Related
I'm trying to get data from the database using Room, I want to get the data in the format {registration_number, List, List} but I'm getting an error:
"Cannot find the parent entity column area_name in ... and my intermediate class"
and in fact I hide that maybe I am taking the wrong approach, please guide me, because I am new in this area
to extract the data I use an intermediate class
my class is:
data class LastConfiscats(
#ColumnInfo(name = "registration_number")
var slaugh_num: String,
// #ColumnInfo(name = "area_name",
#Relation(entity = Area::class, parentColumn = "area_name", entityColumn = "name")
var areaName: List<String>,
// #ColumnInfo(name = "confiscation_name")
#Relation(entity = Confiscation::class, parentColumn = "confiscation_name", entityColumn = "name")
var confiscationName: List<String>
and DAO method to select data:
#Query("SELECT registration_number, area.[name] AS area_name, confiscations.[name] AS confiscation_name " +
"FROM car_body, car_body_confiscations" +
"INNER JOIN area ON car_body_confiscations.area_id == area.id " +
"INNER JOIN confiscations ON car_body_confiscations.confiscation_id == confiscations.id " +
"WHERE car_body.id == car_body_confiscations.car_body_id ORDER BY car_body.id DESC LIMIT :row_count")
fun getLastConfiscats(row_count: Int): LiveData<List<LastConfiscats>>
The linkage scheme between the tables that I am trying to implement is as follows:
There are examples on the internet how to make a relationship between 2 tables but I need to create a relationship between 4 tables.
Please help me to get the data in the right way
UPDATE :
My Area entity is:
#Entity(tableName = "area")
data class Area( #PrimaryKey(autoGenerate = true) var id: Int?, var name: String? )
but in my Confiscation entity I also have "name" column:
#Entity(tableName = "confiscations")
data class Confiscation( #PrimaryKey(autoGenerate = true) var id: Int?, var name: String? )
The actual message you are getting is because when you use #Relation the parent MUST exist and be annotated with #Embedded.
The parent and entity columns MUST be columns in the respective classes.
As an example the following will enable you to get a List of Confiscations, with the related CarBody and the respective Areas (note colum names based upon the screen shots):-
data class LastConfiscats(
#Embedded
var carBodyConfiscations: Car_Body_Confiscations,
#Relation(entity = CarBody::class, parentColumn = "car_body_id", entityColumn = "id")
var carBody: CarBody,
#Relation(entity = Area::class, parentColumn = "areaId", entityColumn = "id")
var area: List<Area>
)
You could use the above with a query such as:-
#Query("SELECT * FROM Car_Body_Confiscations")
fun getCardBodyJoinedWithStuff(): List<LastConfiscats>
No JOINS needed. That is because Room builds the underlying SQL. First is basically the copy of the supplied query. After retrieving the Car_Body_Confiscations it then uses queries based upon the field names/#ColumnInfo and runs queries for each Car_Body_Connfiscation.
For each #Relationship it populates the respective fields (1 carBody and the List of Areas) using queries that it builds. Here's and example of part of the code, for the above from the java(generated) for the query above :-
Main (parent query)
#Override
public List<LastConfiscats> getCardBodyJoinedWithStuff() {
final String _sql = "SELECT * FROM Car_Body_Confiscations";
final RoomSQLiteQuery _statement = RoomSQLiteQuery.acquire(_sql, 0);
....
Later Nn (getting the CarBody(s) there will only be 1)
StringBuilder _stringBuilder = StringUtil.newStringBuilder();
_stringBuilder.append("SELECT `id`,`registrationNumber`,`datetime`,`userId`,`testId` FROM `CarBody` WHERE `id` IN (");
final int _inputSize = _map.size();
Even Later On (Areas)
StringBuilder _stringBuilder = StringUtil.newStringBuilder();
_stringBuilder.append("SELECT `id`,`name` FROM `Area` WHERE `id` IN (");
Now if you want to code your own JOINS etc and alias columns then you will have to consider a few things.
The receiving class MUST be able to be built from the result set and thus column names MUST match the fields in the POJO (unless using #Prefix annotation).
You also need to be aware that the result set will be the cartesian product, thus in the case of doing the above, bypassing how Room does it, the for each combination/permutation of confiscation/carbody/area you get a row (unless grouped/excluded by where clause). So if you have 1 confiscation joined to 1 car but with 10 areas then you would get 10 rows all with the same confiscation and carbody.
You may wish to consider having a look at Room #Relation annotation with a One To Many relationship. Which explains this a little more and includes an example of using a JOINs
Additional - User and TestLists
You may well want to include the CarBody's User and the Test_Lists so you have a result with all of the related data.
This needs to be looked at from a hierarchical perspective. That is the confiscation has a direct link/reference/map to the CarBody but underneath that are the links/references/mappings to the User from the CarBody and to the Test_Lists.
So to incorporate this you need a POJO for a CarBody with it's User and it's Test_Lists. So, for example:-
data class CarBodyWithUserAndWithTestList(
#Embedded
var carBody: CarBody,
#Relation(
entity = Users::class,
parentColumn = "userId",
entityColumn = "id"
)
var users: Users,
#Relation(
entity = Test_List::class,
parentColumn = "testId",
entityColumn = "id"
)
var testList: List<Test_List>
)
With this you can then amend the LastConfiscats to include a CarBodyWithUserAndWithTestList instead of just a CarBody e.g.:
data class LastConfiscats(
#Embedded
var carBodyConfiscations: Car_Body_Confiscations,
#Relation(entity = CarBody::class, parentColumn = "car_body_id", entityColumn = "id")
//var carBody: CarBody, /* REMOVED */
var carBodyWithUserAndWithTestList: CarBodyWithUserAndWithTestList, /* ADDED */
#Relation(entity = Area::class, parentColumn = "areaId", entityColumn = "id")
var area: List<Area>
)
Note that the #Relation has the CarBody class as the entity. That is because the CarBody is the class that needs to be inspected in order for Room to ascertain the columns used for the links/references/,mappings.
*Working Example/Demo
Here's the entire code for a Working example that inserts some data into all the tables and then extracts the data using the getCardBodyJoinedWithStuff query, it then writes the data to the Log.
the code includes ForeignKey constraints which enforces and helps to maintain referential integrity.
for id's Long rather than Int has been used as Long properly reflects the potential size of the field/value.
autoGenerate = true has not been used as this is inefficient and not needed see https://sqlite.org/autoinc.html, which includes as the very first statement The AUTOINCREMENT keyword imposes extra CPU, memory, disk space, and disk I/O overhead and should be avoided if not strictly needed. It is usually not needed. (autoGenerate = true results in AUTOINCREMENT)
So all the classes/interfaces :-
#Entity(
foreignKeys = [
ForeignKey(
Users::class,
parentColumns = ["id"],
childColumns = ["userId"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
),
ForeignKey(
Test_List::class,
parentColumns = ["id"],
childColumns = ["testId"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
)
]
)
data class CarBody(
#PrimaryKey
var id: Long?=null,
var registrationNumber: Int,
var datetime: String,
#ColumnInfo(index = true)
var userId: Long,
#ColumnInfo(index = true)
var testId: Long
)
#Entity
data class Users(
#PrimaryKey
var id:Long?=null,
var name: String,
var lastName: String,
var email: String,
var password: String
)
#Entity
data class Test_List(
#PrimaryKey
var id: Long?=null,
var date: String,
var is_saved: Boolean
)
#Entity(
foreignKeys = [
ForeignKey(
entity = CarBody::class,
parentColumns = ["id"],
childColumns = ["car_body_id"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
),
ForeignKey(
entity = Confiscation::class,
parentColumns = ["id"],
childColumns = ["confiscation_id"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
),
ForeignKey(
entity = Area::class,
parentColumns = ["id"],
childColumns = ["areaId"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
)
]
)
data class Car_Body_Confiscations(
#PrimaryKey
var id: Long?=null,
#ColumnInfo(index = true)
var car_body_id: Long,
#ColumnInfo(index = true)
var confiscation_id: Long,
#ColumnInfo(index = true)
var areaId: Long
)
#Entity
data class Area(
#PrimaryKey
var id: Long?=null,
var name: String
)
#Entity
data class Confiscation(
#PrimaryKey
var id: Long?=null,
var name: String
)
#Dao
interface AllDao {
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(area: Area): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(carBodyConfiscations: Car_Body_Confiscations): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(carBody: CarBody): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(confiscation: Confiscation): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(users: Users): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(testList: Test_List): Long
#Transaction
#Query("SELECT * FROM Car_Body_Confiscations")
fun getCardBodyJoinedWithStuff(): List<LastConfiscats>
}
#Database(entities = [
Area::class,
Car_Body_Confiscations::class,
CarBody::class,
Confiscation::class,
Users::class,
Test_List::class
],
exportSchema = false, version = 1)
abstract class TheDatabase: RoomDatabase() {
abstract fun getAllDao(): AllDao
companion object {
private var instance: TheDatabase?=null
fun getInstance(context: Context): TheDatabase {
if (instance==null) {
instance = Room.databaseBuilder(context,TheDatabase::class.java,"the_database.db")
.allowMainThreadQueries()
.build()
}
return instance as TheDatabase
}
}
}
data class LastConfiscats(
#Embedded
var carBodyConfiscations: Car_Body_Confiscations,
#Relation(entity = Confiscation::class, parentColumn = "confiscation_id", entityColumn = "id")
var confiscation: Confiscation,
#Relation(entity = CarBody::class, parentColumn = "car_body_id", entityColumn = "id")
//var carBody: CarBody, /* REMOVED */
var carBodyWithUserAndWithTestList: CarBodyWithUserAndWithTestList, /* ADDED */
#Relation(entity = Area::class, parentColumn = "areaId", entityColumn = "id")
var area: List<Area>
)
data class CarBodyWithUserAndWithTestList(
#Embedded
var carBody: CarBody,
#Relation(
entity = Users::class,
parentColumn = "userId",
entityColumn = "id"
)
var users: Users,
#Relation(
entity = Test_List::class,
parentColumn = "testId",
entityColumn = "id"
)
var testList: List<Test_List>
)
The following activity code (note that main thread used for brevity and convenience):-
class MainActivity : AppCompatActivity() {
lateinit var db: TheDatabase
lateinit var dao: AllDao
override fun onCreate(savedInstanceState: Bundle?) {
super.onCreate(savedInstanceState)
setContentView(R.layout.activity_main)
db = TheDatabase.getInstance(this)
dao = db.getAllDao()
dao.insert(Users(100,"Fred","Bloggs","fredBloggs#mail.com","password"))
dao.insert(Users(200,"Jane","Doe","janeDoe#email.org","password"))
/* example where id is autogenerated */
val marySmithId = dao.insert(Users(name = "Mary", lastName = "Smith", email = "marysmith#mailit.co.uk", password = "1234567890"))
dao.insert(Test_List(1,"2022-01-01",false))
dao.insert(Test_List(2,"2022-02-02",true))
dao.insert(CarBody(1000,1234,"2022-01-01",100 /* Fred Bloggs*/,2 ))
dao.insert(CarBody(2000,4321,"2021-12-05",100,1))
dao.insert(CarBody(3000,1111,"2021-09-10",200,2))
dao.insert(Area(100,"Area100"))
dao.insert(Area(200,"Area200"))
dao.insert(Area(300,"Area300"))
dao.insert(Area(400,"Area400"))
dao.insert(Confiscation(901,"C1"))
dao.insert(Confiscation(902,"C2"))
dao.insert(Confiscation(903,"C3"))
dao.insert(Confiscation(904,"C4"))
dao.insert(Car_Body_Confiscations(500,1000,901,100))
dao.insert(Car_Body_Confiscations(510,2000,904,400))
dao.insert(Car_Body_Confiscations(520,3000,902,300))
/* Extract the data and output to the Log */
for(cbc in dao.getCardBodyJoinedWithStuff()) {
val areaList = StringBuilder()
for (a in cbc.area) {
areaList.append("\n\t\tArea is ${a.name} ID is ${a.id}")
}
val testList = StringBuilder()
testList.append("\n\t\tThere are ${cbc.carBodyWithUserAndWithTestList.testList.size} TestLists, they are:")
for (t in cbc.carBodyWithUserAndWithTestList.testList) {
testList.append("\n\t\t\t${t.date} Save is ${t.is_saved} ID is ${t.id}")
}
Log.d(
"DBINFO",
"CBC ID =${cbc.carBodyConfiscations.id}" +
"\n\tConfiscation Name is ${cbc.confiscation.name}" +
"\n\tAreas (there is/are ${cbc.area.size}) they are $areaList}" +
"\n\tCarBody Reg is ${cbc.carBodyWithUserAndWithTestList.carBody.registrationNumber} " +
"Date is ${cbc.carBodyWithUserAndWithTestList.carBody.datetime}" +
"\n\t\tUser is ${cbc.carBodyWithUserAndWithTestList.users.name}" +
",${cbc.carBodyWithUserAndWithTestList.users.lastName} " +
"email is ${cbc.carBodyWithUserAndWithTestList.users.email}" +
"$testList"
)
}
}
}
Result
The Log after running:-
D/DBINFO: CBC ID =500
Confiscation Name is C1
Areas (there is/are 1) they are
Area is Area100 ID is 100}
CarBody Reg is 1234 Date is 2022-01-01
User is Fred,Bloggs email is fredBloggs#mail.com
There are 1 TestLists, they are:
2022-02-02 Save is true ID is 2
D/DBINFO: CBC ID =510
Confiscation Name is C4
Areas (there is/are 1) they are
Area is Area400 ID is 400}
CarBody Reg is 4321 Date is 2021-12-05
User is Fred,Bloggs email is fredBloggs#mail.com
There are 1 TestLists, they are:
2022-01-01 Save is false ID is 1
D/DBINFO: CBC ID =520
Confiscation Name is C2
Areas (there is/are 1) they are
Area is Area300 ID is 300}
CarBody Reg is 1111 Date is 2021-09-10
User is Jane,Doe email is janeDoe#email.org
There are 1 TestLists, they are:
2022-02-02 Save is true ID is 2
Re the Comment
I actually have a Cartesian product, I had to process it somehow, although I do not know how yet.
You may find that the above is fine and processes the product pretty easily.
Where Room's relationship handling can become restrictive is if you want to selectively retrieve related data. The way Room handles #Relation means that it retrieves ALL children irrespective of any JOINS and WHERE clauses. They are only effective if they affect the result of the topmost parent.
In your case, where you don't actually cater for lists (such as multiple users per carbody) then Room should suffice.
The original Query - revisited
Changing your query a little to (largely to suit the previous classes ) to:-
#Query("SELECT " +
"registrationNumber, " +
"area.[name] AS area_name, " +
"confiscation.[name] AS confiscation_name " +
"FROM carbody, car_body_confiscations " +
"INNER JOIN area ON car_body_confiscations.areaId == area.id " +
"INNER JOIN confiscation ON car_body_confiscations.confiscation_id == confiscation.id " +
"WHERE carbody.id == car_body_confiscations.car_body_id " +
"ORDER BY carbody.id DESC " +
"LIMIT :row_count"
)
fun getLastConfiscats(row_count: Int): /*LiveData<*/List<MyQueryPOJO>/*>*/
see the following re MyQueryPOJO
And adding a suitable class (no #Embeddeds or #Relations needed, so Room doesn't get confused with column names) :-
data class MyQueryPOJO(
/* The output columns of the query */
var registrationNumber: Int,
#ColumnInfo(name = "area_name")
var not_the_area_name: String,
var confiscation_name: String
)
note how the not_the_area_name field has the #ColumnInfo annotation to tell it to use the area_name output column
In the activity, using:-
for (mqo in dao.getLastConfiscats(10)) {
Log.d("DBINFO","Reg = ${mqo.registrationNumber} Confiscation = ${mqo.confiscation_name} Area Name = ${mqo.not_the_area_name}")
}
Results in (with the same data) :-
D/DBINFO: Reg = 1111 Confiscation = C2 Area Name = Area300
D/DBINFO: Reg = 4321 Confiscation = C4 Area Name = Area400
D/DBINFO: Reg = 1234 Confiscation = C1 Area Name = Area100
as the relationships all all basically 1-1 (the references are back to front for a 1-many) the cartesian product is fine as there will not be any duplicates.
I'm trying to save Lists and Books data in a local database using Room.
where each list contains books.
but have a problem implementing this.
Well, I used type converters & it worked but the google docs suggest that that's not an ideal way of saving complex objects and suggest using Relationships
So I tried to define a relationship between the two objects but I couldn't get it to work.
can someone please help out.
#Entity(tableName = "lists_table")
data class Lists(
#PrimaryKey
#ColumnInfo(name = "display_name")
#Json(name = "display_name") val displayName: String,
val books: List<Books>
)
#Entity(tableName = "books_table")
data class Books(
#PrimaryKey
#ColumnInfo(name = "author") val author: String,
#ColumnInfo(name = "book_image") #Json(name = "book_image") val bookImage: String,
#ColumnInfo(name = "width") #Json(name = "book_image_width") val imageWidth: Int,
#ColumnInfo(name = "height") #Json(name = "book_image_height") val imageHeight: Int,
#ColumnInfo(name = "contributor") val contributor: String,
#ColumnInfo(name = "description") val description: String,
#ColumnInfo(name = "publisher") val publisher: String,
#ColumnInfo(name = "rank") val rank: Int,
#ColumnInfo(name = "rank_last_week") #Json(name = "rank_last_week") val rankLastWeek: Int,
#ColumnInfo(name = "title") val title: String,
#ColumnInfo(name = "weeks_on_list") #Json(name = "weeks_on_list") val weeksOnList: Int,
)
Assuming that a Book can exist in many lists, not just one list (a many-many relationship) then you have an table (know by many names such as a reference table, mapping table, associative table ....). Such a table has a map to one of the objects (list or book) an a map to the other (book or list).
To map an object then the map must uniquely identify the mapped object. As PRIMARY KEYS must be UNIQUE then in your case the display_name column is the primary key for a Lists object and the author column is the primary key for a book object.
You may wish to reconsider, as the implication is that every book must have an author unique to the book, so an author could not have multiple books.
So the mapping table would consist of these two columns (display_name and author). The display_name together with the author would be the primary key.
You cannot use the #PrimaryKey annotation to specify a composite primary key, so you use the primaryKeys parameter of the #Entity annotation, which takes a value that is an array of Strings (the respective column names) .
A consideration, is whether or not to enforce the relationships actually being true relationships. ForeignKey constraints (rules) can not only enforce referential integrity they can also help maintain the integrity by specifying that children or updated/deleted in line with the parent.
So you could have:-
#Entity(
primaryKeys = ["display_name_map","bookId_map"],
/* Optional but suggested */
foreignKeys = [
ForeignKey(
entity = Lists::class,
parentColumns = ["display_name"],
childColumns = ["display_name_map"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
),
ForeignKey(
entity = Books::class,
parentColumns = ["bookId"],
childColumns = ["bookId_map"],
onDelete = ForeignKey.CASCADE,
onUpdate = ForeignKey.CASCADE
)
]
)
data class ListsBookMap(
val display_name_map: String,
#ColumnInfo(index = true) /* have an index associated with the table according to bookId */
val bookId_map: Long
)
Note that bookId is the mapped column, thus an author can author many books
i.e. Books has changed to:-
#PrimaryKey
val bookId: Long?=null, /* ADDED as the primary key so that an author can have many books */
#ColumnInfo(name = "author") val author: String,
To facilitate retrieving the parent (Lists) with it's related children (Books) then you utilise a POJO with the parent annotated with #Embedded and the list of children annotated with the #Relation annotation that includes the associateBy parameter to specify the mapping table and the columns therein, i.e. the Junction.
So this could be:-
data class ListWithBooks(
#Embedded
val list: Lists,
#Relation(
entity = Books::class,
parentColumn = "display_name",
entityColumn = "bookId",
associateBy = Junction(
ListsBookMap::class,
parentColumn = "display_name_map",
entityColumn = "bookId_map"
)
)
val booksList: List<Books>
)
You would ned to be able to insert data, including the mappings so you could have an #Dao annotated interface such as :-
#Dao
interface ListAndBookDao {
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(books: Books): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(lists: Lists): Long
#Insert(onConflict = OnConflictStrategy.IGNORE)
fun insert(listsBookMap: ListsBookMap): Long
#Transaction
#Query("SELECT * FROM lists_table")
fun getAllListsWithListOfBooks(): List<ListsWithBooks>
}
Note the #Query this will retrieve Lists with the books for the current Lists.
Demo
Using the above, and a suitable #Database annotated abstract class, and the following in an activity (note run on the main thread for convenience and brevity):-
class MainActivity : AppCompatActivity() {
lateinit var db: TheDatabase
lateinit var dao: ListAndBookDao
override fun onCreate(savedInstanceState: Bundle?) {
super.onCreate(savedInstanceState)
setContentView(R.layout.activity_main)
db = TheDatabase.getInstance(this)
dao = db.getListAndBookDao()
val b1Id = dao.insert(Books(null,"A1","B1IMAGE",10,12,"C","Desc","P",5,4,"The Book 1",10))
val b2Id = dao.insert(Books(null,"A2","B2IMAGE",10,12,"C","Desc","P",5,4,"The Book 2",10))
val b3Id = dao.insert(Books(null,"A3","B3IMAGE",10,12,"C","Desc","P",5,4,"The Book 3",10))
val b4Id = dao.insert(Books(null,"A1","B4IMAGE",10,12,"C","Desc","P",5,4,"The Book 4",10))
val b5Id = dao.insert(Books(null,"A4","B5IMAGE",10,12,"C","Desc","P",5,4,"The Book 5",10))
val list1Name = "BL001"
val list2Name = "BL002"
val list3Name = "BL003"
val list4Name = "BL004"
val list5Name ="BL005"
dao.insert(Lists(list1Name))
dao.insert(Lists(list2Name))
dao.insert(Lists(list3Name))
dao.insert(Lists(list4Name))
dao.insert(Lists(list5Name))
/* Add books to the lists (i.e. the mappings)*/
dao.insert(ListsBookMap(list1Name,b1Id))
dao.insert(ListsBookMap(list1Name,b3Id))
dao.insert(ListsBookMap(list1Name,b5Id))
dao.insert(ListsBookMap(list2Name,b2Id))
dao.insert(ListsBookMap(list2Name,b4Id))
dao.insert(ListsBookMap(list3Name,b1Id))
dao.insert(ListsBookMap(list3Name,b2Id))
dao.insert(ListsBookMap(list3Name,b3Id))
dao.insert(ListsBookMap(list3Name,b4Id))
dao.insert(ListsBookMap(list3Name,b5Id))
dao.insert(ListsBookMap(list4Name,b3Id))
val sb = StringBuilder()
for (lwb in dao.getAllListsWithListOfBooks()) {
sb.clear()
for (b in lwb.booksList) {
sb.append("\n\tBook is ${b.title}, Author is ${b.author} Image is ${b.bookImage} ....")
}
Log.d("DBINFO","List is ${lwb.list.displayName} it has ${lwb.booksList.size} books. They are:-$sb")
}
}
}
When run (only designed to run once) then the output to the log includes:-
D/DBINFO: List is BL001 it has 3 books. They are:-
Book is The Book 1, Author is A1 Image is B1IMAGE ....
Book is The Book 3, Author is A3 Image is B3IMAGE ....
Book is The Book 5, Author is A4 Image is B5IMAGE ....
D/DBINFO: List is BL002 it has 2 books. They are:-
Book is The Book 2, Author is A2 Image is B2IMAGE ....
Book is The Book 4, Author is A1 Image is B4IMAGE ....
D/DBINFO: List is BL003 it has 5 books. They are:-
Book is The Book 1, Author is A1 Image is B1IMAGE ....
Book is The Book 2, Author is A2 Image is B2IMAGE ....
Book is The Book 3, Author is A3 Image is B3IMAGE ....
Book is The Book 4, Author is A1 Image is B4IMAGE ....
Book is The Book 5, Author is A4 Image is B5IMAGE ....
D/DBINFO: List is BL004 it has 1 books. They are:-
Book is The Book 3, Author is A3 Image is B3IMAGE ....
D/DBINFO: List is BL005 it has 0 books. They are:-
i.e. as expected
I'm using Room for persistence in my app. I have a Player table, which has a name and matchesPlayed column.
When a match is played by a Player, I would like to increment matchesPlayed if it was played by an existing Player, or create a new Player otherwise. In order to determine this, I created a SQL query getPlayerByName(name: String) which should return us the Player if it exists. When I run the query in the Database Inspector in Android Studio, everything works fine. However, when I run the same query in the app, it fails. Does anybody know what I'm doing wrong?
Player.kt
#Entity(tableName = "players")
data class Player(
#PrimaryKey val id: String,
val name: String,
#ColumnInfo(name = "matches_played") val matchesPlayed: String,
#ColumnInfo(name = "matches_won") val matchesWon: String
)
PlayerDao.kt
#Dao
interface PlayerDao {
#Query("SELECT * FROM $PLAYER_TABLE_NAME")
fun getAll(): Single<List<Player>>
#Query("SELECT * FROM $PLAYER_TABLE_NAME WHERE name = :name LIMIT 1")
fun getPlayerByName(name: String): Maybe<Player>
#Insert(onConflict = OnConflictStrategy.REPLACE)
fun insert(player: Player): Completable
}
PlayerViewModel.kt
#HiltViewModel
class PlayerViewModel #Inject constructor(
private val playerRepository: PlayerRepositoryImpl
) : ViewModel() {
...
private fun findAndInsertPlayer(
playerName: String,
playerWon: Boolean
) {
viewModelScope.launch {
val increment = if (playerWon) 1 else 0
playerRepository.getPlayerByName(playerName)
.subscribe(
{ player ->
// Handle Success
},
{
// Somehow, this block always gets called
}
)
}
)
}
I have the following object:
#Entity(tableName = "Section", primaryKeys = ["sectionID","number","numberOfServers"])
data class Section(
#ColumnInfo(name = "number")
var number: Int,
#ColumnInfo(name = "numberOfServer")
var numberOfServers: Int
){
#ColumnInfo(name = "sectionID")
var id: Long = 0
}
then I have the DAO method to insert into room as such:
#Insert(onConflict = OnConflictStrategy.IGNORE)
suspend fun insertSection(section: Section): Long
My goal is for an insert operation to be ignored if a Section object with the same pair of number and numberOfServers already exists in the database. In addition I want the section to have an id which is autogenerated by room. I have seen posts that use indexes and the unique attribute to achieve the composite primary key part of my goal and then use the #PrimaryKey(autogenerate = true) to generate the id. However I am not sure that using indexes works the same as primaryKeys does. As far as I see it if I use indexes and set those to unique room will check
if(index1 != unique){
ignore
}else if(index2 != unique){
ignore
...
etc.
while what I want is more of a
if((index1 && index2 combination) != unique){
ignore
}
It seems that indexes do offer the functionality I need according to the documentation and so I ended up with this:
#Entity(tableName = "Section", indices = [Index(value = ["number","numberOfServer"], unique = true)])
data class Section(
#ColumnInfo(name = "number")
var number: Int,
#ColumnInfo(name = "numberOfServer")
var numberOfServers: Int
){
#PrimaryKey(autoGenerate = true)
#ColumnInfo(name = "sectionID")
var id: Long = 0
}
Thank you for participating
TL;DR: #Depth(value = -1) throws nullpointer and other values above 1 are ignored
In my Spring Boot with Neo4j project I have 3 simple entities with relationships:
#NodeEntity
data class Metric(
#Id #GeneratedValue val id: Long = -1,
val name: String = "",
val description: String = "",
#Relationship(type = "CALCULATES")
val calculates: MutableSet<Calculable> = mutableSetOf()
) {
fun calculates(calculable: Calculus) = calculates.add(calculable)
fun calculate() = calculates.map { c -> c.calculate() }.sum()
}
interface Calculable {
fun calculate(): Double
}
#NodeEntity
data class Calculus(
#Id #GeneratedValue val id: Long = -1,
val name: String = "",
#Relationship(type = "LEFT")
var left: Calculable? = null,
#Relationship(type = "RIGHT")
var right: Calculable? = null,
var operator: Operator? = null
) : Calculable {
override fun calculate(): Double =
operator!!.apply(left!!.calculate(), right!!.calculate())
}
#NodeEntity
data class Value(
#Id #GeneratedValue val id: Long = -1,
val name: String = "",
var value: Double = 0.0
) : Calculable {
override fun calculate(): Double = value
}
enum class Operator : BinaryOperator<Double>, DoubleBinaryOperator {//not relevant}
I create a simple graph like this one:
With the following repositories:
#Repository
interface MetricRepository : Neo4jRepository<Metric, Long>{
#Depth(value = 2)
fun findByName(name: String): Metric?
}
#Repository
interface CalculusRepository : Neo4jRepository<Calculus, Long>{
fun findByName(name: String): Calculus?
}
#Repository
interface ValueRepository : Neo4jRepository<Value, Long>{
fun findByName(name: String): Value?
}
And the following code:
// calculus
val five = valueRepository.save(Value(
name = "5",
value = 5.0
))
val two = valueRepository.save(Value(
name = "2",
value = 2.0
))
val fiveTimesTwo = calculusRepository.save(Calculus(
name = "5 * 2",
operator = Operator.TIMES,
left = five,
right = two
))
println("---")
println(fiveTimesTwo)
val fromRepository = calculusRepository.findByName("5 * 2")!!
println(fromRepository) // sometimes has different id than fiveTimesTwo
println("5 * 2 = ${fromRepository.calculate()}")
println("--- \n")
// metric
val metric = metricRepository.save(Metric(
name = "Metric 1",
description = "Measures a calculus",
calculates = mutableSetOf(fromRepository)
))
metricRepository.save(metric)
println("---")
println(metric)
val metricFromRepository = metricRepository.findByName("Metric 1")!!
println(metricFromRepository) // calculates node is partially empty
println("--- \n")
To retrieve the same graph as shown in the picture above (taken from the actual neo4j dashboard), I do metricRepository.findByName("Metric 1") which has #Depth(value = 2) and then print the saved metric and the retrieved metric:
Metric(id=9, name=Metric 1, description=Measures a calculus, calculates=[Calculus(id=2, name=5 * 2, left=Value(id=18, name=5, value=5.0), right=Value(id=1, name=2, value=2.0), operator=TIMES)])
Metric(id=9, name=Metric 1, description=Measures a calculus, calculates=[Calculus(id=2, name=5 * 2, left=null, right=null, operator=TIMES)])
No matter the value of the depth, I can't get the Metric node with all his children nodes, it retrieves one level deep max and returns null on the leaf nodes.
I've read in the docs that depth=-1 retrieves the fully-resolved node but doing so causes the findByName() method to fail with a null pointer: kotlin.KotlinNullPointerException: null
Here is a list of resources I've consulted and a working GitHub repository with the full code:
GitHub Repo
Spring Data Neo4j Reference Documentation
Neo4j-OGM Docs
Final notes:
The entities all have default parameters because Kotlin then makes an empty constructor, I think the OGM needs it
I've also tried making custom queries but couldn't specify the depth value because there are different relationships and can be at different levels
To use the GitHub repository I linked you must have Neo4j installed, the repo has a stackoverflow-question branch with all the code.
Versions:
Spring boot: 2.3.0.BUILD-SNAPSHOT
spring-boot-starter-data-neo4j: 2.3.0.BUILD-SNAPSHOT
Thank you for helping and all feedback is welcomed!
The problem is not with the query depth but with the model. The Metric entity has a relation with Calculable, but Calculable itself has no relationships defined. Spring Data cannot scan all possible implementations of the Calculable interface for their relationships. If you changed Metrics.calculates type to MutableSet<Calculus>, it would work as expected.
To see Cypher requests send to the server you can add logging.level.org.neo4j.ogm.drivers.bolt=DEBUG to the application.properties
Request before the change:
MATCH (n:`Metric`) WHERE n.`name` = $`name_0` WITH n RETURN n,[ [ (n)->[r_c1:`CALCULATES`]->(x1) | [ r_c1, x1 ] ] ], ID(n) with params {name_0=Metric 1}
Request after the change:
MATCH (n:`Metric`) WHERE n.`name` = $`name_0` WITH n RETURN n,[ [ (n)->[r_c1:`CALCULATES`]->(c1:`Calculus`) | [ r_c1, c1, [ [ (c1)-[r_l2:`LEFT`]-(v2:`Value`) | [ r_l2, v2 ] ], [ (c1)-[r_r2:`RIGHT`]-(v2:`Value`) | [ r_r2, v2 ] ] ] ] ] ], ID(n) with params {name_0=Metric 1}