Skip to main content

Authoring Languages for LionWeb

LionWeb is an open initiative to enable interoperability among language engineering tools.

Therefore, typically one would:

  • Use one of the tools compatible with LionWeb to author a language
  • Export the language into LionWeb format and import it in other tools

Alternatively, a language can also be defined programmatically using the API provided by LionWeb Python.

Using LionWeb-compatible tools to author languages​

In most real-world use cases, language definitions (or metamodels) are created using dedicated language workbenches or modeling tools. These tools provide expressive, user-friendly environments to author, maintain, and evolve languages.

You may want to consider

  • JetBrains MPS: A powerful projectional editor with LionWeb export support provided through LionWeb MPS.
  • Freon: A lightweight web-based projectional editor, with support for LionWeb provided through LionWeb-Freon-M3.
  • Starlasu: A cross-platform framework for language engineering framework developed by Strumenta.

These tools allow engineers to create languages using their built-in mechanisms and then export them to LionWeb-compatible formats. Once exported, these languages can be:

  • Used in other LionWeb-aware tools.
  • Serialized to formats like JSON or Protobuf.
  • Re-imported across the ecosystem.

This workflow maximizes interoperability and reuse, allowing language definitions to move seamlessly across platforms.

Authoring Languages Programmatically​

While most users rely on external tools, it is also possible to author languages programmatically using LionWeb-Python.

Using the API in the core module, you can define metamodels directly in Python code. This gives you the flexibility to:

  • Build metamodels dynamically.
  • Serialize and persist them.
  • Use them in Python libraries and programs.
  • Export them to LionWeb formats for use elsewhere.

Supported Serialization Formats​

The LionWeb Python implementation supports serialization in JSON (standard and human-readable). It also support ProtoBuf.

A series of models can also be store in a single LionWeb Archive

Example: Defining a Language Programmatically​

The following example shows how to define a minimal language with a single concept Task that has a name property.

from lionweb.language import Language, Concept, Property, LionCoreBuiltins

from pathlib import Path

from lionweb.serialization import SerializationProvider
from lionweb.utils.language_validator import LanguageValidator

# Define the 'Task' concept
task_concept = Concept(name="Task", key="Task", id="Task-id", abstract=False, partition=False)

# Add a 'name' property
name_property = Property(name="name", key="task-name", id="task-name-id", type=LionCoreBuiltins.get_string())
task_concept.add_feature(name_property)

# Define the language container
task_language = Language(
    name="Task Language",
    key="task",
    id="task-id",
    version="1.0",
)
task_language.add_element(task_concept)

# Validate the language model
res = LanguageValidator().validate(task_language)
if res.has_errors():
    raise ValueError(f"Let's fix these errors: {res.issues}")

# Serialize to JSON
serialization = SerializationProvider.create_standard_json_serialization()
json_output = serialization.serialize_tree_to_json_string(task_language)

# Write to file
Path("task-language.json").write_text(json_output, encoding="utf-8")