chemicalml 0.2.0 → 0.2.1

data/lib/chemicalml/cml/base/string_array.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module StringArray
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::StringArray
+            attribute :builtin, :string
+            attribute :convention, :string
+            attribute :dict_ref, :string
+            attribute :id, :string
+            attribute :title, :string
+            attribute :min, :string
+            attribute :max, :string
+            attribute :size, :string
+            attribute :delimiter, :string
+            attribute :content, :string
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root "stringArray"
+              map_attribute "builtin", to: :builtin
+              map_attribute "convention", to: :convention
+              map_attribute "dictRef", to: :dict_ref
+              map_attribute "id", to: :id
+              map_attribute "title", to: :title
+              map_attribute "min", to: :min
+              map_attribute "max", to: :max
+              map_attribute "size", to: :size
+              map_attribute "delimiter", to: :delimiter
+              map_content to: :content
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/base/substance.rb

@@ -13,6 +13,12 @@ module Chemicalml
             attribute :names, :name, collection: true
             attribute :identifiers, :identifier, collection: true
 
+            attribute :dict_ref, :string
+            attribute :convention, :string
+            attribute :type, :string
+            attribute :ref, :string
+            attribute :count, :string
+            attribute :state, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "substance"
@@ -21,6 +27,12 @@ module Chemicalml
               map_element "molecule", to: :molecule
               map_element "name", to: :names
               map_element "identifier", to: :identifiers
+              map_attribute "dictRef", to: :dict_ref
+              map_attribute "convention", to: :convention
+              map_attribute "type", to: :type
+              map_attribute "ref", to: :ref
+              map_attribute "count", to: :count
+              map_attribute "state", to: :state
             end
           end
         end

data/lib/chemicalml/cml/base/substance_list.rb

@@ -14,6 +14,9 @@ module Chemicalml
 
                         attribute :substances, :substance, collection: true
 
+            attribute :substance_list_type, :string
+            attribute :role, :string
+            attribute :ref, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               map_element "substance", to: :substances
@@ -22,6 +25,9 @@ module Chemicalml
               map_attribute "title", to: :title
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
+              map_attribute "substanceListType", to: :substance_list_type
+              map_attribute "role", to: :role
+              map_attribute "ref", to: :ref
             end
           end
         end

data/lib/chemicalml/cml/base/symmetry.rb

@@ -14,6 +14,8 @@ module Chemicalml
             attribute :pointGroup, :string
             attribute :spaceGroup, :string
 
+            attribute :irreducible_representation, :string
+            attribute :number, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "symmetry"
@@ -23,6 +25,8 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_attribute "pointGroup", to: :pointGroup
               map_attribute "spaceGroup", to: :spaceGroup
+              map_attribute "irreducibleRepresentation", to: :irreducible_representation
+              map_attribute "number", to: :number
             end
           end
         end

data/lib/chemicalml/cml/base/system.rb

@@ -15,6 +15,8 @@ module Chemicalml
                         attribute :atom_array, :atomArray
             attribute :molecules, :molecule, collection: true
 
+            attribute :dimensionality, :string
+            attribute :periodicity, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               map_element "atomArray", to: :atom_array
@@ -24,6 +26,8 @@ module Chemicalml
               map_attribute "title", to: :title
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
+              map_attribute "dimensionality", to: :dimensionality
+              map_attribute "periodicity", to: :periodicity
             end
           end
         end

data/lib/chemicalml/cml/base/table.rb

@@ -15,6 +15,11 @@ module Chemicalml
                         attribute :table_content, :tableContent
             attribute :table_header, :tableHeader
 
+            attribute :rows, :string
+            attribute :columns, :string
+            attribute :units, :string
+            attribute :table_type, :string
+            attribute :data_type, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               map_element "tableContent", to: :table_content
@@ -24,6 +29,11 @@ module Chemicalml
               map_attribute "title", to: :title
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
+              map_attribute "rows", to: :rows
+              map_attribute "columns", to: :columns
+              map_attribute "units", to: :units
+              map_attribute "tableType", to: :table_type
+              map_attribute "dataType", to: :data_type
             end
           end
         end

data/lib/chemicalml/cml/base/table_content.rb

@@ -15,6 +15,7 @@ module Chemicalml
                         attribute :table_rows, :tableRow, collection: true
             attribute :table_row_lists, :tableRowList, collection: true
 
+            attribute :delimiter, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               map_element "tableRow", to: :table_rows
@@ -24,6 +25,7 @@ module Chemicalml
               map_attribute "title", to: :title
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
+              map_attribute "delimiter", to: :delimiter
             end
           end
         end

data/lib/chemicalml/cml/base/table_header_cell.rb

@@ -13,6 +13,11 @@ module Chemicalml
             attribute :convention, :string
             attribute :content, :string
 
+            attribute :data_type, :string
+            attribute :units, :string
+            attribute :constant_to_s_i, :string
+            attribute :multiplier_to_s_i, :string
+            attribute :unit_type, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "tableHeaderCell"
@@ -21,6 +26,11 @@ module Chemicalml
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
               map_content to: :content
+              map_attribute "dataType", to: :data_type
+              map_attribute "units", to: :units
+              map_attribute "constantToSI", to: :constant_to_s_i
+              map_attribute "multiplierToSI", to: :multiplier_to_s_i
+              map_attribute "unitType", to: :unit_type
             end
           end
         end

data/lib/chemicalml/cml/base/tcell.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Tcell
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Tcell
+            attribute :units, :string
+            attribute :data_type, :string
+            attribute :title, :string
+            attribute :id, :string
+            attribute :convention, :string
+            attribute :dict_ref, :string
+            attribute :content, :string
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root "tcell"
+              map_attribute "units", to: :units
+              map_attribute "dataType", to: :data_type
+              map_attribute "title", to: :title
+              map_attribute "id", to: :id
+              map_attribute "convention", to: :convention
+              map_attribute "dictRef", to: :dict_ref
+              map_content to: :content
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/base/torsion.rb

@@ -14,6 +14,12 @@ module Chemicalml
             attribute :atomRefs4, :string
             attribute :units, :string
 
+            attribute :angle_units, :string
+            attribute :error_value, :string
+            attribute :error_basis, :string
+            attribute :min, :string
+            attribute :max, :string
+            attribute :ref, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "torsion"
@@ -23,6 +29,12 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_attribute "atomRefs4", to: :atomRefs4
               map_attribute "units", to: :units
+              map_attribute "angleUnits", to: :angle_units
+              map_attribute "errorValue", to: :error_value
+              map_attribute "errorBasis", to: :error_basis
+              map_attribute "min", to: :min
+              map_attribute "max", to: :max
+              map_attribute "ref", to: :ref
             end
           end
         end

data/lib/chemicalml/cml/base/trow.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Trow
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Trow
+            attribute :data_type, :string
+            attribute :title, :string
+            attribute :id, :string
+            attribute :convention, :string
+            attribute :dict_ref, :string
+            attribute :content, :string
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root "trow"
+              map_attribute "dataType", to: :data_type
+              map_attribute "title", to: :title
+              map_attribute "id", to: :id
+              map_attribute "convention", to: :convention
+              map_attribute "dictRef", to: :dict_ref
+              map_content to: :content
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/base/unit.rb

@@ -17,6 +17,12 @@ module Chemicalml
             attribute :definition, :string
             attribute :description, :string
 
+            attribute :units, :string
+            attribute :name, :string
+            attribute :is_s_i, :string
+            attribute :multiplier_to_data, :string
+            attribute :power, :string
+            attribute :abbreviation, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "unit"
@@ -29,6 +35,12 @@ module Chemicalml
               map_attribute "unitType", to: :unit_type
               map_element "definition", to: :definition
               map_element "description", to: :description
+              map_attribute "units", to: :units
+              map_attribute "name", to: :name
+              map_attribute "isSI", to: :is_s_i
+              map_attribute "multiplierToData", to: :multiplier_to_data
+              map_attribute "power", to: :power
+              map_attribute "abbreviation", to: :abbreviation
             end
           end
         end

data/lib/chemicalml/cml/base/unit_list.rb

@@ -13,6 +13,12 @@ module Chemicalml
             attribute :description, :string
             attribute :units, :unit, collection: true
 
+            attribute :id, :string
+            attribute :dict_ref, :string
+            attribute :si_namespace, :string
+            attribute :dictionary_prefix, :string
+            attribute :href, :string
+            attribute :unit_list_type, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "unitList"
@@ -21,6 +27,12 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_element "description", to: :description
               map_element "unit", to: :units
+              map_attribute "id", to: :id
+              map_attribute "dictRef", to: :dict_ref
+              map_attribute "siNamespace", to: :si_namespace
+              map_attribute "dictionaryPrefix", to: :dictionary_prefix
+              map_attribute "href", to: :href
+              map_attribute "unitListType", to: :unit_list_type
             end
           end
         end

data/lib/chemicalml/cml/base/unit_type.rb

@@ -12,6 +12,11 @@ module Chemicalml
             attribute :definition, :string
             attribute :description, :string
 
+            attribute :title, :string
+            attribute :parent_s_i, :string
+            attribute :abbreviation, :string
+            attribute :preserve, :string
+            attribute :symbol, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "unitType"
@@ -19,6 +24,11 @@ module Chemicalml
               map_attribute "name", to: :name
               map_element "definition", to: :definition
               map_element "description", to: :description
+              map_attribute "title", to: :title
+              map_attribute "parentSI", to: :parent_s_i
+              map_attribute "abbreviation", to: :abbreviation
+              map_attribute "preserve", to: :preserve
+              map_attribute "symbol", to: :symbol
             end
           end
         end

data/lib/chemicalml/cml/base/unit_type_list.rb

@@ -13,6 +13,11 @@ module Chemicalml
             attribute :description, :string
             attribute :unit_types, :unitType, collection: true
 
+            attribute :id, :string
+            attribute :dict_ref, :string
+            attribute :si_namespace, :string
+            attribute :dictionary_prefix, :string
+            attribute :href, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "unitTypeList"
@@ -21,6 +26,11 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_element "description", to: :description
               map_element "unitType", to: :unit_types
+              map_attribute "id", to: :id
+              map_attribute "dictRef", to: :dict_ref
+              map_attribute "siNamespace", to: :si_namespace
+              map_attribute "dictionaryPrefix", to: :dictionary_prefix
+              map_attribute "href", to: :href
             end
           end
         end

data/lib/chemicalml/cml/base/vector3.rb

@@ -13,6 +13,7 @@ module Chemicalml
             attribute :convention, :string
             attribute :content, :string
 
+            attribute :units, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "vector3"
@@ -21,6 +22,7 @@ module Chemicalml
               map_attribute "dictRef", to: :dict_ref
               map_attribute "convention", to: :convention
               map_content to: :content
+              map_attribute "units", to: :units
             end
           end
         end

data/lib/chemicalml/cml/base/xaxis.rb

@@ -14,6 +14,9 @@ module Chemicalml
             attribute :units, :string
             attribute :multiplier, :string
 
+            attribute :ref, :string
+            attribute :multiplier_to_data, :string
+            attribute :constant_to_data, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "xaxis"
@@ -23,6 +26,9 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_attribute "units", to: :units
               map_attribute "multiplier", to: :multiplier
+              map_attribute "ref", to: :ref
+              map_attribute "multiplierToData", to: :multiplier_to_data
+              map_attribute "constantToData", to: :constant_to_data
             end
           end
         end

data/lib/chemicalml/cml/base/yaxis.rb

@@ -14,6 +14,9 @@ module Chemicalml
             attribute :units, :string
             attribute :multiplier, :string
 
+            attribute :ref, :string
+            attribute :multiplier_to_data, :string
+            attribute :constant_to_data, :string
             xml do
               namespace Chemicalml::Cml::Namespace
               root "yaxis"
@@ -23,6 +26,9 @@ module Chemicalml
               map_attribute "convention", to: :convention
               map_attribute "units", to: :units
               map_attribute "multiplier", to: :multiplier
+              map_attribute "ref", to: :ref
+              map_attribute "multiplierToData", to: :multiplier_to_data
+              map_attribute "constantToData", to: :constant_to_data
             end
           end
         end

data/lib/chemicalml/cml/base.rb

@@ -131,6 +131,23 @@ module Chemicalml
       autoload :Xaxis, "chemicalml/cml/base/xaxis"
       autoload :Yaxis, "chemicalml/cml/base/yaxis"
       autoload :ZMatrix, "chemicalml/cml/base/z_matrix"
+      autoload :Alternative, "chemicalml/cml/base/alternative"
+      autoload :Arg, "chemicalml/cml/base/arg"
+      autoload :ComplexObject, "chemicalml/cml/base/complex_object"
+      autoload :Expression, "chemicalml/cml/base/expression"
+      autoload :Float, "chemicalml/cml/base/float"
+      autoload :FloatArray, "chemicalml/cml/base/float_array"
+      autoload :Integer, "chemicalml/cml/base/integer"
+      autoload :IntegerArray, "chemicalml/cml/base/integer_array"
+      autoload :Operator, "chemicalml/cml/base/operator"
+      autoload :RelatedEntry, "chemicalml/cml/base/related_entry"
+      autoload :String, "chemicalml/cml/base/string"
+      autoload :StringArray, "chemicalml/cml/base/string_array"
+      autoload :Tcell, "chemicalml/cml/base/tcell"
+      autoload :Trow, "chemicalml/cml/base/trow"
+      autoload :Annotation, "chemicalml/cml/base/annotation"
+      autoload :Appinfo, "chemicalml/cml/base/appinfo"
+      autoload :Enumeration, "chemicalml/cml/base/enumeration"
     end
   end
 end

data/lib/chemicalml/cml/elements.rb

@@ -133,13 +133,48 @@ module Chemicalml
         ZMatrix:             :zMatrix,
       }.freeze
 
-      # Elements that exist in Schema 3 but NOT in Schema 2.4.
-      SCHEMA3_ONLY = %i[Module].freeze
+      # Elements declared in the Schema 3 XSD but absent from the
+      # Schema 2.4 XSD. Verified by diffing the actual XSDs in
+      # `reference-docs/schemas/`: Schema 3 declares `<anyCml>` while
+      # Schema 2.4 does not. Conversely, Schema 2.4 declares 17 legacy
+      # elements (`alternative`, `annotation`, `appinfo`, `arg`,
+      # `complexObject`, `enumeration`, `expression`, `float`,
+      # `floatArray`, `integer`, `integerArray`, `operator`,
+      # `relatedEntry`, `string`, `stringArray`, `tcell`, `trow`) that
+      # are either redundant with Schema 3's unified types or outside
+      # the gem's scope.
+      SCHEMA3_ONLY = %i[AnyCml].freeze
+
+      # Subset of the Schema-2.4-only legacy elements that the gem
+      # explicitly models. Maps Ruby class name → XML element id so
+      # `Schema24::Configuration` can register them without polluting
+      # `Elements::ALL` (which mirrors Schema 3's element set).
+      SCHEMA24_ONLY = {
+        Alternative:   :alternative,
+        Annotation:    :annotation,
+        Appinfo:       :appinfo,
+        Arg:           :arg,
+        ComplexObject: :complexObject,
+        Enumeration:   :enumeration,
+        Expression:    :expression,
+        Float:         :float,
+        FloatArray:    :floatArray,
+        Integer:       :integer,
+        IntegerArray:  :integerArray,
+        Operator:      :operator,
+        RelatedEntry:  :relatedEntry,
+        String:        :string,
+        StringArray:   :stringArray,
+        Tcell:         :tcell,
+        Trow:          :trow
+      }.freeze
 
       # Reverse index: XML element name (String) → Ruby class name
       # (Symbol). O(1) lookup for VersionedParser root detection.
-      XML_TO_CLASS = ALL.each_with_object({}) do |(cls, xml_id), h|
-        h[xml_id.to_s] = cls
+      # Includes both Schema 3 elements and the modeled Schema 2.4
+      # legacy elements.
+      XML_TO_CLASS = [ALL, SCHEMA24_ONLY].each_with_object({}) do |src, h|
+        src.each { |cls, xml_id| h[xml_id.to_s] = cls }
       end.freeze
     end
   end

data/lib/chemicalml/cml/role/alternative.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Alternative
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/annotation.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Annotation
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/appinfo.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Appinfo
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/arg.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Arg
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/cml_module.rb

@@ -4,9 +4,8 @@ module Chemicalml
   module Cml
     module Role
       # Type marker for any wire class implementing the CML
-      # <module> element. Included by Schema3::Module.
-      # (Schema24::Module does not exist - schema 2.4 lacks the
-      # generic <module> element.)
+      # <module> element. Included by both Schema3::Module and
+      # Schema24::Module.
       module Module
       end
     end

data/lib/chemicalml/cml/role/complex_object.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module ComplexObject
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/enumeration.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Enumeration
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/expression.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Expression
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/float.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Float
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/float_array.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module FloatArray
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/integer.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module Integer
+      end
+    end
+  end
+end

data/lib/chemicalml/cml/role/integer_array.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Role
+      module IntegerArray
+      end
+    end
+  end
+end