MySQL 多表连接

Question

I. Intro

I am building a support system, where a user from a certain country raises an issue on a certain category, and experts from that country, administrative division, and category get assigned the issue.

Ex User from country Germany with zip code 1000 rises an issue on category Software . Experts from country Germany , and/or province with zip code boundaries MIN_PROVINCE_ZIPCODE <= 1000 >= MAX_PROVINCE_ZIPCODE , and/or region with zip code boundaries MIN_REGION_ZIPCODE <= 1000 >= MAX_REGION_ZIPCODE , and category Software get assigned the issue.

Ie: Select all issues where issue country is equal to expert country, and issue category is equal to one of experts' categories, and/or issue zip code is greater or equal to minimum province zip code and smaller or equal to maximum province zip code, and/or issue zip code is greater or equal to region minimum zip code and issue code is smaller or equal to zip codes.

"and/or" means if experts are assigned to take issues from specific administrative division(s), if they are not, then assign them everything that matches their country and category

II. Database Schemas & Records

*Keep in mind!*

a) Experts can be part of...

1. one, multiple, or no category(ies)
2. one, multiple, or no province(s)
3. one, multiple, or no region(s)

b) If experts are NOT part of...

1. a category, then no issues will be assigned to them
2. a province, then all issues for their country and category will be assigned to them
3. a region, then all issues for their province(s) will be assigned to them

1. Schemas

CREATE TABLE IF NOT EXISTS `categories` (
  `id` int(11) unsigned NOT NULL AUTO_INCREMENT,
  `name` varchar(300) NOT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `provinces` (
  `id` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
  `country` varchar(300) NOT NULL,
  `province` varchar(300) NOT NULL,
  `min_zipcode` int(5) unsigned NOT NULL,
  `max_zipcode` int(5) unsigned NOT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `regions` (
  `id` int(11) unsigned NOT NULL AUTO_INCREMENT,
  `provinceid` int(11) unsigned NOT NULL,
  `region` varchar(300) NOT NULL,
  `min_zipcode` int(5) unsigned NOT NULL,
  `max_zipcode` int(5) unsigned NOT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `issues` (
  `id` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
  `categoryid` int(11) unsigned NOT NULL,
  `country` varchar(150) NOT NULL,
  `zipcode` int(5) NOT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `experts` (
  `id` int(11) unsigned NOT NULL AUTO_INCREMENT,
  `country` varchar(150) NOT NULL DEFAULT 'none',
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `experts_categories` (
  `expertid` int(11) unsigned NOT NULL,
  `categoryid` int(11) unsigned NOT NULL,
  PRIMARY KEY (`expertid`,`categoryid`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `experts_provinces` (
  `expertid` int(11) unsigned NOT NULL,
  `provinceid` int(11) unsigned NOT NULL,
  PRIMARY KEY (`expertid`,`provinceid`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `experts_regions` (
  `expertid` int(11) NOT NULL,
  `regionid` int(11) NOT NULL,
  PRIMARY KEY (`expertid`,`regionid`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

2. Records

INSERT INTO `categories` (`id`, `name`) VALUES
(1, 'Software'),
(2, 'Hardware');

INSERT INTO `experts` (`id`, `country`) VALUES
(1, 'Germany'),
(2, 'France'),
(3, 'Germany');

INSERT INTO `experts_categories` (`expertid`, `categoryid`) VALUES
(1, 1),
(1, 2),
(2, 1),
(3, 1);

INSERT INTO `experts_provinces` (`expertid`, `provinceid`) VALUES
(1, 4),
(2, 6),
(2, 7);

INSERT INTO `experts_regions` (`expertid`, `regionid`) VALUES
(1, 8),
(1, 10);

INSERT INTO `issues` (`id`, `categoryid`, `country`, `zipcode`) VALUES
(1, 2, 'Germany', 2100),
(2, 1, 'France', 1900),
(3, 1, 'Germany', 1500),
(4, 2, 'Germany', 2800),
(5, 2, 'France', 1850);

INSERT INTO `provinces` (`id`, `country`, `province`, `min_zipcode`, `max_zipcode`) VALUES
(1, 'Germany', 'Province One', 1000, 1299),
(2, 'Germany', 'Province Two', 1300, 1499),
(3, 'Germany', 'Province Three', 1500, 1999),
(4, 'Germany', 'Province Four', 2000, 2899),
(5, 'France', 'Province One', 1000, 1799),
(6, 'France', 'Province Two', 1800, 2199),
(7, 'France', 'Province Three', 2200, 2399);

INSERT INTO `regions` (`id`, `provinceid`, `region`, `min_zipcode`, `max_zipcode`) VALUES
(1, 1, 'Region One', 1000, 1099),
(2, 1, 'Region Two', 1100, 1159),
(3, 1, 'Region Three', 1160, 1299),
(4, 2, 'Region One', 1300, 1400),
(5, 2, 'Region Two', 1401, 1499),
(6, 3, 'Region One', 1500, 1699),
(7, 3, 'Region Two', 1700, 1999),
(8, 4, 'Region One', 2000, 2299),
(9, 4, 'Region Two', 2300, 2599),
(10, 4, 'Region Three', 2600, 2699),
(11, 4, 'Region Four', 2700, 2899),
(12, 5, 'Region One', 1000, 1699),
(13, 5, 'Region Two', 1700, 1799),
(14, 6, 'Region One', 1800, 2000),
(15, 6, 'Region Two', 2001, 2199),
(16, 7, 'Region One', 2200, 2299),
(17, 7, 'Region Two', 2300, 2399);

3. Visual Schemas

mysql> DESC `categories`;
+-------+------------------+------+-----+---------+----------------+
| Field | Type             | Null | Key | Default | Extra          |
+-------+------------------+------+-----+---------+----------------+
| id    | int(11) unsigned | NO   | PRI | NULL    | auto_increment |
| name  | varchar(300)     | NO   |     | NULL    |                |
+-------+------------------+------+-----+---------+----------------+

mysql> DESC `provinces`;
+-------------+---------------------+------+-----+---------+----------------+
| Field       | Type                | Null | Key | Default | Extra          |
+-------------+---------------------+------+-----+---------+----------------+
| id          | bigint(11) unsigned | NO   | PRI | NULL    | auto_increment |
| country     | varchar(300)        | NO   |     | NULL    |                |
| province    | varchar(300)        | NO   |     | NULL    |                |
| min_zipcode | int(5) unsigned     | NO   |     | NULL    |                |
| max_zipcode | int(5) unsigned     | NO   |     | NULL    |                |
+-------------+---------------------+------+-----+---------+----------------+

mysql> DESC `regions`;
+-------------+------------------+------+-----+---------+----------------+
| Field       | Type             | Null | Key | Default | Extra          |
+-------------+------------------+------+-----+---------+----------------+
| id          | int(11) unsigned | NO   | PRI | NULL    | auto_increment |
| provinceid  | int(11) unsigned | NO   |     | NULL    |                |
| region      | varchar(300)     | NO   |     | NULL    |                |
| min_zipcode | int(5) unsigned  | NO   |     | NULL    |                |
| max_zipcode | int(5) unsigned  | NO   |     | NULL    |                |
+-------------+------------------+------+-----+---------+----------------+

mysql> DESC `issues`;
+------------+---------------------+------+-----+---------+----------------+
| Field      | Type                | Null | Key | Default | Extra          |
+------------+---------------------+------+-----+---------+----------------+
| id         | bigint(11) unsigned | NO   | PRI | NULL    | auto_increment |
| categoryid | int(11) unsigned    | NO   |     | NULL    |                |
| country    | varchar(150)        | NO   |     | NULL    |                |
| zipcode    | int(5)              | NO   |     | NULL    |                |
+------------+---------------------+------+-----+---------+----------------+

mysql> DESC `experts`;
+---------+------------------+------+-----+---------+----------------+
| Field   | Type             | Null | Key | Default | Extra          |
+---------+------------------+------+-----+---------+----------------+
| id      | int(11) unsigned | NO   | PRI | NULL    | auto_increment |
| country | varchar(150)     | NO   |     | none    |                |
+---------+------------------+------+-----+---------+----------------+

mysql> DESC `experts_categories`;
+------------+------------------+------+-----+---------+-------+
| Field      | Type             | Null | Key | Default | Extra |
+------------+------------------+------+-----+---------+-------+
| expertid   | int(11) unsigned | NO   | PRI | NULL    |       |
| categoryid | int(11) unsigned | NO   | PRI | NULL    |       |
+------------+------------------+------+-----+---------+-------+

mysql> DESC `experts_provinces`;
+------------+------------------+------+-----+---------+-------+
| Field      | Type             | Null | Key | Default | Extra |
+------------+------------------+------+-----+---------+-------+
| expertid   | int(11) unsigned | NO   | PRI | NULL    |       |
| provinceid | int(11) unsigned | NO   | PRI | NULL    |       |
+------------+------------------+------+-----+---------+-------+

mysql> DESC `experts_regions`;
+----------+---------+------+-----+---------+-------+
| Field    | Type    | Null | Key | Default | Extra |
+----------+---------+------+-----+---------+-------+
| expertid | int(11) | NO   | PRI | NULL    |       |
| regionid | int(11) | NO   | PRI | NULL    |       |
+----------+---------+------+-----+---------+-------+

4. Visual Records

mysql> SELECT * FROM `categories`;
+----+----------+
| id | name     |
+----+----------+
|  1 | Software |
|  2 | Hardware |
+----+----------+

mysql> SELECT * FROM `provinces`;
+----+---------+----------------+-------------+-------------+
| id | country | province       | min_zipcode | max_zipcode |
+----+---------+----------------+-------------+-------------+
|  1 | Germany | Province One   |        1000 |        1299 |
|  2 | Germany | Province Two   |        1300 |        1499 |
|  3 | Germany | Province Three |        1500 |        1999 |
|  4 | Germany | Province Four  |        2000 |        2899 |
|  5 | France  | Province One   |        1000 |        1799 |
|  6 | France  | Province Two   |        1800 |        2199 |
|  7 | France  | Province Three |        2200 |        2399 |
+----+---------+----------------+-------------+-------------+

mysql> SELECT * FROM `regions`;
+----+------------+--------------+-------------+-------------+
| id | provinceid | region       | min_zipcode | max_zipcode |
+----+------------+--------------+-------------+-------------+
|  1 |          1 | Region One   |        1000 |        1099 |
|  2 |          1 | Region Two   |        1100 |        1159 |
|  3 |          1 | Region Three |        1160 |        1299 |
|  4 |          2 | Region One   |        1300 |        1400 |
|  5 |          2 | Region Two   |        1401 |        1499 |
|  6 |          3 | Region One   |        1500 |        1699 |
|  7 |          3 | Region Two   |        1700 |        1999 |
|  8 |          4 | Region One   |        2000 |        2299 |
|  9 |          4 | Region Two   |        2300 |        2599 |
| 10 |          4 | Region Three |        2600 |        2699 |
| 11 |          4 | Region Four  |        2700 |        2899 |
| 12 |          5 | Region One   |        1000 |        1699 |
| 13 |          5 | Region Two   |        1700 |        1799 |
| 14 |          6 | Region One   |        1800 |        2000 |
| 15 |          6 | Region Two   |        2001 |        2199 |
| 16 |          7 | Region One   |        2200 |        2299 |
| 17 |          7 | Region Two   |        2300 |        2399 |
+----+------------+--------------+-------------+-------------+

mysql> SELECT * FROM `issues`;
+----+------------+---------+---------+
| id | categoryid | country | zipcode |
+----+------------+---------+---------+
|  1 |          2 | Germany |    2100 |
|  2 |          1 | France  |    1900 |
|  3 |          1 | Germany |    1500 |
|  4 |          2 | Germany |    2800 |
|  5 |          2 | France  |    1850 |
+----+------------+---------+---------+

mysql> SELECT * FROM `experts`;
+----+---------+
| id | country |
+----+---------+
|  1 | Germany |
|  2 | France  |
|  3 | Germany |
+----+---------+

mysql> SELECT * FROM `experts_categories`;
+----------+------------+
| expertid | categoryid |
+----------+------------+
|        1 |          1 |
|        1 |          2 |
|        2 |          1 |
|        3 |          1 |
+----------+------------+

mysql> SELECT * FROM `experts_provinces`;
+----------+------------+
| expertid | provinceid |
+----------+------------+
|        1 |          4 |
|        2 |          6 |
|        2 |          7 |
+----------+------------+

mysql> SELECT * FROM `experts_regions`;
+----------+----------+
| expertid | regionid |
+----------+----------+
|        1 |        8 |
|        1 |       10 |
+----------+----------+

III. Solution

I have managed to come up with half of the solution.

1. My half solution

a) Query:

SELECT 
        `i`.`id` `issue_id`, 
        `e`.`id` `expert_id`
FROM `issues` `i` 
INNER JOIN `experts` `e` 
        ON `i`.`country` = `e`.`country` 
INNER JOIN `experts_categories` `ec` 
        ON `e`.`id` = `ec`.`expertid` 
        AND `i`.`categoryid` = `ec`.`categoryid`
ORDER BY `e`.`id`, `ec`.`categoryid` ASC

b) Result:

+----------+-----------+
| issue_id | expert_id |
+----------+-----------+
|        3 |         1 |
|        1 |         1 |
|        4 |         1 |
|        2 |         2 |
|        3 |         3 |
+----------+-----------+

c) Accurate result would have been:

+----------+-----------+
| issue_id | expert_id |
+----------+-----------+
|        1 |         1 |
|        2 |         2 |
|        3 |         3 |
+----------+-----------+

Explanation, as to why the above visual result is the accurate one.

First, let's get a "full join" so we can make the comparisons:

SELECT 
        `i`.`id` `issue_id`, `e`.`id` `expert_id`, `i`.`categoryid` `issue_category_id`, `ec`.`categoryid` `expert_category_id`, 
        `i`.`country` `issue_country`, `e`.`country` `expert_country`,
        `i`.`zipcode` `issue_zipcode`,
        `p`.`id` `province_id`, `p`.`min_zipcode` `province_min_zipcode`, `p`.`max_zipcode` `province_max_zipcode`,
        `r`.`id` `region_id`, `r`.`min_zipcode` `region_min_zipcode`, `r`.`max_zipcode` `region_max_zipcode`
FROM `issues` `i`
INNER JOIN `experts` `e` 
        ON `i`.`country` = `e`.`country`
INNER JOIN `experts_categories` `ec` 
        ON `ec`.`expertid` = `e`.`id`
        AND `i`.`categoryid` = `ec`.`categoryid`
LEFT JOIN `experts_provinces` `ep`
        ON `e`.`id` = `ep`.`expertid`
LEFT JOIN `provinces` `p`
        ON `ep`.`provinceid` = `p`.`id`
LEFT JOIN `experts_regions` `er`
        ON `e`.`id` = `er`.`expertid`
LEFT JOIN `regions` `r`
        ON `er`.`regionid` = `r`.`id`
        AND `p`.`id` = `r`.`provinceid`
ORDER BY `e`.`id`,`ec`.`categoryid` ASC

+----------+-----------+-------------------+--------------------+---------------+----------------+---------------+-------------+----------------------+----------------------+-----------+--------------------+--------------------+
| issue_id | expert_id | issue_category_id | expert_category_id | issue_country | expert_country | issue_zipcode | province_id | province_min_zipcode | province_max_zipcode | region_id | region_min_zipcode | region_max_zipcode |
+----------+-----------+-------------------+--------------------+---------------+----------------+---------------+-------------+----------------------+----------------------+-----------+--------------------+--------------------+
|        3 |         1 |                 1 |                  1 | Germany       | Germany        |          1500 |           4 |                 2000 |                 2899 |        10 |               2600 |               2699 |
|        3 |         1 |                 1 |                  1 | Germany       | Germany        |          1500 |           4 |                 2000 |                 2899 |         8 |               2000 |               2299 |
|        1 |         1 |                 2 |                  2 | Germany       | Germany        |          2100 |           4 |                 2000 |                 2899 |        10 |               2600 |               2699 |
|        1 |         1 |                 2 |                  2 | Germany       | Germany        |          2100 |           4 |                 2000 |                 2899 |         8 |               2000 |               2299 |
|        4 |         1 |                 2 |                  2 | Germany       | Germany        |          2800 |           4 |                 2000 |                 2899 |        10 |               2600 |               2699 |
|        4 |         1 |                 2 |                  2 | Germany       | Germany        |          2800 |           4 |                 2000 |                 2899 |         8 |               2000 |               2299 |
|        2 |         2 |                 1 |                  1 | France        | France         |          1900 |           7 |                 2200 |                 2399 |      NULL |               NULL |               NULL |
|        2 |         2 |                 1 |                  1 | France        | France         |          1900 |           6 |                 1800 |                 2199 |      NULL |               NULL |               NULL |
|        3 |         3 |                 1 |                  1 | Germany       | Germany        |          1500 |        NULL |                 NULL |                 NULL |      NULL |               NULL |               NULL |
+----------+-----------+-------------------+--------------------+---------------+----------------+---------------+-------------+----------------------+----------------------+-----------+--------------------+--------------------+

So comparing (b) query result with (c), manually fixed result, we can notice...

1. issue_id number 3 can NOT be assigned to expert_id number 1 , because issue_id number 1 is from country Germany , just like the expert, is assigned on the category_id number 2 , just like the expert, BUT has a zip code 1500 , and expert_id number 1 is assigned to take issues only from province_id number 4 and regions_id number 8 and 10 within that province . So the regions zip codes range from 2000 to 2299 and from 2600 to 2699 , where our issue zip code does not belong.
2. issue_id number 1 can be assigned to expert_id number 1 , because issue_id number 1 is from country Germany , just like the expert, is assigned on the category_id number 2 , just like the expert, has a zip code 2100 , which is between the boundaries of the province_id number 4 and region_id number 8 within the province, which is assigned to be covered by expert_id number 1 .
3. issue_id number 4 can NOT be assigned to expert_id number 1 , because issue_id number 4 is from country Germany , just like the expert, is assigned on the category_id number 4 , BUT has a zip code 2800 , which is within the boundaries of the province_id number 4 , but it is not within the boundaries of the region_id number 8 and 10 , which is assigned to be covered by expert_id number 1
4. issue_id number 2 can be assigned to expert_id number 2 , because issue_id number 2 is from country France , just like the expert, is assigned on the category_id number 1 , just like the expert, has a zip code 1900 , which is within the boundaries of the province_id number 6 , assigned to be covered by this expert.
5. issue_id number 3 can be assigned to expert_id number 3 , because issue_id number 3 is from country Germany , just like the expert, is assigned on the category_id number 1 , just like the expert. Furthermore, this expert does not have any administrative division restriction. That is, this expert can take issues from all Germany

Thus, we have listed all issues that can be assigned to experts .

2. Missing half solution

As you can see, my half solution does not take into account the administrative division restrictions.
I can not use procedure(s) or views to achieve this, though, I can split it up in multiple queries if that would help.

Database is MySQL (5.0.1 - MySQL Community Server (GPL)) and programming language is PHP (5.1).

Answer 1

I just modify the answer from @krubo.

If you want sub-query, the query would be:

SELECT 
    tis.id AS issue_id, 
    tex.id AS expert_id, 
    tis.categoryid AS issue_category_id,
    tex.categoryid AS expert_category_id,
    tis.country AS issue_country,
    tex.country AS expert_country,
    tis.zipcode AS issue_zipcode,
    tis.provinceid AS province_id,
    tis.province_minzipcode AS province_minzipcode,
    tis.province_maxzipcode AS province_maxzipcode,
    tis.regionid AS region_id,
    tis.region_minzipcode AS region_minzipcode,
    tis.region_maxzipcode AS region_maxzipcode
FROM 
(
    SELECT 
        i.id, categoryid, i.country, zipcode, 
        provinces.id AS provinceid, provinces.min_zipcode AS province_minzipcode,
        provinces.max_zipcode AS province_maxzipcode, regions.id AS regionid, 
        regions.min_zipcode AS region_minzipcode, 
        regions.max_zipcode AS region_maxzipcode
    FROM 
        issues AS i 
    LEFT JOIN provinces ON i.country=provinces.country
      AND i.zipcode BETWEEN provinces.min_zipcode AND provinces.max_zipcode
    LEFT JOIN regions on provinces.id=regions.provinceid
      AND i.zipcode BETWEEN regions.min_zipcode AND regions.max_zipcode
) AS tis 
JOIN 
( 
    SELECT 
       e.id, country, categoryid, provinceid, regionid
    FROM 
       experts e
    JOIN experts_categories ON e.id=experts_categories.expertid
    LEFT JOIN experts_provinces ON e.id=experts_provinces.expertid
    LEFT JOIN experts_regions ON e.id=experts_regions.expertid
) AS tex  
WHERE 
    tis.country=tex.country
    AND tis.categoryid=tex.categoryid
    AND (tis.provinceid IS NULL
        OR tex.provinceid IS NULL
        OR tis.provinceid=tex.provinceid)
    AND (tis.regionid IS NULL
        OR tex.regionid IS NULL
        OR tis.regionid=tex.regionid);

The result is:

+----------+-----------+-------------------+--------------------+---------------+----------------+---------------+-------------+----------------------+----------------------+-----------+--------------------+--------------------+
| issue_id | expert_id | issue_category_id | expert_category_id | issue_country | expert_country | issue_zipcode | province_id | province_min_zipcode | province_max_zipcode | region_id | region_min_zipcode | region_max_zipcode |
+----------+-----------+-------------------+--------------------+---------------+----------------+---------------+-------------+----------------------+----------------------+-----------+--------------------+--------------------+
|        1 |         1 |                 2 |                  2 | Germany       | Germany        |          2100 |           4 |                 2000 |                 2899 |         8 |               2000 |               2299 |
|        2 |         2 |                 1 |                  1 | France        | France         |          1900 |           6 |                 2000 |                 2199 |        14 |               1800 |               2000 |
|        3 |         3 |                 1 |                  1 | Germany       | Germany        |          1500 |           3 |                 2000 |                 1999 |         6 |               1500 |               1699 |

Answer 2

Your SQL will be very complex if you don't use at least one view to organize it. Try this view to match each issue with a province and region:

create view viewissues as
   select issues.id, categoryid, issues.country, zipcode,
      provinces.id as provinceid, regions.id as regionid
   from issues
   left join provinces on issues.country=provinces.country
      and issues.zipcode between provinces.min_zipcode and provinces.max_zipcode
   left join regions on provinces.id=regions.provinceid
      and issues.zipcode between regions.min_zipcode and regions.max_zipcode;

and this view to list experts according to their categories, provinces if any, and regions if any:

create view viewexperts as
   select experts.id, country, categoryid, provinceid, regionid
   from experts
   join experts_categories on experts.id=experts_categories.expertid
   left join experts_provinces on experts.id=experts_provinces.expertid
   left join experts_regions on experts.id=experts_regions.expertid;

Now your final query can be more manageable, by selecting based on these views:

select distinct viewissues.id, viewexperts.id
from viewissues join viewexperts
where viewissues.country=viewexperts.country
and viewissues.categoryid=viewexperts.categoryid
and (viewissues.provinceid is null
  or viewexperts.provinceid is null
  or viewissues.provinceid=viewexperts.provinceid)
and (viewissues.regionid is null
  or viewexperts.regionid is null
  or viewissues.regionid=viewexperts.regionid);