Iterate over table rows to find corrupted one

README.md

Taken from here.

Query iterates over rows of table my_table. The last printed ctid - is the last before corrupted row. After getting this value ('(78498,1)' for example) you should check the next one ('(78498,2)' for example):

SELECT * FROM my_table WHERE ctid = '(78498,2)';

And then delete it.

Find corrupted row in PostgreSQL

Naming file

find.sql

$$ DO
DECLARE
    result tid;
    curs refcursor;
    ctids record;
    tmp_row record;
    count bigint;
BEGIN
    OPEN curs FOR EXECUTE 'SELECT ctid FROM tbl_las_tin';

    count = 1;
    FETCH curs INTO ctids;

    WHILE ctids.ctid IS NOT NULL LOOP
        count = count + 1;

        result = ctids.ctid;
        FETCH curs INTO ctids;

        EXECUTE 'SELECT (each(hstore(my_table))).* FROM my_table WHERE ctid = $1'
            INTO tmp_row
            USING ctids.ctid;

        IF count % 100 = 0 THEN
            RAISE NOTICE 'rows processed: %', count;
        END IF;
    END LOOP;

    CLOSE curs;
    RAISE NOTICE 'No errors found';

    EXCEPTION
        WHEN OTHERS THEN
            RAISE NOTICE 'LAST CTID: %', result;
            RAISE NOTICE '%: %', SQLSTATE, SQLERRM;
END
$$ LANGUAGE plpgsql;