Merge pull request jamis#40 from sobstel/master

Allow to update duplicates on conflict in PostgreSQL

Author: mberlanda

README.md

@@ -149,22 +149,29 @@ Book.bulk_insert(*destination_columns, ignore: true) do |worker|
 end
 ```
 
-### Update Duplicates (MySQL)
+### Update Duplicates (MySQL, PostgreSQL)
 
 If you don't want to ignore duplicate rows but instead want to update them
 then you can use the _update_duplicates_ option. Set this option to true
-and when a duplicate row is found the row will be updated with your new
-values. Default value for this option is false.
+(MySQL) or list unique column names (PostgreSQL) and when a duplicate row
+is found the row will be updated with your new values.
+Default value for this option is false.
 
 ```ruby
 destination_columns = [:title, :author]
 
-# Update duplicate rows
+# Update duplicate rows (MySQL)
 Book.bulk_insert(*destination_columns, update_duplicates: true) do |worker|
   worker.add(...)
   worker.add(...)
   # ...
 end
+
+# Update duplicate rows (PostgreSQL)
+Book.bulk_insert(*destination_columns, update_duplicates: %w[title]) do |worker|
+  worker.add(...)
+  # ...
+end
 ```
 
 ### Return Primary Keys (PostgreSQL, PostGIS)

lib/bulk_insert/worker.rb

@@ -150,8 +150,14 @@ def primary_key_return_statement
     end
 
     def on_conflict_statement
-      if (adapter_name =~ /\APost(?:greSQL|GIS)/i && ignore )
+      is_postgres = adapter_name =~ /\APost(?:greSQL|GIS)/i
+      if is_postgres && ignore
         ' ON CONFLICT DO NOTHING'
+      elsif is_postgres && update_duplicates
+        update_values = @columns.map do |column|
+          "#{column.name}=EXCLUDED.#{column.name}"
+        end.join(', ')
+        ' ON CONFLICT(' + update_duplicates.join(', ') + ') DO UPDATE SET ' + update_values
       elsif adapter_name =~ /^mysql/i && update_duplicates
         update_values = @columns.map do |column|
           "`#{column.name}`=VALUES(`#{column.name}`)"

test/bulk_insert/worker_test.rb

@@ -342,6 +342,23 @@ class BulkInsertWorkerTest < ActiveSupport::TestCase
     assert_equal pgsql_worker.compose_insert_query, "INSERT  INTO \"testings\" (\"greeting\",\"age\",\"happy\",\"created_at\",\"updated_at\",\"color\") VALUES ('Yo',15,0,NULL,NULL,'chartreuse') ON CONFLICT DO NOTHING RETURNING id"
   end
 
+  test "adapter dependent postgresql methods (with update_duplicates)" do
+    pgsql_worker = BulkInsert::Worker.new(
+      Testing.connection,
+      Testing.table_name,
+      'id',
+      %w(greeting age happy created_at updated_at color),
+      500, # batch size
+      false, # ignore
+      %w(greeting age happy), # update duplicates
+      true # return primary keys
+    )
+    pgsql_worker.adapter_name = 'PostgreSQL'
+    pgsql_worker.add ["Yo", 15, false, nil, nil]
+
+    assert_equal pgsql_worker.compose_insert_query, "INSERT  INTO \"testings\" (\"greeting\",\"age\",\"happy\",\"created_at\",\"updated_at\",\"color\") VALUES ('Yo',15,0,NULL,NULL,'chartreuse') ON CONFLICT(greeting, age, happy) DO UPDATE SET greeting=EXCLUDED.greeting, age=EXCLUDED.age, happy=EXCLUDED.happy, created_at=EXCLUDED.created_at, updated_at=EXCLUDED.updated_at, color=EXCLUDED.color RETURNING id"
+  end
+
   test "adapter dependent PostGIS methods" do
     pgsql_worker = BulkInsert::Worker.new(
       Testing.connection,