MySQL 联合和重复

Question

This is a short snippet of a very long query with about 6 unions but they are all roughly the same logic:

(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = 'Common'
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT 2)

UNION 

(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = ?
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT 1) 

In the second query, the rarity is set depending on chance calculated in PHP. So I get a random number and if it's under 5 the rarity is " Rare " but if it's over 5 the rarity is " Common ".

The problem now arises when it gets " Common ", sometimes the query will choose a common that was already selected in the first query ( I don't want duplicates ). Is it possible to work around this? How can I make it so that it avoids picking a duplicate which will get stripped by the UNION ?

MySQL Version 5.7.30

DB Fiddle Example: https://www.db-fiddle.com/f/7wFDnYuVPt15qPorQvemrw/4

Further explanation :

What I need: Always return 3 unique rows using this UNION .

Problem: It sometimes returns 2 rows as the second query can end up picking a row that was already chosen in the first query.

UNION ALL will not work as it will return duplicates.

Answer 1

I would change the variable parts of the query. Right now your parameter is set_rarity ( cr.set_rarity =? ). I would let it be fixed to "Rare", but make the LIMITs variable.

Assuming you have something like this in your PHP code:

if ($randomNumber > 5) {
    $rarityParam = 'Common';
} else {
    $rarityParam = 'Rare';
}

I would change it to:

$commonLimitParam = 2;
$rareLimitParam = 0;

if ($randomNumber > 5) {
    $commonLimitParam += 1;
} else {
    $rareLimitParam   += 1;
}

and bind the above parameters into tghe following query:

(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = 'Common'
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT ?)

UNION 

(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = 'Rare'
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT ?) 

Answer 2

Look for

SELECT *
FROM (
-- your query
(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = 'Common'
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT 2)

UNION 

(SELECT cr.id, cr.price, cd.times, cd.name, cr.set_rarity, cd.type, cd.currentPrice_tcg 
FROM card_sets_rarity cr
JOIN card_database cd
ON cr.id = cd.id
WHERE cr.set_rarity = ?
AND cr.set_name = 'Dark Beginning 1'
AND cd.type = 'Spell Card'
GROUP BY cd.id ORDER BY RAND() LIMIT 3)  -- LIMIT enlarged to 3
-- end of your query
     ) subquery
ORDER BY set_rarity != 'Common', RAND() LIMIT 3

If parameter is not 'Common' then you'll get 2 'Common' rows and 3 non-'Common' row from subquery, sorting will posess 'Common' rows first and 3rd will be random of 3 non-'Common' rows.

If parameter is 'Common' then you'll get 3-5 rows in subquery and select 3 random from them.

---

fiddle (based on OP's fiddle) - always returns 3 different records (if exists in source table, of course).