JSON Object Generator¶

The JSON Object generator serialises an object model — the instances of your classes, not the classes themselves — into a single JSON document. Use it to export a concrete system state for fixtures, seed data, test payloads, or to hand a worked example of your model to another tool.

It is the instance-level counterpart of the JSON Schema generator: JSON Schema describes the shape your data must have, the JSON Object generator emits data that has that shape.

Usage¶

Create a JSONObjectGenerator, pass the ObjectModel, and call generate:

from besser.BUML.metamodel.structural import (
    Class, DomainModel, Property, StringType, IntegerType,
    BinaryAssociation, Multiplicity
)
from besser.BUML.metamodel.object import (
    ObjectModel, Object, DataValue, AttributeLink, LinkEnd, Link
)
from besser.generators.json import JSONObjectGenerator

# --- Structural model ------------------------------------------------
name_prop = Property(name="name", type=StringType)
age_prop = Property(name="age", type=IntegerType)
author_cls = Class(name="Author", attributes={name_prop, age_prop})

title_prop = Property(name="title", type=StringType)
book_cls = Class(name="Book", attributes={title_prop})

author_end = Property(name="author", type=author_cls, multiplicity=Multiplicity(1, 1))
book_end = Property(name="books", type=book_cls, multiplicity=Multiplicity(0, "*"))
writes = BinaryAssociation(name="writes", ends={author_end, book_end})

domain_model = DomainModel(
    name="Library", types={author_cls, book_cls}, associations={writes}
)

# --- Object model (the instances) ------------------------------------
alice = Object(name="alice", classifier=author_cls, slots=[
    AttributeLink(
        value=DataValue(classifier=StringType, value="Alice", name="v1"),
        attribute=name_prop,
    ),
    AttributeLink(
        value=DataValue(classifier=IntegerType, value=30, name="v2"),
        attribute=age_prop,
    ),
])
book1 = Object(name="book1", classifier=book_cls, slots=[
    AttributeLink(
        value=DataValue(classifier=StringType, value="Python 101", name="v3"),
        attribute=title_prop,
    ),
])
link = Link(name="writes_link", association=writes, connections=[
    LinkEnd(name="author", association_end=author_end, object=alice),
    LinkEnd(name="books", association_end=book_end, object=book1),
])

object_model = ObjectModel(name="LibraryObjects", objects={alice, book1})

# --- Generate ---------------------------------------------------------
generator = JSONObjectGenerator(model=object_model)
generator.generate()

Constructor parameters:

model: the ObjectModel to serialise. Passing anything else raises TypeError — this generator does not accept a DomainModel.
output_dir: output directory (default: <<current_directory>>/output).

The output file is named after the object model, with spaces replaced by underscores: the example above writes LibraryObjects.json. A model with no name falls back to object_model.json.

Output format¶

{
  "name": "LibraryObjects",
  "objects": [
    {
      "id": "alice",
      "class": "Author",
      "attributes": {
        "name": "Alice",
        "age": 30
      },
      "relationships": {
        "author": [
          "book1"
        ]
      }
    },
    {
      "id": "book1",
      "class": "Book",
      "attributes": {
        "title": "Python 101"
      },
      "relationships": {
        "books": [
          "alice"
        ]
      }
    }
  ]
}

The document is a flat list rather than a nested tree, so a cycle in your instance graph cannot produce infinite output:

name — the object model’s name. A description key is added when the model carries one in its metadata.
objects — every object, sorted case-insensitively by name.
id — the object’s name, and the key other objects reference it by.
class — the name of the object’s classifier.
attributes — one entry per attribute slot. Omitted when the object has none.
relationships — links to other objects, by id. Omitted when the object participates in no link.

Note

Each entry under relationships is keyed by the name of the object’s own association end (falling back to the association’s name when the end is unnamed), not by the name of the end it points at. In the example above, alice is attached to the author end of the writes association, so her link to book1 is listed under "author".

Value serialisation¶

Attribute values are converted to JSON-native types:

B-UML value	JSON
`date` / `datetime` / `time`	ISO-8601 string
`timedelta`	number of seconds (float)
Enumeration literal	the literal’s name, as a string
`set` / `list` / `tuple`	JSON array (sets are sorted for determinism)
Any other object with a `name_`	that name, as a string
Everything else	passed through to `json.dump` unchanged

Output is written with indent=2 and ensure_ascii=False, so non-ASCII text stays readable rather than being escaped.

Using it from the Web Modeling Editor¶

The generator is registered as jsonobject in the editor’s generator registry. It reads the project’s object diagram rather than the class diagram, and the editor downloads the result as object_model.json.