nextmv 0.40.0__py3-none-any.whl → 1.0.0__py3-none-any.whl

nextmv/input.py

@@ -38,7 +38,6 @@ from enum import Enum
 from typing import Any
 
 from nextmv._serialization import serialize_json
-from nextmv.deprecated import deprecated
 from nextmv.options import Options
 
 INPUTS_KEY = "inputs"
@@ -77,8 +76,6 @@ class InputFormat(str, Enum):
     """JSON format, utf-8 encoded."""
     TEXT = "text"
     """Text format, utf-8 encoded."""
-    CSV = "csv"
-    """CSV format, utf-8 encoded."""
     CSV_ARCHIVE = "csv-archive"
     """CSV archive format: multiple CSV files."""
     MULTI_FILE = "multi-file"
@@ -376,8 +373,6 @@ class Input:
        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.
@@ -466,12 +461,6 @@ class Input:
                 "input_format InputFormat.TEXT, supported type is `str`"
             )
 
-        elif self.input_format == InputFormat.CSV and not isinstance(self.data, list):
-            raise ValueError(
-                f"unsupported Input.data type: {type(self.data)} with "
-                "input_format InputFormat.CSV, supported type is `list`"
-            )
-
         elif self.input_format == InputFormat.CSV_ARCHIVE and not isinstance(self.data, dict):
             raise ValueError(
                 f"unsupported Input.data type: {type(self.data)} with "
@@ -607,8 +596,6 @@ class LocalInputLoader(InputLoader):
     >>> loader = LocalInputLoader()
     >>> # Load JSON from stdin or file
     >>> input_obj = loader.load(input_format=InputFormat.JSON, path="data.json")
-    >>> # Load CSV from a file
-    >>> input_obj = loader.load(input_format=InputFormat.CSV, path="data.csv")
     """
 
     def _read_text(path: str, _) -> str:
@@ -672,7 +659,6 @@ class LocalInputLoader(InputLoader):
     STDIN_READERS = {
         InputFormat.JSON: lambda _: json.load(sys.stdin),
         InputFormat.TEXT: lambda _: sys.stdin.read().rstrip("\n"),
-        InputFormat.CSV: lambda csv_configurations: list(csv.DictReader(sys.stdin, **csv_configurations)),
     }
     """
     Dictionary of functions to read from standard input.
@@ -687,7 +673,7 @@ class LocalInputLoader(InputLoader):
     FILE_READERS = {
         InputFormat.JSON: _read_json,
         InputFormat.TEXT: _read_text,
-        InputFormat.CSV: _read_csv,
+        "CSV": _read_csv,
     }
     """
     Dictionary of functions to read from files.
@@ -706,23 +692,23 @@ class LocalInputLoader(InputLoader):
     ) -> Input:
         """
         Load the input data. The input data can be in various formats. For
-        `InputFormat.JSON`, `InputFormat.TEXT`, and `InputFormat.CSV`, the data
-        can be streamed from stdin or read from a file. When the `path`
-        argument is provided (and valid), the input data is read from the file
-        specified by `path`, otherwise, it is streamed from stdin. For
-        `InputFormat.CSV_ARCHIVE`, the input data is read from the directory
-        specified by `path`. If the `path` is not provided, the default
-        location `input` is used. The directory should contain one or more
-        files, where each file in the directory is a CSV file.
+        `InputFormat.JSON` and `InputFormat.TEXT`, the data can be streamed
+        from stdin or read from a file. When the `path` argument is provided
+        (and valid), the input data is read from the file specified by `path`,
+        otherwise, it is streamed from stdin. For `InputFormat.CSV_ARCHIVE`,
+        the input data is read from the directory specified by `path`. If the
+        `path` is not provided, the default location `input` is used. The
+        directory should contain one or more files, where each file in the
+        directory is a CSV file.
 
         The `Input` that is returned contains the `data` attribute. This data
         can be of different types, depending on the provided `input_format`:
 
         - `InputFormat.JSON`: the data is a `dict[str, Any]`.
         - `InputFormat.TEXT`: the data is a `str`.
-        - `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.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
@@ -745,11 +731,11 @@ class LocalInputLoader(InputLoader):
             `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.
+            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
         -------
@@ -766,7 +752,7 @@ class LocalInputLoader(InputLoader):
         if csv_configurations is None:
             csv_configurations = {}
 
-        if input_format in [InputFormat.JSON, InputFormat.TEXT, InputFormat.CSV]:
+        if input_format in [InputFormat.JSON, InputFormat.TEXT]:
             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)
@@ -785,7 +771,7 @@ class LocalInputLoader(InputLoader):
         self,
         csv_configurations: dict[str, Any] | None,
         path: str | None = None,
-        input_format: InputFormat | None = InputFormat.JSON,
+        input_format: InputFormat | str | None = InputFormat.JSON,
         use_file_reader: bool = False,
     ) -> dict[str, Any] | str | list[dict[str, Any]]:
         """
@@ -871,7 +857,7 @@ class LocalInputLoader(InputLoader):
                 stripped = file.removesuffix(csv_ext)
                 data[stripped] = self._load_utf8_encoded(
                     path=os.path.join(dir_path, file),
-                    input_format=InputFormat.CSV,
+                    input_format="CSV",
                     use_file_reader=True,
                     csv_configurations=csv_configurations,
                 )
@@ -954,57 +940,6 @@ class LocalInputLoader(InputLoader):
         return data
 
 
-def load_local(
-    input_format: InputFormat | None = InputFormat.JSON,
-    options: Options | None = None,
-    path: str | None = None,
-    csv_configurations: dict[str, Any] | None = None,
-) -> Input:
-    """
-    !!! warning
-        `load_local` is deprecated, use `load` instead.
-
-    Load input data from local sources.
-
-    This is a convenience function for instantiating a `LocalInputLoader`
-    and calling its `load` method.
-
-    Parameters
-    ----------
-    input_format : InputFormat, optional
-        Format of the input data. Default is `InputFormat.JSON`.
-    options : Options, optional
-        Options for loading the input data.
-    path : str, optional
-        Path to the input data.
-    csv_configurations : dict[str, Any], optional
-        Configurations for loading CSV files. Custom kwargs for
-        Python's `csv.DictReader`.
-
-    Returns
-    -------
-    Input
-        The loaded input data in an Input object.
-
-    Raises
-    ------
-    ValueError
-        If the path is invalid or data format is incorrect.
-
-    See Also
-    --------
-    load : The recommended function to use instead.
-    """
-
-    deprecated(
-        name="load_local",
-        reason="`load_local` is deprecated, use `load` instead",
-    )
-
-    loader = LocalInputLoader()
-    return loader.load(input_format, options, path, csv_configurations)
-
-
 _LOCAL_INPUT_LOADER = LocalInputLoader()
 """Default instance of LocalInputLoader used by the load function."""
 
@@ -1034,7 +969,6 @@ def load(
 
     - `InputFormat.JSON`: the data is a `dict[str, Any]`
     - `InputFormat.TEXT`: the data is a `str`
-    - `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]`
@@ -1119,8 +1053,6 @@ def load(
     >>> from nextmv.input import load, InputFormat
     >>> # Load JSON from stdin
     >>> input_obj = load(input_format=InputFormat.JSON)
-    >>> # Load CSV from a file
-    >>> input_obj = load(input_format=InputFormat.CSV, path="data.csv")
     >>> # Load CSV archive from a directory
     >>> input_obj = load(input_format=InputFormat.CSV_ARCHIVE, path="input_dir")
     """

nextmv/local/application.py

@@ -256,8 +256,7 @@ class Application:
             `nextmv.InputFormat.MULTI_FILE`, you should use the
             `input_dir_path` argument instead. This argument takes precedence
             over the `input`. If `input_dir_path` is specified, this function
-            looks for files in that directory and tars them, to later be
-            uploaded using the `upload_large_input` method. If both the
+            looks for files in that directory and tars them. If both the
             `input_dir_path` and `input` arguments are provided, the `input`
             is ignored.
 
@@ -273,9 +272,6 @@ class Application:
 
             When working with JSON or text data, use the `input` argument
             directly.
-
-            In general, if an input is too large, it will be uploaded with the
-            `upload_large_input` method.
         name: Optional[str]
             Name of the local run.
         description: Optional[str]
@@ -399,8 +395,7 @@ class Application:
             `nextmv.InputFormat.MULTI_FILE`, you should use the
             `input_dir_path` argument instead. This argument takes precedence
             over the `input`. If `input_dir_path` is specified, this function
-            looks for files in that directory and tars them, to later be
-            uploaded using the `upload_large_input` method. If both the
+            looks for files in that directory and tars them. If both the
             `input_dir_path` and `input` arguments are provided, the `input` is
             ignored.
 
@@ -416,9 +411,6 @@ class Application:
 
             When working with JSON or text data, use the `input` argument
             directly.
-
-            In general, if an input is too large, it will be uploaded with the
-            `upload_large_input` method.
         name: Optional[str]
             Name of the local run.
         description: Optional[str]
@@ -699,11 +691,7 @@ class Application:
 
         def polling_func() -> tuple[Any, bool]:
             run_information = self.run_metadata(run_id=run_id)
-            if run_information.metadata.status_v2 in {
-                StatusV2.succeeded,
-                StatusV2.failed,
-                StatusV2.canceled,
-            }:
+            if run_information.metadata.run_is_finalized():
                 return run_information, True
 
             return None, False

nextmv/local/runner.py

@@ -204,7 +204,7 @@ def new_run(
     if description is None:
         description = f"Local run created at {created_at.isoformat().replace('+00:00', 'Z')}"
 
-    if name is None:
+    if name is None or name == "":
         name = f"local run {run_id}"
 
     information = RunInformation(

nextmv/model.py

@@ -18,6 +18,8 @@ deployed to Nextmv Cloud for execution.
 import logging
 import os
 import shutil
+import sqlite3
+import time
 import warnings
 from dataclasses import dataclass
 from typing import Any
@@ -84,6 +86,9 @@ def _custom_showwarning(message, category, filename, lineno, file=None, line=Non
     if "mlflow/pyfunc/__init__.py" in filename:
         return
 
+    if "/mlflow/tracing/provider.py" in filename:
+        return
+
     _original_showwarning(message, category, filename, lineno, file, line)
 
 
@@ -288,7 +293,6 @@ class Model:
             ) from e
 
         finally:
-            from mlflow.models import infer_signature
             from mlflow.pyfunc import PythonModel, save_model
 
         class MLFlowModel(PythonModel):
@@ -342,12 +346,16 @@ class Model:
 
         _cleanup_python_model(model_dir, configuration, verbose=False)
 
-        signature = None
-        if configuration.options is not None:
-            options_dict = configuration.options.to_dict()
-            signature = infer_signature(
-                params=options_dict,
-            )
+        # Removing this seems to make the "apps from models" experience once
+        # again. I am not removing it entirely because we might want to
+        # re-introduce it later on.
+        # signature = None
+        # if configuration.options is not None:
+        #     options_dict = configuration.options.to_dict()
+        #     signature = infer_signature(
+        #         model_input={},
+        #         params=options_dict,
+        #     )
 
         # We use mlflow to save the model to the local filesystem, to be able to
         # load it later on.
@@ -356,7 +364,7 @@ class Model:
             path=model_path,  # Customize the name of the model location.
             infer_code_paths=True,  # Makes the imports portable.
             python_model=MLFlowModel(),
-            signature=signature,  # Allows us to work with our own `Options` class.
+            # signature=signature,  # Please see comment above about keeping this in case we need to go back.
         )
 
         # Create an auxiliary requirements file with the model dependencies.
@@ -415,9 +423,7 @@ def _cleanup_python_model(
     if os.path.exists(model_path):
         shutil.rmtree(model_path)
 
-    mlruns_path = os.path.join(model_dir, "mlruns")
-    if os.path.exists(mlruns_path):
-        shutil.rmtree(mlruns_path)
+    _cleanup_mlflow_db(model_dir)
 
     requirements_file = os.path.join(model_dir, _REQUIREMENTS_FILE)
     if os.path.exists(requirements_file):
@@ -429,3 +435,36 @@ def _cleanup_python_model(
 
     if verbose:
         log("🧹 Cleaned up Python model artifacts.")
+
+
+def _cleanup_mlflow_db(model_dir: str) -> None:
+    """
+    Clean up the mlflow.db file created during model packaging.
+
+    Parameters
+    ----------
+    model_dir : str
+        The directory where the model was saved.
+    """
+    mlflow_db_path = os.path.join(model_dir, "mlflow.db")
+    if not os.path.exists(mlflow_db_path):
+        return
+
+    # Try to close any open SQLite connections and retry deletion
+    max_retries = 5
+    for attempt in range(max_retries):
+        try:
+            # Attempt to connect and close to release any locks
+            try:
+                conn = sqlite3.connect(mlflow_db_path)
+                conn.close()
+            except Exception:
+                pass
+            os.remove(mlflow_db_path)
+            break
+        except PermissionError:
+            if attempt < max_retries - 1:
+                time.sleep(0.5)
+            else:
+                log(f"Could not delete {mlflow_db_path} after {max_retries} retries due to file lock.")
+                break