rambling-trie-opal 2.1.1

data/lib/rambling/trie/readers.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+# %w(plain_text).each do |file|
+#   require File.join('rambling', 'trie', 'readers', file)
+# end
+
+require 'rambling/trie/readers/plain_text'
+
+module Rambling
+  module Trie
+    # Namespace for all readers.
+    module Readers
+    end
+  end
+end

data/lib/rambling/trie/readers/plain_text.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    module Readers
+      # File reader for .txt files.
+      class PlainText
+        # Yields each word read from a .txt file.
+        # @param [String] filepath the full path of the file to load the words
+        #   from.
+        # @yield [String] Each line read from the file.
+        def each_word filepath
+          File.foreach(filepath) { |line| yield line.chomp! }
+        end
+      end
+    end
+  end
+end

data/lib/rambling/trie/serializers.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+# %w(file marshal yaml zip).each do |file|
+#   require File.join('rambling', 'trie', 'serializers', file)
+# end
+
+require 'rambling/trie/serializers/file'
+require 'rambling/trie/serializers/marshal'
+require 'rambling/trie/serializers/yaml'
+# require 'rambling/trie/serializers/zip'
+
+module Rambling
+  module Trie
+    # Namespace for all serializers.
+    module Serializers
+    end
+  end
+end

data/lib/rambling/trie/serializers/file.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    module Serializers
+      # Basic file serializer. Dumps/loads string contents from files.
+      class File
+        # Loads contents from a specified filepath.
+        # @param [String] filepath the filepath to load contents from.
+        # @return [String] all contents of the file.
+        def load filepath
+          ::File.read filepath
+        end
+
+        # Dumps contents into a specified filepath.
+        # @param [String] contents the contents to dump.
+        # @param [String] filepath the filepath to dump the contents to.
+        # @return [Numeric] number of bytes written to disk.
+        def dump contents, filepath
+          ::File.open filepath, 'w+' do |f|
+            f.write contents
+          end
+        end
+      end
+    end
+  end
+end

data/lib/rambling/trie/serializers/marshal.rb

@@ -0,0 +1,48 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    module Serializers
+      # Serializer for Ruby marshal format (.marshal) files.
+      class Marshal
+        # Creates a new Marshal serializer.
+        # @param [Serializer] serializer the serializer responsible to write to
+        #   and read from disk.
+        def initialize serializer = nil
+          @serializer = serializer || Rambling::Trie::Serializers::File.new
+        end
+
+        # Loads marshaled object from contents in filepath and deserializes it
+        # into a {Nodes::Node Node}.
+        # @param [String] filepath the full path of the file to load the
+        #   marshaled object from.
+        # @return [Nodes::Node] The deserialized {Nodes::Node Node}.
+        # @see https://ruby-doc.org/core-2.5.0/Marshal.html#method-c-load
+        #   Marshal.load
+        # @note Use of
+        #   {https://ruby-doc.org/core-2.5.0/Marshal.html#method-c-load
+        #   Marshal.load} is generally discouraged. Only use this with trusted
+        #   input.
+        def load filepath
+          ::Marshal.load serializer.load filepath
+        end
+
+        # Serializes a {Nodes::Node Node} and dumps it as a marshaled object
+        # into filepath.
+        # @param [Nodes::Node] node the node to serialize
+        # @param [String] filepath the full path of the file to dump the
+        #   marshaled object into.
+        # @return [Numeric] number of bytes written to disk.
+        # @see https://ruby-doc.org/core-2.5.0/Marshal.html#method-c-dump
+        #   Marshal.dump
+        def dump node, filepath
+          serializer.dump ::Marshal.dump(node), filepath
+        end
+
+        private
+
+        attr_reader :serializer
+      end
+    end
+  end
+end

data/lib/rambling/trie/serializers/yaml.rb

@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    module Serializers
+      # Serializer for Ruby yaml format (.yaml) files.
+      class Yaml
+        # Creates a new Yaml serializer.
+        # @param [Serializer] serializer the serializer responsible to write to
+        #   and read from disk.
+        def initialize serializer = nil
+          @serializer = serializer || Rambling::Trie::Serializers::File.new
+        end
+
+        # Loads serialized object from YAML file in filepath and deserializes
+        # it into a {Nodes::Node Node}.
+        # @param [String] filepath the full path of the file to load the
+        #   serialized YAML object from.
+        # @return [Nodes::Node] The deserialized {Nodes::Node Node}.
+        # @see https://ruby-doc.org/stdlib-2.5.0/libdoc/psych/rdoc/Psych.html#method-c-safe_load
+        #   Psych.safe_load
+        def load filepath
+          require 'yaml'
+          ::YAML.safe_load(
+            serializer.load(filepath),
+            [
+              Symbol,
+              Rambling::Trie::Nodes::Raw,
+              Rambling::Trie::Nodes::Compressed,
+            ],
+            [],
+            true,
+          )
+        end
+
+        # Serializes a {Nodes::Node Node} and dumps it as a YAML object into
+        # filepath.
+        # @param [Nodes::Node] node the node to serialize
+        # @param [String] filepath the full path of the file to dump the YAML
+        #   object into.
+        # @return [Numeric] number of bytes written to disk.
+        # @see https://ruby-doc.org/stdlib-2.5.0/libdoc/psych/rdoc/Psych.html#method-c-dump
+        #   Psych.dump
+        def dump node, filepath
+          require 'yaml'
+          serializer.dump ::YAML.dump(node), filepath
+        end
+
+        private
+
+        attr_reader :serializer
+      end
+    end
+  end
+end

data/lib/rambling/trie/serializers/zip.rb

@@ -0,0 +1,74 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    module Serializers
+      # Zip file serializer. Dumps/loads contents from zip files. Automatically
+      # detects if zip file contains `.marshal` or `.yml` file
+      class Zip
+        # Creates a new Zip serializer.
+        # @param [Configuration::Properties] properties the configuration
+        #   properties set up so far.
+        def initialize properties
+          @properties = properties
+        end
+
+        # Unzip contents from specified filepath and load in contents from
+        # unzipped files.
+        # @param [String] filepath the filepath to load contents from.
+        # @return [String] all contents of the unzipped loaded file.
+        # @see https://github.com/rubyzip/rubyzip#reading-a-zip-file Zip
+        #   reading a file
+        def load filepath
+          require 'zip'
+
+          ::Zip::File.open filepath do |zip|
+            entry = zip.entries.first
+            entry_path = path entry.name
+            entry.extract entry_path
+
+            serializer = serializers.resolve entry.name
+            serializer.load entry_path
+          end
+        end
+
+        # Dumps contents and zips into a specified filepath.
+        # @param [String] contents the contents to dump.
+        # @param [String] filepath the filepath to dump the contents to.
+        # @return [Numeric] number of bytes written to disk.
+        # @see https://github.com/rubyzip/rubyzip#basic-zip-archive-creation
+        #   Zip archive creation
+        def dump contents, filepath
+          require 'zip'
+
+          ::Zip::File.open filepath, ::Zip::File::CREATE do |zip|
+            filename = ::File.basename filepath, '.zip'
+
+            entry_path = path filename
+            serializer = serializers.resolve filename
+            serializer.dump contents, entry_path
+
+            zip.add filename, entry_path
+          end
+        end
+
+        private
+
+        attr_reader :properties
+
+        def serializers
+          properties.serializers
+        end
+
+        def tmp_path
+          properties.tmp_path
+        end
+
+        def path filename
+          require 'securerandom'
+          ::File.join tmp_path, "#{SecureRandom.uuid}-#{filename}"
+        end
+      end
+    end
+  end
+end

data/lib/rambling/trie/stringifyable.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    # Provides the String representation behavior for the trie data structure.
+    module Stringifyable
+      # String representation of the current node, if it is a terminal node.
+      # @return [String] the string representation of the current node.
+      # @raise [InvalidOperation] if node is not terminal or is root.
+      def as_word
+        if letter && !terminal?
+          raise Rambling::Trie::InvalidOperation,
+            'Cannot represent branch as a word'
+        end
+
+        to_s
+      end
+
+      # String representation of the current node.
+      # @return [String] the string representation of the current node.
+      def to_s
+        parent.to_s + letter.to_s
+      end
+    end
+  end
+end

data/lib/rambling/trie/version.rb

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+
+module Rambling
+  module Trie
+    # Current version of the rambling-trie.
+    VERSION = '2.1.1'
+  end
+end

data/rambling-trie-opal.gemspec

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+$LOAD_PATH.push File.expand_path('../lib', __FILE__)
+require 'rambling/trie/version'
+
+Gem::Specification.new do |gem|
+  gem.authors = ['Ribose Inc.']
+  gem.email = ['open.source@ribose.com']
+
+  gem.description = <<~DESCRIPTION.gsub(%r{\s+}, ' ')
+    The original Rambling Trie gem but Opal-compatible.
+  DESCRIPTION
+
+  gem.summary = 'A Ruby implementation (Opal Compatible) of the trie data structure.'
+  gem.homepage = 'http://github.com/interscript/rambling-trie'
+  gem.date = Time.now.strftime '%Y-%m-%d'
+
+  executables = `git ls-files -- bin/*`.split "\n"
+  files = `git ls-files -- {lib,*file,*.gemspec,LICENSE*,README*}`.split "\n"
+  test_files = `git ls-files -- {test,spec,features}/*`.split "\n"
+
+  gem.executables = executables.map { |f| File.basename f }
+  gem.files = files
+  gem.test_files = test_files
+  gem.require_paths = %w(lib)
+
+  gem.name = 'rambling-trie-opal'
+  gem.license = 'MIT'
+  gem.version = Rambling::Trie::VERSION
+  gem.platform = Gem::Platform::RUBY
+  gem.required_ruby_version = '~> 2.4'
+
+  gem.add_development_dependency 'rake', '~> 13.0'
+  gem.add_development_dependency 'rspec', '~> 3.9'
+  gem.add_development_dependency 'yard', '~> 0.9.25'
+end

data/spec/assets/test_words.en_US.txt

@@ -0,0 +1,23 @@
+are
+beautiful
+course
+false
+hello
+hi
+is
+it
+mine
+my
+no
+of
+ours
+today
+true
+truth
+whatever
+where
+who
+yeah
+you
+your
+yours

data/spec/assets/test_words.es_DO.txt

@@ -0,0 +1,24 @@
+claro
+donde
+dónde
+eres
+es
+eso
+falso
+hermoso
+hola
+hoy
+mi
+mío
+no
+nuestro
+ola
+quien
+quién
+si
+tu
+tus
+tuyos
+verdad
+verdadero
+🙃

data/spec/integration/rambling/trie_spec.rb

@@ -0,0 +1,87 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+require 'zip'
+
+describe Rambling::Trie do
+  let(:assets_path) { File.join ::SPEC_ROOT, 'assets' }
+
+  context 'when providing words directly' do
+    it_behaves_like 'a compressible trie' do
+      let(:trie) { Rambling::Trie.create }
+      let(:words) { %w(a couple of words for our full trie integration test) }
+
+      before do
+        trie.concat words
+      end
+    end
+  end
+
+  context 'when provided with words with unicode characters' do
+    it_behaves_like 'a compressible trie' do
+      let(:trie) { Rambling::Trie.create }
+      let(:words) do
+        %w(poquísimas palabras para nuestra prueba de integración completa 🙃)
+      end
+
+      before do
+        trie.concat words
+      end
+    end
+  end
+
+  context 'when provided with a filepath' do
+    let(:trie) { Rambling::Trie.create filepath }
+    let(:words) { File.readlines(filepath).map(&:chomp) }
+
+    context 'with english words' do
+      it_behaves_like 'a compressible trie' do
+        let(:filepath) { File.join assets_path, 'test_words.en_US.txt' }
+      end
+    end
+
+    context 'with unicode characters' do
+      it_behaves_like 'a compressible trie' do
+        let(:filepath) { File.join assets_path, 'test_words.es_DO.txt' }
+      end
+    end
+  end
+
+  describe 'dump and load' do
+    let(:words_filepath) { File.join assets_path, 'test_words.en_US.txt' }
+    let(:words) { File.readlines(words_filepath).map(&:chomp) }
+
+    context 'when serialized with Ruby marshal format (default)' do
+      it_behaves_like 'a serializable trie' do
+        let(:trie_to_serialize) { Rambling::Trie.create words_filepath }
+        let(:format) { :marshal }
+      end
+    end
+
+    context 'when serialized with YAML' do
+      it_behaves_like 'a serializable trie' do
+        let(:trie_to_serialize) { Rambling::Trie.create words_filepath }
+        let(:format) { :yml }
+      end
+    end
+
+    # context 'when serialized with zipped Ruby marshal format' do
+    #   before do
+    #     @original_on_exists_proc = ::Zip.on_exists_proc
+    #     @original_continue_on_exists_proc = ::Zip.continue_on_exists_proc
+    #     ::Zip.on_exists_proc = true
+    #     ::Zip.continue_on_exists_proc = true
+    #   end
+
+    #   after do
+    #     ::Zip.on_exists_proc = @original_on_exists_proc
+    #     ::Zip.continue_on_exists_proc = @original_continue_on_exists_proc
+    #   end
+
+    #   it_behaves_like 'a serializable trie' do
+    #     let(:trie_to_serialize) { Rambling::Trie.create words_filepath }
+    #     let(:format) { 'marshal.zip' }
+    #   end
+    # end
+  end
+end