Watermarking for Spark structured streaming with three way joins - scala

I have 3 streams of data: foo, bar and baz.
There's a necessity to join these streams with LEFT OUTER JOIN in a following chain: foo -> bar -> baz.
Here's an attempt to mimic these streams with built-in rate stream:
val rateStream = session.readStream
.format("rate")
.option("rowsPerSecond", 5)
.option("numPartitions", 1)
.load()
val fooStream = rateStream
.select(col("value").as("fooId"), col("timestamp").as("fooTime"))
val barStream = rateStream
.where(rand() < 0.5) // Introduce misses for ease of debugging
.select(col("value").as("barId"), col("timestamp").as("barTime"))
val bazStream = rateStream
.where(rand() < 0.5) // Introduce misses for ease of debugging
.select(col("value").as("bazId"), col("timestamp").as("bazTime"))
That's the first approach to join all together these streams, with assumption that potential delays for foo, bar and baz are small (~ 5 seconds):
val foobarStream = fooStream
.withWatermark("fooTime", "5 seconds")
.join(
barStream.withWatermark("barTime", "5 seconds"),
expr("""
barId = fooId AND
fooTime >= barTime AND
fooTime <= barTime + interval 5 seconds
"""),
joinType = "leftOuter"
)
val foobarbazQuery = foobarStream
.join(
bazStream.withWatermark("bazTime", "5 seconds"),
expr("""
bazId = fooId AND
bazTime >= fooTime AND
bazTime <= fooTime + interval 5 seconds
"""),
joinType = "leftOuter")
.writeStream
.format("console")
.start()
With setup from above, I'm able to observe following tuples of data:
(some_foo, some_bar, some_baz)
(some_foo, some_bar, null)
but still missing (some_foo, null, some_baz) and (some_foo, null, null).
Any ideas, how to properly configure watermarks in order to get all combinations?
UPDATE:
After adding additional watermark for foobarStream surprisingly on barTime:
val foobarbazQuery = foobarStream
.withWatermark("barTime", "1 minute")
.join(/* ... */)`
I'm able to get this (some_foo, null, some_baz) combination, but still missing (some_foo, null, null)...

I'm leaving some information just for reference.
Chaining stream-stream joins doesn't work correctly because Spark only supports global watermark (instead of operator-wise watermark) which may lead to drop intermediate outputs between joins.
Apache Spark community indicated this issue and discussed while ago. Here's a link for more details:
https://lists.apache.org/thread.html/cc6489a19316e7382661d305fabd8c21915e5faf6a928b4869ac2b4a#%3Cdev.spark.apache.org%3E
(Disclaimer: I'm the author initiated the mail thread.)

Related

Spark DataFrames Scala - jump to next group during a loop

I have a dataframe as below and it records the quarter and date during which same incident occurs to which IDs.
I would like to mark an ID and date if the incident happen at two consecutive quarters. And this is how I did it.
val arrArray = dtf.collect.map(x => (x(0).toString, x(1).toString, x(2).toString.toInt))
if (arrArray.length > 0) {
var bufBAQDate = ArrayBuffer[Any]()
for (a <- 1 to arrArray.length - 1) {
val (strBAQ1, strDate1, douTime1) = arrArray(a - 1)
val (strBAQ2, strDate2, douTime2) = arrArray(a)
if (douTime2 - douTime1 == 15 && strBAQ1 == strBAQ2 && strDate1 == strDate2) {
bufBAQDate = (strBAQ2, strDate2) +: bufBAQDate
//println(strBAQ2+" "+strDate2+" "+douTime2)
}
}
val vecBAQDate = bufBAQDate.distinct.toVector.reverse
Is there a better way of doing it? As the same insident can happen many times to one ID during a single day, it is better to jump to the next ID and/or date once an ID and a date is marked. I dont want to create nested loops to filter dataframe.
Note that you current solution misses 20210103 as 1400 - 1345 = 55
I think this does the trick
val windowSpec = Window.partitionBy("ID")
.orderBy("datetime_seconds")
val consecutiveIncidents = dtf.withColumn("raw_datetime", concat($"Date", $"Time"))
.withColumn("datetime", to_timestamp($"raw_datetime", "yyyyMMddHHmm"))
.withColumn("datetime_seconds", $"datetime".cast(LongType))
.withColumn("time_delta", ($"datetime_seconds" - lag($"datetime_seconds", 1).over(windowSpec)) / 60)
.filter($"time_delta" === lit(15))
.select("ID", "Date")
.distinct()
.collect()
.map { case Row(id, date) => (id, date) }
.toList
Basically - convert the datetimes to timestamps, then look for records with the same ID and consecutive times, with their times separated by 15 minutes.
This is done by using lag over a window grouped by ID and ordered by the time.
In order to calculate the time difference The timestamp is converted to unix epoch seconds.
If you don't want to count day-crossing incidients, you can add the date to the groupyBy clause of the window

Saving a Spark-SQL file as csv

I am trying to save the output of SparkSQL to a path but not sure what function to use. I want to do this without using spark data frames. I was trying using write.mode("overwrite").csv("file:///home/user204943816622/Task_3a-out") but not successful. Can someone tell how to do it?
Note: spark SQL will give the output in multiple files. Need to ensure that the data is sorted globally across all the files (parts). So, all words in part 0, will be alphabetically before the words in part 1.
case class Docword(docId: Int, vocabId: Int, count: Int)
case class VocabWord(vocabId: Int, word: String)
// Read the input data
val docwords = spark.read.
schema(Encoders.product[Docword].schema).
option("delimiter", " ").
csv("hdfs:///user/bdc_data/t3/docword.txt").
as[Docword]
val vocab = spark.read.
schema(Encoders.product[VocabWord].schema).
option("delimiter", " ").
csv("hdfs:///user/bdc_data/t3/vocab.txt").
as[VocabWord]
docwords.createOrReplaceTempView("docwords")
vocab.createOrReplaceTempView("vocab")
spark.sql("""SELECT vocab.word AS word1, SUM(count) count1 FROM
docwords INNER JOIN vocab
ON docwords.vocabId = vocab.vocabId
GROUP BY word
ORDER BY count1 DESC""").show(10)
write.mode("overwrite").csv("file:///home/user204943816622/Task_3a-out")
// Required to exit the spark-shell
sys.exit(0)
.show() returns void you should dp something like below
val writeDf = spark.sql("""SELECT vocab.word AS word1, SUM(count) count1 FROM
docwords INNER JOIN vocab
ON docwords.vocabId = vocab.vocabId
GROUP BY word
ORDER BY count1 DESC""")
writeDf.write.mode("overwrite").csv("file:///home/user204943816622/Task_3a-out")
writeDf.show() // this should not be used in prod environment

Exporting PySpark Dataframe to Azure Data Lake Takes Forever

The code below ran perfectly well on the standalone version of PySpark 2.4 on Mac OS (Python 3.7) when the size of the input data (around 6 GB) was small. However, when I ran the code on HDInsight cluster (HDI 4.0, i.e. Python 3.5, PySpark 2.4, 4 worker nodes and each has 64 cores and 432 GB of RAM, 2 header nodes and each has 4 cores and 28 GB of RAM, 2nd generation of data lake) with larger input data (169 GB), the last step, which is, writing data to the data lake, took forever (I killed it after 24 hours of execution) to complete. Given the fact that HDInsight is not popular in the cloud computing community, I could only reference posts that complained about the low speed when writing dataframe to S3. Some suggested to repartition the dataset, which I did, but it did not help.
from pyspark.sql import SparkSession
from pyspark.sql.types import ArrayType, StringType, IntegerType, BooleanType
from pyspark.sql.functions import udf, regexp_extract, collect_set, array_remove, col, size, asc, desc
from pyspark.ml.fpm import FPGrowth
import os
os.environ["PYSPARK_PYTHON"] = "/usr/bin/python3.5"
os.environ["PYSPARK_DRIVER_PYTHON"] = "/usr/bin/python3.5"
def work(order_path, beer_path, corpus_path, output_path, FREQ_THRESHOLD=1000, LIFT_THRESHOLD=1):
print("Creating Spark Environment...")
spark = SparkSession.builder.appName("Menu").getOrCreate()
print("Spark Environment Created!")
print("Working on Checkpoint1...")
orders = spark.read.csv(order_path)
orders.createOrReplaceTempView("orders")
orders = spark.sql(
"SELECT _c14 AS order_id, _c31 AS in_menu_id, _c32 AS in_menu_name FROM orders"
)
orders.createOrReplaceTempView("orders")
beer = spark.read.csv(
beer_path,
header=True
)
beer.createOrReplaceTempView("beer")
beer = spark.sql(
"""
SELECT
order_id AS beer_order_id,
in_menu_id AS beer_in_menu_id,
'-999' AS beer_in_menu_name
FROM beer
"""
)
beer.createOrReplaceTempView("beer")
orders = spark.sql(
"""
WITH orders_beer AS (
SELECT *
FROM orders
LEFT JOIN beer
ON orders.in_menu_id = beer.beer_in_menu_id
)
SELECT
order_id,
in_menu_id,
CASE
WHEN beer_in_menu_name IS NOT NULL THEN beer_in_menu_name
WHEN beer_in_menu_name IS NULL THEN in_menu_name
END AS menu_name
FROM orders_beer
"""
)
print("Checkpoint1 Completed!")
print("Working on Checkpoint2...")
corpus = spark.read.csv(
corpus_path,
header=True
)
keywords = corpus.select("Food_Name").rdd.flatMap(lambda x: x).collect()
orders = orders.withColumn(
"keyword",
regexp_extract(
"menu_name",
"(?=^|\s)(" + "|".join(keywords) + ")(?=\s|$)",
0
)
)
orders.createOrReplaceTempView("orders")
orders = spark.sql("""
SELECT order_id, in_menu_id, keyword
FROM orders
WHERE keyword != ''
""")
orders.createOrReplaceTempView("orders")
orders = orders.groupBy("order_id").agg(
collect_set("keyword").alias("items")
)
print("Checkpoint2 Completed!")
print("Working on Checkpoint3...")
fpGrowth = FPGrowth(
itemsCol="items",
minSupport=0,
minConfidence=0
)
model = fpGrowth.fit(orders)
print("Checkpoint3 Completed!")
print("Working on Checkpoint4...")
frequency = model.freqItemsets
frequency = frequency.filter(col("freq") > FREQ_THRESHOLD)
frequency = frequency.withColumn(
"items",
array_remove("items", "-999")
)
frequency = frequency.filter(size(col("items")) > 0)
frequency = frequency.orderBy(asc("items"), desc("freq"))
frequency = frequency.dropDuplicates(["items"])
frequency = frequency.withColumn(
"antecedent",
udf(
lambda x: "|".join(sorted(x)), StringType()
)(frequency.items)
)
frequency.createOrReplaceTempView("frequency")
lift = model.associationRules
lift = lift.drop("confidence")
lift = lift.filter(col("lift") > LIFT_THRESHOLD)
lift = lift.filter(
udf(
lambda x: x == ["-999"], BooleanType()
)(lift.consequent)
)
lift = lift.drop("consequent")
lift = lift.withColumn(
"antecedent",
udf(
lambda x: "|".join(sorted(x)), StringType()
)(lift.antecedent)
)
lift.createOrReplaceTempView("lift")
result = spark.sql(
"""
SELECT lift.antecedent, freq AS frequency, lift
FROM lift
INNER JOIN frequency
ON lift.antecedent = frequency.antecedent
"""
)
print("Checkpoint4 Completed!")
print("Writing Result to Data Lake...")
result.repartition(1024).write.mode("overwrite").parquet(output_path)
print("All Done!")
def main():
work(
order_path=169.1 GB of txt,
beer_path=4.9 GB of csv,
corpus_path=210 KB of csv,
output_path="final_result.parquet"
)
if __name__ == "__main__":
main()
I first thought this was caused by the file format parquet. However, when I tried csv, I met with the same problem. I tried result.count() to see how many rows the table result has. It took forever to get the row number, just like writing the data to the data lake.
There was a suggestion to use broadcast hash join instead of the default sort-merge join if a large dataset is joined with a small dataset. I thought it was worth trying because the smaller samples in the pilot study told me the row number of frequency is roughly 0.09% of that of lift (See the query below if you have difficulty tracking frequency and lift).
SELECT lift.antecedent, freq AS frequency, lift
FROM lift
INNER JOIN frequency
ON lift.antecedent = frequency.antecedent
With that in mind, I revised my code:
from pyspark.sql import SparkSession
from pyspark.sql.types import ArrayType, StringType, IntegerType, BooleanType
from pyspark.sql.functions import udf, regexp_extract, collect_set, array_remove, col, size, asc, desc
from pyspark.ml.fpm import FPGrowth
import os
os.environ["PYSPARK_PYTHON"] = "/usr/bin/python3.5"
os.environ["PYSPARK_DRIVER_PYTHON"] = "/usr/bin/python3.5"
def work(order_path, beer_path, corpus_path, output_path, FREQ_THRESHOLD=1000, LIFT_THRESHOLD=1):
print("Creating Spark Environment...")
spark = SparkSession.builder.appName("Menu").getOrCreate()
print("Spark Environment Created!")
print("Working on Checkpoint1...")
orders = spark.read.csv(order_path)
orders.createOrReplaceTempView("orders")
orders = spark.sql(
"SELECT _c14 AS order_id, _c31 AS in_menu_id, _c32 AS in_menu_name FROM orders"
)
orders.createOrReplaceTempView("orders")
beer = spark.read.csv(
beer_path,
header=True
)
beer.createOrReplaceTempView("beer")
beer = spark.sql(
"""
SELECT
order_id AS beer_order_id,
in_menu_id AS beer_in_menu_id,
'-999' AS beer_in_menu_name
FROM beer
"""
)
beer.createOrReplaceTempView("beer")
orders = spark.sql(
"""
WITH orders_beer AS (
SELECT *
FROM orders
LEFT JOIN beer
ON orders.in_menu_id = beer.beer_in_menu_id
)
SELECT
order_id,
in_menu_id,
CASE
WHEN beer_in_menu_name IS NOT NULL THEN beer_in_menu_name
WHEN beer_in_menu_name IS NULL THEN in_menu_name
END AS menu_name
FROM orders_beer
"""
)
print("Checkpoint1 Completed!")
print("Working on Checkpoint2...")
corpus = spark.read.csv(
corpus_path,
header=True
)
keywords = corpus.select("Food_Name").rdd.flatMap(lambda x: x).collect()
orders = orders.withColumn(
"keyword",
regexp_extract(
"menu_name",
"(?=^|\s)(" + "|".join(keywords) + ")(?=\s|$)",
0
)
)
orders.createOrReplaceTempView("orders")
orders = spark.sql("""
SELECT order_id, in_menu_id, keyword
FROM orders
WHERE keyword != ''
""")
orders.createOrReplaceTempView("orders")
orders = orders.groupBy("order_id").agg(
collect_set("keyword").alias("items")
)
print("Checkpoint2 Completed!")
print("Working on Checkpoint3...")
fpGrowth = FPGrowth(
itemsCol="items",
minSupport=0,
minConfidence=0
)
model = fpGrowth.fit(orders)
print("Checkpoint3 Completed!")
print("Working on Checkpoint4...")
frequency = model.freqItemsets
frequency = frequency.filter(col("freq") > FREQ_THRESHOLD)
frequency = frequency.withColumn(
"antecedent",
array_remove("items", "-999")
)
frequency = frequency.drop("items")
frequency = frequency.filter(size(col("antecedent")) > 0)
frequency = frequency.orderBy(asc("antecedent"), desc("freq"))
frequency = frequency.dropDuplicates(["antecedent"])
frequency = frequency.withColumn(
"antecedent",
udf(
lambda x: "|".join(sorted(x)), StringType()
)(frequency.antecedent)
)
lift = model.associationRules
lift = lift.drop("confidence")
lift = lift.filter(col("lift") > LIFT_THRESHOLD)
lift = lift.filter(
udf(
lambda x: x == ["-999"], BooleanType()
)(lift.consequent)
)
lift = lift.drop("consequent")
lift = lift.withColumn(
"antecedent",
udf(
lambda x: "|".join(sorted(x)), StringType()
)(lift.antecedent)
)
result = lift.join(
frequency.hint("broadcast"),
["antecedent"],
"inner"
)
print("Checkpoint4 Completed!")
print("Writing Result to Data Lake...")
result.repartition(1024).write.mode("overwrite").parquet(output_path)
print("All Done!")
def main():
work(
order_path=169.1 GB of txt,
beer_path=4.9 GB of csv,
corpus_path=210 KB of csv,
output_path="final_result.parquet"
)
if __name__ == "__main__":
main()
The code ran perfectly well with the same sample data on my Mac OS and as expected took less time (34 seconds vs. 26 seconds). Then I decided to run the code to HDInsight with full datasets. In the last step, which is writing data to the data lake, the task failed and I was told Job cancelled because SparkContext was shut down. I am rather new to big data and have no idea with this means. Posts on the internet said there could be many reasons behind it. Whatever the method I should use, how to optimize my code so I can get the desired output in the data lake within bearable amount of time?
I would try several things, ordered by the amount of energy they require:
Check if the ADL storage is in the same region as your HDInsight cluster.
Add calls for df = df.cache() after heavy calculations, or even write and then read the dataframes into and from a cache storage in between these calculations.
Replace your UDFs with "native" Spark code, since UDFs are one of the performance bad practices of Spark.
I have figured it out after five days' struggle. Here are the approaches that I took to optimize the code. The time of code execution dropped from more than 24 hours to around 10 minutes. Code optimization is really really important.
As David Taub below pointed out, use df.cache() after heavy computation or before feeding the data to the model. I used df.cache().count() since calling .cache() on its own is lazily evaluated but the following .count() forces an evaluation of the entire dataset.
Use flashtext instead of regular expression to extract keywords. This greatly improves code performance.
Be careful with joins / merge. It might get extremely slow due to data skewness. Always think about ways to avoid unnecessary joins.
Set minSupport for FPGrowth. This significantly reduces the time when calling model.freqItemsets.

Spark reading from Postgres JDBC table slow

I am trying to load about 1M rows from a PostgreSQL database into Spark. When using Spark it takes about 10s. However, loading the same query using psycopg2 driver takes 2s. I am using postgresql jdbc driver version 42.0.0
def _loadFromPostGres(name):
url_connect = "jdbc:postgresql:"+dbname
properties = {"user": "postgres", "password": "postgres"}
df = SparkSession.builder.getOrCreate().read.jdbc(url=url_connect, table=name, properties=properties)
return df
df = _loadFromPostGres("""
(SELECT "seriesId", "companyId", "userId", "score"
FROM user_series_game
WHERE "companyId"=655124304077004298) as
user_series_game""")
print measure(lambda : len(df.collect()))
The output is -
--- 10.7214591503 seconds ---
1076131
Using psycopg2 -
import psycopg2
conn = psycopg2.connect(conn_string)
cur = conn.cursor()
def _exec():
cur.execute("""(SELECT "seriesId", "companyId", "userId", "score"
FROM user_series_game
WHERE "companyId"=655124304077004298)""")
return cur.fetchall()
print measure(lambda : len(_exec()))
cur.close()
conn.close()
The output is -
--- 2.27961301804 seconds ---
1076131
The measure function -
def measure(func) :
start_time = time.time()
x = func()
print("--- %s seconds ---" % (time.time() - start_time))
return x
Kindly help me find the cause of this problem.
Edit 1
I did a few more benchmarks. Using Scala and JDBC -
import java.sql._;
import scala.collection.mutable.ArrayBuffer;
def exec() {
val url = ("jdbc:postgresql://prod.caumccqvmegm.ap-southeast-1.rds.amazonaws.com/prod"+
"?tcpKeepAlive=true&prepareThreshold=-1&binaryTransfer=true&defaultRowFetchSize=10000")
val conn = DriverManager.getConnection(url,"postgres","postgres");
val sqlText = """SELECT "seriesId", "companyId", "userId", "score"
FROM user_series_game
WHERE "companyId"=655124304077004298"""
val t0 = System.nanoTime()
val stmt = conn.prepareStatement(sqlText, ResultSet.TYPE_FORWARD_ONLY, ResultSet.CONCUR_READ_ONLY)
val rs = stmt.executeQuery()
val list = new ArrayBuffer[(Long, Long, Long, Double)]()
while (rs.next()) {
val seriesId = rs.getLong("seriesId")
val companyId = rs.getLong("companyId")
val userId = rs.getLong("userId")
val score = rs.getDouble("score")
list.append((seriesId, companyId, userId, score))
}
val t1 = System.nanoTime()
println("Elapsed time: " + (t1 - t0) * 1e-9 + "s")
println(list.size)
rs.close()
stmt.close()
conn.close()
}
exec()
The output was -
Elapsed time: 1.922102285s
1143402
When I did collect() in Spark + Scala -
import org.apache.spark.sql.SparkSession
def exec2() {
val spark = SparkSession.builder().getOrCreate()
val url = ("jdbc:postgresql://prod.caumccqvmegm.ap-southeast-1.rds.amazonaws.com/prod"+
"?tcpKeepAlive=true&prepareThreshold=-1&binaryTransfer=true&defaultRowFetchSize=10000")
val sqlText = """(SELECT "seriesId", "companyId", "userId", "score"
FROM user_series_game
WHERE "companyId"=655124304077004298) as user_series_game"""
val t0 = System.nanoTime()
val df = spark.read
.format("jdbc")
.option("url", url)
.option("dbtable", sqlText)
.option("user", "postgres")
.option("password", "postgres")
.load()
val list = df.collect()
val t1 = System.nanoTime()
println("Elapsed time: " + (t1 - t0) * 1e-9 + "s")
print (list.size)
}
exec2()
The output was
Elapsed time: 1.486141076s
1143445
So 4x amount of extra time is spent within Python serialisation. I understand there will be some penalty, but this seems too much.
The reason is really simple and have two simultaneous reasons.
First I will give you a perpective of how psycopg2 works.
This lib psycopg2 works like any other lib to connect to a RDMS. This lib will send the query to the engine of your postgres and it will return the data to you. Straight foward like this.
Conn -> Query -> ReturnData -> FetchData
When you use spark is a little bit different in two ways. Spark is not like a programatic language that run in one single thread. It has a Distributed System to work. Even if you are running in a local machine. See Spark has a basic concept of Driver(Master) and Workers.
The Driver recieve the request to execute the query to the Postgres, the Driver will not request the data for each worker request the information from your Postgres.
If you see the documentation here you will se a note like this:
Don’t create too many partitions in parallel on a large cluster; otherwise Spark might crash your external database systems.
This note means that each worker will be responsible to request the data for your postgres. This is a small overhead of starting this process but nothing really big. But have a overhead here, to send the data to each worker.
Seccond point, your collect in this part of code:
print measure(lambda : len(df.collect()))
The collect function will send a command for all of your workers to send the data to your Driver. To store in the memory of your driver, it is like a Reduce, it creates Shuffle in the middle of the process. Shuffle is the step of the process that the data is send to other workers. In the case of collect each worker will send that to your Driver.
So the steps of Spark in JDBC of your code is:
(Workers)Conn -> (Workers)Query -> (Workers)FetchData -> (Driver)
Request the Data -> (Workers) Shuffle -> (Driver) Collect
Well there in a bunch of other stuffs that happens with the Spark, like the QueryPlan, build the DataFrame and other stuffs.
That is the reason that you have faster response in your simple code of Python than Spark.

combining slick queries into single query

Using Slick 3.1, how do I combine multiple Queries into a single query for the same type? This is not a join or a union, but combining query "segments" to create a single query request. These "segments" can be any individually valid query.
val query = TableQuery[SomeThingValid]
// build up pieces of the query in various parts of the application logic
val q1 = query.filter(_.value > 10)
val q2 = query.filter(_.value < 40)
val q3 = query.sortBy(_.date.desc)
val q4 = query.take(5)
// how to combine these into a single query ?
val finalQ = ??? q1 q2 q3 q4 ???
// in order to run in a single request
val result = DB.connection.run(finalQ.result)
EDIT:
the expected sql should be something like:
SELECT * FROM "SomeThingValid" WHERE "SomeThingValid"."value" > 10 AND "SomeThingValid"."valid" < 40 ORDER BY "MemberFeedItem"."date" DESC LIMIT 5
val q1 = query.filter(_.value > 10)
val q2 = q1.filter(_.value < 40)
val q3 = q2.sortBy(_.date.desc)
val q4 = q3.take(5)
I think you should do something like the above (and pass around Querys) but if you insist on passing around query "segments", something like this could work:
type QuerySegment = Query[SomeThingValid, SomeThingValid, Seq] => Query[SomeThingValid, SomeThingValid, Seq]
val q1: QuerySegment = _.filter(_.value > 10)
val q2: QuerySegment = _.filter(_.value < 40)
val q3: QuerySegment = _.sortBy(_.date.desc)
val q4: QuerySegment = _.take(5)
val finalQ = Function.chain(Seq(q1, q2, q3, q4))(query)
I've used this "pattern" with slick2.0
val query = TableQuery[SomeThingValid]
val flag1, flag3 = false
val flag2, flag4 = true
val bottomFilteredQuery = if(flag1) query.filter(_.value > 10) else query
val topFilteredQuery = if(flag2) bottomFilteredQuery.filter(_.value < 40) else bottomFilteredQuery
val sortedQuery = if(flag3) topFilteredQuery.soryBy(_.date.desc) else topFilteredQuery
val finalQ = if(flag4) sortedQuery.take(5) else sortedQuery
It's Just a worth remark to mention here from the slick essential book, it seems that you might need to avoid combining multiple queries in one single query.
Combining actions to sequence queries is a powerful feature of Slick.
However, you may be able to reduce multiple queries into a single
database query. If you can do that, you’re probably better off doing
it.
I think, it should work. But I didn't test it yet.
val users = TableQuery[Users]
val filter1: Query[Users, User, Seq] = users.filter(condition1)
val filter2: Query[Users, User, Seq] = users.filter(condition2)
(filter1 ++ filter2).result.headOption