MySQL JOIN->即使没有匹配项，也从右表中获取所有数据

Question

I have two tables, one with data (A) and other with categories (B) which is in one of those categories or uncategorized. I need to make simple stats, so I am using JOIN to get all data from table A and join it with table B.

What I need is, that even if there is no match in A, I want to show all categories from B on graph (with count = 0, but I need them to be shown).

So if there are no items with that category id, it does not show that category at all, but I need to show it, even if it has 0 items.

I tried LEFT JOIN, RIGHT JOIN, both of them with UNION and nothing worked.

Table A

id | some | irrelevant | data | category
-----------------------------------------
1  |      |            |      | 0
2  |      |            |      | 0
3  |      |            |      | 1
4  |      |            |      | 2
5  |      |            |      | 3
6  |      |            |      | 3
. . . . . . . . . . .

Table B

id | title  | color
---------------------
1  | Cat 1  | #FFFFFF
2  | Cat 2  | #FF00FF
3  | Cat 3  | #FF0000
4  | Cat 4  | #00FF00
5  | Cat 5  | #00FFFF
6  | Cat 6  | #000000
7  | Cat 7  | #FFFF00

With those data I want to get something like:

Array ( [category] => 0 [count] => 3 [title] => [color] => ) 
Array ( [category] => 1 [count] => 1 [title] => Cat 1 [color] => #FFFFFF ) 
Array ( [category] => 2 [count] => 1 [title] => Cat 2 [color] => #FF00FF ) 
Array ( [category] => 3 [count] => 2 [title] => Cat 3 [color] => #FF0000 ) 
Array ( [category] => 4 [count] => 0 [title] => Cat 4 [color] => #00FF00 ) 
Array ( [category] => 5 [count] => 0 [title] => Cat 5 [color] => #00FFFF ) 
Array ( [category] => 6 [count] => 0 [title] => Cat 6 [color] => #000000 ) 
Array ( [category] => 7 [count] => 0 [title] => Cat 7 [color] => #FFFF00 ) 

But thats what I actually get:

Array ( [category] => 0 [count] => 3 [title] => [color] => ) 
Array ( [category] => 1 [count] => 1 [title] => Cat 1 [color] => #FFFFFF ) 
Array ( [category] => 2 [count] => 1 [title] => Cat 2 [color] => #FF00FF ) 
Array ( [category] => 3 [count] => 2 [title] => Cat 3 [color] => #FF0000 )

This is the call :

"SELECT category, COUNT(*) AS count, title, color FROM A LEFT JOIN B ON B.id=A.category GROUP BY category"

PS.: Once more, I trued RIGHT join too and UNION of RIGHT and LEFT join. Also, the names of tables are really long, cant really use them in call as A.category etc.

Is there some way to do that without using two SQL calls and two loops? I really don't want to do that.

Thanks.

Answer 1

And what about this :

SQL Fiddle

MySQL 5.5.32 Schema Setup :

CREATE TABLE TableA
    (`id` int, `category` int)
;

INSERT INTO TableA
    (`id`, `category`)
VALUES
    (1, 0),
    (2, 0),
    (3, 1),
    (4, 2),
    (5, 3),
    (6, 3)
;

CREATE TABLE TableB
    (`id` int, `title` varchar(5), `color` varchar(7))
;

INSERT INTO TableB
    (`id`, `title`, `color`)
VALUES
    (1, 'Cat 1', '#FFFFFF'),
    (2, 'Cat 2', '#FF00FF'),
    (3, 'Cat 3', '#FF0000'),
    (4, 'Cat 4', '#00FF00'),
    (5, 'Cat 5', '#00FFFF'),
    (6, 'Cat 6', '#000000'),
    (7, 'Cat 7', '#FFFF00')
;

Query 1 :

SELECT TableB.id as category, count(distinct TableA.id) as count,title,color
FROM
(SELECT * FROM TableB
UNION
SELECT 0 as id, '' as title, '' as color) AS TableB
LEFT OUTER JOIN TableA on TableB.id = TableA.category
GROUP BY TableB.id, title, color
ORDER BY TableB.id

Results :

| CATEGORY | COUNT | TITLE |   COLOR |
|----------|-------|-------|---------|
|        0 |     2 |       |         |
|        1 |     1 | Cat 1 | #FFFFFF |
|        2 |     1 | Cat 2 | #FF00FF |
|        3 |     2 | Cat 3 | #FF0000 |
|        4 |     0 | Cat 4 | #00FF00 |
|        5 |     0 | Cat 5 | #00FFFF |
|        6 |     0 | Cat 6 | #000000 |
|        7 |     0 | Cat 7 | #FFFF00 |

Also, instead of using UNION SELECT 0 as id, '' as title, '' as color , you can add a category with id 0 in your TableB

Answer 2

[UPDATE] Adding both create data SQL and the SQL query below to prove the data:

To Add Test Data run the following SQL commands:

DROP TABLE junk_category;

CREATE TABLE junk_category (
    id INT PRIMARY KEY,
    name VARCHAR(32) NOT NULL
);

INSERT INTO junk_category (id,name) VALUES (1,'Cat 1'), (2,'Cat 2'), (3, 'Cat 3'), (4, 'Cat 4');

DROP TABLE junk_data;

CREATE TABLE junk_data (
    id INT PRIMARY KEY,
    name VARCHAR(32) NOT NULL,
    category INT NULL,
    INDEX byCategory (category)
);

INSERT INTO junk_data (id, name, category)
VALUES
(1,'one',1),
(2,'two',1),
(3,'three',1),
(4,'four',2),
(5,'five',2),
(6,'six',3),
(7,'seven',NULL),
(8,'eight',NULL);

To Query Test Data:

SELECT 'Uncategorized' AS category, COUNT(*) AS count
    FROM junk_data A
    WHERE A.category IS NULL
UNION
SELECT B.name AS category, COUNT(DISTINCT A.id) AS count
    FROM junk_category B
    LEFT JOIN junk_data A ON A.category=B.id
    GROUP BY category

The only change needed from the query I had given you before was the DISTINCT A.id. I missed that in the previous paste I had done. Without the DISTINCT A.id, the category itself counts as 1. The DISTINCT limits it to the data table. However, you said you were seeing 3...but the query doesn't "copy" and it wouldn't be 3 where it should be none...so if you are seeing 3 where there should be none, there is some other issue.

[UPDATE 2015-03-03]

If you are not getting all your categories from this query:

SELECT B.name AS category, COUNT(DISTINCT A.id) AS count
    FROM junk_category B
    LEFT JOIN junk_data A ON A.category=B.id
    GROUP BY category

There is only one reason that this wouldn't return all your categories: if they are not uniquely named . The only filter in the above is the GROUP BY .

However, I realize that I may have confused you because I am naming B.name AS category which is the name of one of your columns. The GROUP BY is really GROUP BY B.name . Thus, the only way to lose any entries is if they have the same B.name . So, in any case, your GROUP BY needs to be on your B table, not your A table.

There is not much else I can do without seeing your actual query and seeing if your GROUP BY or WHERE are placing additional restrictions. Or utilize my install-junk sql and make sure you can get the same results and then try to figure out the differences between my queries and what your actual queries are doing.

Answer 3

This answer is not the best variant, but I want to provide it in case nobody here knows how to do what I want with single query and loop.

---

Right know I "solved" it with while loops (which is exactly what I did not want to use), but it actually looks like what I want is impossible...

$get_stats = dbquery("SELECT COUNT(*) AS count, category FROM XXX GROUP BY category");
$get_cats = dbquery("SELECT * FROM YYYYY");
$categories = array();
$stats_arr = array();

while ($cat = dbarray($get_cats)){
    $categories[$cat['id']] = array($cat['title'], $cat['color']);
    $stats_arr[$cat['id']] = array('count' => 0, 'title' => $cat['title'], 'color' => $cat['color']);
}

while ($stats = dbarray($get_stats)){
    if ($stats['category'] === "0"){
        $stats_arr[0] = array('count' => $stats['count'], 'title' => 'Uncategorized', 'color' => '#FFFFFF');
    }else{
        $stats_arr[$stats['category']] = array('count' => $stats['count'], 'title' => $categories[$stats['category']][0], 'color' => $categories[$stats['category']][1]);
    }
}
ksort($stats_arr);

foreach ($stats_arr as $id => $data) {
    // Code here
}

I still want to make it work without need for all of those loops tho.

Its OK for now, as its on a page where I already was looping trough categories, but it still added 2 loops and I want to use it on separate page, so I need to solve it with single query + loop.