Line of Dict of Elements

[ax3l]

Overwrite validation and model dump.

Alternate implementation of #13 that does not make the Python API harder to use.

[ax3l]

Also creates:

kind: Line
line:
- drift1:
    kind: Drift
    length: 0.25
- quad1:
    MagneticMultipoleP:
      Bn1: 1.0
    kind: Quadrupole
    length: 1.0
- drift2:
    kind: Drift
    length: 0.5
- quad2:
    MagneticMultipoleP:
      Bn1: -1.0
    kind: Quadrupole
    length: 1.0
- drift3:
    kind: Drift
    length: 0.5

but keeps the Python API user friendly :)

[EZoni]

Due to the fact that we override model_dump only for Line, creating and serializing an element outside a line, e.g.,

element = DriftElement(name="drift1", length=0.25)
yaml_data = yaml.dump(element.model_dump(), default_flow_style=False)
print(yaml_data)

produces

kind: Drift
length: 0.25
name: drift1

which is the default YAML output without custom serialization.

Instead, do we want this to also produce the PALS-like output,

- drift1:
    kind: Drift
    length: 0.25

as in https://github.com/campa-consortium/pals/blob/main/examples/fodo.pals.yaml?

I think it would work if we override model_dump of BaseElement with

    def model_dump(self, *args, **kwargs):
        """This makes sure the element name property is moved out and up to a one-key dictionary"""
        elem_dict = super().model_dump(*args, **kwargs)
        name = elem_dict.pop("name", None)
        if name is None:
            raise ValueError("Element missing 'name' attribute")
        data = [{name: elem_dict}]
        return data

and also override model_dump of Line with

    def model_dump(self, *args, **kwargs):
        """This makes sure the element name property is moved out and up to a one-key dictionary"""
        # Use default dump for non-line fields
        data = super().model_dump(*args, **kwargs)
        # Reformat 'line' field as list of single-key dicts
        new_line = []
        for elem in self.line:
            #  Use custom dump for each line element
            elem_dict = elem.model_dump(**kwargs)[0]
            new_line.append(elem_dict)
        data["line"] = new_line
        return data

Here's the git diff that seems to be working for me, what do you think?

$ git diff
diff --git a/schema/BaseElement.py b/schema/BaseElement.py
index 43b63b6..9b361b2 100644
--- a/schema/BaseElement.py
+++ b/schema/BaseElement.py
@@ -14,3 +14,12 @@ class BaseElement(BaseModel):
 
     # element name
     name: str
+
+    def model_dump(self, *args, **kwargs):
+        """This makes sure the element name property is moved out and up to a one-key dictionary"""
+        elem_dict = super().model_dump(*args, **kwargs)
+        name = elem_dict.pop("name", None)
+        if name is None:
+            raise ValueError("Element missing 'name' attribute")
+        data = [{name: elem_dict}]
+        return data
diff --git a/schema/Line.py b/schema/Line.py
index 48c1580..d8472ee 100644
--- a/schema/Line.py
+++ b/schema/Line.py
@@ -69,17 +69,12 @@ class Line(BaseModel):
         """This makes sure the element name property is moved out and up to a one-key dictionary"""
         # Use default dump for non-line fields
         data = super().model_dump(*args, **kwargs)
-
         # Reformat 'line' field as list of single-key dicts
         new_line = []
         for elem in self.line:
-            # The element's name is the dict key; dump all other fields except 'name'
-            elem_dict = elem.model_dump(exclude={"name"}, **kwargs)
-            name = getattr(elem, "name", None)
-            if name is None:
-                raise ValueError("Element missing 'name' attribute")
-            new_line.append({name: elem_dict})
-
+            #  Use custom dump for each line element
+            elem_dict = elem.model_dump(**kwargs)[0]
+            new_line.append(elem_dict)
         data["line"] = new_line
         return data

[ax3l]

Yes, that would be perfect! I did initially to to implement something along these lines but did not find a way, let's do your update!