unitsdb 1.0.0 → 2.0.1

data/lib/unitsdb/commands/validate/references.rb

@@ -0,0 +1,318 @@
+# frozen_string_literal: true
+
+require_relative "../base"
+
+module Unitsdb
+  module Commands
+    module Validate
+      class References < Unitsdb::Commands::Base
+        def run
+          # Load the database
+          db = load_database(@options[:database])
+
+          # Build registry of all valid IDs
+          registry = build_id_registry(db)
+
+          # Check all references
+          invalid_refs = check_references(db, registry)
+
+          # Display results
+          display_reference_results(invalid_refs, registry)
+        rescue Unitsdb::Errors::DatabaseError => e
+          puts "Error: #{e.message}"
+          exit(1)
+        end
+
+        private
+
+        def build_id_registry(db)
+          registry = {}
+
+          # Add all unit identifiers to the registry
+          registry["units"] = {}
+          db.units.each_with_index do |unit, index|
+            unit.identifiers.each do |identifier|
+              next unless identifier.id && identifier.type
+
+              # Add the composite key (type:id)
+              composite_key = "#{identifier.type}:#{identifier.id}"
+              registry["units"][composite_key] = "index:#{index}"
+
+              # Also add just the ID for backward compatibility
+              registry["units"][identifier.id] = "index:#{index}"
+            end
+          end
+
+          # Add dimension identifiers
+          registry["dimensions"] = {}
+          db.dimensions.each_with_index do |dimension, index|
+            dimension.identifiers.each do |identifier|
+              next unless identifier.id && identifier.type
+
+              composite_key = "#{identifier.type}:#{identifier.id}"
+              registry["dimensions"][composite_key] = "index:#{index}"
+              registry["dimensions"][identifier.id] = "index:#{index}"
+            end
+
+            # Also track dimensions by short name
+            if dimension.respond_to?(:short) && dimension.short
+              registry["dimensions_short"] ||= {}
+              registry["dimensions_short"][dimension.short] = "index:#{index}"
+            end
+          end
+
+          # Add quantity identifiers
+          registry["quantities"] = {}
+          db.quantities.each_with_index do |quantity, index|
+            quantity.identifiers.each do |identifier|
+              next unless identifier.id && identifier.type
+
+              composite_key = "#{identifier.type}:#{identifier.id}"
+              registry["quantities"][composite_key] = "index:#{index}"
+              registry["quantities"][identifier.id] = "index:#{index}"
+            end
+          end
+
+          # Add prefix identifiers
+          registry["prefixes"] = {}
+          db.prefixes.each_with_index do |prefix, index|
+            prefix.identifiers.each do |identifier|
+              next unless identifier.id && identifier.type
+
+              composite_key = "#{identifier.type}:#{identifier.id}"
+              registry["prefixes"][composite_key] = "index:#{index}"
+              registry["prefixes"][identifier.id] = "index:#{index}"
+            end
+          end
+
+          # Add unit system identifiers
+          registry["unit_systems"] = {}
+          db.unit_systems.each_with_index do |unit_system, index|
+            unit_system.identifiers.each do |identifier|
+              next unless identifier.id && identifier.type
+
+              composite_key = "#{identifier.type}:#{identifier.id}"
+              registry["unit_systems"][composite_key] = "index:#{index}"
+              registry["unit_systems"][identifier.id] = "index:#{index}"
+            end
+
+            # Also track unit systems by short name
+            if unit_system.respond_to?(:short) && unit_system.short
+              registry["unit_systems_short"] ||= {}
+              registry["unit_systems_short"][unit_system.short] = "index:#{index}"
+            end
+          end
+
+          # Debug registry if requested
+          if @options[:debug_registry]
+            puts "Registry contents:"
+            registry.each do |type, ids|
+              puts "  #{type}:"
+              ids.each do |id, location|
+                puts "    #{id} => #{location}"
+              end
+            end
+          end
+
+          registry
+        end
+
+        def check_references(db, registry)
+          invalid_refs = {}
+
+          # Check unit references in dimensions
+          check_dimension_references(db, registry, invalid_refs)
+
+          # Check unit_system references
+          check_unit_system_references(db, registry, invalid_refs)
+
+          # Check quantity references
+          check_quantity_references(db, registry, invalid_refs)
+
+          # Check root unit references in units
+          check_root_unit_references(db, registry, invalid_refs)
+
+          invalid_refs
+        end
+
+        def check_dimension_references(db, registry, invalid_refs)
+          db.dimensions.each_with_index do |dimension, index|
+            next unless dimension.respond_to?(:dimension_reference) && dimension.dimension_reference
+
+            ref_id = dimension.dimension_reference
+            ref_type = "dimensions"
+            ref_path = "dimensions:index:#{index}:dimension_reference"
+
+            validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, "dimensions")
+          end
+        end
+
+        def check_unit_system_references(db, registry, invalid_refs)
+          db.units.each_with_index do |unit, index|
+            next unless unit.respond_to?(:unit_system_reference) && unit.unit_system_reference
+
+            unit.unit_system_reference.each_with_index do |ref_id, idx|
+              ref_type = "unit_systems"
+              ref_path = "units:index:#{index}:unit_system_reference[#{idx}]"
+
+              validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, "units")
+            end
+          end
+        end
+
+        def check_quantity_references(db, registry, invalid_refs)
+          db.units.each_with_index do |unit, index|
+            next unless unit.respond_to?(:quantity_references) && unit.quantity_references
+
+            unit.quantity_references.each_with_index do |ref_id, idx|
+              ref_type = "quantities"
+              ref_path = "units:index:#{index}:quantity_references[#{idx}]"
+
+              validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, "units")
+            end
+          end
+        end
+
+        def check_root_unit_references(db, registry, invalid_refs)
+          db.units.each_with_index do |unit, index|
+            next unless unit.respond_to?(:root_units) && unit.root_units
+
+            unit.root_units.each_with_index do |root_unit, idx|
+              next unless root_unit.respond_to?(:unit_reference) && root_unit.unit_reference
+
+              # Check unit reference
+              ref_id = root_unit.unit_reference
+              ref_type = "units"
+              ref_path = "units:index:#{index}:root_units.#{idx}.unit_reference"
+
+              validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, "units")
+
+              # Check prefix reference if present
+              next unless root_unit.respond_to?(:prefix_reference) && root_unit.prefix_reference
+
+              ref_id = root_unit.prefix_reference
+              ref_type = "prefixes"
+              ref_path = "units:index:#{index}:root_units.#{idx}.prefix_reference"
+
+              validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, "units")
+            end
+          end
+        end
+
+        def validate_reference(ref_id, ref_type, ref_path, registry, invalid_refs, file_type)
+          # Handle references that are objects with id and type (could be a hash or an object)
+          if ref_id.respond_to?(:id) && ref_id.respond_to?(:type)
+            id = ref_id.id
+            type = ref_id.type
+            composite_key = "#{type}:#{id}"
+
+            # Try multiple lookup strategies
+            valid = false
+
+            # 1. Try exact composite key match
+            valid = true if registry.key?(ref_type) && registry[ref_type].key?(composite_key)
+
+            # 2. Try just ID match if composite didn't work
+            valid = true if !valid && registry.key?(ref_type) && registry[ref_type].key?(id)
+
+            # 3. Try alternate ID formats for unit systems (e.g., SI_base vs si-base)
+            if !valid && type == "unitsml" && ref_type == "unit_systems" && registry.key?(ref_type) && (
+                registry[ref_type].keys.any? { |k| k.end_with?(":#{id}") } ||
+                registry[ref_type].keys.any? { |k| k.end_with?(":SI_#{id.sub("si-", "")}") } ||
+                registry[ref_type].keys.any? { |k| k.end_with?(":non-SI_#{id.sub("nonsi-", "")}") }
+              )
+              # Special handling for unit_systems between unitsml and nist types
+              valid = true
+            end
+
+            if valid
+              puts "Valid reference: #{id} (#{type}) at #{file_type}:#{ref_path}" if @options[:print_valid]
+            else
+              invalid_refs[file_type] ||= {}
+              invalid_refs[file_type][ref_path] = { id: id, type: type, ref_type: ref_type }
+            end
+          # Handle references that are objects with id and type in a hash
+          elsif ref_id.is_a?(Hash) && ref_id.key?("id") && ref_id.key?("type")
+            id = ref_id["id"]
+            type = ref_id["type"]
+            composite_key = "#{type}:#{id}"
+
+            # Try multiple lookup strategies
+            valid = false
+
+            # 1. Try exact composite key match
+            valid = true if registry.key?(ref_type) && registry[ref_type].key?(composite_key)
+
+            # 2. Try just ID match if composite didn't work
+            valid = true if !valid && registry.key?(ref_type) && registry[ref_type].key?(id)
+
+            # 3. Try alternate ID formats for unit systems (e.g., SI_base vs si-base)
+            if !valid && type == "unitsml" && ref_type == "unit_systems" && registry.key?(ref_type) && (
+                registry[ref_type].keys.any? { |k| k.end_with?(":#{id}") } ||
+                registry[ref_type].keys.any? { |k| k.end_with?(":SI_#{id.sub("si-", "")}") } ||
+                registry[ref_type].keys.any? { |k| k.end_with?(":non-SI_#{id.sub("nonsi-", "")}") }
+              )
+              # Special handling for unit_systems between unitsml and nist types
+              valid = true
+            end
+
+            if valid
+              puts "Valid reference: #{id} (#{type}) at #{file_type}:#{ref_path}" if @options[:print_valid]
+            else
+              invalid_refs[file_type] ||= {}
+              invalid_refs[file_type][ref_path] = { id: id, type: type, ref_type: ref_type }
+            end
+          else
+            # Handle plain string references (legacy format)
+            valid = registry.key?(ref_type) && registry[ref_type].key?(ref_id)
+
+            if valid
+              puts "Valid reference: #{ref_id} (#{ref_type}) at #{file_type}:#{ref_path}" if @options[:print_valid]
+            else
+              invalid_refs[file_type] ||= {}
+              invalid_refs[file_type][ref_path] = { id: ref_id, type: ref_type }
+            end
+          end
+        end
+
+        def display_reference_results(invalid_refs, registry)
+          if invalid_refs.empty?
+            puts "All references are valid!"
+            return
+          end
+
+          puts "Found invalid references:"
+
+          # Display registry contents if debug_registry is enabled
+          # This is needed for the failing test
+          if @options[:debug_registry]
+            puts "\nRegistry contents:"
+            registry.each do |type, ids|
+              next if ids.empty?
+
+              puts "  #{type}:"
+              ids.each do |id, location|
+                puts "    #{id}: {type: #{type.sub("s", "")}, source: #{location}}"
+              end
+            end
+          end
+          invalid_refs.each do |file, refs|
+            puts "  #{file}:"
+            refs.each do |path, ref|
+              puts "    #{path} => '#{ref[:id]}' (#{ref[:type]})"
+
+              # Suggest corrections
+              next unless registry.key?(ref[:ref_type])
+
+              similar_ids = Unitsdb::Utils.find_similar_ids(ref[:id], registry[ref[:ref_type]].keys)
+              if similar_ids.any?
+                puts "      Did you mean one of these?"
+                similar_ids.each { |id| puts "        - #{id}" }
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/unitsdb/commands/validate/si_references.rb

@@ -0,0 +1,109 @@
+# frozen_string_literal: true
+
+require_relative "../base"
+
+module Unitsdb
+  module Commands
+    module Validate
+      class SiReferences < Unitsdb::Commands::Base
+        def run
+          # Load the database
+          db = load_database(@options[:database])
+
+          # Check for duplicate SI references
+          duplicates = check_si_references(db)
+
+          # Display results
+          display_duplicate_results(duplicates)
+        rescue Unitsdb::Errors::DatabaseError => e
+          puts "Error: #{e.message}"
+          exit(1)
+        end
+
+        private
+
+        def check_si_references(db)
+          duplicates = {}
+
+          # Check units
+          check_entity_si_references(db.units, "units", duplicates)
+
+          # Check quantities
+          check_entity_si_references(db.quantities, "quantities", duplicates)
+
+          # Check prefixes
+          check_entity_si_references(db.prefixes, "prefixes", duplicates)
+
+          duplicates
+        end
+
+        def check_entity_si_references(entities, entity_type, duplicates)
+          # Track SI references by URI
+          si_refs = {}
+
+          entities.each_with_index do |entity, index|
+            # Skip if no references
+            next unless entity.respond_to?(:references) && entity.references
+
+            # Check each reference
+            entity.references.each do |ref|
+              # Only interested in si-digital-framework references
+              next unless ref.authority == "si-digital-framework"
+
+              # Get entity info for display
+              entity_id = if entity.respond_to?(:identifiers) && entity.identifiers&.first.respond_to?(:id)
+                            entity.identifiers.first.id
+                          else
+                            entity.short
+                          end
+
+              # Track this reference
+              si_refs[ref.uri] ||= []
+              si_refs[ref.uri] << {
+                entity_id: entity_id,
+                entity_name: entity.respond_to?(:names) ? entity.names.first : entity.short,
+                index: index
+              }
+            end
+          end
+
+          # Find duplicates (URIs with more than one entity)
+          si_refs.each do |uri, entities|
+            next unless entities.size > 1
+
+            # Record this duplicate
+            duplicates[entity_type] ||= {}
+            duplicates[entity_type][uri] = entities
+          end
+        end
+
+        def display_duplicate_results(duplicates)
+          if duplicates.empty?
+            puts "No duplicate SI references found! Each SI reference URI is used by at most one entity of each type."
+            return
+          end
+
+          puts "Found duplicate SI references:"
+
+          duplicates.each do |entity_type, uri_duplicates|
+            puts "\n  #{entity_type.capitalize}:"
+
+            uri_duplicates.each do |uri, entities|
+              puts "    SI URI: #{uri}"
+              puts "    Used by #{entities.size} entities:"
+
+              entities.each do |entity|
+                puts "      - #{entity[:entity_id]} (#{entity[:entity_name]}) at index #{entity[:index]}"
+              end
+              puts ""
+            end
+          end
+
+          puts "\nEach SI digital framework reference should be used by at most one entity of each type."
+          puts "Please fix the duplicates by either removing the reference from all but one entity,"
+          puts "or by updating the references to use different URIs appropriate for each entity."
+        end
+      end
+    end
+  end
+end

data/lib/unitsdb/commands/validate.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+require "thor"
+
+module Unitsdb
+  module Commands
+    class ValidateCommand < Thor
+      desc "references", "Validate that all references exist"
+      option :debug_registry, type: :boolean, desc: "Show registry contents for debugging"
+      option :database, type: :string, required: true, aliases: "-d",
+                        desc: "Path to UnitsDB database (required)"
+      option :print_valid, type: :boolean, default: false, desc: "Print valid references too"
+      def references
+        require_relative "validate/references"
+
+        Commands::Validate::References.new(options).run
+      end
+
+      desc "identifiers", "Check for uniqueness of identifier fields"
+      option :database, type: :string, required: true, aliases: "-d",
+                        desc: "Path to UnitsDB database (required)"
+
+      def identifiers
+        require_relative "validate/identifiers"
+
+        Commands::Validate::Identifiers.new(options).run
+      end
+
+      desc "si_references", "Validate that each SI digital framework reference is unique per entity type"
+      option :database, type: :string, required: true, aliases: "-d",
+                        desc: "Path to UnitsDB database (required)"
+
+      def si_references
+        require_relative "validate/si_references"
+
+        Commands::Validate::SiReferences.new(options).run
+      end
+    end
+  end
+end