MySQL Closure Table 分层数据库 - 如何以正确的顺序提取信息

Posted 2023-03-31

【中文标题】MySQL Closure Table 分层数据库 - 如何以正确的顺序提取信息

【英文标题】：MySQL Closure Table hierarchical database - How to pull information out in the correct order

【发布时间】：2011-11-24 04:45:54

【问题描述】：

我有一个 mysql 数据库，它使用 Closure Table 方法保存分层数据。一个简单的示例数据库创建脚本遵循问题。我目前的问题是如何以正确的顺序从数据库中提取数据？我目前正在使用以下 select 语句。

SELECT `TreeData`.`iD`, `TreeData`.`subsectionOf`,
       CONCAT(REPEAT('-', `TreePaths`.`len`),`TreeData`.`name`),
       `TreePaths`.`len`,`TreePaths`.`ancestor`,`TreePaths`.`descendant`
FROM `TreeData`
LEFT JOIN `TreePaths` ON `TreeData`.`iD` = `TreePaths`.`descendant`
WHERE `TreePaths`.`ancestor` = 1
ORDER BY `TreeData`.`subsectionOrder`

它提取了正确的信息，但顺序不正确。

带有示例数据的示例数据库创建脚本。

-- Simple Sample
SET FOREIGN_KEY_CHECKS=0;
DROP TRIGGER IF EXISTS Tree_Insert;
DROP TRIGGER IF EXISTS Tree_Update;
DROP TABLE IF EXISTS TreePaths;
DROP TABLE IF EXISTS TreeData;
SET FOREIGN_KEY_CHECKS=1;


CREATE TABLE `TreeData` (
    `iD`              INT NOT NULL,             -- PK
    `subsectionOf`    INT,                      -- Parent ID & FK
    `subsectionOrder` INT,                      -- Oder of Subsections 
    `name`            NVARCHAR(500) NOT NULL,   -- Name for the entry
    PRIMARY KEY (`iD`),
    FOREIGN KEY (`subsectionOf`) REFERENCES TreeData(`iD`) ON DELETE CASCADE,
    INDEX(`name`)
) ENGINE = MYISAM;

-- Trigger to update the EntryPaths table for new entries
DELIMITER //
CREATE TRIGGER `Tree_Insert` AFTER INSERT ON `TreeData` FOR EACH ROW 
BEGIN
    INSERT INTO `TreePaths` (`ancestor`, `descendant`, `len`)
        SELECT `ancestor`, NEW.`iD`, len + 1 FROM `TreePaths`
            WHERE `descendant` = NEW.`subsectionOf`
            UNION ALL SELECT NEW.`iD`, NEW.`iD`, 0;
END; //
DELIMITER ;


DELIMITER //
CREATE TRIGGER `Tree_Update` BEFORE UPDATE ON `TreeData` FOR EACH ROW 
BEGIN
    -- From http://www.mysqlperformanceblog.com/2011/02/14/moving-subtrees-in-closure-table/
    IF OLD.`subsectionOf` != NEW.`subsectionOf` THEN
        -- Remove the node from its current parent
        DELETE a FROM `TreePaths` AS a
        JOIN `TreePaths` AS d ON a.`descendant` = d.`descendant`
        LEFT JOIN `TreePaths` AS x
        ON x.`ancestor` = d.`ancestor` AND x.`descendant` = a.`ancestor`
        WHERE d.`ancestor` = OLD.`iD` AND x.`ancestor` IS NULL;

        -- Add the node to its new parent
        INSERT `TreePaths` (`ancestor`, `descendant`, `len`)
        SELECT supertree.`ancestor`, subtree.`descendant`, supertree.`len`+subtree.`len`+1
        FROM `TreePaths` AS supertree JOIN `TreePaths` AS subtree
        WHERE subtree.`ancestor` = OLD.`iD`
        AND supertree.`descendant` = NEW.`subsectionOf`;
    END IF;
END; //
DELIMITER ;


CREATE TABLE `TreePaths` (
    `ancestor`      INT NOT NULL,
    `descendant`    INT NOT NULL,
    `len`           INT NOT NULL,
    PRIMARY KEY (`ancestor`, `descendant`),
    FOREIGN KEY (`ancestor`) REFERENCES TreeData(`iD`) ON DELETE CASCADE,
    FOREIGN KEY (`descendant`) REFERENCES TreeData(`iD`) ON DELETE CASCADE
) ENGINE = MYISAM;

INSERT INTO `TreeData` VALUES(1, NULL, NULL, 'Root A');
INSERT INTO `TreeData` VALUES(2, 1, 1, 'Item 1');
INSERT INTO `TreeData` VALUES(3, 1, 2, 'Item 2');
INSERT INTO `TreeData` VALUES(4, 1, 3, 'Item 3');
INSERT INTO `TreeData` VALUES(5, 2, 2, 'Item 1 Sub Item 2');
INSERT INTO `TreeData` VALUES(6, 2, 1, 'Item 1 Sub Item 1');
INSERT INTO `TreeData` VALUES(7, 1, 3, 'Item 4');
INSERT INTO `TreeData` VALUES(8, 4, 1, 'Item 3 Sub Item 1');
INSERT INTO `TreeData` VALUES(9, 4, 2, 'Item 3 Sub Item 2');
INSERT INTO `TreeData` VALUES(10, NULL, NULL, 'Root B');
INSERT INTO `TreeData` VALUES(11, 10, 1, 'Item A');
INSERT INTO `TreeData` VALUES(12, 10, 2, 'Item B');
INSERT INTO `TreeData` VALUES(13, 10, 3, 'Item C');

【参考方案1】：

SELECT d.`iD`, d.`subsectionOf`,
       CONCAT(REPEAT('-', p.`len`), d.`name`) as hier,
       p.`len`, p.`ancestor`, p.`descendant`,
       GROUP_CONCAT(crumbs.`ancestor`) AS breadcrumbs
FROM `TreeData` AS d
JOIN `TreePaths` AS p ON d.`iD` = p.`descendant`
JOIN `TreePaths` AS crumbs ON crumbs.`descendant` = p.`descendant`
WHERE p.`ancestor` = 1
GROUP BY d.`iD`
ORDER BY breadcrumbs;

+----+--------------+---------------------+-----+----------+------------+-------------+
| iD | subsectionOf | hier                | len | ancestor | descendant | breadcrumbs |
+----+--------------+---------------------+-----+----------+------------+-------------+
|  1 |         NULL | Root A              |   0 |        1 |          1 | 1           | 
|  2 |            1 | -Item 1             |   1 |        1 |          2 | 1,2         | 
|  5 |            2 | --Item 1 Sub Item 2 |   2 |        1 |          5 | 1,2,5       | 
|  6 |            2 | --Item 1 Sub Item 1 |   2 |        1 |          6 | 1,2,6       | 
|  3 |            1 | -Item 2             |   1 |        1 |          3 | 1,3         | 
|  4 |            1 | -Item 3             |   1 |        1 |          4 | 1,4         | 
|  8 |            4 | --Item 3 Sub Item 1 |   2 |        1 |          8 | 1,4,8       | 
|  9 |            4 | --Item 3 Sub Item 2 |   2 |        1 |          9 | 1,4,9       | 
|  7 |            1 | -Item 4             |   1 |        1 |          7 | 1,7         | 
+----+--------------+---------------------+-----+----------+------------+-------------+

【讨论】：

这仅适用于您的树是按顺序创建的。想象一下，您的根项目不是 1，而是 12，这将不再有效。在此处移动几个节点以删除序列，此打印将中断。有人有其他解决方案吗？

如果您的闭包表还包含pathlength 列，您可以使用GROUP_CONCAT(crumbs.ancestor ORDER BY pathlength)。

@Thomas GROUP_CONCAT(DISTINCT crumbs.ancestor ORDER BY crumbs.ancestor) 也会完成这项工作

我不明白为什么，但我必须 ORDER BY pathlength DESC 才能使面包屑的顺序正确（即以祖先的 ID 开头）。

我得到了这个工作，但是如果我有多个根项（根 A、根 B、根 C，...）并且我想要从根开始的完整树，我该如何实现项目。根项目的 subsectionOf 为 NULL