huginn_acumen_product_agent 1.7.0 → 2.1.0

data/lib/huginn_acumen_product_agent/concerns/alternate_products_query_concern.rb

@@ -0,0 +1,98 @@
+# frozen_string_literal: true
+
+# This module is responsible for reading/processing data from the Product_Link
+# table. This table defines the link between product records.
+#
+# It's important to note that this table defines links between related products
+# as well as product formats. The `Alt_Format` field determines whether the linked
+# product is an alternate format for a given title.
+#
+# For the purposes of this concern, we only care about Product_Link records where
+# the `Alt_Format` field is `true`
+module AlternateProductsQueryConcern
+  extend AcumenQueryConcern
+
+  # This function returns two data elements in a hash object.
+  # The `id_set` is the full collection of IDs to be fetched including the input
+  # product_ids and all of their alternate format ids. The goal of this value
+  # is to reduce the resource requirements of running each "bundle" individually
+  #
+  # The alternate_ids_map contains arrays of product IDs mapped to their master
+  # product id. This map will be used to assemble fetched product data into bundles
+  def fetch_alternate_format_ids(acumen_client, product_ids)
+    begin
+      link_data = acumen_client.get_linked_products(product_ids)
+
+      links = process_alternate_format_response(link_data)
+
+      mapped_ids = map_alternate_format_links(links, product_ids)
+
+      id_set = [] + product_ids
+      mapped_ids.each_value { |bundle| id_set += bundle }
+
+      return {id_set: id_set, alternate_ids_map: mapped_ids }
+    rescue => error
+      issue_error(AcumenAgentError.new(
+        'fetch_alternate_format_ids',
+        'Failed attempting to lookup alternate products',
+        product_ids,
+        error
+      ))
+    end
+  end
+
+  # This function parses the raw data returned from the Product_Link table
+  # The resulting array contains the set alternate format IDs associated with a
+  # single product
+  def process_alternate_format_response(raw_data)
+    results = []
+    raw_data.map do |link|
+
+      begin
+        mapped = response_mapper(link, {
+          'Product_Link.Link_From_ID' => 'from_id',
+          'Product_Link.Link_To_ID' => 'to_id',
+          'Product_Link.Alt_Format' => 'alt_format',
+        })
+
+        if mapped['alt_format'].to_s != '0' && !mapped.in?(results)
+          results.push(mapped)
+        end
+
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'process_alternate_format_response',
+          'Failed while processing alternate format links',
+          raw_data,
+          error
+        ))
+      end
+    end
+
+    return results
+  end
+
+  # Returns a map that ties each provided `product_id` to an array of IDs for its
+  # other formats
+  def map_alternate_format_links(links, product_ids)
+    results = {}
+
+    product_ids.each do |id|
+
+      begin
+        alternates = links.select { |l| l['from_id'] == id }
+        results[id] = alternates.map { |l| l['to_id'] }
+
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'map_alternate_format_links',
+          'Failed while mapping alternate format links',
+          { id: id, links: links },
+          error
+        ))
+      end
+    end
+
+    return results
+  end
+end

data/lib/huginn_acumen_product_agent/concerns/inv_product_query_concern.rb

@@ -0,0 +1,140 @@
+# frozen_string_literal: true
+
+# This module is responsible for reading/processing data recrods from the Inv_Product
+# table in Acumen. This table contains the baseline product information -- name, sku,
+# format, etc.
+module InvProductQueryConcern
+  extend AcumenQueryConcern
+
+  # Fetch/Process the Acumen data,
+  def fetch_inv_product_data(acumen_client, product_ids, digital_format_list)
+    product_data = acumen_client.get_products(product_ids)
+
+    return process_inv_product_response(product_data, digital_format_list)
+  end
+
+  # This function returns an array of Acumen products mapped to Schema.org/Product
+  # objects. We've added additional fields of:
+  #
+  #     *  `isAvailableForPurchase` -- used to control product deletion in external systems
+  #     *  `acumenAttributes` -- Additional acumen data that doesn't have a direct 1:1 field
+  #        on the Product, but may be useful in other platforms
+  def process_inv_product_response(raw_data, digital_format_list)
+    raw_data.map do |p|
+
+      begin
+        product = response_mapper(p, {
+            'Inv_Product.ID' => 'identifier',
+            'Inv_Product.ProdCode' => 'sku',
+            'Inv_Product.Full_Title' => 'name',
+            'Inv_Product.SubTitle' => 'disambiguatingDescription',
+            'Inv_Product.ISBN_UPC' => 'isbn',
+            'Inv_Product.Pub_Date' => 'datePublished',
+            'Inv_Product.Next_Release' => 'releaseDate',
+        })
+
+        # Nullify blank dates
+        if product['datePublished'] == '0000-00-00T00:00:00'
+          product['datePublished'] = nil
+        end
+
+        if product['releaseDate'] == '0000-00-00T00:00:00'
+          product['releaseDate'] = nil
+        end
+
+        product['@type'] = 'Product'
+        product['isTaxable'] = get_field_value(p, 'Inv_Product.Taxable') == '1'
+        product['productAvailability'] = get_field_value(p, 'Inv_Product.Not_On_Website') == '0'
+        product['noBackorderFill'] = get_field_value(p, 'Inv_Product.No_Backorder_Fill') == '1'
+        product['acumenAttributes'] = {
+          'info_alpha_1' => get_field_value(p, 'Inv_Product.Info_Alpha_1'),
+          'info_boolean_1' => get_field_value(p, 'Inv_Product.Info_Boolean_1'), # is_available_on_formed
+        }
+        product['additionalProperty'] = [
+          {
+              '@type' => 'PropertyValue',
+              'propertyID' => 'is_master',
+              'value' => get_field_value(p, 'Inv_Product.OnWeb_LinkOnly') == '0',
+          },
+          {
+              # NOTE: This is different than isAvailableForPurchase. This
+              '@type' => 'PropertyValue',
+              'propertyID' => 'disable_web_purchase',
+              'value' => get_field_value(p, 'Inv_Product.Disable_Web_Purchase'),
+          }
+        ]
+
+        product['offers'] = [{
+          '@type' => 'Offer',
+          'price' => get_field_value(p, 'Inv_Product.Price_1'),
+          'availability' => get_field_value(p, 'Inv_Product.BO_Reason')
+        }]
+
+        if get_field_value(p, 'Inv_Product.Price_2')
+          product['offers'].push({
+            '@type' => 'Offer',
+            'price' => get_field_value(p, 'Inv_Product.Price_2'),
+            'availability' => get_field_value(p, 'Inv_Product.BO_Reason')
+          })
+        end
+
+        not_on_website = get_field_value(p, 'Inv_Product.Not_On_Website')
+        disable_web_purchase = get_field_value(p, 'Inv_Product.Disable_Web_Purchase')
+
+        product_availability = 'available'
+
+
+        if (disable_web_purchase == '1')
+          product_availability = 'disabled'
+        end
+
+        if (not_on_website == '1')
+          product_availability = 'not available'
+        end
+
+        product['productAvailability'] = product_availability
+
+        weight = get_field_value(p, 'Inv_Product.Weight')
+        product['weight'] = get_quantitative_value(weight, 'oz.')
+
+        # The category used here is the Acumen Product category. Functionally, this
+        # serves as the product's _format_. This field is different from Web Categories
+        # which behave more like traditional category taxonomies.
+        category = get_field_value(p, 'Inv_Product.Category')
+        if category
+          product['acumenAttributes']['category'] = category
+          product['isDigital'] = digital_format_list.find { |f| f == category } ? true : false
+
+          if category == 'Paperback'
+            product['additionalType'] = 'Book'
+            product['bookFormat'] = "http://schema.org/Paperback"
+            product['accessMode'] = "textual"
+          elsif category == 'Hardcover'
+            product['additionalType'] = 'Book'
+            product['bookFormat'] = "http://schema.org/Hardcover"
+            product['accessMode'] = "textual"
+          elsif category == 'eBook'
+            product['additionalType'] = 'Book'
+            product['bookFormat'] = "http://schema.org/EBook"
+            product['accessMode'] = "textual"
+          elsif category == 'CD'
+            product['additionalType'] = 'CreativeWork'
+            product['accessMode'] = "auditory"
+          end
+        end
+
+        product
+
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'process_inv_product_response',
+          'Failed to load Inventory Product Records',
+          { raw_data: raw_data },
+          error,
+        ))
+
+        return
+      end
+    end
+  end
+end

data/lib/huginn_acumen_product_agent/concerns/inv_status_query_concern.rb

@@ -0,0 +1,76 @@
+# frozen_string_literal: true
+
+# This module is responsible for reading/processing the Inv_Status table. This table
+# contains stock information for the product and is tied to product availability.
+module ProdMktQueryConcern
+  extend AcumenQueryConcern
+
+  # Update the provided products with their associated marketing data
+  # NOTE: The `products` here are Shema.org/Product records mapped from Inv_Product
+  # data
+  def fetch_inv_status(acumen_client, products)
+
+    product_skus = products.map { |p| p['sku'] }
+    inventory_data = acumen_client.get_inv_status(product_skus)
+    inventory_data = process_inv_status_response(inventory_data)
+
+    return map_inv_status_data(products, inventory_data)
+  end
+
+  # This function parses the raw data returned from the Prod_Mkt table
+  def process_inv_status_response(raw_data)
+    results = []
+    raw_data.map do |inv_status|
+
+      begin
+        mapped = response_mapper(inv_status, {
+          'Inv_Status.Warehouse' => 'warehouse',
+          'Inv_Status.ProdCode' => 'sku',
+          'Inv_Status.Available' => 'quantity',
+        })
+
+        results << mapped
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'process_inv_status_response',
+          'Failed while processing Prod_Mkt record',
+          inv_status,
+          error,
+        ))
+      end
+    end
+
+    results
+  end
+
+  # This function maps parsed Prod_Mkt records to their matching product record
+  # and updates the product object with the additional data
+  def map_inv_status_data(products, inventory)
+    products.map do |product|
+      inventory_data = inventory.select { |i| i['sku'] == product['sku'] }
+      begin
+
+        if inventory_data
+          quantity = 0
+          inventory_data.each do |i|
+            quantity = quantity + i['quantity'].to_i if quantity.present?
+          end
+
+          product['acumenAttributes']['stock_quantity'] = quantity
+        end
+
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'map_inv_status_data',
+          'Failed to map inventory data for product',
+          { product: product, inventory_data: inventory_data },
+          error,
+        ))
+      end
+
+      product
+    end
+
+    return products
+  end
+end

data/lib/huginn_acumen_product_agent/concerns/prod_mkt_query_concern.rb

@@ -0,0 +1,163 @@
+# frozen_string_literal: true
+
+# This module is responsible for reading/processing the ProdMkt table. This table
+# contains more detailed product meta information: Publisher, page counts, age range,
+# description, etc.
+module ProdMktQueryConcern
+  extend AcumenQueryConcern
+
+  # Update the provided products with their associated marketing data
+  # NOTE: The `products` here are Shema.org/Product records mapped from Inv_Product
+  # data
+  def fetch_product_marketing(acumen_client, products)
+
+    product_ids = products.map { |p| p['identifier'] }
+    marketing_data = acumen_client.get_products_marketing(product_ids)
+    marketing_data = process_prod_mkt_response(marketing_data)
+
+    return map_marketing_data(products, marketing_data)
+  end
+
+  # This function parses the raw data returned from the Prod_Mkt table
+  def process_prod_mkt_response(raw_data)
+    results = {}
+    raw_data.map do |product_marketing|
+
+      begin
+        mapped = response_mapper(product_marketing, {
+          'ProdMkt.Product_ID' => 'product_id',
+          'ProdMkt.Product_Code' => 'sku',
+          'ProdMkt.ID' => 'id',
+          'ProdMkt.Pages' => 'pages',
+          'ProdMkt.Publisher' => 'publisher',
+          'ProdMkt.Description_Short' => 'description_short',
+          'ProdMkt.Description_Long' => 'description_long',
+          'ProdMkt.Height' => 'height',
+          'ProdMkt.Width' => 'width',
+          'ProdMkt.Thickness' => 'thickness',
+          'ProdMkt.Meta_Keywords' => 'meta_keywords',
+          'ProdMkt.Meta_Description' => 'meta_description',
+          'ProdMkt.Extent_Unit' => 'extent_unit',
+          'ProdMkt.Extent_Value' => 'extent_value',
+          'ProdMkt.Age_Highest' => 'age_highest',
+          'ProdMkt.Age_Lowest' => 'age_lowest',
+          'ProdMkt.Awards' => 'awards',
+          'ProdMkt.Dimensions_Unit_Measure' => 'dimensions_unit_measure',
+          'ProdMkt.Excerpt' => 'excerpt',
+          'ProdMkt.Grade_Highest' => 'grade_highest',
+          'ProdMkt.Grade_Lowest' => 'grade_lowest',
+          'ProdMkt.Status' => 'status',
+          'ProdMkt.UPC' => 'upc',
+          'ProdMkt.Weight_Unit_Measure' => 'weight_unit_measure',
+          'ProdMkt.Weight' => 'weight',
+          'ProdMkt.Info_Text_01' => 'info_text_01',
+          'ProdMkt.Info_Text_02' => 'info_text_02',
+          'ProdMkt.Religious_Text_Identifier' => 'religious_text_identifier',
+          'ProdMkt.Info_Alpha_07' => 'info_alpha_07',
+        })
+
+        results[mapped['product_id']] = mapped
+        # NOTE: In this case, product_id matches the Inv_Product.ID fields
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'process_prod_mkt_response',
+          'Failed while processing Prod_Mkt record',
+          product_marketing,
+          error,
+        ))
+      end
+    end
+
+    results
+  end
+
+  # This function maps parsed Prod_Mkt records to their matching product record
+  # and updates the product object with the additional data
+  def map_marketing_data(products, marketing_data)
+    products.map do |product|
+      marketing = marketing_data[product['identifier']]
+
+      begin
+
+        if marketing
+          product['acumenAttributes']['product_marketing_id'] = marketing['id']
+
+          product['publisher'] = {
+            '@type': 'Organization',
+            'name' => marketing['publisher']
+          };
+          product['description'] = marketing['description_long']
+          product['abstract'] = marketing['description_short']
+          product['keywords'] = marketing['meta_keywords']
+          product['text'] = marketing['excerpt']
+
+          if marketing['age_lowest'] && marketing['age_highest']
+            product['typicalAgeRange'] = "#{marketing['age_lowest']}-#{marketing['age_highest']}"
+          end
+
+          #----------  Product Page Attributes  ----------#
+          if marketing['grade_lowest'] || marketing['grade_highest']
+            # educationalUse? educationalAlignment?
+            product['additionalProperty'].push({
+              '@type' => 'PropertyValue',
+              'name' => 'Grade',
+              'propertyID' => 'grade_range',
+              'minValue' => marketing['grade_lowest'],
+              'maxValue' => marketing['grade_highest'],
+              'value' => "#{marketing['grade_lowest']}-#{marketing['grade_highest']}",
+            })
+          end
+
+          if marketing['awards']
+            product['additionalProperty'].push({
+              '@type' => 'PropertyValue',
+              'propertyID' => 'awards',
+              'name' => 'Awards',
+              'value' => marketing['awards'],
+            })
+          end
+
+          #----------  Acumen Specific Properties  ----------#
+          product['acumenAttributes']['extent_unit'] = marketing['extent_unit']
+          product['acumenAttributes']['extent_value'] = marketing['extent_value']
+          product['acumenAttributes']['info_text_01'] = marketing['info_text_01'] # editorial_reviews
+          product['acumenAttributes']['info_text_02'] = marketing['info_text_02'] # product_samples
+          product['acumenAttributes']['info_alpha_07'] = marketing['info_alpha_07'] # video_urls
+          product['acumenAttributes']['meta_description'] = marketing['meta_description']
+          product['acumenAttributes']['religious_text_identifier'] = marketing['religious_text_identifier']
+          product['acumenAttributes']['status'] = marketing['status']
+
+          product['gtin12'] = marketing['upc']
+          product['numberOfPages'] = marketing['pages']
+
+          product['height'] = get_quantitative_value(
+            marketing['height'], marketing['dimensions_unit_measure']
+          )
+          product['width'] = get_quantitative_value(
+            marketing['width'], marketing['dimensions_unit_measure']
+          )
+          product['depth'] = get_quantitative_value(
+            marketing['thickness'], marketing['dimensions_unit_measure']
+          )
+          if product['weight']['value'] == '0'
+              product['weight'] = get_quantitative_value(
+                marketing['weight'], marketing['weight_unit_measure']
+              )
+          end
+        end
+
+      rescue => error
+        issue_error(AcumenAgentError.new(
+          'map_marketing_data',
+          'Failed to map marketing data for product',
+          { product: product, marketing: marketing },
+          error,
+        ))
+      end
+
+      product
+    end
+
+    return products
+  end
+end