oso-oso 0.2.2 → 0.4.0

data/lib/oso/polar/polar.rb

@@ -1,19 +1,44 @@
 # frozen_string_literal: true
 
 require 'json'
+require 'pp'
 require 'set'
+require 'digest/md5'
+
+# Missing Ruby type.
+module PolarBoolean; end
+# Monkey-patch Ruby true type.
+class TrueClass; include PolarBoolean; end
+# Monkey-patch Ruby false type.
+class FalseClass; include PolarBoolean; end
 
 module Oso
   module Polar
     # Create and manage an instance of the Polar runtime.
     class Polar # rubocop:disable Metrics/ClassLength
+      # @return [Host]
+      attr_reader :host
+
       def initialize
-        @ffi_instance = FFI::Polar.create
-        @calls = {}
-        @classes = {}
-        @constructors = {}
-        @instances = {}
-        @load_queue = Set.new
+        @ffi_polar = FFI::Polar.create
+        @host = Host.new(ffi_polar)
+        @loaded_names = {}
+        @loaded_contents = {}
+
+        # Register built-in classes.
+        register_class PolarBoolean, name: 'Boolean'
+        register_class Integer
+        register_class Float
+        register_class Array, name: 'List'
+        register_class Hash, name: 'Dictionary'
+        register_class String
+      end
+
+      # Replace the current Polar instance but retain all registered classes and constructors.
+      def clear
+        loaded_names.clear
+        loaded_contents.clear
+        @ffi_polar = FFI::Polar.create
       end
 
       # Enqueue a Polar policy file for loading into the KB.
@@ -21,11 +46,25 @@ module Oso
       # @param name [String]
       # @raise [PolarFileExtensionError] if provided filename has invalid extension.
       # @raise [PolarFileNotFoundError] if provided filename does not exist.
-      def load_file(name)
-        raise PolarFileExtensionError unless ['.pol', '.polar'].include? File.extname(name)
-        raise PolarFileNotFoundError, name unless File.file?(name)
+      def load_file(name) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+        raise PolarFileExtensionError, name unless File.extname(name) == '.polar'
 
-        load_queue << name
+        file_data = File.open(name, &:read)
+        hash = Digest::MD5.hexdigest(file_data)
+
+        if loaded_names.key?(name)
+          raise PolarFileAlreadyLoadedError, name if loaded_names[name] == hash
+
+          raise PolarFileContentsChangedError, name
+        elsif loaded_contents.key?(hash)
+          raise PolarFileNameChangedError, name, loaded_contents[hash]
+        else
+          load_str(file_data, filename: name)
+          loaded_names[name] = hash
+          loaded_contents[hash] = name
+        end
+      rescue Errno::ENOENT
+        raise PolarFileNotFoundError, name
       end
 
       # Load a Polar string into the KB.
@@ -38,344 +77,118 @@ module Oso
       def load_str(str, filename: nil) # rubocop:disable Metrics/MethodLength
         raise NullByteInPolarFileError if str.chomp("\0").include?("\0")
 
-        ffi_instance.load_str(str, filename: filename)
+        ffi_polar.load_str(str, filename: filename)
         loop do
-          next_query = ffi_instance.next_inline_query
+          next_query = ffi_polar.next_inline_query
           break if next_query.nil?
 
           begin
-            Query.new(next_query, polar: self).results.next
+            Query.new(next_query, host: host).results.next
           rescue StopIteration
             raise InlineQueryFailedError
           end
         end
       end
 
-      # Replace the current Polar instance but retain all registered classes and
-      # constructors.
-      def clear
-        @ffi_instance = FFI::Polar.create
+      # Query for a predicate, parsing it if necessary.
+      #
+      # @overload query(query)
+      #   @param query [String]
+      #   @return [Enumerator] of resulting bindings
+      #   @raise [Error] if the FFI call raises one.
+      # @overload query(query)
+      #   @param query [Predicate]
+      #   @return [Enumerator] of resulting bindings
+      #   @raise [Error] if the FFI call raises one.
+      def query(query)
+        new_host = host.dup
+        case query
+        when String
+          ffi_query = ffi_polar.new_query_from_str(query)
+        when Predicate
+          ffi_query = ffi_polar.new_query_from_term(new_host.to_polar_term(query))
+        else
+          raise InvalidQueryTypeError
+        end
+        Query.new(ffi_query, host: new_host).results
       end
 
-      # Query for a predicate.
+      # Query for a rule.
       #
       # @param name [String]
       # @param args [Array<Object>]
+      # @return [Enumerator] of resulting bindings
       # @raise [Error] if the FFI call raises one.
-      def query_pred(name, args:)
-        clear_query_state
-        load_queued_files
-        pred = Predicate.new(name, args: args)
-        query_ffi_instance = ffi_instance.new_query_from_term(to_polar_term(pred))
-        Query.new(query_ffi_instance, polar: self).results
+      def query_rule(name, *args)
+        query(Predicate.new(name, args: args))
       end
 
       # Start a REPL session.
       #
       # @raise [Error] if the FFI call raises one.
-      def repl # rubocop:disable Metrics/MethodLength
-        clear_query_state
-        load_queued_files
-        loop do
-          query = Query.new(ffi_instance.new_query_from_repl, polar: self)
-          results = query.results.to_a
+      def repl(load: false) # rubocop:disable Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity, Metrics/AbcSize
+        ARGV.map { |f| load_file(f) } if load
+
+        loop do # rubocop:disable Metrics/BlockLength
+          print 'query> '
+          begin
+            query = $stdin.readline.chomp.chomp(';')
+          rescue EOFError
+            return
+          end
+
+          begin
+            ffi_query = ffi_polar.new_query_from_str(query)
+          rescue ParseError => e
+            puts "Parse error: #{e}"
+            next
+          end
+
+          begin
+            results = Query.new(ffi_query, host: host).results.to_a
+          rescue PolarRuntimeError => e
+            puts e
+            next
+          end
+
           if results.empty?
-            puts 'False'
+            pp false
           else
             results.each do |result|
-              puts result
+              if result.empty?
+                pp true
+              else
+                pp result
+              end
             end
           end
         end
       end
 
-      # Get a unique ID from Polar.
-      #
-      # @return [Integer]
-      # @raise [Error] if the FFI call raises one.
-      def new_id
-        ffi_instance.new_id
-      end
-
       # Register a Ruby class with Polar.
       #
       # @param cls [Class]
       # @param name [String]
       # @param from_polar [Proc]
       # @raise [InvalidConstructorError] if provided an invalid 'from_polar' constructor.
-      def register_class(cls, name: nil, from_polar: nil) # rubocop:disable Naming/MethodParameterName
-        # TODO(gj): should this take 3 args: cls (Class), constructor_cls
-        # (Option<Class>) that defaults to cls, and constructor_method
-        # (Option<Symbol>) that defaults to :new?
-        name = cls.name if name.nil?
-        raise DuplicateClassAliasError, name: name, old: get_class(name), new: cls if classes.key? name
-
-        classes[name] = cls
-        if from_polar.nil?
-          constructors[name] = :new
-        elsif from_polar.respond_to? :call
-          constructors[name] = from_polar
-        else
-          raise InvalidConstructorError
-        end
-
+      def register_class(cls, name: nil, from_polar: nil)
+        from_polar = Proc.new if block_given?
+        name = host.cache_class(cls, name: name, constructor: from_polar)
         register_constant(name, value: cls)
       end
 
       def register_constant(name, value:)
-        ffi_instance.register_constant(name, value: to_polar_term(value))
-      end
-
-      # Register a Ruby method call, wrapping the call result in a generator if
-      # it isn't already one.
-      #
-      # @param method [#to_sym]
-      # @param call_id [Integer]
-      # @param instance [Hash]
-      # @param args [Array<Hash>]
-      # @raise [InvalidCallError] if the method doesn't exist on the instance or
-      #   the args passed to the method are invalid.
-      def register_call(method, call_id:, instance:, args:)
-        return if calls.key?(call_id)
-
-        args = args.map { |a| to_ruby(a) }
-        if instance["value"].has_key? "ExternalInstance"
-          instance_id = instance["value"]["ExternalInstance"]["instance_id"]
-          instance = get_instance(instance_id)
-        else
-          instance = to_ruby(instance)
-        end
-        result = instance.__send__(method, *args)
-        result = [result].to_enum unless result.is_a? Enumerator # Call must be a generator.
-        calls[call_id] = result.lazy
-      rescue ArgumentError, NoMethodError
-        raise InvalidCallError
-      end
-
-      # Retrieve the next result from a registered call and pass it to {#to_polar_term}.
-      #
-      # @param id [Integer]
-      # @return [Hash]
-      # @raise [StopIteration] if the call has been exhausted.
-      def next_call_result(id)
-        to_polar_term(calls[id].next)
-      end
-
-      # Construct and cache a Ruby instance.
-      #
-      # @param cls_name [String]
-      # @param fields [Hash<String, Hash>]
-      # @param id [Integer]
-      # @raise [PolarRuntimeError] if instance construction fails.
-      def make_instance(cls_name, fields:, id:) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-        constructor = get_constructor(cls_name)
-        fields = Hash[fields.map { |k, v| [k.to_sym, to_ruby(v)] }]
-        instance = if constructor == :new
-                     if fields.empty?
-                       get_class(cls_name).__send__(:new)
-                     else
-                       get_class(cls_name).__send__(:new, **fields)
-                     end
-                   elsif fields.empty?
-                     constructor.call
-                   else
-                     constructor.call(**fields)
-                   end
-        cache_instance(instance, id: id)
-      rescue StandardError => e
-        raise PolarRuntimeError, "Error constructing instance of #{cls_name}: #{e}"
-      end
-
-      # Check if an instance has been cached.
-      #
-      # @param id [Integer]
-      # @return [Boolean]
-      def instance?(id)
-        instances.key? id
-      end
-
-      # Fetch a Ruby instance from the {#instances} cache.
-      #
-      # @param id [Integer]
-      # @return [Object]
-      # @raise [UnregisteredInstanceError] if the ID has not been registered.
-      def get_instance(id)
-        raise UnregisteredInstanceError, id unless instance? id
-
-        instances[id]
-      end
-
-      # Check if the left class is more specific than the right class for the
-      # given instance.
-      #
-      # @param instance_id [Integer]
-      # @param left_tag [String]
-      # @param right_tag [String]
-      # @return [Boolean]
-      def subspecializer?(instance_id, left_tag:, right_tag:)
-        mro = get_instance(instance_id).class.ancestors
-        mro.index(get_class(left_tag)) < mro.index(get_class(right_tag))
-      rescue StandardError
-        false
-      end
-
-      # Check if instance is an instance of class.
-      #
-      # @param instance_id [Integer]
-      # @param class_tag [String]
-      # @return [Boolean]
-      def isa?(instance_id, class_tag:)
-        instance = get_instance(instance_id)
-        cls = get_class(class_tag)
-        instance.is_a? cls
-      rescue PolarRuntimeError
-        false
-      end
-
-      # Check if two instances unify
-      #
-      # @param left_instance_id [Integer]
-      # @param right_instance_id [Integer]
-      # @return [Boolean]
-      def unify?(left_instance_id, right_instance_id)
-        left_instance = get_instance(left_instance_id)
-        right_instance = get_instance(right_instance_id)
-        left_instance == right_instance
-      rescue PolarRuntimeError
-        false
-      end
-
-      # Turn a Ruby value into a Polar term that's ready to be sent across the
-      # FFI boundary.
-      #
-      # @param value [Object]
-      # @return [Hash<String, Object>]
-      def to_polar_term(value) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength
-        value = case true # rubocop:disable Lint/LiteralAsCondition
-                when value.instance_of?(TrueClass) || value.instance_of?(FalseClass)
-                  { 'Boolean' => value }
-                when value.instance_of?(Integer)
-                  { 'Number' => { 'Integer' => value } }
-                when value.instance_of?(Float)
-                  { 'Number' => { 'Float' => value } }
-                when value.instance_of?(String)
-                  { 'String' => value }
-                when value.instance_of?(Array)
-                  { 'List' => value.map { |el| to_polar_term(el) } }
-                when value.instance_of?(Hash)
-                  { 'Dictionary' => { 'fields' => value.transform_values { |v| to_polar_term(v) } } }
-                when value.instance_of?(Predicate)
-                  { 'Call' => { 'name' => value.name, 'args' => value.args.map { |el| to_polar_term(el) } } }
-                when value.instance_of?(Variable)
-                  # This is supported so that we can query for unbound variables
-                  { 'Variable' => value }
-                else
-                  { 'ExternalInstance' => { 'instance_id' => cache_instance(value) } }
-                end
-        { 'value' => value }
-      end
-
-      # Turn a Polar term passed across the FFI boundary into a Ruby value.
-      #
-      # @param data [Hash<String, Object>]
-      # @option data [Integer] :id
-      # @option data [Integer] :offset Character offset of the term in its source string.
-      # @option data [Hash<String, Object>] :value
-      # @return [Object]
-      # @raise [UnexpectedPolarTypeError] if type cannot be converted to Ruby.
-      def to_ruby(data) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength
-        tag, value = data['value'].first
-        case tag
-        when 'String', 'Boolean'
-          value
-        when 'Number'
-          value.values.first
-        when 'List'
-          value.map { |el| to_ruby(el) }
-        when 'Dictionary'
-          value['fields'].transform_values { |v| to_ruby(v) }
-        when 'ExternalInstance'
-          get_instance(value['instance_id'])
-        when 'Call'
-          Predicate.new(value['name'], args: value['args'].map { |a| to_ruby(a) })
-        else
-          raise UnexpectedPolarTypeError, tag
-        end
-      end
-
-      # Load all queued files, flushing the {#load_queue}.
-      def load_queued_files
-        load_queue.reject! do |filename|
-          File.open(filename) { |file| load_str(file.read, filename: filename) }
-          true
-        end
+        ffi_polar.register_constant(name, value: host.to_polar_term(value))
       end
 
       private
 
-      # @return [Hash<Integer, Enumerator>]
-      attr_reader :calls
-      # @return [Hash<String, Class>]
-      attr_reader :classes
-      # @return [Hash<String, Object>]
-      attr_reader :constructors
       # @return [FFI::Polar]
-      attr_reader :ffi_instance
-      # @return [Hash<Integer, Object>]
-      attr_reader :instances
-      # @return [Array<String>]
-      attr_reader :load_queue
-
-      # Clear the instance and call caches.
-      def clear_query_state
-        calls.clear
-        instances.clear
-      end
-
-      # Query for a Polar string.
-      #
-      # @param str [String]
-      # @return [Enumerator]
-      def query_str(str)
-        clear_query_state
-        load_queued_files
-        query_ffi_instance = ffi_instance.new_query_from_str(str)
-        Query.new(query_ffi_instance, polar: self).results
-      end
-
-      # Cache a Ruby instance, fetching a {#new_id} if one isn't provided.
-      #
-      # @param instance [Object]
-      # @param id [Integer]
-      # @return [Integer]
-      def cache_instance(instance, id: nil)
-        id = new_id if id.nil?
-        instances[id] = instance
-        id
-      end
-
-      # Fetch a Ruby class from the {#classes} cache.
-      #
-      # @param name [String]
-      # @return [Class]
-      # @raise [UnregisteredClassError] if the class has not been registered.
-      def get_class(name)
-        raise UnregisteredClassError, name unless classes.key? name
-
-        classes[name]
-      end
-
-      # Fetch a constructor from the {#constructors} cache.
-      #
-      # @param name [String]
-      # @return [Symbol] if constructor is the default of `:new`.
-      # @return [Proc] if a custom constructor was registered.
-      # @raise [UnregisteredConstructorError] if the constructor has not been registered.
-      def get_constructor(name)
-        raise MissingConstructorError, name unless constructors.key? name
-
-        constructors[name]
-      end
+      attr_reader :ffi_polar
+      # @return [Hash<String, String>]
+      attr_reader :loaded_names
+      # @return [Hash<String, String>]
+      attr_reader :loaded_contents
     end
   end
 end