在 Postgresql 的 where 子句中使用别名列

Question

I have a query like this:

SELECT
    jobs.*, 
    (
        CASE
            WHEN lead_informations.state IS NOT NULL THEN lead_informations.state
            ELSE 'NEW'
        END
    ) AS lead_state
FROM
    jobs
    LEFT JOIN lead_informations ON
        lead_informations.job_id = jobs.id
        AND
        lead_informations.mechanic_id = 3
WHERE
    lead_state = 'NEW'

Which gives the following error:

PGError: ERROR:  column "lead_state" does not exist
LINE 1: ...s.id AND lead_informations.mechanic_id = 3 WHERE (lead_state...

In MySql this is valid, but apparently not in Postgresql. From what I can gather, the reason is that the SELECT part of the query is evaluated later than the WHERE part. Is there a common workaround for this problem?

Answer 1

I struggled on the same issue and "mysql syntax is non-standard" is not a valid argument in my opinion. PostgreSQL adds handy non-standard extensions as well, for example "INSERT ... RETURNING ..." to get auto ids after inserts. Also, repeating large queries is not an elegant solution.

WITH jobs_refined AS (
    SELECT
        jobs.*,
        (SELECT CASE WHEN lead_informations.state IS NOT NULL THEN lead_informations.state ELSE 'NEW' END) AS lead_state
    FROM jobs
    LEFT JOIN lead_informations
        ON lead_informations.job_id = jobs.id
        AND lead_informations.mechanic_id = 3
)
SELECT *
FROM jobs_refined
WHERE lead_state = 'NEW'

Answer 2

You would need to either duplicate the case statement in the where clause, or my preference is to do something like the following:

SELECT *
FROM (
  SELECT 
      jobs.*, 
      (CASE WHEN lead_informations.state IS NOT NULL THEN lead_informations.state ELSE 'NEW' END) as lead_state
  FROM 
      "jobs"
      LEFT JOIN lead_informations ON lead_informations.job_id = jobs.id
      AND lead_informations.mechanic_id = 3
) q1
WHERE (lead_state = 'NEW')

Answer 3

MySQL's support is, as you experienced, non-standard. The correct way is to reprint the same expression used in the SELECT clause:

SELECT
    jobs.*, 
    CASE 
         WHEN lead_informations.state IS NOT NULL THEN lead_informations.state 
         ELSE 'NEW' 
    END AS lead_state
FROM
    jobs
    LEFT JOIN lead_informations ON
        lead_informations.job_id = jobs.id
        AND
        lead_informations.mechanic_id = 3
WHERE
    lead_informations.state IS NULL

Answer 4

I believe the common solution is to use an inner SELECT for the calculation (or CASE statement in this case) so that the result of the inner SELECT is available to the entire outer query by the time the execution gets to that query. Otherwise, the WHERE clause is evaluated first and knows nothing about the SELECT clause.

Answer 5

I used alias in where like this. (Sub Query).

Select "Vendors"."VendorId", "Vendors"."Name","Result"."Total" 
From (Select "Trans"."VendorId", ("Trans"."A"+"Trans"."B"+"Trans"."C")    AS "Total"
        FROM "Trans"
    WHERE "Trans"."Year"=2014                                                
    ) As "Result"
JOIN "Vendors" ON "Result"."VendorId"="Vendors"."VendorId" 
WHERE "Vendors"."Class"='I' AND "Result"."Total" > 200

Answer 6

Subquery:

SELECT "tab_1"."BirthDate", "tab_1"."col_1" FROM (
   SELECT BirthDate, DATEADD(year, 18, BirthDate) AS "col_1" FROM Employees
) AS "tab_1"
WHERE "tab_1"."col_1" >= '2000-12-31';