比较运算符在 MySQL 查询中不起作用

Question

i have this MySQL query:

SELECT * FROM forex_pair_signals AS SENALES 
JOIN forex_pair_price AS PRECIO 
ON SENALES.forex_pair_price_id = PRECIO.forex_pair_price_id 
JOIN forex_pair AS PARES 
ON PRECIO.forex_pair_id = PARES.forex_pair_id
WHERE 
PRECIO.forex_pair_price_time >= '2019-02-01 00:00' AND
PRECIO.forex_pair_price_time <= '2019-02-07 01:00' AND 
TIME (PRECIO.forex_pair_price_time) BETWEEN '06:00' AND '13:00' AND 
PARES.forex_pair_id = 13 AND SENALES.forex_pair_signals_result = 1 AND
PRECIO.forex_pair_price_rsi >= '80.00' 
OR PRECIO.forex_pair_price_rsi <= '20.00' 
ORDER BY SENALES.forex_pair_signals_id DESC

This query works fine except when add the following filter:

AND PRECIO.forex_pair_price_rsi >= '80.00' OR 
PRECIO.forex_pair_price_rsi <= '20.00'

The problem is that the last filter do ignores all other filters and the result is get all data that does comply with the last filter.

Answer 1

AND is evaluated before OR because AND has a higher precedence. Use parentheses to override the operator precendence :

WHERE
    PRECIO.forex_pair_price_time >= '2019-02-01 00:00' AND
    PRECIO.forex_pair_price_time <= '2019-02-07 01:00' AND 
    TIME (PRECIO.forex_pair_price_time) BETWEEN '06:00' AND '13:00' AND 
    PARES.forex_pair_id = 13 AND SENALES.forex_pair_signals_result = 1 AND
    (
        PRECIO.forex_pair_price_rsi >= '80.00' 
        OR
        PRECIO.forex_pair_price_rsi <= '20.00'
    )

Or you could rewrite your condition so it's just one expression:

    PRECIO.forex_pair_price_rsi not between '20.01' and '79.99'

Answer 2

Operator precedence When you use more than one logical operator in an expression, MySQL evaluates the OR operators after the AND operators. This is called operator precedence. 1

 SELECT true OR false AND false;

Result

true OR false AND false ----------------------- 1

 SELECT (true OR false) AND false;

Result

(true OR false) AND false ------------------------- 0

thus you need parantheses for such case

WHERE 
PRECIO.forex_pair_price_time >= '2019-02-01 00:00' AND
PRECIO.forex_pair_price_time <= '2019-02-07 01:00' AND 
TIME (PRECIO.forex_pair_price_time) BETWEEN '06:00' AND '13:00' AND 
PARES.forex_pair_id = 13 AND SENALES.forex_pair_signals_result = 1 AND
(
    PRECIO.forex_pair_price_rsi >= '80.00' OR
    PRECIO.forex_pair_price_rsi <= '20.00' 
)

Answer 3

You should use parenthesis while using OR function, as it corresponds to either one or two WHERE Clause defined within it by differentiating in brackets.

SELECT * FROM forex_pair_signals AS SENALES 
JOIN forex_pair_price AS PRECIO 
ON SENALES.forex_pair_price_id = PRECIO.forex_pair_price_id 
JOIN forex_pair AS PARES 
ON PRECIO.forex_pair_id = PARES.forex_pair_id
WHERE 
PRECIO.forex_pair_price_time >= '2019-02-01 00:00' AND
PRECIO.forex_pair_price_time <= '2019-02-07 01:00' AND 
TIME (PRECIO.forex_pair_price_time) BETWEEN '06:00' AND '13:00' AND 
PARES.forex_pair_id = 13 AND SENALES.forex_pair_signals_result = 1 AND
(PRECIO.forex_pair_price_rsi >= '80.00' 
OR PRECIO.forex_pair_price_rsi <= '20.00') 
ORDER BY SENALES.forex_pair_signals_id DESC;

Answer 4

I think I would try to tidy up the query by consolidating the ranged logic in the WHERE clause.

SELECT *
FROM forex_pair_signals AS senales
INNER JOIN forex_pair_price AS precio ON senales.forex_pair_price_id = precio.forex_pair_price_id 
INNER JOIN forex_pair AS pares ON precio.forex_pair_id = pares.forex_pair_id
WHERE pares.forex_pair_id = 13
  AND senales.forex_pair_signals_result = 1
  AND precio.forex_pair_price_time BETWEEN '2019-02-01 00:00' AND '2019-02-07 01:00' 
  AND precio.forex_pair_price_rsi NOT BETWEEN '20.00' AND '80.00'
  AND TIME(precio.forex_pair_price_time) BETWEEN '06:00' AND '13:00'
ORDER BY senales.forex_pair_signals_id DESC

Your final WHERE clause condition can be rewritten as NOT BETWEEN which totally avoids the issue.

I also took the liberty to make your table aliases lowercase and I rearranged the WHERE clause conditions to position the simpler/lighter conditions first.

I think this query is much tidier now.

(ps Try to avoid using * in your SELECT claues; you should only ask for data that you intend to use.)