Mesh 2016

Gemfile.lock

@@ -211,6 +211,3 @@ DEPENDENCIES
   spring
   uglifier (>= 1.3.0)
   web-console (~> 2.0)
-
-BUNDLED WITH
-   1.11.2

app/models/adrg.rb

@@ -8,7 +8,7 @@ class Adrg < ActiveRecord::Base
   include MultiLanguageText
 
   searchkick word_middle: [:text_de, :text_fr, :text_it],
-             callbacks: false, language: 'german', batch_size: 50,
+             callbacks: false, language: 'german', batch_size: 10,
              synonyms: -> { CSV.read('data/mesh_2016/synonyms.csv', {col_sep: ';'}) }
 
   def code_display

lib/tasks/db.rake

@@ -268,4 +268,37 @@ namespace :db do
     Rake::Task['db:seed_numcase_data'].reenable
     Rake::Task['db:seed_numcase_data'].invoke(File.join(args.directory, '2013'), '2013')
   end
-end
+
+  desc 'Reads synonyms.csv and writes relevant synonyms to synonyms_reduced.csv.'
+  task :reduce_synonyms, [:directory] => :environment do |t, args|
+    Searchkick.timeout = 20
+
+    file_name = File.join(args.directory, 'mesh_2016/synonyms.csv')
+    csv_contents = CSV.read(file_name, col_sep: ';')
+    count = csv_contents.length
+    pg = ProgressBar.create(total: count, title: "Outputting relevant synonyms...")
+
+    CSV.open(File.join(args.directory, 'mesh_2016/synonyms_reduced.csv',), "w", col_sep: ';') do |csv|
+    csv_contents.each do |row|
+        pg.increment
+        found = false
+        row.each do |item|
+          [Adrg, Drg, Mdc].each do |model|
+            ['de'].each do |locale| # at the moment ['de'] is enough because the synonyms list only contains german synonyms
+              result = model.search item,
+                           fields: ['code^5', {'text_' + locale.to_s + '^2' => :word_middle}, 'relevant_codes_' + locale.to_s],
+                           limit: 1, highlight: {tag: '<mark>'},
+                           misspellings: {edit_distance: 1}, execute: false
+              found = (found or result.length > 0)
+            end
+          end
+        end
+        if found
+          csv << row
+        else
+        end
+     end
+    end
+    pg.finish
+  end
+end