How to group by / Order by with logical sequence

For PostgreSQL:

SELECT section,
       ARRAY_AGG("number" ORDER BY "number") AS "numbers",
       CONCAT_WS(' ', 'from', MIN("number"), 'to', MAX("number")) AS "sequence"
FROM   (
  SELECT *,
         SUM(incr) OVER(ORDER BY section, "Condition", "number") AS grp
  FROM   (
    SELECT *,
           CASE WHEN LAG("Condition") OVER(ORDER BY section, "number") = "Condition"
             THEN NULL
             ELSE 1
           END AS incr
    FROM   <your_table>
  ) q
  WHERE NOT "Condition" = 'No'
) q
GROUP BY
       section, grp
ORDER BY
       section
;


  section | numbers |   sequence    
 ---------+---------+-------------
  A       | {1,2}   | from 1 to 2
  A       | {4,5,6} | from 4 to 6
  B       | {1,2}   | from 1 to 2
 (3 rows)

Here,

• the CASE statement will assign and incrementor (incr = 1) to rows whose "Condition" is different from the previous row
• the SUM creates a running sum to serve as group value (grp)

I give you part of the answer with some thoughts, the path to adopt according to me.

First, you need to create a new field that you can use to group your values.

Without this field, it's complicated to express your need:

Aggregate me the numbers by section, but when you meet the condition "No" you aggregate in another group.

You must arrive at the following result:

Secondly, you can execute the following sql query using the method array_agg to aggregate your values in an array.

SELECT section, array_agg(number) as number -- use the array_agg method
FROM test  -- name of your layer
WHERE "group" != '0'  -- exclusion of the no condition
GROUP BY section, "group" -- group by with the field previously created
ORDER BY section, number

The result is the following :

Not exactly what you asked for, but if you can live with the section being only one row, sqlite has a group_concat function making it very simple:

select section, group_concat(number) numbers
from mytest
where condition <> 'No'
group by section
order by number