In MySQL, how is it possible to do a SELECT on an array of constant values? Something similar to this:
SELECT ['1', '2', '3'] AS ID;
Where the desired output is:
+-----+
| ID |
+-----+
| 1 |
| 2 |
| 3 |
You can use joins to generate a series of rows.
SELECT 1 AS ID UNION SELECT 2 UNION SELECT 3
For small amounts of data this works well. Down side is that there is no index to check these on.
A temp table is likely far better for this as at least any SQL that joins against these values can then use indexes.
Another possibility if the values are from a limited pool is to just have a table which contains all possible values and select from it:-
SELECT ID
FROM all_poss_values
WHERE ID IN (1,2,3)
or using a generated range of values (which again loses the availability of indexes):-
SELECT 1 + units.i + tens.i * 10 AS ID
FROM (SELECT 0 AS i UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) units,
(SELECT 0 AS i UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) tens
WHERE 1 + units.i + tens.i * 10 IN (1,2,3)
Related
I'm trying to create a stored procedure to quickly insert some data into my DB and just having a little trouble figuring out the quickest way to do this.
Into table one I insert a bunch of data about a file and return the new ID of that file.
I then have a separate table that tracks which versions of the app that file is compatible with.
This table has two columns
FileID, Version
Let's say the returned FileID = 1 in this instance.
The Version String will look like this
"1.1, 1.2, 2.1, 2.2"
Essentially I want to split that string and loop it so it ends up as 4 rows like this
FileID
Version
1
1.1
1
1.2
1
2.1
1
2.2
You can accomplish this goal with a little bit of creative splitting:
SELECT zz.`file_id`, zz.`word`
FROM (SELECT DISTINCT `file_id`, LOWER(TRIM(SUBSTRING_INDEX(SUBSTRING_INDEX('1.1,1.2,1.3,2.0,2.1,2.1.5,2.2', ',', num.`id`), ',', -1))) as `word`
FROM (SELECT (h*1000+t*100+u*10+v+1) as `id`
FROM (SELECT 0 h UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) a,
(SELECT 0 t UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) b,
(SELECT 0 u UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) c,
(SELECT 0 v UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) d) num
WHERE num.`id` >= 0) zz
WHERE zz.`word` NOT IN ('')
ORDER BY zz.`word`;
This will give you up to 10,000 versions … if you need that many.
My next database table will be set up more optimally. Unfortunately this one was already set up where one column [data] contains checkbox array values that were saved the following way:
value 1|~|value 1 value 2|~|value 2 value 3|~|value 3
Not optimal, I know.
What I need is a mysql query that select only the values in [data] column in front of the |~|. Basically think I need to select the only odd values.
Any help pointing me in the right direction is greatly appreciated. I tried an if statement in a query and it did not work. Of course I deleted that by mistake.
What I need is a mysql query that select only the
values in [data] column in front of the |~|.
One thing to note the numbers before |~| must be unique.
It will not show the same number twice.
Query
SELECT
DISTINCT
SUBSTRING (
record_data.column
, LOCATE('|~|', record_data.`column` , number_generator.number) - 1
, 1
) AS number
FROM (
SELECT
#row := #row + 1 AS number
FROM (
SELECT 0 UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9
) record_1
CROSS JOIN (
SELECT 0 UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9
) record_2
CROSS JOIN (
SELECT 0 UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9
) record_4
CROSS JOIN (
SELECT 0 UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9
) record_5
CROSS JOIN (
SELECT #row := 0
) AS init_user_params
) AS number_generator
CROSS JOIN (
SELECT
*
FROM (
SELECT 'value 1|~|value 1 value 2|~|value 2 value 3|~|value 3' AS `column`
) AS record
) AS record_data
WHERE
LOCATE('|~|', record_data.`column` , number_generator.number) <> 0
Result
| number |
| ------ |
| 1 |
| 2 |
| 3 |
demo
My database is like
Name | IC | Item
--------------------------
lee | xxx | pear,bear
--------------------------
ron | xxx | apple,dog
what should I do to retrieve the 4 values contained in the column "Item" and then separate them?
Do you have only two items separated by comma in Item? Or it may vary?
LE: you can use this SQL split comma separated row
LLE: just played around with that and this what I've done:
create table myTable(name varchar(7), ic varchar(7), item varchar(200));
insert into myTable(name,ic,item) values ('lee','xxx','pear,bear');
insert into myTable(name,ic,item) values ('ron','xxx','apple,dog');
insert into myTable(name,ic,item) values ('a','xxx','gamma');
insert into myTable(name,ic,item) values ('b','xxx','a,b,c,d');
SELECT SUBSTRING_INDEX(SUBSTRING_INDEX(t.item, ',', n.n), ',', -1) value
FROM myTable t CROSS JOIN
(
SELECT a.N + b.N * 10 + 1 n
FROM
(SELECT 0 AS N UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9) a
,(SELECT 0 AS N UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9) b
ORDER BY n
) n
WHERE n.n <= 1 + (LENGTH(t.item) - LENGTH(REPLACE(t.item, ',', '')))
ORDER BY value;
I want to read about 8000 files, each containing the daily stock prices of a distinct stock, into a single table and select the latest price in each 2-minute interval and write a Null if no record available in an interval. My idea is add a column called bucketNumber to indicate which interval the record falls into, create another table containing one column of values 1, 2, ..., 195 repeating 8000 times and then joining the two tables. At last select the record with largest timestamps for records with the same bucketNumber.
Is this a good way to do the job? If it is, then how to efficiently generate a table with one column of values 1, 2, ..., 195 repeating 8000 times.
Here's a query that will return you a column of integer values from 1 to 8000
SELECT thousands.d*1000 + hundreds.d*100 + tens.d*10 + ones.d + 1 AS num
FROM ( SELECT 0 AS d UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3
UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6
UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9
) ones
CROSS
JOIN ( SELECT 0 AS d UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3
UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6
UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9
) tens
CROSS
JOIN ( SELECT 0 AS d UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3
UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6
UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9
) hundreds
CROSS
JOIN ( SELECT 0 AS d UNION ALL SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3
UNION ALL SELECT 4 UNION ALL SELECT 5 UNION ALL SELECT 6
UNION ALL SELECT 7 UNION ALL SELECT 8 UNION ALL SELECT 9
) thousands
HAVING num <= 8000
ORDER BY num
Seems like a stored procedure would be the easiest approach. Just loop through each 2 minute interval and select the price from the record having the maximum time in the interval. You could include arguments for the start and end time, which would provide a more general solution.
I have a query like this:
SELECT * FROM mytable WHERE
((num=8198747 AND class='A') OR
(num=1646463 AND class='B') OR
(num=4099442 AND class='C') OR
(num=1176312 AND class='A') OR
(num=2146847 AND class='B') OR
(num=7000296 AND class='F') OR
--...about 400 more clauses like this
)
SHOW INDEXES FROM mytable;
+---------+------------+----------+--------------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name |
+---------+------------+----------+--------------+---------------+
| mytable | 0 | PRIMARY | 1 | id |
| mytable | 1 | nc_idx | 1 | num |
| mytable | 1 | nc_idx | 2 | class |
+---------+------------+----------+--------------+---------------+
3 rows in set (0.00 sec)
My understanding is that for each of the ~400 clauses in the query, it will do a separate BTREE lookup on num=XXXXXXXX. Is there any value in changing the query to:
SELECT * FROM mytable WHERE
((class='A' AND num IN (8198747, 1176312, ...)) OR
(class='B' AND num IN (1247910, 1248192, ...)) OR
(class='F' AND num IN (7244626, 9084903, ...)) OR
--...for each class in the query
)
after adding a new index cn_idx on class and num, in that order?
I think it won't be much faster, since class is just one char, and thus the number of BTREE lookups will be the same. But, each subtree will be shorter. Thoughts?
In short
just stick to OR
the index on num is the only index that will help this query
an composite index on (num,class) has negligible effect, since num is already very selective
Two other ways to write it, using UNION ALL
SELECT * FROM mytable
WHERE (num=8198747 AND class='A')
UNION ALL
SELECT * FROM mytable
WHERE (num=1646463 AND class='B')
UNION ALL
SELECT * FROM mytable
WHERE (num=4099442 AND class='C')
UNION ALL
... etc ...
This may work well if you don't have too many num/class pairs, as each one will individually perform an index seek. (A composite index on num/class will work better than num and class indexes single-field indexes. class/num is less selective)
The second way uses JOIN mechanics, by making a virtual table out of the num/class pairs:
SELECT t.*
FROM mytable t
JOIN (
select 8198747 as num, 'A' as class union all
select 1646463, 'B' union all
select 4099442, 'C' union all
... etc ...
) v on v.num=t.num and v.class=t.class
Performance comparison
Create a table
create table mytable (
id int auto_increment primary key,
num int,
class char(1),
other varchar(10),
date timestamp default current_timestamp) ENGINE InnoDB;
Fill it up with 1 million records
(Note: data properties - selectivity: num ~ 1, class ~ 1/26)
insert into mytable(num, class, other)
select rand()*100000000, char(rand()*26+65), concat('',rand()*10000000)
from
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) a,
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) b,
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) c,
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) d,
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) e,
(select 1 a union all select 2 union all select 3 union all select 4
union all select 5 union all select 6 union all select 7
union all select 8 union all select 9 union all select 0) f
Create the indexes
create index nc_num on mytable(num);
create index nc_class on mytable(class);
Select using OR
select * from mytable
WHERE
(num=38142659 and class='T') OR
(num=42476845 and class='E') OR
(num=45205882 and class='B') OR
(num=84861596 and class='K') OR
..... 100 in total
Output of Show profiles: (run set profiling=1; once. then run the queries. run show profiles to see the last timings)
Duration: 0.00003025
Explain extended (add explain extended before the query)
"id";"select_type";"table";"type";"possible_keys";"key";"key_len";"ref";"rows";"Extra"
"1";"SIMPLE";"mytable";"range";"nc_num,nc_class";"nc_num";"5";NULL;"125";"Using where"
Select using UNION ALL between num/class
SELECT * FROM mytable WHERE (num=38142659 AND class='T') UNION ALL
SELECT * FROM mytable WHERE (num=42476845 AND class='E') UNION ALL
SELECT * FROM mytable WHERE (num=45205882 AND class='B') UNION ALL
SELECT * FROM mytable WHERE (num=84861596 AND class='K') UNION ALL
.... 100 in total
Show Profile
Duration: 0.00069525
Explain extended
"id";"select_type";"table";"type";"possible_keys";"key";"key_len";"ref";"rows";"Extra"
"1";"PRIMARY";"mytable";"ref";"nc_num,nc_class";"nc_num";"5";"const";"1";"Using where"
"2";"UNION";"mytable";"ref";"nc_num,nc_class";"nc_num";"5";"const";"1";"Using where"
"3";"UNION";"mytable";"ref";"nc_num,nc_class";"nc_num";"5";"const";"1";"Using where"
"4";"UNION";"mytable";"ref";"nc_num,nc_class";"nc_num";"5";"const";"1";"Using where"
... etc
Select using Union All to build up a virtual table
SELECT t.*
FROM mytable t
JOIN (
select 41805446 num, 'X' collate utf8_general_ci class union all
select 84867135, 'P' union all
select 52747446, 'R' union all
.... etc...
) v on v.num=t.num and v.class=t.class
Show profile
Duration: 0.00026100
Explain extended
"id";"select_type";"table";"type";"possible_keys";"key";"key_len";"ref";"rows";"Extra"
"1";"PRIMARY";"<derived2>";"ALL";NULL;NULL;NULL;NULL;"100";""
"1";"PRIMARY";"t";"ref";"nc_num";"nc_num";"5";"v.num";"1";"Using where"
"2";"DERIVED";NULL;NULL;NULL;NULL;NULL;NULL;NULL;"No tables used"
"3";"UNION";NULL;NULL;NULL;NULL;NULL;NULL;NULL;"No tables used"
"4";"UNION";NULL;NULL;NULL;NULL;NULL;NULL;NULL;"No tables used"
....
"101";"UNION";NULL;NULL;NULL;NULL;NULL;NULL;NULL;"No tables used"
NULL;"UNION RESULT";"<union2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,...>";"ALL";NULL;NULL;NULL;NULL;NULL;""
Select using IN
SELECT * FROM mytable WHERE
(class='A' and num in (28538065
)) or (class='B' and num in (70851926
,90457823
,94804149
)) or (class='C' and num in (74179050
,43280101
,24562525
,56859448
,38226813
,33532373
,93501613
,28634136
,8204338
,15636810
)) or (class='D' and num in (26672499
.... etc
Show profile
Duration: 0.00003125
Explain extended
"id";"select_type";"table";"type";"possible_keys";"key";"key_len";"ref";"rows";"Extra"
"1";"SIMPLE";"mytable";"range";"nc_num,nc_class";"nc_num";"5";NULL;"136";"Using where"