@lionweb/utilities NPM

The `utilities` package

An NPM package that can be added to a Node.js/NPM codebase as follows:

$ npm add @lionweb/utilities

It contains utilities on top of the core package, such as:

Diagram generation (PlantUML and Mermaid) from an M2.
“Textualization”, i.e. render a serialization chunk as text.
Sort the contents of a serialization chunk as much as possible, for the sake of comparing serialization chunks.
Compute metrics for a serialization chunk (also relative to a given M2).
Infer a (partial) M2 from a serialization chunk.
Generation of TypeScript type definitions from an M2.

Make withoutAnnotations not modify the original serialization chunk.
(Use the littoral-templates package for textualization — of M2s, so far. This is a technical change, not a functional one, except for maybe some extra whitespace.)
Add reference utilities that all return ReferenceValue objects:
- referenceValues(<source node>, <extraction facade>): all references within the given scope.
- incomingReferences(<target node(s)>, <search scope>, <extraction facade>): all (unique) references coming into the given target node(s) from the search scope.
- referencesToOutOfScopeNodes(<scope>, <extraction facade>) all reference targets that are not in the given scope.

Add a withoutAnnotations function that removes all annotations, and all their descendants, from a serialization chunk.
The inferLanguagesFromSerializationChunk function now detects whether a property is optional from null values in the serialization.

TS types generator now generates 1 file per language. The generator also has several fixes, and assumes that the generated code will be embedded in an ES6 module.

Rename inferLanguagesFromChunk → inferLanguagesFromSerializationChunk (for consistency).
Extend the functionality of the measure metrics function.

Add a inferLanguagesFromChunk function that infers a language from a serialization chunk.
Renaming for consistency and clarity:
- readChunk → readSerializationChunk
- tryLoad[All]AsLanguages → tryRead[All]AsLanguages
tryReadAllAsLanguages combines all serialization chunks (that are the result of serializing Languages) into one before attempting to deserialize, in order to have cross-language references resolve.

Improve PlantUML and Mermaid diagram generation:
- Primitive types now rendered as classes marked as "<>".
- Concepts or Annotations that extend the "Node" concept built into LionCore don't show that.
- Optional (single-valued) features are now shown with a question mark (?) directly after the type, instead of being marked as "<>".
- Fix/add support for annotations.
Remove all Ecore functionality (=issue #121)

Add support for annotations in diagram generation (PlantUML and Mermaid), and textualization.
Remove the shortenedSerializationChunk function, as it isn't used, nor very useful compared to the sorting and textualization.
Breaking name changes: rename asText → languageAsText, shortenedSerialization → shortenedSerializationChunk, sortedSerialization → sortedSerializationChunk.
Expose an isSerializedLanguages function.
Expose an orderedSerializationChunk function that aligns the JSON key-value pairs in a serialization chunk according to the specification of the serialization format. That means that key-value pairs appear in precisely the same order as they do in the specification, and that missing key-value pairs are put in and get their default values.
Expose a generic textual syntax for serialization chunks, through the genericTreeAsText function. This function is optionally language-aware: by providing an array of Languages as its second argument, it will try and look up LanguageEntity and Feature names instead of rendering their keys.