Generate Manifold

lib/manifold/api/schema_generator.rb

@@ -0,0 +1,84 @@
+# frozen_string_literal: true
+
+module Manifold
+  module API
+    # Handles schema generation for Manifold tables
+    class SchemaGenerator
+      VALID_OPERATORS = %w[AND OR NOT NAND NOR XOR XNOR].freeze
+
+      def initialize(dimensions_fields, manifold_yaml)
+        @dimensions_fields = dimensions_fields
+        @manifold_yaml = manifold_yaml
+      end
+
+      def dimensions_schema
+        [
+          { "type" => "STRING", "name" => "id", "mode" => "REQUIRED" },
+          { "type" => "RECORD", "name" => "dimensions", "mode" => "REQUIRED",
+            "fields" => @dimensions_fields }
+        ]
+      end
+
+      def manifold_schema
+        [
+          { "type" => "STRING", "name" => "id", "mode" => "REQUIRED" },
+          { "type" => "TIMESTAMP", "name" => "timestamp", "mode" => "REQUIRED" },
+          { "type" => "RECORD", "name" => "dimensions", "mode" => "REQUIRED",
+            "fields" => @dimensions_fields },
+          { "type" => "RECORD", "name" => "metrics", "mode" => "REQUIRED",
+            "fields" => metrics_fields }
+        ]
+      end
+
+      private
+
+      def metrics_fields
+        return [] unless @manifold_yaml["contexts"] && @manifold_yaml["metrics"]
+
+        @manifold_yaml["contexts"].map do |context_name, _context_config|
+          {
+            "name" => context_name,
+            "type" => "RECORD",
+            "mode" => "NULLABLE",
+            "fields" => context_metrics_fields
+          }
+        end
+      end
+
+      def context_metrics_fields
+        [
+          *countif_fields,
+          *sumif_fields
+        ]
+      end
+
+      def countif_fields
+        return [] unless @manifold_yaml.dig("metrics", "countif")
+
+        [{
+          "name" => @manifold_yaml["metrics"]["countif"],
+          "type" => "INTEGER",
+          "mode" => "NULLABLE"
+        }]
+      end
+
+      def sumif_fields
+        return [] unless @manifold_yaml.dig("metrics", "sumif")
+
+        @manifold_yaml["metrics"]["sumif"].keys.map do |metric_name|
+          {
+            "name" => metric_name,
+            "type" => "INTEGER",
+            "mode" => "NULLABLE"
+          }
+        end
+      end
+
+      def validate_operator!(operator)
+        return if VALID_OPERATORS.include?(operator)
+
+        raise ArgumentError, "Invalid operator: #{operator}. Valid operators are: #{VALID_OPERATORS.join(", ")}"
+      end
+    end
+  end
+end

lib/manifold/api/workspace.rb

@@ -2,6 +2,26 @@
 
 module Manifold
   module API
+    # Handles terraform configuration generation
+    class TerraformGenerator
+      def initialize(name, vectors, vector_service, manifold_yaml)
+        @name = name
+        @vectors = vectors
+        @vector_service = vector_service
+        @manifold_yaml = manifold_yaml
+      end
+
+      def generate(path)
+        config = Terraform::WorkspaceConfiguration.new(@name)
+        @vectors.each do |vector|
+          vector_config = @vector_service.load_vector_config(vector)
+          config.add_vector(vector_config)
+        end
+        config.merge_config = @manifold_yaml["dimensions"]&.fetch("merge", nil) if @manifold_yaml["dimensions"]
+        config.write(path)
+      end
+    end
+
     # Encapsulates a single manifold.
     class Workspace
       attr_reader :name, :template_path, :logger
@@ -27,11 +47,17 @@ def add
       end
 
       def generate(with_terraform: false)
-        return unless manifold_exists? && any_vectors?
+        return nil unless manifold_exists? && any_vectors?
 
+        tables_directory.mkpath
         generate_dimensions
-        generate_terraform if with_terraform
+        generate_manifold
         logger.info("Generated BigQuery dimensions table schema for workspace '#{name}'.")
+
+        return unless with_terraform
+
+        generate_terraform
+        logger.info("Generated Terraform configuration for workspace '#{name}'.")
       end
 
       def tables_directory
@@ -74,16 +100,28 @@ def generate_dimensions
         dimensions_path.write(dimensions_schema_json.concat("\n"))
       end
 
+      def generate_manifold
+        manifold_schema_path.write(manifold_schema_json.concat("\n"))
+      end
+
+      def manifold_schema_path
+        tables_directory.join("manifold.json")
+      end
+
+      def schema_generator
+        @schema_generator ||= SchemaGenerator.new(dimensions_fields, manifold_yaml)
+      end
+
+      def manifold_schema
+        schema_generator.manifold_schema
+      end
+
       def dimensions_schema
-        [
-          { "type" => "STRING", "name" => "id", "mode" => "REQUIRED" },
-          { "type" => "RECORD", "name" => "dimensions", "mode" => "REQUIRED",
-            "fields" => dimensions_fields }
-        ]
+        schema_generator.dimensions_schema
       end
 
       def dimensions_fields
-        vectors.filter_map do |vector|
+        @dimensions_fields ||= vectors.filter_map do |vector|
           logger.info("Loading vector schema for '#{vector}'.")
           @vector_service.load_vector_schema(vector)
         end
@@ -106,13 +144,12 @@ def vectors
       end
 
       def generate_terraform
-        config = Terraform::WorkspaceConfiguration.new(name)
-        vectors.each do |vector|
-          vector_config = @vector_service.load_vector_config(vector)
-          config.add_vector(vector_config)
-        end
-        config.merge_config = manifold_yaml["dimensions"]&.fetch("merge", nil) if manifold_yaml["dimensions"]
-        config.write(terraform_main_path)
+        terraform_generator = TerraformGenerator.new(name, vectors, @vector_service, manifold_yaml)
+        terraform_generator.generate(terraform_main_path)
+      end
+
+      def manifold_schema_json
+        JSON.pretty_generate(manifold_schema)
       end
     end
   end

lib/manifold/templates/workspace_template.yml

@@ -8,23 +8,24 @@ dimensions:
   merge:
     source: lib/views/select_my_vector.sql
 
-metrics:
-  - name: # Add your metric name here, e.g. Pageviews
-
-    id:
-      field: # Identify the field that uniquely identifies each manifold vector
-      type: # Specify the type of that field, e.g. INTEGER
+timestamp:
+  interval: HOUR
+  field: timestamp
 
-    interval:
-      type: # Specify the interval type, e.g. TIMESTAMP or DATE
-      expression: # Compute the interval for the entry, e.g. TIMESTAMP_TRUNC(timestamp, HOUR)
+contexts:
+  paid: IS_PAID(context.location)
+  organic: IS_ORGANIC(context.location)
+  paidOrganic:
+    fields:
+      - paid
+      - organic
+    operator: AND
 
-    aggregations:
-      # Add any aggregations this metric should present
+metrics:
+  countif: tapCount
+  sumif:
+    sequenceSum:
+      field: context.sequence
 
-    source:
-      type: BIGQUERY_TABLE
-      project: # Add your project name here
-      dataset: # Add your dataset name here
-      table: # Add your table name
-      filter: # (optional) Add your filter condition here
+source: my_project.my_dataset.my_table
+filter: timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 90 DAY)

lib/manifold/terraform/workspace_configuration.rb

@@ -53,13 +53,28 @@ def dataset_config
 
       def table_config
         {
-          "dimensions" => {
-            "dataset_id" => name,
-            "project" => "${var.project_id}",
-            "table_id" => "Dimensions",
-            "schema" => "${file(\"${path.module}/tables/dimensions.json\")}",
-            "depends_on" => ["google_bigquery_dataset.#{name}"]
-          }
+          "dimensions" => dimensions_table_config,
+          "manifold" => manifold_table_config
+        }
+      end
+
+      def dimensions_table_config
+        {
+          "dataset_id" => name,
+          "project" => "${var.project_id}",
+          "table_id" => "Dimensions",
+          "schema" => "${file(\"${path.module}/tables/dimensions.json\")}",
+          "depends_on" => ["google_bigquery_dataset.#{name}"]
+        }
+      end
+
+      def manifold_table_config
+        {
+          "dataset_id" => name,
+          "project" => "${var.project_id}",
+          "table_id" => "Manifold",
+          "schema" => "${file(\"${path.module}/tables/manifold.json\")}",
+          "depends_on" => ["google_bigquery_dataset.#{name}"]
         }
       end
 

lib/manifold/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Manifold
-  VERSION = "0.0.14"
+  VERSION = "0.0.15"
 end