improve local es setup

app/models/elastic_search.rb

@@ -46,15 +46,15 @@ def es_host(config)
   # Create elastic search document by id and body. Eg: id: "1234567", body: {id: "1234567", title: "test"}
   def create_document(id, body)
     response = @client.create index: @index, type: @type, id: id, body: body
-    LogWrapper.log('DEBUG', {'message' => "ES document successfully created. Id: #{id}", 
+    LogWrapper.log('DEBUG', {'message' => "ES document successfully created. Id: #{id}",
                              'method' => 'create_document'})
     response
   end
 
   # Delete elastic search document by id. Eg: id: "1234567"
   def delete_document_by_id(id)
     response = @client.delete index: @index, type: @type, id: id
-    LogWrapper.log('DEBUG', {'message' => "ES document successfully deleted. Id: #{id}", 
+    LogWrapper.log('DEBUG', {'message' => "ES document successfully deleted. Id: #{id}",
                              'method' => 'delete_document_by_id'})
 
     response
@@ -169,7 +169,7 @@ def group_by_facets_query(aggregation_hash)
     # aggregation_hash["availability"] = { "terms": { "field": "availability.raw", :size => 10, :order => {:_key => "asc"} } }
     aggregation_hash["area of study"] = { terms: { field: "area_of_study", :size => 100, :order => {:_key => "asc"} } }
 
-    aggregation_hash["subjects"] = {:nested => {:path => "subjects"}, 
+    aggregation_hash["subjects"] = {:nested => {:path => "subjects"},
     :aggregations => {:subjects => {:composite => {:size => 3000, :sources => [{:id => {:terms => {:field => "subjects.id"}}},
                                                                                {:title => {:terms => {:field => "subjects.title.keyword"}}}]}}}}
     aggregation_hash
@@ -292,10 +292,10 @@ def teacher_sets_sort_order(sort_order = 0)
     end
     query
   end
-  
+
   # Search elastic documents based on the query.Eg: body: {id: "1234567", title: "test"}
   def search_by_query(body)
-    LogWrapper.log('INFO', {'message' => "Elastic search query: #{body}", 
+    LogWrapper.log('INFO', {'message' => "Elastic search query: #{body}",
                             'method' => 'search_by_query'})
     results = {}
     resp = @client.search(index: @index, body: body)
@@ -332,4 +332,33 @@ def update_document_by_id(id, query)
   def delete_by_query(query)
     @client.delete_by_query(index: @index, body: query)
   end
+
+  def create_or_update_index(index_name, body)
+    begin
+      if @client.indices.exists?(index: index_name)
+        puts "Index #{index_name} already exists. Updating..."
+        @client.indices.put_mapping(index: index_name, body: body[:mappings])
+      else
+        puts "Creating index #{index_name}..."
+        @client.indices.create(index: index_name, body: body)
+      end
+      puts "Index #{index_name} created/updated successfully."
+    rescue StandardError => e
+      puts "Error: #{e.message}"
+    end
+  end
+
+  def delete_index(index_name)
+    begin
+      if @client.indices.exists?(index: index_name)
+        puts "Deleting index #{index_name}..."
+        @client.indices.delete(index: index_name)
+        puts "Index #{index_name} deleted successfully."
+      else
+        puts "Index #{index_name} does not exist."
+      end
+    rescue StandardError => e
+      puts "Error: #{e.message}"
+    end
+  end
 end

docker-compose.yml

@@ -19,6 +19,9 @@ services:
       bash -c "
       RAILS_ENV=test bundle exec rake db:set_up_for_local;
       RAILS_ENV=development bundle exec rake db:set_up_for_local;
+      RAILS_ENV=development bundle exec rake elasticsearch:delete_teacherset_index;
+      RAILS_ENV=development bundle exec rake elasticsearch:create_teacherset_index;
+      RAILS_ENV=development bundle exec rake seeds:teacher_sets;
       bundle exec rails server -b 0.0.0.0
       "
 

lib/tasks/elasticsearch.rake

@@ -0,0 +1,138 @@
+namespace :elasticsearch do
+  desc 'Create Elasticsearch mappings and index for teachersets'
+  task create_teacherset_index: :environment do
+    require 'elasticsearch'
+
+    body = {
+      settings: {
+        analysis: {
+          analyzer: {
+            ts_analyzer: {
+              tokenizer: 'standard',
+              filter: %w[lowercase stop asciifolding]
+            }
+          }
+        }
+      },
+      mappings: {
+        properties: {
+          title: {
+            type: 'text',
+            fields: {
+              keyword: {
+                type: 'keyword'
+              }
+            },
+            analyzer: 'default'
+          },
+          description: {
+            type: 'text',
+            fields: {
+              keyword: {
+                type: 'keyword'
+              }
+            },
+            analyzer: 'default'
+          },
+          contents: {
+            type: 'text',
+            fields: {
+              keyword: {
+                type: 'keyword'
+              }
+            },
+            analyzer: 'default'
+          },
+          grade_begin: { type: 'long' },
+          grade_end: { type: 'long' },
+          id: { type: 'long' },
+          details_url: { type: 'text' },
+          availability: {
+            type: 'text',
+            fields: {
+              raw: { type: 'keyword' }
+            }
+          },
+          total_copies: { type: 'long' },
+          call_number: {
+            type: 'text',
+            fields: {
+              keyword: { type: 'keyword' }
+            }
+          },
+          language: {
+            type: 'text',
+            fields: {
+              keyword: { type: 'keyword' }
+            }
+          },
+          physical_description: {
+            type: 'text',
+            fields: {
+              keyword: { type: 'keyword' }
+            }
+          },
+          primary_language: {
+            type: 'text',
+            fielddata: true
+          },
+          available_copies: { type: 'long' },
+          bnumber: {
+            type: 'text',
+            fields: {
+              keyword: { type: 'keyword' }
+            }
+          },
+          set_type: {
+            type: 'text',
+            fielddata: true
+          },
+          area_of_study: {
+            type: 'text',
+            fielddata: true
+          },
+          created_at: {
+            type: 'date',
+            format: 'strict_date_optional_time||epoch_millis'
+          },
+          updated_at: {
+            type: 'date',
+            format: 'strict_date_optional_time||epoch_millis'
+          },
+          subjects: {
+            type: 'nested',
+            properties: {
+              id: { type: 'long' },
+              title: {
+                type: 'text',
+                fields: {
+                  keyword: { type: 'keyword' }
+                }
+              },
+              created_at: {
+                type: 'date',
+                format: 'strict_date_optional_time||epoch_millis'
+              },
+              updated_at: {
+                type: 'date',
+                format: 'strict_date_optional_time||epoch_millis'
+              }
+            }
+          }
+        }
+      }
+    }
+
+    index_name = 'teacherset'
+    ElasticSearch.new.create_or_update_index(index_name, body)
+  end
+
+  desc 'Create Elasticsearch mappings and index for teacherset subjects'
+  task delete_teacherset_index: :environment do
+    require 'elasticsearch'
+
+    index_name = 'teacherset'
+    ElasticSearch.new.delete_index(index_name)
+  end
+end
+