added support of inputs parameters that are recognised by the API.

Models are now loaded in separate endpoints for the inputs to be easier to recognise

source/utils/__init__.py

@@ -0,0 +1 @@
+from . import parameters

source/utils/parameters.py

@@ -0,0 +1,54 @@
+import inspect
+from datetime import datetime
+
+import fastapi
+
+
+# the list of types and their name that can be used by the API
+types: dict[str, type] = {
+    "bool": bool,
+    "int": int,
+    "float": float,
+    "str": str,
+    "bytes": bytes,
+    "list": list,
+    "tuple": tuple,
+    "set": set,
+    "dict": dict,
+    "datetime": datetime,
+    "file": fastapi.UploadFile,
+}
+
+
+def load(parameters_definition: dict[str, dict]) -> list[inspect.Parameter]:
+    """
+    Load a list python function parameters from their definitions.
+    :param parameters_definition: the definitions of the parameters
+    :return: the python function parameters
+
+    Examples:
+        >>> parameters_definition = {
+        ...     "boolean": {"type": "bool", "default": False},
+        ...     "list": {"type": "list", "default": [1, 2, 3]},
+        ...     "datetime": {"type": "datetime"},
+        ...     "file": {"type": "file"},
+        ... }
+        >>> parameters = load_parameters(parameters_definition)
+    """
+
+    parameters: list[inspect.Parameter] = []
+
+    for name, definition in parameters_definition.items():
+        # deserialize the parameter
+        parameter = inspect.Parameter(
+            name,
+            inspect.Parameter.POSITIONAL_OR_KEYWORD,
+            default=definition.get("default", inspect.Parameter.empty),
+            annotation=types[definition["type"]],
+        )
+        parameters.append(parameter)
+
+    # sort the parameters so that non-default arguments always end up before default ones
+    parameters.sort(key=lambda parameter: parameter.default is inspect.Parameter.empty, reverse=True)
+
+    return parameters