新题定义:该用户之前从没做过该题，对该用户来说，为新题

CREATE TABLE log_table1 (
  id INT,
  user_id INT,
  date DATETIME,
  question_id INT
);
INSERT INTO log_table1 (id, user_id, date, question_id) VALUES
  (1, 1001, '2022-09-01 00:00:00', 1),
  (2, 1001, '2022-09-02 00:00:00', 2),
  (3, 1001, '2022-09-03 00:00:00', 3),
  (4, 1002, '2022-09-04 00:00:00', 3),
  (5, 1002, '2022-09-05 00:00:00', 8),
  (6, 1002, '2022-09-19 00:00:00', 1),
  (7, 1002, '2022-09-19 00:00:00', 4),
  (8, 1002, '2022-09-20 00:00:00', 1),
  (9, 1002, '2022-09-20 00:00:00', 5),
  (10, 1003, '2022-09-19 00:00:00', 1),
  (11, 1003, '2022-09-19 00:00:00', 2),
  (12, 1003, '2022-09-19 00:00:00', 3),
  (13, 1003, '2022-09-20 00:00:00', 3),
  (14, 1003, '2022-09-20 00:00:00', 3),
  (15, 1003, '2022-09-20 00:00:00', 3),
  (16, 1003, '2022-09-21 00:00:00', 7),
  (17, 1003, '2022-09-21 00:00:00', 8),
  (18, 1003, '2022-09-22 00:00:00', 1),
  (19, 1003, '2022-09-22 00:00:00', 9),
  (20, 1003, '2022-09-23 00:00:00', 10),
  (21, 1004, '2022-09-20 00:00:00', 2),
  (22, 1004, '2022-09-21 00:00:00', 3),
  (23, 1004, '2022-09-22 00:00:00', 4),
  (24, 1004, '2022-09-23 00:00:00', 5);

如上表log_table，表存储的数据是用户的做题记录。用户没做过的题为新题，写SQL找出2022-09-19到2022-09-23期间内，每天都做过新题的用户ID。
id为索引值，question id是题目的编号，不同的题目id不同新题定义:该用户之前从没做过该题，对该用户来说，为新题。

SELECT user_id
FROM log_table1 lt
WHERE date BETWEEN '2022-09-19' AND '2022-09-23'
  AND NOT EXISTS (
    SELECT 1
    FROM log_table1
    WHERE user_id = lt.user_id
      AND date < lt.date
      AND question_id = lt.question_id
  )
GROUP BY user_id
HAVING COUNT(DISTINCT date) = 5;

此查询首先在WHERE子句中筛选出日期在2022-09-19到2022-09-23之间的记录。然后，使用NOT EXISTS子查询排除每个用户已经在当天之前做过的题目。最后，通过GROUP BY用户ID分组，并通过HAVING子句筛选出每个用户在时间范围内的独立日期数量等于5（表示每天都出现）。

SELECT user_id
FROM (
  SELECT
    user_id,
    date,
    question_id,
    COUNT(*) OVER (PARTITION BY user_id, question_id) AS cnt
  FROM log_table
  WHERE DATE(date) BETWEEN '2022-09-19' AND '2022-09-23'
) subquery
WHERE cnt = 1
GROUP BY user_id
HAVING COUNT(DISTINCT DATE(date)) = 5;