mysql - How to use a subquery for dbtable option in jdbc data source?

Question

I want to use Spark to process some data from a JDBC source. But to begin with, instead of reading original tables from JDBC, I want to run some queries on the JDBC side to filter columns and join tables, and load the query result as a table in Spark SQL.

The following syntax to load raw JDBC table works for me:

df_table1 = sqlContext.read.format('jdbc').options(
    url="jdbc:mysql://foo.com:3306",
    dbtable="mydb.table1",
    user="me",
    password="******",
    driver="com.mysql.jdbc.Driver" # mysql JDBC driver 5.1.41
).load() 
df_table1.show() # succeeded

According to Spark documentation (I'm using PySpark 1.6.3):

dbtable: The JDBC table that should be read. Note that anything that is valid in a FROM clause of a SQL query can be used. For example, instead of a full table you could also use a subquery in parentheses.

So just for experiment, I tried something simple like this:

df_table1 = sqlContext.read.format('jdbc').options(
    url="jdbc:mysql://foo.com:3306",
    dbtable="(SELECT * FROM mydb.table1) AS table1",
    user="me",
    password="******",
    driver="com.mysql.jdbc.Driver"
).load() # failed

It threw the following exception:

com.mysql.jdbc.exceptions.jdbc4.MySQLSyntaxErrorException: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'table1 WHERE 1=0' at line 1

I also tried a few other variations of the syntax (add / remove parentheses, remove 'as' clause, switch case, etc) without any luck. So what would be the correct syntax? Where can I find more detailed documentation for the syntax? Besides, where does this weird "WHERE 1=0" in error message come from? Thanks!

Answer 1

For reading data from JDBC source using sql query in Spark SQL, you can try something like this:

val df_table1 = sqlContext.read.format("jdbc").options(Map(
    ("url" -> "jdbc:postgresql://localhost:5432/mydb"),
    ("dbtable" -> "(select * from table1) as table1"),
    ("user" -> "me"),
    ("password" -> "******"),
    ("driver" -> "org.postgresql.Driver"))
).load()

I tried it using PostgreSQL. You can modify it according to MySQL.