manifold-cli 0.0.16 → 0.0.18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 38a614204bb12682a02b29ea889a9f44a6aab5af58b746322ba607252f9ff272
-  data.tar.gz: 1ee65656d387eabc11fe6eb9effcf6119ec28210d32efc7bf0b8ae57ea37c43b
+  metadata.gz: 26f9112132a14f5bbb3cd123dc4d399ed9d618ec41638eaf43b5bbb0f870b6dd
+  data.tar.gz: cc8bc857ca5163f2e86ffe8656fe48b8fb46c3295bfa7f3ffad531597e0b15df
 SHA512:
-  metadata.gz: 762a4575c8423177ccbd69622da95d034aa311f51733f4f6d123199ec35af390acee194ce543c907f127939ed6dae8c15de97d11183b2bb508b0b3a8f1195e02
-  data.tar.gz: 359ae735a6be1fec84b46a0410395feaaef0b12fee3df69f887cfc7723b57d58341dc1684ea93f401b91b5e7510e1bb085175af2bc193578471f72f4b85c955a
+  metadata.gz: dba055e83e3ef141fd49c8af5d7079d33a81ac6a54539c32270e86ecbd73840ee183f611e99f7c985c1b7c5cbb7bd32b2ac650abc8faa85946600a2ddf10af1b
+  data.tar.gz: d44010ad67f25f6ee9bd4762ee34a319062ebb82d68af81192bdba8becebad6850fd65e3e5e12f628114dd19ba17378bbecf196320e226613bb29e4062b3a80d

data/lib/manifold/api/schema_generator.rb

@@ -4,8 +4,6 @@ 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
@@ -33,39 +31,52 @@ module Manifold
       private
 
       def metrics_fields
-        return [] unless @manifold_yaml["contexts"] && @manifold_yaml["metrics"]
+        return [] unless @manifold_yaml["metrics"]
+
+        @manifold_yaml["metrics"].map do |group_name, group_config|
+          {
+            "name" => group_name,
+            "type" => "RECORD",
+            "mode" => "NULLABLE",
+            "fields" => group_metrics_fields(group_config)
+          }
+        end
+      end
+
+      def group_metrics_fields(group_config)
+        return [] unless group_config["breakouts"] && group_config["aggregations"]
 
-        @manifold_yaml["contexts"].map do |context_name, _context_config|
+        group_config["breakouts"].map do |breakout_name, _breakout_config|
           {
-            "name" => context_name,
+            "name" => breakout_name,
             "type" => "RECORD",
             "mode" => "NULLABLE",
-            "fields" => context_metrics_fields
+            "fields" => breakout_metrics_fields(group_config)
           }
         end
       end
 
-      def context_metrics_fields
+      def breakout_metrics_fields(group_config)
         [
-          *countif_fields,
-          *sumif_fields
+          *countif_fields(group_config),
+          *sumif_fields(group_config)
         ]
       end
 
-      def countif_fields
-        return [] unless @manifold_yaml.dig("metrics", "countif")
+      def countif_fields(group_config)
+        return [] unless group_config.dig("aggregations", "countif")
 
         [{
-          "name" => @manifold_yaml["metrics"]["countif"],
+          "name" => group_config["aggregations"]["countif"],
           "type" => "INTEGER",
           "mode" => "NULLABLE"
         }]
       end
 
-      def sumif_fields
-        return [] unless @manifold_yaml.dig("metrics", "sumif")
+      def sumif_fields(group_config)
+        return [] unless group_config.dig("aggregations", "sumif")
 
-        @manifold_yaml["metrics"]["sumif"].keys.map do |metric_name|
+        group_config["aggregations"]["sumif"].keys.map do |metric_name|
           {
             "name" => metric_name,
             "type" => "INTEGER",
@@ -73,12 +84,6 @@ module Manifold
           }
         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

data/lib/manifold/api/workspace.rb

@@ -19,12 +19,91 @@ module Manifold
           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.dimensions_config = @manifold_yaml["dimensions"]&.fetch("merge", nil) if @manifold_yaml["dimensions"]
         config.manifold_config = @manifold_yaml
         config.write(path)
       end
     end
 
+    # Handles SQL generation for manifold workspaces
+    class SqlGenerator
+      def initialize(name, manifold_yaml)
+        @name = name
+        @manifold_yaml = manifold_yaml
+      end
+
+      def generate_dimensions_merge_sql(source_sql)
+        return unless valid_dimensions_config?
+
+        sql_builder = Terraform::SQLBuilder.new(@name, @manifold_yaml)
+        sql_builder.build_dimensions_merge_sql(source_sql)
+      end
+
+      private
+
+      def valid_dimensions_config?
+        return false unless @manifold_yaml
+
+        !@manifold_yaml["dimensions"]&.dig("merge", "source").nil?
+      end
+    end
+
+    # Handles schema file generation for manifold workspaces
+    class SchemaWriter
+      def initialize(name, vectors, vector_service, manifold_yaml, logger)
+        @name = name
+        @vectors = vectors
+        @vector_service = vector_service
+        @manifold_yaml = manifold_yaml
+        @logger = logger
+      end
+
+      def write_schemas(tables_directory)
+        tables_directory.mkpath
+        write_dimensions_schema(tables_directory)
+        write_manifold_schema(tables_directory)
+      end
+
+      private
+
+      def write_dimensions_schema(tables_directory)
+        dimensions_path = tables_directory.join("dimensions.json")
+        dimensions_path.write(dimensions_schema_json.concat("\n"))
+      end
+
+      def write_manifold_schema(tables_directory)
+        manifold_path = tables_directory.join("manifold.json")
+        manifold_path.write(manifold_schema_json.concat("\n"))
+      end
+
+      def schema_generator
+        @schema_generator ||= SchemaGenerator.new(dimensions_fields, @manifold_yaml)
+      end
+
+      def manifold_schema
+        schema_generator.manifold_schema
+      end
+
+      def dimensions_schema
+        schema_generator.dimensions_schema
+      end
+
+      def dimensions_fields
+        @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 manifold_schema_json
+        JSON.pretty_generate(manifold_schema)
+      end
+    end
+
     # Encapsulates a single manifold.
     class Workspace
       attr_reader :name, :template_path, :logger
@@ -52,13 +131,13 @@ module Manifold
       def generate(with_terraform: false)
         return nil unless manifold_exists? && any_vectors?
 
-        tables_directory.mkpath
-        generate_dimensions
-        generate_manifold
+        generate_schemas
         logger.info("Generated BigQuery dimensions table schema for workspace '#{name}'.")
 
         return unless with_terraform
 
+        write_manifold_merge_sql
+        write_dimensions_merge_sql
         generate_terraform
         logger.info("Generated Terraform configuration for workspace '#{name}'.")
       end
@@ -89,53 +168,54 @@ module Manifold
         directory.join("main.tf.json")
       end
 
-      private
+      def write_manifold_merge_sql
+        return unless manifold_file
 
-      def directory
-        Pathname.pwd.join("workspaces", name)
+        sql_builder = Terraform::SQLBuilder.new(name, manifold_yaml)
+        sql = sql_builder.build_manifold_merge_sql
+        routines_directory.join("merge_manifold.sql").write(sql)
       end
 
-      def manifold_yaml
-        @manifold_yaml ||= YAML.safe_load_file(manifold_path)
-      end
+      def write_dimensions_merge_sql
+        return unless dimensions_merge_source_exists?
 
-      def generate_dimensions
-        dimensions_path.write(dimensions_schema_json.concat("\n"))
-      end
+        sql = generate_dimensions_merge_sql
+        return unless sql
 
-      def generate_manifold
-        manifold_schema_path.write(manifold_schema_json.concat("\n"))
+        write_dimensions_merge_sql_file(sql)
       end
 
-      def manifold_schema_path
-        tables_directory.join("manifold.json")
+      def dimensions_merge_source_exists?
+        manifold_yaml["dimensions"]&.dig("merge", "source")
       end
 
-      def schema_generator
-        @schema_generator ||= SchemaGenerator.new(dimensions_fields, manifold_yaml)
+      def generate_dimensions_merge_sql
+        source_sql = File.read(Pathname.pwd.join(manifold_yaml["dimensions"]["merge"]["source"]))
+        SqlGenerator.new(name, manifold_yaml).generate_dimensions_merge_sql(source_sql)
       end
 
-      def manifold_schema
-        schema_generator.manifold_schema
+      def write_dimensions_merge_sql_file(sql)
+        routines_directory.mkpath
+        dimensions_merge_sql_path.write(sql)
       end
 
-      def dimensions_schema
-        schema_generator.dimensions_schema
+      def dimensions_merge_sql_path
+        routines_directory.join("merge_dimensions.sql")
       end
 
-      def dimensions_fields
-        @dimensions_fields ||= vectors.filter_map do |vector|
-          logger.info("Loading vector schema for '#{vector}'.")
-          @vector_service.load_vector_schema(vector)
-        end
+      private
+
+      def directory
+        Pathname.pwd.join("workspaces", name)
       end
 
-      def dimensions_schema_json
-        JSON.pretty_generate(dimensions_schema)
+      def manifold_yaml
+        @manifold_yaml ||= YAML.safe_load_file(manifold_path)
       end
 
-      def dimensions_path
-        tables_directory.join("dimensions.json")
+      def generate_schemas
+        SchemaWriter.new(name, vectors, @vector_service, manifold_yaml, logger)
+                    .write_schemas(tables_directory)
       end
 
       def any_vectors?
@@ -151,10 +231,6 @@ module Manifold
         terraform_generator.manifold_config = manifold_yaml
         terraform_generator.generate(terraform_main_path)
       end
-
-      def manifold_schema_json
-        JSON.pretty_generate(manifold_schema)
-      end
     end
   end
 end

data/lib/manifold/cli.rb

@@ -4,7 +4,7 @@ module Manifold
   # CLI provides command line interface functionality
   # for creating and managing umbrella projects for data management.
   class CLI < Thor
-    attr_accessor :logger, :bq_service
+    attr_accessor :logger
 
     def initialize(*args, logger: Logger.new($stdout))
       super(*args)

data/lib/manifold/templates/workspace_template.yml

@@ -12,20 +12,17 @@ timestamp:
   interval: HOUR
   field: timestamp
 
-contexts:
-  paid: IS_PAID(context.location)
-  organic: IS_ORGANIC(context.location)
-  paidOrganic:
-    fields:
-      - paid
-      - organic
-    operator: AND
-
 metrics:
-  countif: tapCount
-  sumif:
-    sequenceSum:
-      field: context.sequence
+  renders:
+    breakouts:
+      paid: IS_PAID(context.location)
+      organic: IS_ORGANIC(context.location)
+
+    aggregations:
+      countif: renderCount
+      sumif:
+        sequenceSum:
+          field: context.sequence
 
-source: my_project.my_dataset.my_table
-filter: timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 90 DAY)
+    source: my_project.render_metrics
+    filter: timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 90 DAY)

data/lib/manifold/terraform/workspace_configuration.rb

@@ -3,82 +3,40 @@
 module Manifold
   module Terraform
     # Handles building metrics SQL for manifold routines
-    class MetricsBuilder
-      def initialize(manifold_config)
+    class MetricsSQLBuilder
+      def initialize(name, manifold_config)
+        @name = name
         @manifold_config = manifold_config
       end
 
-      def build_metrics_struct
-        return "" unless @manifold_config&.dig("contexts") && @manifold_config&.dig("metrics")
-
-        context_structs = @manifold_config["contexts"].map do |name, config|
-          condition = build_context_condition(name, config)
-          metrics = build_context_metrics(condition)
-          "STRUCT(#{metrics}) AS #{name}"
-        end
-
-        context_structs.join(",\n")
+      def build_metrics_select
+        <<~SQL
+          SELECT
+              id,
+              timestamp,
+              #{build_metrics_struct}
+            FROM #{build_metric_joins}
+        SQL
       end
 
       private
 
-      def build_context_metrics(condition)
-        metrics = []
-        add_count_metrics(metrics, condition)
-        add_sum_metrics(metrics, condition)
-        metrics.join(",\n")
-      end
-
-      def add_count_metrics(metrics, condition)
-        return unless @manifold_config.dig("metrics", "countif")
-
-        metrics << "COUNTIF(#{condition}) AS #{@manifold_config["metrics"]["countif"]}"
-      end
-
-      def add_sum_metrics(metrics, condition)
-        @manifold_config.dig("metrics", "sumif")&.each do |name, config|
-          metrics << "SUM(IF(#{condition}, #{config["field"]}, 0)) AS #{name}"
-        end
-      end
-
-      def build_context_condition(_name, config)
-        return config unless config.is_a?(Hash)
-
-        operator = config["operator"]
-        fields = config["fields"]
-        build_operator_condition(operator, fields)
-      end
-
-      def build_operator_condition(operator, fields)
-        conditions = fields.map { |f| @manifold_config["contexts"][f] }
-        case operator
-        when "AND", "OR" then join_conditions(conditions, operator)
-        when "NOT" then negate_condition(conditions.first)
-        when "NAND", "NOR" then negate_joined_conditions(conditions, operator[1..])
-        when "XOR" then build_xor_condition(conditions)
-        when "XNOR" then build_xnor_condition(conditions)
-        else config
-        end
-      end
-
-      def join_conditions(conditions, operator)
-        conditions.join(" #{operator} ")
-      end
-
-      def negate_condition(condition)
-        "NOT (#{condition})"
+      def build_metrics_struct
+        metric_groups = @manifold_config["metrics"].keys
+        metric_groups.map { |group| "#{group}.metrics #{group}" }.join(",\n    ")
       end
 
-      def negate_joined_conditions(conditions, operator)
-        "NOT (#{join_conditions(conditions, operator)})"
-      end
+      def build_metric_joins
+        metric_groups = @manifold_config["metrics"]
+        joins = metric_groups.map { |group, config| "#{config["source"]} AS #{group}" }
+        first = joins.shift
+        return first if joins.empty?
 
-      def build_xor_condition(conditions)
-        "(#{conditions[0]} AND NOT #{conditions[1]}) OR (NOT #{conditions[0]} AND #{conditions[1]})"
+        "#{first}\n  #{joins.map { |table| "FULL OUTER JOIN #{table} USING (id, timestamp)" }.join("\n  ")}"
       end
 
-      def build_xnor_condition(conditions)
-        "(#{conditions[0]} AND #{conditions[1]}) OR (NOT #{conditions[0]} AND NOT #{conditions[1]})"
+      def timestamp_field
+        @manifold_config&.dig("timestamp", "field")
       end
     end
 
@@ -87,17 +45,17 @@ module Manifold
       def initialize(name, manifold_config)
         @name = name
         @manifold_config = manifold_config
+        @metrics_builder = MetricsSQLBuilder.new(name, manifold_config)
       end
 
-      def build_manifold_merge_sql(_metrics_builder, &)
+      def build_manifold_merge_sql
         return "" unless valid_config?
 
         <<~SQL
           MERGE #{@name}.Manifold AS target USING (
-            #{build_metrics_cte(&)}
-            #{build_final_select}
+            #{build_source_query}
           ) AS source
-          ON source.id = target.id AND source.timestamp = target.timestamp
+          #{build_merge_conditions}
           #{build_merge_actions}
         SQL
       end
@@ -117,37 +75,36 @@ module Manifold
       private
 
       def valid_config?
-        source_table && timestamp_field
+        source_table && timestamp_field && @manifold_config["metrics"]
+      end
+
+      def source_table
+        first_group = @manifold_config["metrics"]&.values&.first
+        first_group&.dig("source")
       end
 
-      def build_metrics_cte(&)
+      def timestamp_field
+        @manifold_config&.dig("timestamp", "field")
+      end
+
+      def build_source_query
         <<~SQL
           WITH Metrics AS (
-            #{build_metrics_select(&)}
+            #{@metrics_builder.build_metrics_select}
           )
-        SQL
-      end
 
-      def build_metrics_select(&block)
-        <<~SQL
           SELECT
-            dimensions.id id,
-            TIMESTAMP_TRUNC(#{timestamp_field}, #{interval}) timestamp,
-            STRUCT(
-              #{block.call}
-            ) AS metrics
-          FROM `#{source_table}`
-          #{where_clause}
-          GROUP BY 1, 2
+            id,
+            timestamp,
+            Dimensions.dimensions,
+            (SELECT AS STRUCT Metrics.* EXCEPT(id, timestamp)) metrics
+          FROM Metrics
+          JOIN #{@name}.Dimensions USING (id)
         SQL
       end
 
-      def build_final_select
-        <<~SQL
-          SELECT id, timestamp, #{@name}.Dimensions.dimensions, Metrics.metrics
-          FROM Metrics
-          LEFT JOIN #{@name}.Dimensions USING (id)
-        SQL
+      def build_merge_conditions
+        "ON source.id = target.id AND source.timestamp = target.timestamp"
       end
 
       def build_merge_actions
@@ -160,24 +117,6 @@ module Manifold
             INSERT ROW;
         SQL
       end
-
-      def source_table
-        @manifold_config["source"]
-      end
-
-      def interval
-        @manifold_config&.dig("timestamp", "interval") || "DAY"
-      end
-
-      def where_clause
-        return "" unless @manifold_config["filter"]
-
-        "WHERE #{@manifold_config["filter"]}"
-      end
-
-      def timestamp_field
-        @manifold_config&.dig("timestamp", "field")
-      end
     end
 
     # Handles building table configurations
@@ -217,13 +156,13 @@ module Manifold
     # Represents a Terraform configuration for a Manifold workspace.
     class WorkspaceConfiguration < Configuration
       attr_reader :name
-      attr_writer :merge_config, :manifold_config
+      attr_writer :dimensions_config, :manifold_config
 
       def initialize(name)
         super()
         @name = name
         @vectors = []
-        @merge_config = nil
+        @dimensions_config = nil
       end
 
       def add_vector(vector_config)
@@ -272,7 +211,7 @@ module Manifold
       end
 
       def dimensions_routine_attributes
-        return nil if @vectors.empty? || @merge_config.nil?
+        return nil if @vectors.empty? || @dimensions_config.nil?
 
         {
           "dataset_id" => name,
@@ -280,51 +219,22 @@ module Manifold
           "routine_id" => "merge_dimensions",
           "routine_type" => "PROCEDURE",
           "language" => "SQL",
-          "definition_body" => dimensions_merge_routine,
+          "definition_body" => "${file(\"${path.module}/routines/merge_dimensions.sql\")}",
           "depends_on" => ["google_bigquery_dataset.#{name}"]
         }
       end
 
-      def dimensions_merge_routine
-        return "" if @vectors.empty? || @merge_config.nil?
-
-        source_sql = File.read(Pathname.pwd.join(@merge_config["source"]))
-        SQLBuilder.new(name, @manifold_config).build_dimensions_merge_sql(source_sql)
-      end
-
       def manifold_routine_attributes
-        return nil unless valid_manifold_config?
-
         {
           "dataset_id" => name,
           "project" => "${var.project_id}",
           "routine_id" => "merge_manifold",
           "routine_type" => "PROCEDURE",
           "language" => "SQL",
-          "definition_body" => manifold_merge_routine,
+          "definition_body" => "${file(\"${path.module}/routines/merge_manifold.sql\")}",
           "depends_on" => ["google_bigquery_dataset.#{name}"]
         }
       end
-
-      def manifold_merge_routine
-        metrics_builder = MetricsBuilder.new(@manifold_config)
-        sql_builder = SQLBuilder.new(name, @manifold_config)
-        sql_builder.build_manifold_merge_sql(metrics_builder) do
-          metrics_builder.build_metrics_struct
-        end
-      end
-
-      def valid_manifold_config?
-        return false unless @manifold_config
-
-        required_fields_present?
-      end
-
-      def required_fields_present?
-        %w[source timestamp.field contexts metrics].all? do |field|
-          @manifold_config&.dig(*field.split("."))
-        end
-      end
     end
   end
 end

data/lib/manifold/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Manifold
-  VERSION = "0.0.16"
+  VERSION = "0.0.18"
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: manifold-cli
 version: !ruby/object:Gem::Version
-  version: 0.0.16
+  version: 0.0.18
 platform: ruby
 authors:
 - claytongentry
 bindir: bin
 cert_chain: []
-date: 2025-02-11 00:00:00.000000000 Z
+date: 2025-02-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor