextmv 0.28.4__py3-none-any.whl → 0.29.0.dev0__py3-none-any.whl

nextmv/input.py

@@ -27,6 +27,7 @@ import csv
 import json
 import os
 import sys
+from collections.abc import Callable
 from dataclasses import dataclass
 from enum import Enum
 from typing import Any, Optional, Union
@@ -58,6 +59,8 @@ class InputFormat(str, Enum):
         CSV format, utf-8 encoded.
     CSV_ARCHIVE : str
         CSV archive format: multiple CSV files.
+    MULTI_FILE : str
+        Multi-file format, used for loading multiple files in a single input.
     """
 
     JSON = "json"
@@ -68,6 +71,233 @@ class InputFormat(str, Enum):
     """CSV format, utf-8 encoded."""
     CSV_ARCHIVE = "csv-archive"
     """CSV archive format: multiple CSV files."""
+    MULTI_FILE = "multi-file"
+    """Multi-file format, used for loading multiple files in a single input."""
+
+
+@dataclass
+class DataFile:
+    """
+    Represents data to be read from a file.
+
+    You can import the `DataFile` class directly from `nextmv`:
+
+    ```python
+    from nextmv import DataFile
+    ```
+
+    This class is used to define data that will be read from a file in the
+    filesystem. It includes the name of the file, and the reader function that
+    will handle the loading, and deserialization of the data from the file.
+    This `DataFile` class is typically used in the `Input`, when the
+    `Input.input_format` is set to `InputFormat.MULTI_FILE`. Given that it is
+    difficul to handle every edge case of how data is deserialized, and read
+    from a file, this class exists so that the user can implement the `reader`
+    callable of their choice and provide it with any `reader_args` and
+    `reader_kwargs` they might need.
+
+    Parameters
+    ----------
+    name : str
+        Name of the data (input) file. The file extension should be included in
+        the name.
+    reader : Callable[[str], Any]
+        Callable that reads the data from the file. This should be a function
+        implemented by the user. There are convenience functions that you can
+        use as a reader as well. The `reader` must receive, at the very minimum,
+        the following arguments:
+
+        - `file_path`: a `str` argument which is the location where this
+          solution will be read from. This includes the dir and name of the
+          file. As such, the `name` parameter of this class is going to be
+          passed to the `reader` function, joined with the directory where the
+          file will be read from.
+
+        The `reader` can also receive additional arguments, and keyword
+        arguments. The `reader_args` and `reader_kwargs` parameters of this
+        class can be used to provide those additional arguments.
+
+        The `reader` function should return the data that will be used in the
+        model.
+    """
+
+    name: str
+    """
+    Name of the data (input) file. The file extension should be included in the
+    name.
+    """
+    loader: Callable[[str], Any]
+    """
+    Callable that reads (loads) the data from the file. This should be a function
+    implemented by the user. There are convenience functions that you can use
+    as a `loader` as well. The `loader` must receive, at the very minimum, the
+    following arguments:
+
+    - `file_path`: a `str` argument which is the location where this
+       solution will be read from. This includes the dir and name of the
+       file. As such, the `name` parameter of this class is going to be
+       passed to the `loader` function, joined with the directory where the
+       file will be read from.
+
+    The `loader` can also receive additional arguments, and keyword arguments.
+    The `loader_args` and `loader_kwargs` parameters of this class can be used
+    to provide those additional arguments.
+
+    The `loader` function should return the data that will be used in the model.
+    """
+    loader_kwargs: Optional[dict[str, Any]] = None
+    """
+    Optional keyword arguments to pass to the loader function. This can be used
+    to customize the behavior of the loader.
+    """
+    loader_args: Optional[list[Any]] = None
+    """
+    Optional positional arguments to pass to the loader function. This can be
+    used to customize the behavior of the loader.
+    """
+
+
+def json_data_file(name: str, json_configurations: Optional[dict[str, Any]] = None) -> DataFile:
+    """
+    This is a convenience function to create a `DataFile` that reads JSON data.
+
+    You can import the `json_data_file` function directly from `nextmv`:
+
+    ```python
+    from nextmv import json_data_file
+    ```
+
+    Parameters
+    ----------
+    name : str
+        Name of the data file. You don't need to include the `.json` extension.
+    json_configurations : dict[str, Any], optional
+        JSON-specific configurations for reading the data.
+
+    Returns
+    -------
+    DataFile
+        A `DataFile` instance that reads JSON data from a file with the given
+        name.
+
+    Examples
+    --------
+    >>> from nextmv import json_data_file
+    >>> data_file = json_data_file("my_data")
+    >>> data = data_file.read()
+    >>> print(data)
+    {
+        "key": "value",
+        "another_key": [1, 2, 3]
+    }
+    """
+
+    if not name.endswith(".json"):
+        name += ".json"
+
+    json_configurations = json_configurations or {}
+
+    def loader(file_path: str) -> Union[dict[str, Any], Any]:
+        with open(file_path, encoding="utf-8") as f:
+            return json.load(f, **json_configurations)
+
+    return DataFile(
+        name=name,
+        loader=loader,
+    )
+
+
+def csv_data_file(name: str, csv_configurations: Optional[dict[str, Any]] = None) -> DataFile:
+    """
+    This is a convenience function to create a `DataFile` that reads CSV data.
+
+    You can import the `csv_data_file` function directly from `nextmv`:
+
+    ```python
+    from nextmv import csv_data_file
+    ```
+
+    Parameters
+    ----------
+    name : str
+        Name of the data file. You don't need to include the `.csv` extension.
+    csv_configurations : dict[str, Any], optional
+        CSV-specific configurations for reading the data.
+
+    Returns
+    -------
+    DataFile
+        A `DataFile` instance that reads CSV data from a file with the given
+        name.
+
+    Examples
+    --------
+    >>> from nextmv import csv_data_file
+    >>> data_file = csv_data_file("my_data")
+    >>> data = data_file.read()
+    >>> print(data)
+    [
+        {"column1": "value1", "column2": "value2"},
+        {"column1": "value3", "column2": "value4"}
+    ]
+    """
+
+    if not name.endswith(".csv"):
+        name += ".csv"
+
+    csv_configurations = csv_configurations or {}
+
+    def loader(file_path: str) -> list[dict[str, Any]]:
+        with open(file_path, encoding="utf-8") as f:
+            return list(csv.DictReader(f, **csv_configurations))
+
+    return DataFile(
+        name=name,
+        loader=loader,
+    )
+
+
+def text_data_file(name: str) -> DataFile:
+    """
+    This is a convenience function to create a `DataFile` that reads utf-8
+    encoded text data.
+
+    You can import the `text_data_file` function directly from `nextmv`:
+
+    ```python
+    from nextmv import text_data_file
+    ```
+
+    You must provide the extension as part of the `name` parameter.
+
+    Parameters
+    ----------
+    name : str
+        Name of the data file. The file extension must be provided in the name.
+
+    Returns
+    -------
+    DataFile
+        A `DataFile` instance that reads text data from a file with the given
+        name.
+
+    Examples
+    --------
+    >>> from nextmv import text_data_file
+    >>> data_file = text_data_file("my_data")
+    >>> data = data_file.read()
+    >>> print(data)
+    This is some text data.
+    """
+
+    def loader(file_path: str) -> str:
+        with open(file_path, encoding="utf-8") as f:
+            return f.read().rstrip("\n")
+
+    return DataFile(
+        name=name,
+        loader=loader,
+    )
 
 
 @dataclass
@@ -83,12 +313,40 @@ class Input:
 
     Parameters
     ----------
-    data : Union[dict[str, Any], str, list[dict[str, Any]], dict[str, list[dict[str, Any]]]]
+    data : Union[Union[dict[str, Any], Any], str, list[dict[str, Any]],
+    dict[str, list[dict[str, Any]]], dict[str, Any]]
         The actual data.
     input_format : InputFormat, optional
         Format of the input data. Default is `InputFormat.JSON`.
     options : Options, optional
         Options that the input was created with.
+
+    Notes
+    -----
+    The `data`'s type must match the `input_format`:
+
+    - `InputFormat.JSON`: the data is `Union[dict[str, Any], Any]`. This just
+       means that the data must be JSON-deserializable, which includes dicts and
+       lists.
+    - `InputFormat.TEXT`: the data is `str`, and it must be utf-8 encoded.
+    - `InputFormat.CSV`: the data is `list[dict[str, Any]]`, where each dict
+       represents a row in the CSV.
+    - `InputFormat.CSV_ARCHIVE`: the data is `dict[str, list[dict[str, Any]]]`,
+       where each key is the name of a CSV file and the value is a list of dicts
+       representing the rows in that CSV file.
+    - `InputFormat.MULTI_FILE`: the data is `dict[str, Any]`, where for each
+       item, the key is the file name (with the extension) and the actual data
+       from the file is the value. When working with multi-file, data is loaded
+       from one or more files in a specific directory. Given that each file can
+       be of different types (JSON, CSV, Excel, etc...), the data captured from
+       each might vary. To reflect this, the data is loaded as a dict of items.
+
+    Raises
+    ------
+    ValueError
+        If the data type doesn't match the expected type for the given format.
+    ValueError
+        If the `input_format` is not one of the supported formats.
     """
 
     data: Union[
@@ -96,6 +354,7 @@ class Input:
         str,  # TEXT
         list[dict[str, Any]],  # CSV
         dict[str, list[dict[str, Any]]],  # CSV_ARCHIVE
+        dict[str, Any],  # MULTI_FILE
     ]
     """
     The actual data.
@@ -106,6 +365,7 @@ class Input:
     - For `TEXT`: `str`
     - For `CSV`: `list[dict[str, Any]]`
     - For `CSV_ARCHIVE`: `dict[str, list[dict[str, Any]]]`
+    - For `MULTI_FILE`: `dict[str, Any]`
     """
 
     input_format: Optional[InputFormat] = InputFormat.JSON
@@ -165,6 +425,12 @@ class Input:
                 "input_format InputFormat.CSV_ARCHIVE, supported type is `dict`"
             )
 
+        elif self.input_format == InputFormat.MULTI_FILE and not isinstance(self.data, dict):
+            raise ValueError(
+                f"unsupported Input.data type: {type(self.data)} with "
+                "input_format InputFormat.MULTI_FILE, supported type is `dict`"
+            )
+
         # Capture a snapshot of the options that were used to create the class
         # so even if they are changed later, we have a record of the original.
         init_options = self.options
@@ -175,8 +441,10 @@ class Input:
         """
         Convert the input to a dictionary.
 
-        This method serializes the Input object to a dictionary format that can be
-        easily converted to JSON or other serialization formats.
+        This method serializes the Input object to a dictionary format that can
+        be easily converted to JSON or other serialization formats. When the
+        `input_type` is set to `InputFormat.MULTI_FILE`, it will not include
+        the `data` field, as it is uncertain how data is deserialized from the file.
 
         Returns
         -------
@@ -201,12 +469,18 @@ class Input:
         {'data': {'key': 'value'}, 'input_format': 'json', 'options': None}
         """
 
-        return {
-            "data": self.data,
+        input_dict = {
             "input_format": self.input_format.value,
             "options": self.options.to_dict() if self.options is not None else None,
         }
 
+        if self.input_format == InputFormat.MULTI_FILE:
+            return input_dict
+
+        input_dict["data"] = self.data
+
+        return input_dict
+
 
 class InputLoader:
     """
@@ -375,6 +649,7 @@ class LocalInputLoader(InputLoader):
         options: Optional[Options] = None,
         path: Optional[str] = None,
         csv_configurations: Optional[dict[str, Any]] = None,
+        data_files: Optional[list[DataFile]] = None,
     ) -> Input:
         """
         Load the input data. The input data can be in various formats. For
@@ -395,6 +670,10 @@ class LocalInputLoader(InputLoader):
         - `InputFormat.CSV`: the data is a `list[dict[str, Any]]`.
         - `InputFormat.CSV_ARCHIVE`: the data is a `dict[str, list[dict[str, Any]]]`.
           Each key is the name of the CSV file, minus the `.csv` extension.
+        - `InputFormat.MULTI_FILE`: the data is a `dict[str, Any]`, where each
+          key is the file name (with extension) and the value is the data read
+          from the file. The data can be of any type, depending on the file
+          type and the reader function provided in the `DataFile` instances.
 
         Parameters
         ----------
@@ -408,6 +687,16 @@ class LocalInputLoader(InputLoader):
             Configurations for loading CSV files. The default `DictReader` is
             used when loading a CSV file, so you have the option to pass in a
             dictionary with custom kwargs for the `DictReader`.
+        data_files : list[DataFile], optional
+            List of `DataFile` instances to read from. This is used when the
+            `input_format` is set to `InputFormat.MULTI_FILE`. Each `DataFile`
+            instance should have a `name` (the file name with extension) and a
+            `loader` function that reads the data from the file. The `loader`
+            function should accept the file path as its first argument and return
+            the data read from the file. The `loader` can also accept additional
+            positional and keyword arguments, which can be provided through the
+            `loader_args` and `loader_kwargs` attributes of the `DataFile`
+            instance.
 
         Returns
         -------
@@ -428,6 +717,14 @@ class LocalInputLoader(InputLoader):
             data = self._load_utf8_encoded(path=path, input_format=input_format, csv_configurations=csv_configurations)
         elif input_format == InputFormat.CSV_ARCHIVE:
             data = self._load_archive(path=path, csv_configurations=csv_configurations)
+        elif input_format == InputFormat.MULTI_FILE:
+            if data_files is None:
+                raise ValueError("data_files must be provided when input_format is InputFormat.MULTI_FILE")
+
+            if not isinstance(data_files, list):
+                raise ValueError("data_files must be a list of DataFile instances")
+
+            data = self._load_multi_file(data_files=data_files, path=path)
 
         return Input(data=data, input_format=input_format, options=options)
 
@@ -528,6 +825,73 @@ class LocalInputLoader(InputLoader):
 
         return data
 
+    def _load_multi_file(
+        self,
+        data_files: list[DataFile],
+        path: Optional[str] = None,
+    ) -> dict[str, Any]:
+        """
+        Load multiple files from a directory.
+
+        This internal method loads all supported files from a specified
+        directory, organizing them into a dictionary where each key is the
+        filename and each value is the parsed file content. Supports CSV files
+        (parsed as list of dictionaries), JSON files (parsed as JSON objects),
+        and any other utf-8 encoded text files (loaded as plain text strings).
+        It also supports Excel files, loading them as DataFrames.
+
+        Parameters
+        ----------
+        data_files : list[DataFile]
+            List of `DataFile` instances to read from.
+        path : str, optional
+            Path to the directory containing files. If None or empty,
+            uses "./inputs" as the default directory.
+
+        Returns
+        -------
+        dict[str, Any]
+            Dictionary mapping filenames to file contents. CSV files are loaded
+            as lists of dictionaries, JSON files as parsed JSON objects, and
+            other utf-8 text files as strings. Excel files are loaded as
+            DataFrames.
+
+        Raises
+        ------
+        ValueError
+            If the path is not a directory or the default directory doesn't exist.
+        """
+
+        dir_path = "inputs"
+        if path is not None and path != "":
+            if not os.path.isdir(path):
+                raise ValueError(f"path {path} is not a directory")
+
+            dir_path = path
+
+        if not os.path.isdir(dir_path):
+            raise ValueError(f'expected input directoy "{dir_path}" to exist as a default location')
+
+        data = {}
+
+        for data_file in data_files:
+            name = data_file.name
+            file_path = os.path.join(dir_path, name)
+
+            if data_file.loader_args is None:
+                data_file.loader_args = []
+            if data_file.loader_kwargs is None:
+                data_file.loader_kwargs = {}
+
+            d = data_file.loader(
+                file_path,
+                *data_file.loader_args,
+                **data_file.loader_kwargs,
+            )
+            data[name] = d
+
+        return data
+
 
 def load_local(
     input_format: Optional[InputFormat] = InputFormat.JSON,
@@ -590,6 +954,7 @@ def load(
     path: Optional[str] = None,
     csv_configurations: Optional[dict[str, Any]] = None,
     loader: Optional[InputLoader] = _LOCAL_INPUT_LOADER,
+    data_files: Optional[list[DataFile]] = None,
 ) -> Input:
     """
     Load input data using the specified loader.
@@ -611,6 +976,36 @@ def load(
     - `InputFormat.CSV`: the data is a `list[dict[str, Any]]`
     - `InputFormat.CSV_ARCHIVE`: the data is a `dict[str, list[dict[str, Any]]]`
         Each key is the name of the CSV file, minus the `.csv` extension.
+    - `InputFormat.MULTI_FILE`: the data is a `dict[str, Any]`
+        where each key is the file name (with extension) and the value is the
+        data read from the file. This is used for loading multiple files in a
+        single input, where each file can be of different types (JSON, CSV,
+        Excel, etc.). The data is loaded as a dict of items, where each item
+        corresponds to a file and its content.
+
+    When specifying `input_format` as `InputFormat.MULTI_FILE`, the
+    `data_files` argument must be provided. This argument is a list of
+    `DataFile` instances, each representing a file to be read. Each `DataFile`
+    instance should have a `name` (the file name with extension) and a `loader`
+    function that reads the data from the file. The `loader` function should
+    accept the file path as its first argument and return the data read from
+    the file. The `loader` can also accept additional positional and keyword
+    arguments, which can be provided through the `loader_args` and
+    `loader_kwargs` attributes of the `DataFile` instance.
+
+    There are convenience functions that can be used to create `DataFile`
+    classes, such as:
+
+    - `json_data_file`: Creates a `DataFile` that reads JSON data.
+    - `csv_data_file`: Creates a `DataFile` that reads CSV data.
+    - `text_data_file`: Creates a `DataFile` that reads utf-8 encoded text
+      data.
+
+    When workiing with data in other formats, such as Excel files, you are
+    encouraged to create your own `DataFile` objects with your own
+    implementation of the `loader` function. This allows you to read data
+    from files in a way that suits your needs, while still adhering to the
+    `DataFile` interface.
 
     Parameters
     ----------
@@ -629,6 +1024,24 @@ def load(
     loader : InputLoader, optional
         The loader to use for loading the input data.
         Default is an instance of `LocalInputLoader`.
+    data_files : list[DataFile], optional
+        List of `DataFile` instances to read from. This is used when the
+        `input_format` is set to `InputFormat.MULTI_FILE`. Each `DataFile`
+        instance should have a `name` (the file name with extension) and a
+        `loader` function that reads the data from the file. The `loader`
+        function should accept the file path as its first argument and return
+        the data read from the file. The `loader` can also accept additional
+        positional and keyword arguments, which can be provided through the
+        `loader_args` and `loader_kwargs` attributes of the `DataFile`
+        instance.
+
+        There are convenience functions that can be used to create `DataFile`
+        classes, such as `json_data_file`, `csv_data_file`, and
+        `text_data_file`. When working with data in other formats, such as
+        Excel files, you are encouraged to create your own `DataFile` objects
+        with your own implementation of the `loader` function. This allows you
+        to read data from files in a way that suits your needs, while still
+        adhering to the `DataFile` interface.
 
     Returns
     -------
@@ -651,4 +1064,4 @@ def load(
     >>> input_obj = load(input_format=InputFormat.CSV_ARCHIVE, path="input_dir")
     """
 
-    return loader.load(input_format, options, path, csv_configurations)
+    return loader.load(input_format, options, path, csv_configurations, data_files)

nextmv/model.py

@@ -24,7 +24,7 @@ from typing import Any, Optional
 
 from nextmv.input import Input
 from nextmv.logger import log
-from nextmv.options import Options
+from nextmv.options import Options, OptionsEnforcement
 from nextmv.output import Output
 
 # The following block of code is used to suppress warnings from mlflow. We
@@ -132,6 +132,9 @@ class ModelConfiguration:
         formatted as they would appear in a requirements.txt file.
     options : Options, optional
         Options that the decision model requires.
+    options_enforcement:
+        Enforcement of options for the model. This controls how options
+        are handled when the model is run.
 
     Examples
     --------
@@ -139,17 +142,23 @@ class ModelConfiguration:
     >>> config = ModelConfiguration(
     ...     name="my_routing_model",
     ...     requirements=["nextroute>=1.0.0"],
-    ...     options=Options({"max_time": 60})
+    ...     options=Options({"max_time": 60}),
+    ...     options_enforcement=OptionsEnforcement(
+                strict=True,
+                validation_enforce=True
+            )
     ... )
     """
 
     name: str
     """The name of the decision model."""
-
     requirements: Optional[list[str]] = None
     """A list of Python dependencies that the decision model requires."""
     options: Optional[Options] = None
     """Options that the decision model requires."""
+    options_enforcement: Optional[OptionsEnforcement] = None
+    """Enforcement of options for the model."""
+
 
 
 class Model: