I am attempting to use Room along with a full-text search (FTS) table. The issue is that Room doesn't recognize or support FTS tables. Though there is a work around for this; you just create the FTS table using execSQL() after the Room database is created. Though you are limited to using RawQuery on tables created this way (instead of Query, Insert, etc.). See my other question on implementing search with Room for more information.

This is works well, but it limits the table to a subset of the functionality normally provided by Room. For instance; you cannot use reactive queries with the RxJava artifact, you can only get a (blocking) result from the database.

So I am trying to mimic this functionality by adding it to my Dao.

To do this, I added a BehaviorProcessor that publishes a new Flowable for each access method. The Processor is then updated each time a value is inserted or deleted into the table.

The issue is that the RawQuery method is a blocking method, and throws an exception if accessed from the main thread. So I wrapped the read method in a Flowable that pushes the result to the Processor upon success. This works fairly well, but I am unsure if this is the best way to go about updating the Processor.

Is there a more "RxJava" way of achieving this effect? Possibly some method I failed to recognize in the Processor that allows me to update it away from the main thread without creating a separate Flowable?

abstract class PlaceDao(private val database: MapDatabase) {

    private val processor = BehaviorProcessor.createDefault<List<Place>>(emptyList())

    protected abstract fun rawReadList(query: SupportSQLiteQuery): Place

    private fun refreshProcessor() {
            emitter.onNext(rawReadList(SimpleSQLiteQuery("SELECT * FROM `places`")))
        }, BackpressureStrategy.LATEST).subscribeOn(Schedulers.io()).observeOn(AndroidSchedulers.mainThread()).subscribe({ places ->

    fun readAll(): Flowable<List<Place>> {
        return processor.publish().autoConnect()

    fun findById(): Flowable<Place> {
        return processor.map { it.first { it.id == id } }.publish().autoConnect()

    fun insert(varargs places: Place) {
        val writableDatabase = database.openHelper.writableDatabase

        for (place in places) {
                "INSERT OR REPLACE INTO `places` " +
                    "(`id`, `title`, `subtitle`, `description`, " +
                    "`latitude`, `longitude`, `type`, `parent`) " +
                    "VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
                arrayOf(place.id, place.title, place.subtitle, place.description,
                    place.location.latitude, place.location.longitude,
                    place.type, place.parent)) 


    fun deleteAll() {
        val writableDatabase = database.openHelper.writableDatabase
        writableDatabase.execSQL("DELETE FROM `places`")



I found out that using a ConnectedFlowable is not the way to go. It seems as if the Flowable returned by publish() is not connected to the BehaviorProcessor and either does not emit the most recent item observed or does not emit any item pushed to the BehaviorSubject after publish() is called.

So I removed any calls to publish() and autoConnect() in my code and just return the BehaviorProcessor itself instead.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.