Streamline Generate

.rubocop.yml

@@ -3,7 +3,8 @@ inherit_from: .rubocop_todo.yml
 require: rubocop-rspec
 
 AllCops:
-  TargetRubyVersion: 3.0
+  NewCops: enable
+  TargetRubyVersion: 3.1
 
 Style/StringLiterals:
   EnforcedStyle: double_quotes

lib/manifold.rb

@@ -1,11 +1,12 @@
 # frozen_string_literal: true
 
+require "json"
 require "logger"
 require "pathname"
 require "thor"
 require "yaml"
 
-Dir[File.join(__dir__, "manifold", "**", "*.rb")].sort.each do |file|
+Dir[File.join(__dir__, "manifold", "**", "*.rb")].each do |file|
   require file
 end
 

lib/manifold/api/project.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+
+module Manifold
+  module API
+    # Projects API
+    class Project
+      attr_reader :name, :logger, :directory
+
+      def initialize(name, logger: Logger.new($stdout), directory: Pathname.pwd.join(name))
+        @name = name
+        @logger = logger
+        @directory = Pathname(directory)
+      end
+
+      def self.create(name, directory: Pathname.pwd.join(name))
+        new(name, directory:).tap do |project|
+          [project.workspaces_directory, project.vectors_directory].each(&:mkpath)
+        end
+      end
+
+      def workspaces
+        @workspaces ||= workspace_directories.map { |dir| Workspace.from_directory(dir, logger:) }
+      end
+
+      def generate
+        workspaces.each(&:generate)
+      end
+
+      def workspaces_directory
+        directory.join("workspaces")
+      end
+
+      def vectors_directory
+        directory.join("vectors")
+      end
+
+      private
+
+      def workspace_directories
+        workspaces_directory.children.select(&:directory?)
+      end
+    end
+  end
+end

lib/manifold/project/vector.rb → lib/manifold/api/vector.rb

@@ -11,8 +11,8 @@ class Vector
       ).freeze
 
       def initialize(name, template_path: DEFAULT_TEMPLATE_PATH)
-        self.name = name
-        self.template_path = Pathname(template_path)
+        @name = name
+        @template_path = Pathname(template_path)
       end
 
       def add
@@ -22,8 +22,6 @@ def add
 
       private
 
-      attr_writer :name, :template_path
-
       def directory
         Pathname.pwd.join("vectors")
       end

lib/manifold/api/workspace.rb

@@ -0,0 +1,104 @@
+# frozen_string_literal: true
+
+module Manifold
+  module API
+    # Encapsulates a single manifold.
+    class Workspace
+      attr_reader :name, :template_path, :logger
+
+      DEFAULT_TEMPLATE_PATH = File.expand_path(
+        "../templates/workspace_template.yml", __dir__
+      ).freeze
+
+      def initialize(name, template_path: DEFAULT_TEMPLATE_PATH, logger: Logger.new($stdout))
+        @name = name
+        @template_path = template_path
+        @logger = logger
+        @vector_service = Services::VectorService.new(logger)
+      end
+
+      def self.from_directory(directory, logger: Logger.new($stdout))
+        new(directory.basename.to_s, logger:)
+      end
+
+      def add
+        [tables_directory, routines_directory].each(&:mkpath)
+        FileUtils.cp(template_path, manifold_path)
+      end
+
+      def generate
+        return unless manifold_exists? && any_vectors?
+
+        generate_dimensions
+        logger.info("Generated BigQuery dimensions table schema for workspace '#{name}'.")
+      end
+
+      def tables_directory
+        directory.join("tables")
+      end
+
+      def routines_directory
+        directory.join("routines")
+      end
+
+      def manifold_file
+        return nil unless manifold_exists?
+
+        File.new(manifold_path)
+      end
+
+      def manifold_exists?
+        manifold_path.file?
+      end
+
+      def manifold_path
+        directory.join("manifold.yml")
+      end
+
+      private
+
+      def directory
+        Pathname.pwd.join("workspaces", name)
+      end
+
+      def manifold_yaml
+        @manifold_yaml ||= YAML.safe_load_file(manifold_path)
+      end
+
+      def generate_dimensions
+        dimensions_path.write(dimensions_schema_json.concat("\n"))
+      end
+
+      def dimensions_schema
+        [
+          { "type" => "STRING", "name" => "id", "mode" => "REQUIRED" },
+          { "type" => "RECORD", "name" => "dimensions", "mode" => "REQUIRED",
+            "fields" => dimensions_fields }
+        ]
+      end
+
+      def dimensions_fields
+        vectors.filter_map do |vector|
+          logger.info("Loading vector schema for '#{vector}'.")
+          @vector_service.load_vector_schema(vector)
+        end
+      end
+
+      def dimensions_schema_json
+        JSON.pretty_generate(dimensions_schema)
+      end
+
+      def dimensions_path
+        tables_directory.join("dimensions.json")
+      end
+
+      def any_vectors?
+        !(vectors.nil? || vectors.empty?)
+      end
+
+      def vectors
+        manifold_yaml["vectors"]
+      end
+    end
+  end
+end

lib/manifold/cli.rb

@@ -11,14 +11,12 @@ def initialize(*args, logger: Logger.new($stdout))
 
       self.logger = logger
       logger.level = Logger::INFO
-
-      self.bq_service = Services::BigQueryService.new(logger)
     end
 
     desc "init NAME", "Generate a new umbrella project for data management"
     def init(name)
       Manifold::API::Project.create(name)
-      logger.info "Created umbrella project '#{name}' with projects and vectors directories."
+      logger.info "Created umbrella project '#{name}' with workspaces and vectors directories."
     end
 
     desc "vectors SUBCOMMAND ...ARGS", "Manage vectors"
@@ -47,14 +45,12 @@ def add(name)
       logger.info "Added workspace '#{name}' with tables and routines directories."
     end
 
-    desc "generate PROJECT_NAME SERVICE", "Generate services for a project"
-    def generate(project_name, service)
-      case service
-      when "bq"
-        bq_service.generate_dimensions_schema(project_name)
-      else
-        logger.error("Unsupported service: #{service}")
-      end
+    desc "generate", "Generate BigQuery schema for all workspaces in the project"
+    def generate
+      name = Pathname.pwd.basename.to_s
+      project = API::Project.new(name, directory: Pathname.pwd, logger:)
+      project.generate
+      logger.info "Generated BigQuery schema for all workspaces in the project."
     end
   end
 end

lib/manifold/services/vector_service.rb

@@ -10,10 +10,7 @@ def initialize(logger)
 
       def load_vector_schema(vector_name)
         path = config_path(vector_name)
-        unless path.file?
-          @logger.error("Vector configuration not found: #{path}")
-          return nil
-        end
+        raise "Vector configuration not found: #{path}" unless path.file?
 
         config = YAML.safe_load_file(path)
         fields = transform_attributes_to_schema(config["attributes"])