lutaml-lml 0.1.7 → 0.2.0

data/lib/lutaml/lml/grammar/lml.parg

@@ -38,7 +38,7 @@ grammar Lml version "1.0.0" {
   boolean = ( "true" / "false" ) as boolean
   number = ( 1*%x30-39 "." 1*%x30-39 ) as float / ( 1*%x30-39 ) as number
   variable = quoted_string / 1*word
-  reference = "reference:(" ( variable *( "." variable ) ) as reference ")"
+  reference = ( "reference:(" / "ref:(" ) ( variable *( "." variable ) ) as reference ")"
   range = ( variable as start ".." variable as end ) as range
   namespaced_identifier = variable *( "::" variable )
   comment_definition = [ ws ] ( "**" / "#" ) ( *( !newline any_char ) ) as comments
@@ -49,8 +49,8 @@ grammar Lml version "1.0.0" {
   cardinality_body_definition = cardchar as min [ ".." [ cardchar as max ] ]
   cardinality = "[" cardinality_body_definition as cardinality "]"
 
-  value = boolean / reference / range / number / quoted_string
-  key_value_pair = variable as key ws [ "=" ] [ ws ] value as value
+  data_value = boolean / reference / range / number / quoted_string
+  key_value_pair = variable as key ws [ "=" ] [ ws ] data_value as value
   key_value_map = "{" [ whitespace ] ( *( key_value_pair whitespace ) ) as key_value_map "}"
   kw_visibility_modifier = "+" / "-" / "#" / "~"
 
@@ -69,8 +69,11 @@ grammar Lml version "1.0.0" {
 
   keyword_type_argument = "type" [ ws ] ( 1*nonspace ) as type [ whitespace ]
   keyword_cardinality_argument = "cardinality" [ ws ] cardinality_body_definition as cardinality [ whitespace ]
-  keyword_any_argument = [ ws ] ( 1*nonspace ) as name ws [ "=" ] [ ws ] attribute_value as value [ whitespace ]
-  keyword_attribute_options = *( keyword_type_argument / keyword_cardinality_argument / ( keyword_any_argument ) as properties )
+  keyword_any_argument = [ ws ] ( 1*nonspace ) as name ws [ "=" / ":" ] [ ws ] attribute_value as value [ whitespace ]
+  keyword_values_argument = "values" ws ( ( value_set_body ) as value_set / ( variable ) as value_set_ref ) [ whitespace ]
+  keyword_pattern_argument = ( "pattern" ws "/" ( *( ( "\\" any_char ) / ( !"/" any_char ) ) ) as pattern "/" [ whitespace ] )
+  value_set_body = "{" [ whitespace ] ( *( ( ( data_value / variable ) as item ) [ "," ] [ whitespace ] ) ) as items "}"
+  keyword_attribute_options = *( keyword_type_argument / keyword_cardinality_argument / keyword_values_argument / keyword_pattern_argument / ( keyword_any_argument ) as properties )
   keyword_attribute_body = "{" [ whitespace ] keyword_attribute_options [ whitespace ] "}"
   keyword_attribute_definition = ( "attribute" ws attribute_name [ ws ] keyword_attribute_body ) as attributes
 
@@ -89,7 +92,43 @@ grammar Lml version "1.0.0" {
 
   # ---- Definitions ----
   class_modifier = [ ( [ whitespace ] ( "abstract" / "interface" ) ) as modifier ws ]
-  class_inner_definitions = definition_body / ( !( "attribute" ws ) attribute_definition ) / keyword_attribute_definition / comment_definition / comment_multiline_definition
+  # ---- Namespaces (RS 3010 extension: lutaml-model XmlNamespace) ----
+  namespace_uri_line = ( [ ws ] "uri" ws ( quoted_string / ( 1*word ) ) as uri [ whitespace ] )
+  namespace_prefix_line = ( [ ws ] "prefix" ws ( quoted_string / ( 1*word ) ) as prefix [ whitespace ] )
+  namespace_body = "{" [ whitespace ] ( *( namespace_uri_line / namespace_prefix_line ) ) as members "}"
+  namespace_inline = ( ws ( quoted_string ) as uri [ ws "prefix" ws ( quoted_string / ( 1*word ) ) as prefix ] )
+  namespace_definition = ( [ whitespace ] "namespace" ws ( 1*word ) as name [ ws ] ( namespace_body / namespace_inline ) [ whitespace ] ) as namespaces
+
+  # ---- Serialization mappings (RS 3010 extension: sibling-linked) ----
+  serialization_body = "{" [ whitespace ] ( *( serialization_line ) ) as rules "}"
+  serialization_line = ( mapping_element_name_line / mapping_element_line / mapping_attribute_line / mapping_content_line / mapping_namespace_line / mapping_mixed_line / legacy_map_line / mapping_map_line / mapping_map_key_line / mapping_map_value_line ) [ whitespace ]
+  legacy_map_line = ( [ ws ] "map" ws dq_string as wire "," ws "to:" ws ( quoted_string / ( 1*word ) ) as to [ whitespace ] ) as legacy_map
+  mapping_element_name_line = ( [ ws ] "element" ws dq_string as element_name [ whitespace ] ) as element_name
+  mapping_element_line = ( [ ws ] "map_element" ws ( element_args ) ) as map_element
+  element_args = ( dq_string as wire [ ws ( 1*word ) as field ] / ( 1*word ) as field )
+  mapping_attribute_line = ( [ ws ] "map_attribute" ws ( attribute_args ) ) as map_attribute
+  attribute_args = ( dq_string as wire [ ws ( 1*word ) as field ] / ( 1*word ) as field )
+  mapping_content_line = ( [ ws ] "map_content" ws ( 1*word ) as field ) as map_content
+  mapping_namespace_line = ( [ ws ] "namespace" ws ( 1*word ) as namespace_ref ) as map_namespace
+  mapping_mixed_line = ( [ ws ] "mixed_content" ) as mixed_content
+  mapping_map_line = ( [ ws ] "map" ws map_args ) as map
+  map_args = ( dq_string as wire ws ( 1*word ) as field / dq_string as wire / ( 1*word ) as field )
+  mapping_map_key_line = ( [ ws ] "map_key" ws "to_instance" ws ( 1*word ) as field ) as map_key
+  mapping_map_value_line = ( [ ws ] "map_value" ws ( "to_instance" / "as_attribute" ) as mode ws ( 1*word ) as field ) as map_value
+  mapping_definition = ( [ whitespace ] "mapping" ws ( 1*word ) as format ws ( 1*word ) as target [ ws ] serialization_body [ whitespace ] ) as serialization_mappings
+
+  # ---- Mapping blocks (lutaml-model wire names) ----
+  map_name = ( quoted_string / 1*( word / "-" / "." ) )
+  map_line = [ ws ] "map" ws ( map_name ) as wire "," ws "to:" ws ( map_name ) as to [ whitespace ]
+  mapping_body = [ ws ] "{" [ whitespace ] ( *( map_line ) ) as maps "}"
+  class_mapping_section = ( [ whitespace ] "mapping" ws ( 1*word ) as format [ ws ] serialization_body [ whitespace ] ) as serialization_mappings
+
+  derive_definition = ( [ whitespace ] "derive" ws attribute_name [ ws ] ":" [ ws ] [ quotes ] ( 1*typechar ) as type [ quotes ] ) as derived_field [ whitespace ]
+  string_format_item = ( [ ws ] ( "artifact" / "root" ) as key ws ( quoted_string / 1*nonspace ) as value [ whitespace ] )
+  string_format_body = "{" [ whitespace ] ( *( string_format_item ) ) as items [ whitespace ] "}"
+  string_format_definition = ( [ whitespace ] "string_format" ws ( variable ) as format [ ws ] string_format_body ) as string_format
+
+  class_inner_definitions = definition_body / definition_line / class_mapping_section / derive_definition / string_format_definition / attribute_comma_definition / ( !( "attribute" ws ) attribute_definition ) / keyword_attribute_definition / comment_definition / comment_multiline_definition
   class_inner_definition = class_inner_definitions [ whitespace ]
   class_body = [ ws ] "{" [ whitespace ] ( *( class_inner_definition ) ) as members "}"
   parent_class = [ ws ] "<" [ ws ] ( class_name_chars ) as parent_class
@@ -98,8 +137,26 @@ grammar Lml version "1.0.0" {
   # Balanced-brace definition text: nested "{...}" and "\"-escapes allowed
   definition_content = *( ( "\\" any_char ) / ( "{" definition_content "}" ) / ( !"}" any_char ) )
   definition_body = [ ws ] "definition" [ whitespace ] "{" ( definition_content ) as definition "}"
-
-  enum_inner_definitions = definition_body / attribute_definition / comment_definition / comment_multiline_definition
+  definition_line = ( [ ws ] "definition" ws ( dq_string / sq_string ) ) as definition [ whitespace ]
+
+  member_field_line = [ ws ] attribute_name [ ws ] ( 1*nonspace ) as member_field_type [ whitespace ]
+  member_fields_definition = ( [ whitespace ] "member_fields" [ whitespace ] "{" [ whitespace ] ( *( member_field_line ) ) as member_fields [ whitespace ] "}" ) as member_fields_decl
+  attribute_comma_definition = ( [ whitespace ] "attribute" ws attribute_name "," [ ws ] ( 1*typechar ) as type [ ws ] [ "{" [ whitespace ] keyword_attribute_options [ whitespace ] "}" ] [ whitespace ] ) as attributes
+  enum_member_with_payload = ( [ whitespace ] attribute_name [ ws ] ( class_body / keyword_attribute_body ) ) as attributes
+
+  # ---- Canonical value construct (RS 3001 par. Value / par. Enums) ----
+  # `value <name> { block }` is the canonical enum member; the bare
+  # identifier member is its documented shortcut. from_table expands to
+  # value members at compile time.
+  value_name = quoted_string / map_name
+  value_inner_definition = ( definition_body / definition_line / instance / attribute_line / comment_definition / comment_multiline_definition ) [ whitespace ]
+  value_body = "{" [ whitespace ] ( *( value_inner_definition ) ) as members "}"
+  value_definition = ( [ whitespace ] "value" ws ( value_name ) as value_name [ ws ] [ value_body ] [ whitespace ] ) as value_decl
+
+  from_table_definition = ( [ whitespace ] "from_table" [ ws ] ( map_name ) as from_table_name [ ws ] [ "{" [ whitespace ] *( "artifact" [ ws ] ( map_name ) as artifact_table [ whitespace ] ) "}" ] [ whitespace ] ) as from_table_decl
+
+  enum_values_shorthand = ( [ whitespace ] "values" ws value_set_body [ whitespace ] ) as value_set_decl
+  enum_inner_definitions = definition_body / definition_line / member_fields_definition / from_table_definition / enum_values_shorthand / value_definition / enum_member_with_payload / attribute_comma_definition / attribute_definition / comment_definition / comment_multiline_definition
   enum_inner_definition = enum_inner_definitions [ whitespace ]
   enum_body = [ ws ] "{" [ whitespace ] ( *( enum_inner_definition ) ) as members "}"
   enum_definition = [ whitespace ] "enum" ws [ quotes ] class_name as name [ quotes ] [ attribute_stereotype ] [ enum_body ]
@@ -107,12 +164,14 @@ grammar Lml version "1.0.0" {
   data_type_definition = [ whitespace ] "data_type" ws [ quotes ] class_name as name [ quotes ] [ attribute_stereotype ] [ data_type_body ]
   primitive_definition = [ whitespace ] "primitive" ws [ quotes ] class_name as name [ quotes ]
 
-  diagram_inner_definitions = title_definition / caption_definition / fontname_definition / ( class_definition ) as classes / ( enum_definition ) as enums / ( primitive_definition ) as primitives / ( data_type_definition ) as data_types / ( association_definition ) as associations / comment_definition / comment_multiline_definition
+  package_body = [ ws ] "{" [ whitespace ] ( *( diagram_inner_definition ) ) as members "}"
+  package_definition = ( [ whitespace ] "package" ws class_name as name [ ws ] package_body [ whitespace ] ) as packages
+  diagram_inner_definitions = namespace_definition / mapping_definition / package_definition / title_definition / caption_definition / fontname_definition / ( class_definition ) as classes / ( enum_definition ) as enums / ( primitive_definition ) as primitives / ( data_type_definition ) as data_types / ( association_definition ) as associations / comment_definition / comment_multiline_definition
   diagram_inner_definition = diagram_inner_definitions [ whitespace ]
   diagram_body = [ ws ] "{" [ whitespace ] ( *( diagram_inner_definition ) ) as members "}"
   diagram_definition = [ whitespace ] "diagram" [ ws ] class_name as name diagram_body [ whitespace ]
   diagram_definitions = diagram_definition [ whitespace ]
-  model_definitions = ( *( diagram_inner_definition ) ) as members
+  model_definitions = ( 1*( diagram_inner_definition ) ) as members
 
   view_only_definitions = ( view_import ) as view_imports / show_directive / hide_directive
   view_inner_definitions = view_only_definitions / diagram_inner_definitions
@@ -141,18 +200,20 @@ grammar Lml version "1.0.0" {
   require_block = ( *( require_stmt [ whitespace ] ) ) as requires
 
   # ---- Instances ----
-  instance = keyword_instance / class_instance
-  keyword_instance = ( [ whitespace ] "instance" [ ws ( namespaced_identifier ) as instance_type ] [ ws ] "{" [ whitespace ] ( ( [ ws ] instance ) / attributes ) "}" ) as instance [ whitespace ]
+  instance = instance_named / keyword_instance / class_instance
+  instance_named = ( [ whitespace ] "instance" ws quoted_string as name [ ws ] [ ( namespaced_identifier ) as instance_type ] [ ws ] "{" [ whitespace ] ( ( [ ws ] instance ) / attributes ) "}" [ whitespace ] ) as instance
+  isa_definition = [ whitespace ] "isa" ws ( namespaced_identifier ) as isa [ whitespace ]
+  keyword_instance = ( [ whitespace ] "instance" [ ws ( namespaced_identifier ) as instance_type ] [ ws ] "{" [ whitespace ] [ isa_definition ] ( ( [ ws ] instance ) / attributes ) "}" ) as instance [ whitespace ]
   class_instance = ( variable as instance_type [ whitespace ] quoted_string as name [ whitespace ] [ [ whitespace ] "extends" [ whitespace ] quoted_string as parent [ whitespace ] ] "{" [ whitespace ] [ lml_instance_body ] "}" ) as instance [ whitespace ]
   lml_instance_body = lml_instance_members [ whitespace ]
   instance_template = [ whitespace ] "template" [ whitespace ] "{" [ whitespace ] attributes [ whitespace ] "}"
   lml_instance_members = ( instance_template ) as template / attributes
-  models = [ whitespace ] "models" [ whitespace ] variable as name [ whitespace ] "{" ( *( model_body ) ) as members "}" [ whitespace ]
-  model_body = ( ( class_definition ) as classes / ( enum_definition ) as enums ) [ whitespace ]
+  models = [ whitespace ] "models" [ whitespace ] variable as name [ whitespace ] "{" ( *( ( ( string_format_definition ) as default_string_formats ) / model_body ) ) as members "}" [ whitespace ]
+  model_body = diagram_inner_definition
 
   # ---- Data structures ----
-  attribute_value = instance / list / key_value_map / value / 1*( !%x0A any_char )
-  list_item = instance / value
+  attribute_value = instance / list / key_value_map / data_value / 1*( !%x0A any_char )
+  list_item = instance / data_value
   list = "[" [ whitespace ] ( *( list_item [ ws ] [ "," ] [ whitespace ] ) ) as list [ whitespace ] "]"
   collection = [ whitespace ] "collection" ws quoted_string as name [ ws ] "{" [ whitespace ] [ includes ] [ whitespace ] [ validation ] [ whitespace ] "}" [ whitespace ]
   includes = [ whitespace ] "includes" [ ws ] ( list ) as includes
@@ -160,14 +221,16 @@ grammar Lml version "1.0.0" {
   condition = [ whitespace ] "condition" ws quoted_string as condition [ whitespace ]
   import = [ whitespace ] "import" [ ws ] "{" [ whitespace ] ( *( import_definition ) ) as imports "}" [ whitespace ]
   import_definition = ( 1*nonspace ) as format_type [ ws ] quoted_string as file [ whitespace ] "{" [ whitespace ] attributes [ whitespace ] "}" [ whitespace ]
-  instances = [ whitespace ] "instances" [ whitespace ] "{" [ whitespace ] [ instances_body ] "}" [ whitespace ]
+  instances = [ whitespace ] "instances" [ ws ] [ ( quoted_string / variable ) as name ] [ whitespace ] "{" [ whitespace ] [ instances_body ] "}" [ whitespace ]
   instances_body = ( *( instances_member [ whitespace ] ) ) as instances
   instances_member = import / ( collection ) as collections / export / instance
   export = [ whitespace ] "export" [ whitespace ] "{" [ whitespace ] ( *( export_format [ whitespace ] ) ) as exports "}" [ whitespace ]
   export_format = [ whitespace ] "format" ws variable as format_type [ whitespace ] "{" [ whitespace ] attributes "}" [ whitespace ]
 
-  # ---- Root (grammar/instances.rb: Full) ----
-  diagram = [ require_block ] ( models / diagram_definitions / view_definitions / instances / instance / model_definitions )
+  # ---- Root (RS 3001: a document interleaves declarations) ----
+  diagram = [ require_block ] ( *( top_level_definition ) ) as members
+
+  top_level_definition = ( models / diagram_definitions / view_definitions / instances / instance / model_definitions ) [ whitespace ]
 }
 
 entry diagram: diagram