chemicalml 0.2.0 → 0.3.0

data/lib/chemicalml/cli/inspect_command.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cli
+    # `chemicalml inspect <file>` — print a tree-style summary of
+    # the document: element names, ids, nested children.
+    class InspectCommand < Command
+      def run(options)
+        path = options[:file]
+        unless path
+          logger.error 'inspect requires a <file> argument'
+          return 2
+        end
+
+        doc = Chemicalml.parse(File.read(path), schema: :schema3)
+        print_tree(doc, 0)
+        0
+      rescue ArgumentError, Lutaml::Model::InvalidFormatError => e
+        logger.error "FAIL: #{e.message}"
+        2
+      end
+
+      private
+
+      def print_tree(node, depth, max_depth: 6)
+        return if depth > max_depth
+        return unless node.is_a?(Lutaml::Model::Serializable)
+
+        indent = '  ' * depth
+        line = "#{indent}#{node.class.name.split('::').last}"
+        line += " [#{node.node_id}]" if node.is_a?(Chemicalml::Cml::Visitable) && node.node_id
+        puts line
+        return unless node.is_a?(Chemicalml::Cml::Visitable)
+
+        node.wire_children.each { |c| print_tree(c, depth + 1, max_depth: max_depth) }
+      end
+    end
+  end
+end

data/lib/chemicalml/cli/validate_command.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cli
+    # `chemicalml validate <file>` — auto-detect convention and
+    # print violations. With `--json` / `-j`, emits machine-readable
+    # JSON.
+    class ValidateCommand < Command
+      def run(options)
+        path = options[:file]
+        unless path
+          logger.error 'validate requires a <file> argument'
+          return 2
+        end
+
+        logger.info "Validating #{path}" unless options[:json]
+        doc = Chemicalml.parse(File.read(path), schema: :schema3)
+        report = Chemicalml.validate(doc)
+
+        if options[:json]
+          puts json_report(report, path)
+        elsif report.ok? && !report.has_warnings?
+          logger.info "OK: #{path}"
+        else
+          logger.error report.summary
+        end
+        report.ok? ? 0 : 1
+      rescue ArgumentError, Lutaml::Model::InvalidFormatError => e
+        logger.error "FAIL: #{e.message}"
+        2
+      end
+
+      private
+
+      def json_report(report, path)
+        require 'json'
+        payload = {
+          file: path,
+          ok: report.ok?,
+          has_warnings: report.has_warnings?,
+          violations: report.violations.map do |v|
+            {
+              severity: v.severity,
+              path: v.path,
+              message: v.message,
+              value: v.value
+            }.compact
+          end
+        }
+        JSON.pretty_generate(payload)
+      end
+    end
+  end
+end

data/lib/chemicalml/cli.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  # Command-line interface. Built on Thor for dispatch; each command is
+  # a dedicated class under `Chemicalml::Cli::*Command` exposing
+  # `.run(args, options)` (and `#run` on an instance).
+  #
+  # Dispatch flow:
+  #
+  #   exe/chemicalml ARGV
+  #     -> Chemicalml::Cli::Base.start(ARGV)  (Thor)
+  #     -> Chemicalml::Cli::<Name>Command.new.run(args, options)
+  #
+  # Adding a new command:
+  #   1. Create `lib/chemicalml/cli/<name>_command.rb` with a class
+  #      extending `Chemicalml::Cli::Command` and implementing `#run`.
+  #   2. Add one `desc + def` block in `Chemicalml::Cli::Base` that
+  #      invokes `<Name>Command.new.run(args, options)`.
+  #   3. Add the autoload entry below.
+  module Cli
+    autoload :Base,           'chemicalml/cli/base'
+    autoload :Command,        'chemicalml/cli/command'
+    autoload :ValidateCommand,    'chemicalml/cli/validate_command'
+    autoload :InspectCommand,     'chemicalml/cli/inspect_command'
+    autoload :ConventionsCommand, 'chemicalml/cli/conventions_command'
+    autoload :DictionariesCommand, 'chemicalml/cli/dictionaries_command'
+    autoload :ElementsCommand,    'chemicalml/cli/elements_command'
+    autoload :ConstraintsCommand, 'chemicalml/cli/constraints_command'
+    autoload :EnumsCommand,       'chemicalml/cli/enums_command'
+    autoload :InfoCommand,        'chemicalml/cli/info_command'
+
+    # Backward-compatible entry point used by `exe/chemicalml` and tests.
+    # @param argv [Array<String>] the command line.
+    # @return [Integer] process exit code.
+    def self.run(argv)
+      Base.start(argv)
+      0
+    rescue StandardError => e
+      warn "FAIL: #{e.message}"
+      2
+    end
+  end
+end

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

@@ -7,18 +7,34 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::Abundance
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
             attribute :convention, :string
 
+            attribute :min, :string
+            attribute :max, :string
+            attribute :units, :string
             xml do
               namespace Chemicalml::Cml::Namespace
-              root "abundance"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
+              root 'abundance'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_attribute 'min', to: :min
+              map_attribute 'max', to: :max
+              map_attribute 'units', to: :units
+            end
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'min', to: :min
+              map 'max', to: :max
+              map 'units', to: :units
             end
           end
         end

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

@@ -7,20 +7,58 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::Action
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
             attribute :convention, :string
             attribute :order, :string
 
+            attribute :units, :string
+            attribute :start, :string
+            attribute :start_condition, :string
+            attribute :duration, :string
+            attribute :end, :string
+            attribute :end_condition, :string
+            attribute :type, :string
+            attribute :action_order, :string
+            attribute :count, :string
+            attribute :ref, :string
             xml do
               namespace Chemicalml::Cml::Namespace
-              root "action"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
-              map_attribute "order", to: :order
+              root 'action'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_attribute 'order', to: :order
+              map_attribute 'units', to: :units
+              map_attribute 'start', to: :start
+              map_attribute 'startCondition', to: :start_condition
+              map_attribute 'duration', to: :duration
+              map_attribute 'end', to: :end
+              map_attribute 'endCondition', to: :end_condition
+              map_attribute 'type', to: :type
+              map_attribute 'actionOrder', to: :action_order
+              map_attribute 'count', to: :count
+              map_attribute 'ref', to: :ref
+            end
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'order', to: :order
+              map 'units', to: :units
+              map 'start', to: :start
+              map 'startCondition', to: :start_condition
+              map 'duration', to: :duration
+              map 'end', to: :end
+              map 'endCondition', to: :end_condition
+              map 'type', to: :type
+              map 'actionOrder', to: :action_order
+              map 'count', to: :count
+              map 'ref', to: :ref
             end
           end
         end

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

@@ -7,23 +7,59 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::ActionList
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
             attribute :convention, :string
             attribute :order, :string
 
-                        attribute :actions, :action, collection: true
+            attribute :actions, :action, collection: true
 
+            attribute :start, :string
+            attribute :start_condition, :string
+            attribute :duration, :string
+            attribute :end, :string
+            attribute :end_condition, :string
+            attribute :units, :string
+            attribute :count, :string
+            attribute :type, :string
+            attribute :action_order, :string
             xml do
               namespace Chemicalml::Cml::Namespace
-              map_element "action", to: :actions
-              root "actionList"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
-              map_attribute "order", to: :order
+              map_element 'action', to: :actions
+              root 'actionList'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_attribute 'order', to: :order
+              map_attribute 'start', to: :start
+              map_attribute 'startCondition', to: :start_condition
+              map_attribute 'duration', to: :duration
+              map_attribute 'end', to: :end
+              map_attribute 'endCondition', to: :end_condition
+              map_attribute 'units', to: :units
+              map_attribute 'count', to: :count
+              map_attribute 'type', to: :type
+              map_attribute 'actionOrder', to: :action_order
+            end
+            key_value do
+              map 'action', to: :actions
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'order', to: :order
+              map 'start', to: :start
+              map 'startCondition', to: :start_condition
+              map 'duration', to: :duration
+              map 'end', to: :end
+              map 'endCondition', to: :end_condition
+              map 'units', to: :units
+              map 'count', to: :count
+              map 'type', to: :type
+              map 'actionOrder', to: :action_order
             end
           end
         end

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

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Alternative
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Alternative
+
+            attribute :id, :string
+            attribute :convention, :string
+            attribute :alternative_type, :string
+            attribute :type, :string
+            attribute :content, :string
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root 'alternative'
+              map_attribute 'id', to: :id
+              map_attribute 'convention', to: :convention
+              map_attribute 'alternativeType', to: :alternative_type
+              map_attribute 'type', to: :type
+              map_content to: :content
+            end
+            key_value do
+              map 'id', to: :id
+              map 'convention', to: :convention
+              map 'alternativeType', to: :alternative_type
+              map 'type', to: :type
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -7,18 +7,28 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::Amount
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
             attribute :convention, :string
 
+            attribute :units, :string
             xml do
               namespace Chemicalml::Cml::Namespace
-              root "amount"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
+              root 'amount'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_attribute 'units', to: :units
+            end
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'units', to: :units
             end
           end
         end

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

@@ -7,6 +7,7 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::Angle
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
@@ -14,15 +15,41 @@ module Chemicalml
             attribute :atomRefs3, :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 "angle"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
-              map_attribute "atomRefs3", to: :atomRefs3
-              map_attribute "units", to: :units
+              root 'angle'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_attribute 'atomRefs3', to: :atomRefs3
+              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
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'atomRefs3', to: :atomRefs3
+              map 'units', to: :units
+              map 'angleUnits', to: :angle_units
+              map 'errorValue', to: :error_value
+              map 'errorBasis', to: :error_basis
+              map 'min', to: :min
+              map 'max', to: :max
+              map 'ref', to: :ref
             end
           end
         end

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

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Annotation
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Annotation
+
+            attribute :id, :string
+            attribute :title, :string
+            attribute :dict_ref, :string
+            attribute :convention, :string
+            attribute :documentation, :documentation
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root 'annotation'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_element 'documentation', to: :documentation
+            end
+            key_value do
+              map 'documentation', to: :documentation
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -7,6 +7,7 @@ module Chemicalml
         def self.included(klass)
           klass.class_eval do
             include Chemicalml::Cml::Role::AnyCml
+
             attribute :id, :string
             attribute :title, :string
             attribute :dict_ref, :string
@@ -14,11 +15,17 @@ module Chemicalml
 
             xml do
               namespace Chemicalml::Cml::Namespace
-              root "anyCml"
-              map_attribute "id", to: :id
-              map_attribute "title", to: :title
-              map_attribute "dictRef", to: :dict_ref
-              map_attribute "convention", to: :convention
+              root 'anyCml'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+            end
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
             end
           end
         end

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

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Appinfo
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Appinfo
+
+            attribute :id, :string
+            attribute :title, :string
+            attribute :dict_ref, :string
+            attribute :convention, :string
+            attribute :content, :string
+
+            attribute :role, :string
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root 'appinfo'
+              map_attribute 'id', to: :id
+              map_attribute 'title', to: :title
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'convention', to: :convention
+              map_content to: :content
+              map_attribute 'role', to: :role
+            end
+            key_value do
+              map 'id', to: :id
+              map 'title', to: :title
+              map 'dictRef', to: :dict_ref
+              map 'convention', to: :convention
+              map 'role', to: :role
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,80 @@
+# frozen_string_literal: true
+
+module Chemicalml
+  module Cml
+    module Base
+      module Arg
+        def self.included(klass)
+          klass.class_eval do
+            include Chemicalml::Cml::Role::Arg
+
+            attribute :title, :string
+            attribute :id, :string
+            attribute :convention, :string
+            attribute :dict_ref, :string
+            attribute :ref, :string
+            attribute :name, :string
+            attribute :data_type, :string
+            attribute :substitute, :string
+            attribute :parameter_name, :string
+            attribute :parent_attribute, :string
+            attribute :delete, :string
+            attribute :eval, :string
+            attribute :content, :string
+
+            attribute :atom, :atom
+            attribute :atom_type, :atomType
+            attribute :expression, :expression
+            attribute :arrays, :array, collection: true
+            attribute :matrices, :matrix, collection: true
+            attribute :scalars, :scalar, collection: true
+
+            xml do
+              namespace Chemicalml::Cml::Namespace
+              root 'arg'
+              map_attribute 'title', to: :title
+              map_attribute 'id', to: :id
+              map_attribute 'convention', to: :convention
+              map_attribute 'dictRef', to: :dict_ref
+              map_attribute 'ref', to: :ref
+              map_attribute 'name', to: :name
+              map_attribute 'dataType', to: :data_type
+              map_attribute 'substitute', to: :substitute
+              map_attribute 'parameterName', to: :parameter_name
+              map_attribute 'parentAttribute', to: :parent_attribute
+              map_attribute 'delete', to: :delete
+              map_attribute 'eval', to: :eval
+              map_content to: :content
+              map_element 'atom', to: :atom
+              map_element 'atomType', to: :atom_type
+              map_element 'expression', to: :expression
+              map_element 'array', to: :arrays
+              map_element 'matrix', to: :matrices
+              map_element 'scalar', to: :scalars
+            end
+            key_value do
+              map 'atom', to: :atom
+              map 'atomType', to: :atom_type
+              map 'expression', to: :expression
+              map 'array', to: :arrays
+              map 'matrix', to: :matrices
+              map 'scalar', to: :scalars
+              map 'title', to: :title
+              map 'id', to: :id
+              map 'convention', to: :convention
+              map 'dictRef', to: :dict_ref
+              map 'ref', to: :ref
+              map 'name', to: :name
+              map 'dataType', to: :data_type
+              map 'substitute', to: :substitute
+              map 'parameterName', to: :parameter_name
+              map 'parentAttribute', to: :parent_attribute
+              map 'delete', to: :delete
+              map 'eval', to: :eval
+            end
+          end
+        end
+      end
+    end
+  end
+end