I have persons, monsters and vampires, all of which can have multiple files referenced from a files table.
Simplified tables:
files: id, name;
character: id, type;
person: id, characterId FK character.id;
monster: id, characterId FK character.id;
vampire: id, characterId FK character.id;
person_files: personId FK person.id, fileId FK file.id, order;
monster_files: monsterId FK monster.id, fileId FK file.id, order;
vampire_files: vampireId FK vampire.id, fileId FK file.id, order;
As input I have a character.id. Am trying to get only one file.id while looking up the lowest ordered files in person_files, vampire_files and monster_files.
If person_files returns a result, we're done. Otherwise look up vampire_files and / or monster_files. Should return null if no files are found.
I could do something like
SELECT *
FROM character
LEFT JOIN person ON person.characterId = character.id
LEFT JOIN monster ON monster.characterId = monster.id
LEFT JOIN vampire ON vampire.characterId = monster.id
LEFT JOIN files as f1 ON person.fileID = f1.fileId
LEFT JOIN files as f2 ON monster.fileID = f2.fileId
LEFT JOIN files as f3 ON monster.fileID = f3.fileId
and filter out the data with where/min/max etc.
However it seems unnecessary to triple join on the files table when I am just looking up one file.id. Is there a subquery possible to prevent joining so many times and return just one myFileId? Would like to do this as efficiently as possible.
ie
LEFT JOIN (SELECT *one file.id*, MIN(order) as myFileId FROM person/monster/vampire...)
One optimization I could think of would be to select one fileId value based on which of person, monster or vampire table has a non-null field. The query will look like this:
SELECT * FROM (
SELECT COALESCE(person.file_id, monster.file_id, vampire.file_id) AS realFileId
FROM character
LEFT JOIN person ON person.characterId = character.id
LEFT JOIN monster ON monster.characterId = monster.id
LEFT JOIN vampire ON vampire.characterId = monster.id
) x
LEFT JOIN files f
ON x.realFileId=f.fileId
If you want to make life simpler for yourself, you may also consider adding a characterType field to character table and use it together with CASE operator to determine which table you need to join:
SELECT * FROM (
SELECT COALESCE(person.file_id, monster.file_id, vampire.file_id) AS realFileId
FROM character
LEFT JOIN person ON person.characterId = character.id
AND character.characterType="person"
LEFT JOIN monster ON monster.characterId = monster.id
AND character.characterType="monster"
LEFT JOIN vampire ON vampire.characterId = monster.id
AND character.characterType="vampire"
) x
LEFT JOIN files f
ON x.realFileId=f.fileId
While this looks more complex than the previous version, the optimizer in this case should optimize away the joins you won't be using (since you'll be literally joining on FALSE) and thus speed up the query quite a bit.
Related
The following query pulls data correctly as expected, however the left join with lnk_cat_isrc table and through that to catalogue table, brings back repeated data if there is more than one item in catalogue which has the same isrcs from isrc table:
SELECT
isrc.ISRC,
isrc.Track_Name,
isrc.ArtistName,
isrc.TitleVersion,
isrc.Track_Time,
`isrc_performer`.`PerformerName` ,
`performer_category`.`PerformerCategory` ,
`isrc_performer`.`PerformerRole` ,
`isrc`.`isrc_ID`,
`isrc_performer`.`Perf_ID`
FROM `isrc`
LEFT JOIN `isrc_performer` ON (isrc.isrc_ID = isrc_performer.isrc_ID)
LEFT JOIN `performer_category` ON (performer_category.PerfCat_ID = isrc_performer.PerfCat_ID)
LEFT JOIN `lnk_cat_isrc` ON (lnk_cat_isrc.isrc_ID = isrc.isrc_ID)
LEFT JOIN `catalogue` ON (catalogue.ID = lnk_cat_isrc.cat_id)
ORDER BY isrc_ID desc LIMIT 0 , 10
";
I cannot use group by on isrc, because the isrc_performer table can have more than one performer to an isrc.
So the relations are like this:
Few items from catalogue table can have several identical items from isrc table. In turn, each isrc can have more than one entry in isrc_performer table.
What I want is to display all corresponding data from isrc_performer in relation to each isrc, but not repeating it for each item from catalogue table.
I also want to display all the rest "empty" isrcs (those which don't have any data in isrc_performer table)
Can you give me any ideas?
P.S. despite I'm not pulling any data from catalogue table itself, I'm using it to search by a catalogue number, when user defines search criteria for $where_condition variable, hence I need to keep it in the query.
i.e. $where_condition = "catalogue.Catalogue LIKE '%test%' OR ISRC LIKE '%test%' OR Track_Name LIKE '%test%' OR ArtistName LIKE '%test%' OR TitleVersion LIKE '%test%' OR PerformerName LIKE '%test%' OR PerformerCategory LIKE '%test%' OR PerformerRole LIKE '%test%'";
------UPD:
trying to graphically represent possible variation in these 3 tables relations:
cat1 - isrc1 - performer1
isrc2 - performer1
- performer2
- performer3
cat2 - isrc2 - performer1
- performer2
- performer3
- isrc3 - performer2
- performer4
cat3 - isrc4
- isrc1 - performer1
UPD (pics added)
Here are screen prints. As you can see on picture 1 there are 9 rows with same isrc number, however there are 3 repeated performers Jason, David, Paul.
This is because 3 different catalogue items have this exact isrc with 3 different performers as per pic 2
= 1(isrc) * 3(catalogue) * 3(performers) = 9 row on output
All I want is that Performers grid would only display 3 rows of this isrc for each performer.
---Rearrange the answer to put the "best" option up top.. .but is all of this for naught.. w/o any data from lnk_cat_isrc or catalogue being returned, why does filtering on catalog make a difference? we're returning all isrc regardless of any filtering because it's a left join...
So this brings into question given sample data what are the expected results.
Possibly more elegant... (but not sure if it would be faster) moving away from exists and simply using a distinct in a subquery so catalog queries always return 1 row per isrc; solving the 1-M problem keeping the left join thereby keeping the isrc records not in the catalog limits.
Return all isrc information performer information if it exists, performer category info if it exists and catalogue information If, and only if it matches the catalog filters.
SELECT isrc.ISRC
, isrc.Track_Name
, isrc.ArtistName
, isrc.TitleVersion
, isrc.Track_Time
,`isrc_performer`.`PerformerName`
,`performer_category`.`PerformerCategory`
,`isrc_performer`.`PerformerRole`
,`isrc`.`isrc_ID`
,`isrc_performer`.`Perf_ID`
FROM `isrc`
LEFT JOIN `isrc_performer`
ON isrc.isrc_ID = isrc_performer.isrc_ID
LEFT JOIN `performer_category`
ON performer_category.PerfCat_ID = isrc_performer.PerfCat_ID
LEFT JOIN (SELECT distinct lnk_cat_isrc.isrc_ID
FROM `lnk_cat_isrc`
INNER JOIN `catalogue`
ON catalogue.ID = lnk_cat_isrc.cat_id
WHERE...) DCat
ON Dcat.isrc_ID = isrc.isrc_ID
ORDER BY isrc_ID desc
LIMIT 0 , 10;
As you pointed out the join is causing the problem. So eliminate the join and use the exists notation. Distinct would also work since you're not selecting any values from catalog; though exists should be faster.
Fast but doesn't include all isrc records... (not sure why the or not exists should bring them back in...)
SELECT isrc.ISRC
, isrc.Track_Name
,isrc.ArtistName
,isrc.TitleVersion
,isrc.Track_Time
,`isrc_performer`.`PerformerName`
,`performer_category`.`PerformerCategory`
,`isrc_performer`.`PerformerRole`
,`isrc`.`isrc_ID`
,`isrc_performer`.`Perf_ID`
FROM `isrc`
LEFT JOIN `isrc_performer`
ON (isrc.isrc_ID = isrc_performer.isrc_ID)
LEFT JOIN `performer_category`
ON (performer_category.PerfCat_ID = isrc_performer.PerfCat_ID)
WHERE EXISTS (SELECT *
FROM `lnk_cat_isrc`
INNER JOIN `catalogue`
ON catalogue.ID = lnk_cat_isrc.cat_id
--and your other criteria
WHERE (lnk_cat_isrc.isrc_ID = isrc.isrc_ID)
)
OR NOT EXISTS (SELECT *
FROM `lnk_cat_isrc`
WHERE lnk_cat_isrc.isrc_ID = isrc.isrc_ID
ORDER BY isrc_ID desc
LIMIT 0 , 10
Or using select distinct simple straight forward; but slow
SELECT isrc.ISRC
, isrc.Track_Name
,isrc.ArtistName
,isrc.TitleVersion
,isrc.Track_Time
,`isrc_performer`.`PerformerName`
,`performer_category`.`PerformerCategory`
,`isrc_performer`.`PerformerRole`
,`isrc`.`isrc_ID`
,`isrc_performer`.`Perf_ID`
FROM `isrc`
LEFT JOIN `isrc_performer`
ON (isrc.isrc_ID = isrc_performer.isrc_ID)
LEFT JOIN `performer_category`
ON (performer_category.PerfCat_ID = isrc_performer.PerfCat_ID)
LEFT JOIN `lnk_cat_isrc`
ON (lnk_cat_isrc.isrc_ID = isrc.isrc_ID)
LEFT JOIN `catalogue`
ON (catalogue.ID = lnk_cat_isrc.cat_id)
--AND (other criteria on catalog here, cause in a where clause you left joins will behave like inner joins)
ORDER BY isrc_ID desc
LIMIT 0 , 10;
I have a MySQL database of contacts with three tables.
person
personContact
personDetails
Each contact shares a primary key called 'ID'
The personContact table contains a value called 'personZip' which happens to be their mailing address zip code.
I'd like to write a SQL query that will give me all the contact data for each person in a specific array of zip codes.
I've written a simple statement to perform an inner join on 2 of the tables:
SELECT * FROM `personContact`
INNER JOIN person
ON personContact.ID=person.ID
I've written a statement to select only the zip codes I need:
SELECT * FROM 'personContact'
WHERE personContact.personZip=12564
OR personContact.personZip=12563
OR personContact.personZip=12522
OR personContact.personZip=12590
OR personContact.personZip=12594
OR personContact.personZip=12533
OR personContact.personZip=12570
OR personContact.personZip=12589
OR personContact.personZip=10509
I'm not sure how to perform two joins, to merge all columns from all three tables.
I'm not sure how to write the query to accommodate both the selection of zip codes and the JOINS.
MySQL errors are not helping me move in the right direction.
You were almost there. Use in which is equivalent to or.
SELECT pc.* --select columns from the other tables as needed.
FROM
`personContact` pc
INNER JOIN person p ON pc.ID = p.ID
INNER JOIN personDetails pd on pd.ID = p.ID
where pc.personzip in (12563, 12522, 10509) -- add more zips as needed
Join all three tables and return all columns while filtering by zip code:
SELECT person.*, personContact.*, personDetails.*
FROM person
INNER JOIN personContact ON personContact.ID = person.ID
INNER JOIN personDetails ON personDetails.ID = person.ID
WHERE personContact.personZip = 12564
OR personContact.personZip = 12563
OR personContact.personZip = 12522
OR personContact.personZip = 12590
OR personContact.personZip = 12594
OR personContact.personZip = 12533
OR personContact.personZip = 12570
OR personContact.personZip = 12589
OR personContact.personZip = 10509
EDIT: Multiple OR statements can be replaced using IN as others have suggested
WHERE personContact.personZip IN (12564, 12563, 12522, 12590 ...)
I have a problem with joining some tables, heres my structure:
tbl_imdb:
fldID fldTitle fldImdbID
1 Moviename 0000001
tbl_genres:
fldID fldGenre
1 Action
2 Drama
tbl_genres_rel:
fldID fldMovieID fldGenreID
1 1 1
2 1 2
What I’m trying to do is a query that will find all movies that is both an action movie and drama, is this possible to do without a subquery, if so, how?
What I'm trying right now is:
SELECT tbl_imdb.*
FROM tbl_imdb
LEFT JOIN tbl_imdb_genres_rel ON ( tbl_imdb.fldID = tbl_imdb_genres_rel.fldMovieID )
LEFT JOIN tbl_imdb_genres ON ( tbl_imdb_genres_rel.fldGenreID = tbl_imdb_genres.fldID )
WHERE tbl_imdb_genres.fldGenre = 'Drama'
AND tbl_imdb_genres.fldGenre = 'Action';
But this dosnt work, however it does work if I only keep one of the two WHERE's, but thats not what I want.
Two ways to do it:
1
SELECT tbl_imdb.*
FROM tbl_imdb
INNER JOIN tbl_genres_rel rel_action
ON tbl_imdb.fldID = rel_action.fldMovieID
INNER JOIN tbl_genres genre_action
ON rel_action.fldGenreId = genre_action.fldID
AND 'Action' = genre_action.fldGenre
INNER JOIN tbl_genres_rel rel_drama
ON tbl_imdb.fldID = rel_drama.fldMovieID
INNER JOIN tbl_genres genre_drama
ON rel_drama.fldGenreId = genre_drama.fldID
AND 'Drama' = genre_drama.fldGenre
This method is on the same path as your original solution. 2 differences:
The join should be inner, not left because you're trying to get movies that certainly have the corresponding genre entry
Since you want to find 2 different generes, you'll have to do the join with tbl_genres_rel and tbl_genres twice, once for each particular genre you're interested in.
2
SELECT tbl_imdb.*
FROM tbl_imdb
INNER JOIN tbl_genres_rel
ON tbl_imdb.fldID = tbl_genres_rel.fldMovieID
INNER JOIN tbl_genres
ON tbl_genres_rel.fldGenreId = tbl_genres.fldID
AND tbl_genres.fldGenre IN ('Action', 'Drama')
GROUP BY tbl_imdb.fldID
HAVING COUNT(*) = 2
Again, the basic join plan is the same. Difference here is that we join to the tbl_genres_rel and tbl_genres path just once. This on itself fetches all genres for one film, and then filters for the one's you're interested in. The ones that qualify will now have 2 rows for each distinct value of tbl_imdb.fldId. The GROUP BY aggregates on that, flattening that into one row. By asserting in the HAVING clause that we have exactly 2 rows, we ensure that we keep only those rows that have both the genres.
(Note that this assumes that there is a unique constraint on tbl_genres_rel over {fldMovieID, fldGenreID}. If such a constraint is not present, you should consider adding it.)
LEFT JOIN is not applicable in your case because records should exist on both tables. And you need to count the instances of the movie
SELECT *
FROM tbl_imdb a
INNER JOIN tbl_genres_rel b
on a.fldID = fldMovieID
INNER JOIN tbl_genres c
on c.fldGenreID = b.fldID
WHERE c.fldGenre IN ('Drama', 'Action')
GROUP BY a.Moviename
HAVING COUNT(*) > 1
i am developing a PHP/MYSQL search module, where i have to search tables based on many different criteria, i have some 11 tables, and i have used multiple joins to create one single MySQL query and based on WHERE clause i intend to search for specific records, here is the MYSQL Query that i am using.
SELECT
prop.id,
prop.serial,
prop.title,
prop.rating,
prop.addDate,
prop.approve,
prop.status,
prop.user_id as userId,
user_det.email as email,
user_det.name as name,
prop.area_id as areaId,
area.name as areaName,
area.zipCode as zipCode,
area.city_id as cityId,
city.name as cityName,
city.state_id as stateId,
state.name as stateName,
state.country_id as countryId,
country.name as countryName,
prop.subCategory_id as subCategoryId,
subCat.name as subCategoryName,
subCat.category_id as categoryId,
cat.name as categoryName,
prop.transaction_id as transactionId,
trans.name as transactionName,
price.area as landArea,
price.price as priceSqFt,
price.total_price as totalPrice,
features.bedroom,
features.bathroom,
features.balcony,
features.furnished,
features.floorNum,
features.totalFloor
FROM properties prop
LEFT JOIN user_details user_det ON (prop.user_id = user_det.user_id)
LEFT JOIN areas area ON (prop.area_id = area.id)
LEFT JOIN cities city ON (area.city_id = city.id)
LEFT JOIN states state ON (city.state_id = state.id)
LEFT JOIN countries country ON (state.country_id = country.id)
LEFT JOIN subCategories subCat ON (prop.subCategory_id = subCat.id)
LEFT JOIN categories cat ON (subCat.category_id = cat.id)
LEFT JOIN transactions trans ON (prop.transaction_id = trans.id)
LEFT JOIN prop_prices price ON (price.property_id = prop.id)
LEFT JOIN prop_features features ON (features.property_id = prop.id)
although all works well here, i have a situation where i have a table called prop_amenities below are the content of this table.
as the table above have multiple property_id if i query it using JOINS then mostly it will return duplicate records or single record omitting others depending on the type of JOIN i use. so instead i would like to deal it this way.
use the table prop_amenities to only deal with conditions not to return the result.
for example i am searching for a property with amenity id 1,5,9,17 and 24, then it should check if all the records exist in the prop_amenities table, i.e 1,5,9,17 and 24 in this case. and return the appropriate records with all above selected columns.
i am clueless on dealing this situation using MySQL. how do i go on this?
thank you..
You said "check if all the records exist in the prop_amenities table" and that's the key word here.
SELECT ...
FROM properties AS prop
LEFT JOIN ...
WHERE EXISTS (SELECT 1 FROM prop_amenities AS pa WHERE pa.property_id = prop.property_id AND pa.amenity_id = 7);
I have entires, equipments, brands, times and seasons.
entries:
id
time
equipment_1
equipment_2
equipments:
id
id_brand
brands:
id
name
times:
id
id_season
seasons:
id
name
My actual SQL query is:
SELECT entries.*, times.id_season AS id_season
FROM entries, seasons
WHERE entries.time = times.id
But in the final query I need the next information that I don't know how to obtain it:
The name for each entries.equipment_ as equipment_1_name and equipment_2_name which is set in brands.name.
The name of the season as season_name.
Thank you in advance!
Assuming you have normalized data. This avoid costly cartesian joins. I never use cartesian joins myself, although there are some cases where they are useful. Not here, though.
SELECT
entries.*,
times.id_seasons AS id_season,
b1.name AS equipment_1_name,
b2.name AS equipment_2_name,
seasons.name AS season_name
FROM entries
LEFT JOIN equipments AS equipments_1
ON equipments_1.id = entries.equipment_1
LEFT JOIN brands AS brands_1
ON brands_1.id = equipments_1.id_brand
LEFT JOIN equipments AS equipments_2
ON equipments_2.id = entries.equipment_2
LEFT JOIN brands AS brands_2
ON brands_2.id = equipments_2.id_brand
LEFT JOIN times
ON times.id = entries.time
LEFT JOIN seasons
ON seasons.id = times.id_season;