Postgresql: add two values to a column separated by a comma?

Question

I'm trying to create a column named geo that has a column that combines two randomly generated points. My table looks like this.

create table data (
    id SERIAL primary key,
    first VARCHAR(50),
    last VARCHAR(50),
    g VARCHAR(50),
    geo VARCHAR(100)
);

Here is my try at generating those random values but get an error because the INSERT has more expression than target columns

INSERT INTO data (first,last,g,geo)
SELECT substr(md5(random()::text), 1, 7),
       substr(md5(random()::text), 1, 10),
       substr(md5(random()::text), 1, 1),
       (random() * (47.606209 - 25.427152)) + 25.427152 , (random() * (-124.389641 - -69.082237)) + -69.082237
FROM generate_series(1, 20);

Does anyone know how to do this? Also, I know geo locations shouldn't be strings but in my case it works perfectly. Thanks

Answer 1

The comma between the values indicates they are two different columns.

INSERT INTO data (first,last,g,geo)
SELECT
  -- first
  substr(md5(random()::text), 1, 7),
  -- last
  substr(md5(random()::text), 1, 10),
  -- g
  substr(md5(random()::text), 1, 1),
  -- geo
  (random() * (47.606209 - 25.427152)) + 25.427152,
  -- ?fifth column?
  (random() * (-124.389641 - -69.082237)) + -69.082237
FROM generate_series(1, 20);

If you want to concatenate strings use || .

INSERT INTO data (first,last,g,geo)
SELECT
  -- first
  substr(md5(random()::text), 1, 7),
  -- last
  substr(md5(random()::text), 1, 10),
  -- g
  substr(md5(random()::text), 1, 1),
  -- geo
  (random() * (47.606209 - 25.427152)) + 25.427152 ||
    ',' ||
    (random() * (-124.389641 - -69.082237)) + -69.082237
FROM generate_series(1, 20);

However, storing these two numeric values as a string will make it slow and difficult to use. Instead, use two numeric columns to store the two numbers.

create table data (
    id bigserial primary key,
    -- Don't put artificial limits on columns, it doesn't save space.
    -- They'll only use as much space as necessary.
    -- Use `text` unless there's a good reason for a limit.
    -- In Postgres, `text` is just an unlimited `varchar`.
    first text,
    last text,
    g text,
    geox numeric,
    geoy numeric
);

INSERT INTO data (first,last,g,geox,geoy)
SELECT
  -- first
  substr(md5(random()::text), 1, 7),
  -- last
  substr(md5(random()::text), 1, 10),
  -- g
  substr(md5(random()::text), 1, 1),
  -- geox
  (random() * (47.606209 - 25.427152)) + 25.427152,
  -- geoy
  (random() * (-124.389641 - -69.082237)) + -69.082237
FROM generate_series(1, 20);

Now they can be compared as numbers and they can be formatted as you need.

select geox || ',' || geoy from data;

Better yet, use the built inpoint type .

create table data (
    id bigserial primary key,
    first text,
    last text,
    g text,
    geo point
);

These can also be generated as x,y strings and cast to a point .

select (random() || ',' || random())::point;

And Postgres has plenty ofbuilt in 2-D geometric functions .

If you want to do real GIS, use PostGIS .