我有一个以两种方式构建的类别树:每个类别都有一个路径和一个parentID。路径由类别的ID(从下到上)构成。 parentID引用另一个类别。所以我的表看起来像这样:
id | name | path | parentID
---+-------------+---------+---------
1 | Root | NULL | NULL
2 | Main | NULL | 1
3 | Electronics | |2| | 2
4 | Computers | |3|2| | 3
5 | PCs | |4|3|2| | 4
6 | Macs | |4|3|2| | 4
7 | Cameras | |3|2| | 3
8 | Canon | |7|3|2| | 7
现在我不需要类别' Root'和' Main'。我试图实现的是这样的输出:
id | resolved_path
---+-----------------------------
3 | Electronics
4 | Electronics_Computers
5 | Electronics_Computers_PCs
6 | Electronics_Computers_Macs
7 | Electronics_Cameras
8 | Electronics_Cameras_Canon
所以我有不同的深度,我需要按相反的顺序排列。我在网上找不到这个。我得到的就是这个片段,它显示了该类别的深度:
SELECT
*,
(ROUND(
(LENGTH(cat.path) - LENGTH(REPLACE(cat.path, '|', ''))) / LENGTH('|')
) - 2) depth
FROM
categories cat
WHERE
cat.path IS NOT NULL
我不知道什么更容易:递归地通过parentID或对路径做一些魔术。
答案 0 :(得分:0)
您可以使用此(未优化)功能作为基线:
DELIMITER //
DROP FUNCTION IF EXISTS extract_path //
CREATE FUNCTION extract_path(idlist VARCHAR(255))
RETURNS LONGTEXT
BEGIN
DECLARE result LONGTEXT;
DECLARE tmplist VARCHAR(255);
DECLARE buffer VARCHAR(255);
DECLARE lastpos INT;
-- reverse and trim last separator (that first of reversed string)
SELECT TRIM(BOTH FROM SUBSTRING(REVERSE(idlist), 2)) INTO tmplist;
mainloop: LOOP
-- split on separator
SELECT LOCATE('|', tmplist) INTO lastpos;
-- detect end
IF lastpos IS NULL OR lastpos < 2
THEN LEAVE mainloop;
END IF;
-- resolve next id
SELECT cat.name INTO buffer
FROM categories cat
WHERE cat.id = TRIM(REVERSE(SUBSTRING(tmplist, 1, lastpos - 1)));
-- append new element
SELECT CONCAT(COALESCE(CONCAT(result, '_'), ''), buffer) INTO result;
-- prepare for next iteration
SELECT TRIM(BOTH FROM SUBSTRING(tmplist, lastpos + 1)) INTO tmplist;
-- detect end (corner case)
IF tmplist IS NULL
THEN LEAVE mainloop;
END IF;
END LOOP;
RETURN result;
END //
SELECT extract_path('|1|2|3|');
-- with categories 1 -> foo ; 2 -> bar ; 3 -> baz
-- output is 'baz_bar_foo'
DELIMITER ;