CDC #198 - Updating to use "uuid" value for Typesense document ID; Up…

…dating indexing process to use "emplace" action instead of "upsert"; Updating indexing process to delete records from the index not included in the batch

README.md

@@ -79,17 +79,19 @@ bundle exec rake typesense:create -- -h host -p port -r protocol -a api_key -c c
 bundle exec rake typesense:delete -- -h host -p port -r protocol -a api_key -c collection_name
 ```
 
-#### Add documents to a collection
+#### Index documents into a collection
 ```bash
 bundle exec rake typesense:index -- -h host -p port -r protocol -a api_key -c collection_name -m model_ids
 ```
 
+**Note:** This task expects the entire collection to be indexed. Any records not included in the batch will be removed from the index.
+
 #### Update a collection
 ```bash
 bundle exec rake typesense:update -- -h host -p port -r protocol -a api_key -c collection_name
 ```
 
-**Note**: This task was added as a workaround for an issue in Typesense indexing nested facetable fields using auto-detection schema. This task should be run _after_ the indexing process to update the "facet" attribute on any fields that should be facetable.
+**Note:** This task was added as a workaround for an issue in Typesense indexing nested facetable fields using auto-detection schema. This task should be run _after_ the indexing process to update the "facet" attribute on any fields that should be facetable.
 
 
 ## Public API

app/services/core_data_connector/search/base.rb

@@ -289,13 +289,11 @@ def search_attributes
             .where(ProjectModelRelationship.arel_table.name => { allow_inverse: true })
         }, as: :related_record, class_name: Relationship.to_s
 
+        # Include the ID attributes as a string by default
+        search_attribute(:id) { uuid }
+        search_attribute(:record_id) { id.to_s }
         search_attribute :uuid
 
-        # Include the ID attribute as a string by default
-        search_attribute(:record_id) do
-          id.to_s
-        end
-
         # Uses the specified attributes to create a JSON object. We'll skip relationships by default as to
         # not create an infinite loop while serializing related records.
         def to_search_json(skip_relationships = true)

lib/typesense/helper.rb

@@ -67,16 +67,23 @@ def index(project_model_ids)
                           hash
                         end
 
+      # Append a unique import_id to all of the documents indexed in this batch
+      import_id = DateTime.now.to_i
+      import_attributes = { import_id: import_id }
+
       # Iterate over the keys and query the records belonging to each project model
       model_classes.keys.each do |model_class|
         klass = model_class.constantize
         ids = model_classes[model_class]
 
         klass.for_search(ids) do |records|
-          documents = records.map { |r| r.to_search_json(false) }
-          collection.documents.import(documents, action: 'upsert')
+          documents = records.map { |r| r.to_search_json(false).merge(import_attributes) }
+          collection.documents.import(documents, action: 'emplace')
         end
       end
+
+      # Delete any records from the index not included in this batch
+      collection.documents.delete(filter_by: "import_id:!=#{import_id}")
     end
 
     def update