fix(Node): Expand the Node union type to include all entity types

python/stencila/schema/json.py

@@ -13,7 +13,7 @@
 from .types import Node, Entity
 
 
-def decode(serialized: str) -> typing.Union[dict, Node]:
+def decode(serialized: str) -> Node:
     """Decode JSON as a `Node`"""
     node = json.loads(serialized)
     return dict_decode(node) if isinstance(node, dict) else node
@@ -24,7 +24,7 @@ def encode(node: Node) -> str:
     return json.dumps(node, default=object_encode, indent=2)
 
 
-def dict_decode(node_dict: dict) -> typing.Union[dict, Entity]:
+def dict_decode(node_dict: dict) -> Node:
     """Convert a dictionary to an `Entity` node (if it has a `type` item)."""
     if "type" not in node_dict:
         return node_dict

python/stencila/schema/types.py

@@ -4412,9 +4412,9 @@ class CitationIntentEnumeration(Enum):
 
 
 """
-Union type for all valid nodes.
+Union type for all schema nodes, including primitives and entities
 """
-Node = Union["Entity", None, bool, int, float, str, Dict[str, Any], Array[Any]]
+Node = Union["ArrayValidator", "Article", "AudioObject", "BooleanValidator", "Brand", "CitationIntentEnumeration", "Cite", "CiteGroup", "Claim", "Code", "CodeBlock", "CodeChunk", "CodeError", "CodeExpression", "CodeFragment", "Collection", "Comment", "ConstantValidator", "ContactPoint", "CreativeWork", "Datatable", "DatatableColumn", "Date", "DefinedTerm", "Delete", "Emphasis", "EnumValidator", "Enumeration", "Figure", "Function", "Grant", "Heading", "ImageObject", "Include", "IntegerValidator", "Link", "List", "ListItem", "Mark", "Math", "MathBlock", "MathFragment", "MediaObject", "MonetaryGrant", "NontextualAnnotation", "Note", "NumberValidator", "Organization", "Paragraph", "Parameter", "Periodical", "Person", "PostalAddress", "Product", "PropertyValue", "PublicationIssue", "PublicationVolume", "Quote", "QuoteBlock", "Review", "SoftwareApplication", "SoftwareEnvironment", "SoftwareSession", "SoftwareSourceCode", "StringValidator", "Strong", "Subscript", "Superscript", "Table", "TableCell", "TableRow", "ThematicBreak", "Thing", "TupleValidator", "Validator", "Variable", "VideoObject", "VolumeMount", None, bool, int, float, str, Dict[str, Any], Array[Any]]
 
 
 """

r/R/types.R

@@ -4508,11 +4508,11 @@ MathTypes <- Union(Math, MathBlock, MathFragment)
 MediaObjectTypes <- Union(MediaObject, AudioObject, ImageObject, VideoObject)
 
 
-#' Union type for all valid nodes.
+#' Union type for all schema nodes, including primitives and entities
 #'
 #' @return A `list` of class `Union` describing valid subtypes of this type
 #' @export
-Node <- Union(Entity, "NULL", "logical", "numeric", "character", "list", Array(Any()))
+Node <- Union(ArrayValidator, Article, AudioObject, BooleanValidator, Brand, CitationIntentEnumeration, Cite, CiteGroup, Claim, Code, CodeBlock, CodeChunk, CodeError, CodeExpression, CodeFragment, Collection, Comment, ConstantValidator, ContactPoint, CreativeWork, Datatable, DatatableColumn, Date, DefinedTerm, Delete, Emphasis, EnumValidator, Enumeration, Figure, Function, Grant, Heading, ImageObject, Include, IntegerValidator, Link, List, ListItem, Mark, Math, MathBlock, MathFragment, MediaObject, MonetaryGrant, NontextualAnnotation, Note, NumberValidator, Organization, Paragraph, Parameter, Periodical, Person, PostalAddress, Product, PropertyValue, PublicationIssue, PublicationVolume, Quote, QuoteBlock, Review, SoftwareApplication, SoftwareEnvironment, SoftwareSession, SoftwareSourceCode, StringValidator, Strong, Subscript, Superscript, Table, TableCell, TableRow, ThematicBreak, Thing, TupleValidator, Validator, Variable, VideoObject, VolumeMount, "NULL", "logical", "numeric", "character", "list", Array(Any()))
 
 
 #' All type schemas that are derived from Thing

rust/src/types.rs

@@ -273,7 +273,8 @@ pub struct CodeExpression {
     pub meta: Option<Object>,
 
     /// The value of the expression when it was last evaluated.
-    pub output: Option<Node>,
+    #[serde(skip)]
+    pub output: Option<Arc<Node>>,
 
     /// The programming language of the code.
     pub programming_language: Option<String>,
@@ -1593,7 +1594,8 @@ pub struct ConstantValidator {
     pub meta: Option<Object>,
 
     /// The value that the node must have.
-    pub value: Option<Node>,
+    #[serde(skip)]
+    pub value: Option<Arc<Node>>,
 }
 impl_type!(ConstantValidator);
 
@@ -2125,7 +2127,8 @@ pub struct ListItem {
     pub is_checked: Option<Bool>,
 
     /// The item represented by this list item.
-    pub item: Option<Node>,
+    #[serde(skip)]
+    pub item: Option<Arc<Node>>,
 
     /// Metadata associated with this item.
     pub meta: Option<Object>,
@@ -2462,7 +2465,8 @@ pub struct Variable {
     pub validator: Option<ValidatorTypes>,
 
     /// The value of the variable.
-    pub value: Option<Node>,
+    #[serde(skip)]
+    pub value: Option<Arc<Node>>,
 }
 impl_type!(Variable);
 
@@ -2480,7 +2484,8 @@ pub struct Parameter {
     pub name: String,
 
     /// The default value of the parameter.
-    pub default: Option<Node>,
+    #[serde(skip)]
+    pub default: Option<Arc<Node>>,
 
     /// The identifier for this item.
     pub id: Option<String>,
@@ -2504,7 +2509,8 @@ pub struct Parameter {
     pub validator: Option<ValidatorTypes>,
 
     /// The value of the variable.
-    pub value: Option<Node>,
+    #[serde(skip)]
+    pub value: Option<Arc<Node>>,
 }
 impl_type!(Parameter);
 
@@ -4969,11 +4975,88 @@ pub enum MediaObjectTypes {
     VideoObject(VideoObject),
 }
 
-/// Union type for all valid nodes.
+/// Union type for all schema nodes, including primitives and entities
 #[derive(Clone, Debug, Serialize, Deserialize)]
 #[serde(untagged)]
 pub enum Node {
-    Entity(Entity),
+    ArrayValidator(ArrayValidator),
+    Article(Article),
+    AudioObject(AudioObject),
+    BooleanValidator(BooleanValidator),
+    Brand(Brand),
+    CitationIntentEnumeration(CitationIntentEnumeration),
+    Cite(Cite),
+    CiteGroup(CiteGroup),
+    Claim(Claim),
+    Code(Code),
+    CodeBlock(CodeBlock),
+    CodeChunk(CodeChunk),
+    CodeError(CodeError),
+    CodeExpression(CodeExpression),
+    CodeFragment(CodeFragment),
+    Collection(Collection),
+    Comment(Comment),
+    ConstantValidator(ConstantValidator),
+    ContactPoint(ContactPoint),
+    CreativeWork(CreativeWork),
+    Datatable(Datatable),
+    DatatableColumn(DatatableColumn),
+    Date(Date),
+    DefinedTerm(DefinedTerm),
+    Delete(Delete),
+    Emphasis(Emphasis),
+    EnumValidator(EnumValidator),
+    Enumeration(Enumeration),
+    Figure(Figure),
+    Function(Function),
+    Grant(Grant),
+    Heading(Heading),
+    ImageObject(ImageObject),
+    Include(Include),
+    IntegerValidator(IntegerValidator),
+    Link(Link),
+    List(List),
+    ListItem(ListItem),
+    Mark(Mark),
+    Math(Math),
+    MathBlock(MathBlock),
+    MathFragment(MathFragment),
+    MediaObject(MediaObject),
+    MonetaryGrant(MonetaryGrant),
+    NontextualAnnotation(NontextualAnnotation),
+    Note(Note),
+    NumberValidator(NumberValidator),
+    Organization(Organization),
+    Paragraph(Paragraph),
+    Parameter(Parameter),
+    Periodical(Periodical),
+    Person(Person),
+    PostalAddress(PostalAddress),
+    Product(Product),
+    PropertyValue(PropertyValue),
+    PublicationIssue(PublicationIssue),
+    PublicationVolume(PublicationVolume),
+    Quote(Quote),
+    QuoteBlock(QuoteBlock),
+    Review(Review),
+    SoftwareApplication(SoftwareApplication),
+    SoftwareEnvironment(SoftwareEnvironment),
+    SoftwareSession(SoftwareSession),
+    SoftwareSourceCode(SoftwareSourceCode),
+    StringValidator(StringValidator),
+    Strong(Strong),
+    Subscript(Subscript),
+    Superscript(Superscript),
+    Table(Table),
+    TableCell(TableCell),
+    TableRow(TableRow),
+    ThematicBreak(ThematicBreak),
+    Thing(Thing),
+    TupleValidator(TupleValidator),
+    Validator(Validator),
+    Variable(Variable),
+    VideoObject(VideoObject),
+    VolumeMount(VolumeMount),
     Null,
     Bool(Bool),
     Integer(Integer),

schema/InlineContent.schema.yaml

@@ -3,11 +3,8 @@ category: text
 status: stable
 description: Union type for valid inline content.
 $comment: |
-  Note that this definition currently does not include
-  `array` and `object` nodes (which are included in `Node`).
-  This seems incongruent, and may be changed in the future.
   The order of these types is important because it determines the
-  order of attempted coercion (particularly important for primitive types).
+  order of attempted coercion (this is particularly important for primitive types).
 anyOf:
   # Entity types in alphabetical order
   - $ref: AudioObject

schema/Node.schema.yaml

@@ -1,18 +1,18 @@
 title: Node
 category: other
 status: unstable
-description: Union type for all valid nodes.
+description: Union type for all schema nodes, including primitives and entities
 $comment: |
+  The entity types in this union are automatically inserted during the build.
   The order of these types is important because it determines the
-  order of attempted coercion (ie. parsing and reshaping to arrays)
+  order of attempted coercion (ie. parsing and reshaping to arrays).
+  Array should come last to avoid single items (e.g. a single string)
+  being coerced into an array.
 anyOf:
-  - $ref: Entity
   - type: 'null'
   - type: boolean
   - type: integer
   - type: number
   - type: string
   - type: object
-  # Array should come last to avoid single items (e.g. a single string)
-  # being coerced into an array.
   - type: array

ts/bindings/rust.ts

@@ -50,9 +50,15 @@ const pointerProperties = [
   'Organization.parentOrganization',
   'ImageObject.publisher', // recursive because publisher has `logo`
   'ImageObject.thumbnail',
+  'ListItem.item',
   'Comment.parentItem',
   'ArrayValidator.contains',
   'ArrayValidator.itemsValidator',
+  'ConstantValidator.value',
+  'CodeExpression.output',
+  'Parameter.default',
+  'Parameter.value',
+  'Variable.value',
 ]
 
 /**

ts/schema.ts

@@ -94,6 +94,9 @@ export async function build(cleanup = true): Promise<void> {
   // Process each of the schemas
   schemata.forEach((schema) => processSchema(schemas, schema))
 
+  // Update the `Node` union schema`
+  updateNodeSchema(schemas)
+
   // Generate additional schemas
   addTypesSchemas(schemas)
 
@@ -440,6 +443,20 @@ const parentSchema = (
   return parent
 }
 
+/**
+ * Add all entity types to the `Node` union schema.
+ */
+const updateNodeSchema = (schemas: Map<string, JsonSchema>): void => {
+  const entitySchema = schemas.get('Entity') as JsonSchema
+  const entityRefs = (entitySchema.descendants ?? []).map((descendant) => ({
+    $ref: `${descendant}.schema.json`,
+  }))
+
+  const nodeSchema = schemas.get('Node') as JsonSchema
+  nodeSchema.anyOf = [...(entityRefs ?? []), ...(nodeSchema.anyOf ?? [])]
+  schemas.set('Node', nodeSchema)
+}
+
 /**
  * Add `*Types` schemas to the map of schemas which
  * are the union (`anyOf`) of any descendant types