60

I am writing a data-mining program, which bulk inserts user data.

The current SQL is just a plain bulk insert:

insert into USERS(
    id, username, profile_picture)
select unnest(array['12345']),
    unnest(array['Peter']),
    unnest(array['someURL']),
on conflict (id) do nothing;

How do I do an update if on conflict? I tried:

...
    unnest(array['Peter']) as a,
    unnest(array['someURL']) as b,
on conflict (id) do 
update set
    username = a,
    profile_picture = b;

But it throws There is a column named "a" in table "*SELECT*", but it cannot be referenced from this part of the query. error.

EDIT:

Table of USERS is very simple:

create table USERS (
    id      text not null primary key,
    username    text,
    profile_picture text
);
MK Yung
  • 4,344
  • 6
  • 30
  • 35

2 Answers2

118

Turns out a special table named excluded contains the row-to-be-inserted (strange name though)

insert into USERS(
    id, username, profile_picture)
select unnest(array['12345']),
    unnest(array['Peter']),
    unnest(array['someURL'])
on conflict (id) do 
update set
    username = excluded.username,
    profile_picture = excluded.profile_picture;

http://www.postgresql.org/docs/9.5/static/sql-insert.html#SQL-ON-CONFLICT

The SET and WHERE clauses in ON CONFLICT DO UPDATE have access to the existing row using the table's name (or an alias), and to rows proposed for insertion using the special excluded table...

Evan Carroll
  • 78,363
  • 46
  • 261
  • 468
MK Yung
  • 4,344
  • 6
  • 30
  • 35
  • 3
    That naming is so strange, I was really confused by the excluded part. Thanks for clarifying. – adnan Jun 08 '17 at 21:18
1

For bulk insert from another table if they are identical you can do it like that :

INSERT INTO table_a (SELECT * FROM table_b)
ON CONFLICT ON CONSTRAINT "pk_guid"
DO UPDATE SET
column1 = excluded.column1, 
column2 = excluded.column2,
column3 = excluded.column3,
......  ;
moaaz salem
  • 141
  • 1
  • 3