aesoptparam 0.3.7__py3-none-any.whl

aesoptparam/__init__.py

@@ -0,0 +1,13 @@
+from param import *
+
+from .parameterized import AESOptParameterized
+from .parameters import (
+    AESOptArray,
+    AESOptBoolean,
+    AESOptInteger,
+    AESOptNumber,
+    AESOptString,
+    ListOfParameterized,
+    SubParameterized,
+)
+from .utils import copy_param, copy_param_ref

aesoptparam/example.py

@@ -0,0 +1,110 @@
+import numpy as np
+
+from aesoptparam import (
+    AESOptArray,
+    AESOptNumber,
+    AESOptParameterized,
+    AESOptString,
+    Dict,
+    ListOfParameterized,
+    String,
+    SubParameterized,
+    copy_param,
+    copy_param_ref,
+)
+
+
+class sub1_class(AESOptParameterized):
+    """Docs sub1"""
+
+    a = AESOptNumber(5.0, doc="Docs for sub1.a", bounds=(0, 10))
+    b = AESOptArray(np.linspace(0, 10, 10), doc="Docs for sub1.b")
+    c = AESOptArray(default_ref=".b", doc="Docs for sub1.c")
+    e = AESOptArray(
+        lambda self: np.full_like(self.b, self.a), shape=".b", doc="Docs for sub1.e"
+    )
+    f = AESOptArray(default_full=(".b", ".a"), doc="Docs for sub1.d")
+    g = AESOptArray(default_full=(".b", 6.0), doc="Docs for sub1.g")
+    h = AESOptArray(
+        default_interp=(np.linspace(0, 10, 5), ".b", ".c"), doc="Docs for sub1.h"
+    )
+    i = AESOptString("sub1.Dummy", doc="Docs for sub1.i")
+    j = AESOptString(".i", doc="Docs for sub1.j")
+    k = AESOptString(lambda self: self.i + "2", doc="Docs for sub1.k")
+
+
+class sub2_class(AESOptParameterized):
+    """Docs sub2"""
+
+    a = copy_param_ref(
+        sub1_class.param.a, "..sub1.a", update=dict(doc="Docs for sub2.a")
+    )
+    b = AESOptNumber(default_ref="..a", doc="Docs for sub2.b")
+    c = AESOptNumber(default_ref="..sub_list[0].a", doc="Docs for sub2.c")
+    d = copy_param_ref(
+        sub1_class.param.b, "..sub1.b", update=dict(doc="Docs for sub2.d")
+    )
+    e = AESOptArray(
+        lambda self: (
+            None if self.parent_object.sub1.b is None else self.parent_object.sub1.b + 1
+        ),
+        doc="Docs for sub2.e",
+    )
+    f = AESOptString("..f", doc="Docs for sub2.f")
+    g = copy_param_ref(
+        sub1_class.param.i, "..sub1.i", update=dict(doc="Docs for sub2.g")
+    )
+    h = AESOptString("..sub_list[0].f", doc="Docs for sub2.h")
+
+
+class sub_list_class(AESOptParameterized):
+    """Docs sub_list"""
+
+    a = AESOptNumber(6.0, doc="Docs for sub_list.a")
+    b = AESOptNumber(default_ref="..a", doc="Docs for sub_list.b")
+    c = AESOptNumber(default_ref="..sub1.a", doc="Docs for sub_list.c")
+    d = AESOptArray(default_ref="..d", doc="Docs for sub_list.d")
+    e = AESOptArray(default_ref="..sub1.b", doc="Docs for sub_list.e")
+    f = AESOptString(default_ref="..f", doc="Docs for sub_list.f")
+    g = AESOptString(default_ref="..sub1.i", doc="Docs for sub_list.g")
+
+
+class main(AESOptParameterized):
+    """Main object"""
+
+    version = String("0.0.0", readonly=True, precedence=0.0)
+    name = String("Dummy", doc="Main dummy", precedence=0.01, constant=False)
+    a = AESOptNumber(4.0, units="rad/s", doc="Docs for .a", bounds=(0, 10))
+    b = AESOptNumber(default_ref=".sub1.a", units="m/s", doc="Docs for .b")
+    c = AESOptNumber(default_ref=".sub_list[0].a", doc="Docs for .c")
+    d = AESOptArray(
+        default_ref=".sub1.b", units="mm/s", doc="Docs for .d", bounds=(0, 10)
+    )
+    e = AESOptArray(
+        np.array([0, 1, 2, 3]),
+        shape=4,
+        units="mm/s",
+        doc="Docs for .d",
+        bounds=(0, 10),
+        dtype=int,
+    )
+    f = AESOptString("Dummy", doc="Docs for .f")
+    g = AESOptString(".f", doc="Docs for .g")
+    h = AESOptString(lambda self: self.f + "2", doc="Docs for .h")
+    i = Dict(doc="Docs for .i")
+    sub1 = SubParameterized(sub1_class)
+    sub2 = SubParameterized(sub2_class)
+    sub_list = ListOfParameterized(sub_list_class)
+    sub_list2 = ListOfParameterized(
+        sub_list_class, default_call=lambda self: [self.add_sub_list2()]
+    )
+
+    def add_sub_list(self, **params):
+        return self.add_ListOfParameterized_item(
+            "sub_list", self.param.sub_list.item_type, **params
+        )
+
+    def add_sub_list2(self, **params):
+        return self.add_ListOfParameterized_item(
+            "sub_list2", self.param.sub_list2.item_type, **params
+        )

aesoptparam/parameterized.py

@@ -0,0 +1,415 @@
+import json
+import re
+
+import numpy as np
+import param as pm
+from scipy.interpolate import PchipInterpolator
+
+from .parameters import Function, ListOfParameterized, Reference
+from .serializer import ASEOptJSONSerialization
+from .utils import read_json, write_json
+from .utils.html_repr import parameterized_repr_html, parameterized_repr_html_class
+from .utils.units import convert_units
+
+
+class AESOptParameterized(pm.Parameterized):
+    """Base object inheriting from `param.Parameterized`. Adds specific methods for interacting with data in AESOpt models."""
+
+    interpolator = PchipInterpolator
+
+    def __init__(self, parent_object=None, **params):
+        """Initialize object with parameters. **params are passed with the .from_dict. The parent_object argument allows nested objects to access data across the full data model"""
+        self.parent_object = parent_object
+        super().__init__()
+        self.from_dict(params)
+
+    def from_dict(self, dict_in):
+        """Set parameters from `dict`"""
+        for key, val in dict_in.items():
+            if key in self.param:
+                p = self.param[key]
+                if hasattr(p, "readonly") and p.readonly:
+                    if np.all(self[key] != val):
+                        raise ValueError(
+                            f"A parameter with readonly has a different value in the dict than currently in the object (given: self.{key}={self[key]}, dict_in['{key}']={val})"
+                        )
+                    continue
+                if hasattr(self[key], "from_dict"):
+                    self[key].from_dict(val)
+                elif (
+                    isinstance(self.param[key], pm.List)
+                    and (self.param[key].item_type is not None)
+                    and hasattr(self.param[key].item_type, "from_dict")
+                ):
+                    for i, el in enumerate(val):
+                        if len(self[key]) > i:
+                            self[key][i].from_dict(el)
+                        else:
+                            self[key].append(
+                                self.param[key].item_type(self).from_dict(el)
+                            )
+                elif isinstance(self.param[key], pm.Array) and isinstance(val, list):
+                    arr = np.asarray(val)
+                    self[key] = arr
+                else:
+                    self[key] = val
+        return self
+
+    def as_dict(self, onlychanged=True, as_refs=False, as_funcs=False) -> dict:
+        """Get object as dict. It also works for nested data structure, references and functions.
+
+        Parameters
+        ----------
+        onlychanged : bool, optional
+            Flag for only returning data that has been changed compared to the default, by default True
+        as_refs : bool, optional
+            Flag for returning the `Reference` instance instead of the values, by default False
+        as_funcs : bool, optional
+            Flag for returning the `Function` instance instead of the values, by default False
+
+        Returns
+        -------
+        dict
+            Object data as a dict
+        """
+        return self._as_dict(
+            onlychanged=onlychanged, serialize=False, as_refs=as_refs, as_funcs=as_funcs
+        )
+
+    def as_serial(self, onlychanged=True, as_refs=True, as_funcs=True) -> dict:
+        """Get instance as a serializable dict. It returns a `dict` with native python data types (e.g. `list`, `float`, `int`, `str`), which can be written to file with `json`, `yaml` and `toml`.
+        It will also convert references (to `$ref *ref*`) and functions (to `$function *function string*`)
+
+        Parameters
+        ----------
+        onlychanged : bool, optional
+            Flag for only returning data that has been changed compared to the default, by default True
+        as_refs : bool, optional
+            Flag for returning the value from references or functions, by default True
+        as_funcs : bool, optional
+            Flag for returning the `Function` instance instead of the values, by default False
+
+        Returns
+        -------
+        dict
+            Object data as a dict
+        """
+        return self._as_dict(
+            onlychanged=onlychanged, serialize=True, as_refs=as_refs, as_funcs=as_funcs
+        )
+
+    def _as_dict(
+        self, onlychanged=True, serialize=False, as_refs=False, as_funcs=False
+    ):
+        out = dict()
+
+        if onlychanged:
+            vals = self._param__private.values
+        else:
+            vals = self.param.values()
+            vals.update(self._param__private.values)
+
+        for key, val in vals.items():
+            p = self.param[key]  # Parameter instance
+            if (key == "name") and p.constant:
+                continue
+            if onlychanged:
+                if (val is None) or (val is p.default):
+                    continue
+                elif (
+                    isinstance(val, np.ndarray)
+                    and isinstance(p.default, np.ndarray)
+                    and np.all(p.default.shape == val.shape)
+                    and np.allclose(val, p.default)
+                ):
+                    continue
+            if hasattr(val, "_as_dict"):
+                out[key] = val._as_dict(
+                    onlychanged=onlychanged,
+                    serialize=serialize,
+                    as_refs=as_refs,
+                    as_funcs=as_funcs,
+                )
+                if not out[key]:
+                    out.pop(key)
+            elif isinstance(val, list) and len(val) > 0 and hasattr(val[0], "_as_dict"):
+                out[key] = val.copy()
+                for iel, el in enumerate(val):
+                    out[key][iel] = el._as_dict(
+                        onlychanged=onlychanged,
+                        serialize=serialize,
+                        as_refs=as_refs,
+                        as_funcs=as_funcs,
+                    )
+                    if not out[key][iel]:
+                        out[key][iel] = None
+                if all([el is None for el in out[key]]):
+                    out.pop(key)
+            elif isinstance(val, Reference) and as_refs:
+                out[key] = val
+            elif isinstance(val, Function) and as_funcs:
+                out[key] = val
+            else:
+                _val = self[key]
+                out[key] = _val.copy() if hasattr(_val, "copy") else _val
+
+            if serialize and (key in out):
+                if isinstance(out[key], np.ndarray):
+                    out[key] = out[key].tolist()
+                elif isinstance(out[key], (Reference, Function)):
+                    out[key] = str(out[key])
+        return out
+
+    def validate_array_shapes(self):
+        """Validate that all arrays with a .shape attribute is consistent with this shape. Will also check nested objects."""
+        for name, param in self.param.objects().items():
+            if hasattr(param, "shape") and (param.shape is not None):
+                shape_ref = self.get_shape_ref(param)
+                if not np.all(self[name].shape == shape_ref):
+                    raise ValueError(
+                        f"For {self.nested_instance_name} the shape of {name} do not match {param.shape} ({name}.shape={self[name].shape}, shape_ref={shape_ref})"
+                    )
+            elif isinstance(param, ListOfParameterized):
+                for el in self[name]:
+                    if hasattr(el, "validate_array_shapes"):
+                        el.validate_array_shapes()
+            elif hasattr(self[name], "validate_array_shapes"):
+                self[name].validate_array_shapes()
+
+    def get_shape_ref(self, param):
+        if isinstance(param.shape, tuple):
+            shape_ref = tuple()
+            for el in param.shape:
+                if isinstance(el, (Reference, str)):
+                    if not isinstance(self[el], np.ndarray):
+                        return None
+                    shape_ref += self[el].shape
+                else:
+                    shape_ref += (el,)
+        elif isinstance(param.shape, Reference):
+            if isinstance(self[param.shape], np.ndarray):
+                shape_ref = self[param.shape].shape
+            else:
+                return None
+        else:
+            # Default to int
+            shape_ref = (param.shape,)
+        return shape_ref
+
+    @property
+    def nested_instance_name(self):
+        """Nested instance name"""
+        if self.has_parent():
+            return self.parent_object.nested_instance_name + "." + self.name
+        return self.name
+
+    def interp(self, x, xp, yp, **interp_kwargs):
+        """1D interpolation using SciPy interpolator (default to [`PChipInterpolator`](https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.PchipInterpolator.html#scipy.interpolate.PchipInterpolator)). Arguments (`x`, `xp`, `yp`) can either be strings (object-paths) or numpy arrays."""
+        _x = x if isinstance(x, np.ndarray) else self[x]
+        _xp = xp if isinstance(xp, np.ndarray) else self[xp]
+        _yp = yp if isinstance(yp, np.ndarray) else self[yp]
+        if any([el is None for el in [_x, _xp, _yp]]):
+            return
+        return self.interpolator(_xp, _yp, **interp_kwargs)(_x)
+
+    def _repr_html_(self):
+        return parameterized_repr_html(self, True, max_arr_size=50)
+
+    def __getitem__(self, key):
+        if isinstance(key, str):
+            if ("." in key) or ("[" in key):
+                path_list = path_to_path_list(key)
+                out = self
+                previous_empty = False
+                for _key in path_list:
+                    if isinstance(_key, str) and _key == "":
+                        if previous_empty:
+                            out = out.parent_object
+                            previous_empty = False
+                        else:
+                            previous_empty = True
+                    else:
+                        if isinstance(_key, int) and len(out) <= _key:
+                            # Return none if array is shorter
+                            return None
+                        out = out[_key]
+                        previous_empty = False
+                return out
+            return getattr(self, key)
+        elif isinstance(key, Reference):
+            return self[key.path]
+        super().__getattribute__(key)
+
+    def __setitem__(self, key, val):
+        if isinstance(key, str):
+            return setattr(self, key, val)
+        super().__setattr__(key, val)
+
+    def get_val(self, name, units=None):
+        """Get a value with other units"""
+        self._validate_has_units(name)
+        return convert_units(self[name], self.param[name].units, units)
+
+    def set_val(self, name, val, units=None):
+        """Get a value with other units"""
+        self._validate_has_units(name)
+        self[name] = convert_units(val, units, self.param[name].units)
+
+    def _validate_has_units(self, name):
+        if not hasattr(self.param[name], "units"):
+            raise RuntimeError(
+                f"parameter with name: {name} do not have units (the parameter do not have `units` attribute, type(.param[name])={type(self.param[name])})"
+            )
+
+    def add_ListOfParameterized_item(self, LOP_name, item_type, **kwargs):
+        """Utility method to add Parameterized entries to a ListOfParameterized. It insure that the parent object is added. It is recommenced that this method is wrapped by a more specific method like `add_XX(self, **kwargs)` where the `LOP_name` and `item_type` is provided statically."""
+        # Instantiate item_type
+        item = item_type(parent_object=self, **kwargs)
+        # Append the item to the list of ListOfParameterized
+        if not LOP_name in self._param__private.values:
+            self._param__private.values[LOP_name] = []
+        self._param__private.values[LOP_name].append(item)
+        # Return the item that is added
+        return item
+
+    def as_json(
+        self, onlychanged=True, as_refs=True, as_funcs=True, **kwargs_json_dumps
+    ):
+        """Get data as a JSON string. `onlychanged` is a flag for only getting non default values."""
+        return json.dumps(
+            self.as_serial(onlychanged=onlychanged, as_refs=as_refs, as_funcs=as_funcs),
+            **kwargs_json_dumps,
+        )
+
+    def write_json(
+        self,
+        filename,
+        onlychanged=True,
+        as_refs=True,
+        as_funcs=True,
+        **kwargs_json_dump,
+    ):
+        """Write data to a JSON file. `onlychanged` is a flag for only getting non default values."""
+        with open(filename, "w") as file:
+            json.dump(
+                self.as_serial(
+                    onlychanged=onlychanged, as_refs=as_refs, as_funcs=as_funcs
+                ),
+                file,
+                **kwargs_json_dump,
+            )
+        return self
+
+    def read_json(self, filename, asarray=True):
+        """Read data from a JSON file."""
+        data = read_json(filename, asarray)
+        data.pop("$schema", None)
+        self.from_dict(data)
+        return self
+
+    def as_json_schema(self, safe=False, subset=None, skip_default_name=True) -> dict:
+        """Get object as a JSON schema
+
+        Parameters
+        ----------
+        safe : bool, optional
+            Flag for getting safe schema, by default False
+        subset : list, optional
+            List of names to skip, by default None
+        skip_default_name : bool, optional
+            Flag for skipping the default name when creating the schema, by default True
+
+        Returns
+        -------
+        dict
+            JSON Schema
+        """
+        return ASEOptJSONSerialization.schema(self, safe, subset, skip_default_name)
+
+    def write_json_schema(
+        self,
+        filename,
+        safe=False,
+        subset=None,
+        skip_default_name=True,
+        **kwargs_json_dump,
+    ):
+        """Write object as a JSON schema file
+
+        Parameters
+        ----------
+        filename : str,Path
+            Filename for the JSON schema
+        safe : bool, optional
+            Flag for getting safe schema, by default False
+        subset : list, optional
+            List of names to skip, by default None
+        skip_default_name : bool, optional
+            Flag for skipping the default name when creating the schema, by default True
+
+        Returns
+        -------
+        self
+        """
+        with open(filename, "w") as file:
+            json.dump(
+                self.as_json_schema(safe, subset, skip_default_name),
+                file,
+                **kwargs_json_dump,
+            )
+        return self
+
+    def has_parent(self):
+        return not self.parent_object is None
+
+    def display(self, open=True, title=None, max_arr_size=50):
+        """Method to display the settings and documentation of the current instance. Similar to just rendering the by last entry of the cell but allows a little more control
+
+        Parameters
+        ----------
+        open : bool, optional
+            Flag for opening or closing root the HTML table, by default True
+        title : str, optional
+            The string that will be shown as the root of the object, by default None
+        max_arr_size : int, optional
+            Maximum size of numeric arrays to render. Using `np.size(array) > max_arr_size`. Default will be determined based on the size of the data, `max_arr_size=[5, 10, 100]` for data with approx. size of [1000, 100, 10] size 100 arrays. Otherwise `max_array_size=-1` which means to render all arrays.
+        """
+        from IPython.display import display
+
+        display(parameterized_repr_html_class(self, open, title, max_arr_size))
+
+
+def path_to_path_list(path):
+    """Converts an object path string to a `path_list`.
+
+    Parameters
+    ----------
+    path : str
+        Object path (e.g. `"key1.key2.key3[0].key4"`)
+
+    Returns
+    -------
+    list
+        List of keys and indices (e.g. `["key1","key2","key3",0,"key4"]`)
+
+    Raises
+    ------
+    ValueError
+        If indices are not integers (e.g. `key1[key3]` will fail)
+    """
+    path_list = []
+    for name in path.split(".") if path.split(".") else [path]:
+        if "[" in name:
+            names = re.findall(r"([^\[\]]+)", name)
+            if len(names) > 1:
+                path_list.append(names[0])
+                names = names[1:]
+            for index in names:
+                if not index.replace("-", "").replace("+", "").isdigit():
+                    raise ValueError(
+                        f"Indices need to integers: index={index} (name={name}, path={path})"
+                    )
+                path_list.append(int(index))
+        else:
+            path_list.append(name)
+    return path_list