manifold-cli 0.0.14 → 0.0.16

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 38d11b038b09c8e1f83ef07ff4b238fcb07edaeeb1a36518a182a0963c5d5589
-  data.tar.gz: 8ad9000c2ed33ff8a8324ce202b2c15b367b894de756099417479e9354750864
+  metadata.gz: 38a614204bb12682a02b29ea889a9f44a6aab5af58b746322ba607252f9ff272
+  data.tar.gz: 1ee65656d387eabc11fe6eb9effcf6119ec28210d32efc7bf0b8ae57ea37c43b
 SHA512:
-  metadata.gz: 79a51d73c4d63c7fc02e13336e1cdd51be4d93023eac402134e1f3be4ecf067d016813b9a533318f37708469ad001a568abdb436b96af3ccace25db0f50ce1d5
-  data.tar.gz: 4143c27f33fbe4b6a7bdc16282080a9c3cd9d5f507c0ded3db633b5a571bbb0a5a39659def1eb9342e8a711875a65c9daf000f15d6e54aaef53e38301de104b8
+  metadata.gz: 762a4575c8423177ccbd69622da95d034aa311f51733f4f6d123199ec35af390acee194ce543c907f127939ed6dae8c15de97d11183b2bb508b0b3a8f1195e02
+  data.tar.gz: 359ae735a6be1fec84b46a0410395feaaef0b12fee3df69f887cfc7723b57d58341dc1684ea93f401b91b5e7510e1bb085175af2bc193578471f72f4b85c955a

data/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

data/lib/manifold/api/workspace.rb

@@ -2,6 +2,29 @@
 
 module Manifold
   module API
+    # Handles terraform configuration generation
+    class TerraformGenerator
+      attr_accessor :manifold_config
+
+      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.manifold_config = @manifold_yaml
+        config.write(path)
+      end
+    end
+
     # Encapsulates a single manifold.
     class Workspace
       attr_reader :name, :template_path, :logger
@@ -27,11 +50,17 @@ module Manifold
       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 +103,28 @@ module Manifold
         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 +147,13 @@ module Manifold
       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.manifold_config = manifold_yaml
+        terraform_generator.generate(terraform_main_path)
+      end
+
+      def manifold_schema_json
+        JSON.pretty_generate(manifold_schema)
       end
     end
   end

data/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)

data/lib/manifold/terraform/workspace_configuration.rb

@@ -2,9 +2,222 @@
 
 module Manifold
   module Terraform
+    # Handles building metrics SQL for manifold routines
+    class MetricsBuilder
+      def initialize(manifold_config)
+        @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")
+      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})"
+      end
+
+      def negate_joined_conditions(conditions, operator)
+        "NOT (#{join_conditions(conditions, operator)})"
+      end
+
+      def build_xor_condition(conditions)
+        "(#{conditions[0]} AND NOT #{conditions[1]}) OR (NOT #{conditions[0]} AND #{conditions[1]})"
+      end
+
+      def build_xnor_condition(conditions)
+        "(#{conditions[0]} AND #{conditions[1]}) OR (NOT #{conditions[0]} AND NOT #{conditions[1]})"
+      end
+    end
+
+    # Handles building SQL for manifold routines
+    class SQLBuilder
+      def initialize(name, manifold_config)
+        @name = name
+        @manifold_config = manifold_config
+      end
+
+      def build_manifold_merge_sql(_metrics_builder, &)
+        return "" unless valid_config?
+
+        <<~SQL
+          MERGE #{@name}.Manifold AS target USING (
+            #{build_metrics_cte(&)}
+            #{build_final_select}
+          ) AS source
+          ON source.id = target.id AND source.timestamp = target.timestamp
+          #{build_merge_actions}
+        SQL
+      end
+
+      def build_dimensions_merge_sql(source_sql)
+        <<~SQL
+          MERGE #{@name}.Dimensions AS TARGET
+          USING (
+            #{source_sql}
+          ) AS source
+          ON source.id = target.id
+          WHEN MATCHED THEN UPDATE SET target.dimensions = source.dimensions
+          WHEN NOT MATCHED THEN INSERT ROW;
+        SQL
+      end
+
+      private
+
+      def valid_config?
+        source_table && timestamp_field
+      end
+
+      def build_metrics_cte(&)
+        <<~SQL
+          WITH Metrics AS (
+            #{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
+        SQL
+      end
+
+      def build_final_select
+        <<~SQL
+          SELECT id, timestamp, #{@name}.Dimensions.dimensions, Metrics.metrics
+          FROM Metrics
+          LEFT JOIN #{@name}.Dimensions USING (id)
+        SQL
+      end
+
+      def build_merge_actions
+        <<~SQL
+          WHEN MATCHED THEN
+            UPDATE SET
+              metrics = source.metrics,
+              dimensions = source.dimensions
+          WHEN NOT MATCHED THEN
+            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
+    class TableConfigBuilder
+      def initialize(name)
+        @name = name
+      end
+
+      def build_table_configs
+        {
+          "dimensions" => dimensions_table_config,
+          "manifold" => manifold_table_config
+        }
+      end
+
+      private
+
+      def dimensions_table_config
+        build_table_config("Dimensions")
+      end
+
+      def manifold_table_config
+        build_table_config("Manifold")
+      end
+
+      def build_table_config(table_id)
+        {
+          "dataset_id" => @name,
+          "project" => "${var.project_id}",
+          "table_id" => table_id,
+          "schema" => "${file(\"${path.module}/tables/#{table_id.downcase}.json\")}",
+          "depends_on" => ["google_bigquery_dataset.#{@name}"]
+        }
+      end
+    end
+
     # Represents a Terraform configuration for a Manifold workspace.
     class WorkspaceConfiguration < Configuration
       attr_reader :name
+      attr_writer :merge_config, :manifold_config
 
       def initialize(name)
         super()
@@ -17,14 +230,12 @@ module Manifold
         @vectors << vector_config
       end
 
-      attr_writer :merge_config
-
       def as_json
         {
           "variable" => variables_block,
           "resource" => {
             "google_bigquery_dataset" => dataset_config,
-            "google_bigquery_table" => table_config,
+            "google_bigquery_table" => TableConfigBuilder.new(name).build_table_configs,
             "google_bigquery_routine" => routine_config
           }.compact
         }
@@ -51,53 +262,68 @@ module Manifold
         }
       end
 
-      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}"]
-          }
-        }
+      def routine_config
+        routines = {
+          "merge_dimensions" => dimensions_routine_attributes,
+          "merge_manifold" => manifold_routine_attributes
+        }.compact
+
+        routines.empty? ? nil : routines
       end
 
-      def routine_config
+      def dimensions_routine_attributes
         return nil if @vectors.empty? || @merge_config.nil?
 
         {
-          "merge_dimensions" => routine_attributes
+          "dataset_id" => name,
+          "project" => "${var.project_id}",
+          "routine_id" => "merge_dimensions",
+          "routine_type" => "PROCEDURE",
+          "language" => "SQL",
+          "definition_body" => dimensions_merge_routine,
+          "depends_on" => ["google_bigquery_dataset.#{name}"]
         }
       end
 
-      def routine_attributes
+      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_dimensions",
+          "routine_id" => "merge_manifold",
           "routine_type" => "PROCEDURE",
           "language" => "SQL",
-          "definition_body" => merge_routine_definition,
+          "definition_body" => manifold_merge_routine,
           "depends_on" => ["google_bigquery_dataset.#{name}"]
         }
       end
 
-      def merge_routine_definition
-        source_sql = read_source_sql(@merge_config["source"])
-        <<~SQL
-          MERGE #{name}.Dimensions AS TARGET
-          USING (
-            #{source_sql}
-          ) AS source
-          ON source.id = target.id
-          WHEN MATCHED THEN UPDATE SET target.dimensions = source.dimensions
-          WHEN NOT MATCHED THEN INSERT ROW;
-        SQL
+      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 read_source_sql(source_path)
-        File.read(Pathname.pwd.join(source_path))
+      def required_fields_present?
+        %w[source timestamp.field contexts metrics].all? do |field|
+          @manifold_config&.dig(*field.split("."))
+        end
       end
     end
   end

data/lib/manifold/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Manifold
-  VERSION = "0.0.14"
+  VERSION = "0.0.16"
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: manifold-cli
 version: !ruby/object:Gem::Version
-  version: 0.0.14
+  version: 0.0.16
 platform: ruby
 authors:
 - claytongentry
 bindir: bin
 cert_chain: []
-date: 2025-02-07 00:00:00.000000000 Z
+date: 2025-02-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -43,6 +43,7 @@ files:
 - lib/manifold.rb
 - lib/manifold/api.rb
 - lib/manifold/api/project.rb
+- lib/manifold/api/schema_generator.rb
 - lib/manifold/api/vector.rb
 - lib/manifold/api/workspace.rb
 - lib/manifold/cli.rb