Skip to content

Latest commit

 

History

History
1111 lines (976 loc) · 57 KB

README.md

File metadata and controls

1111 lines (976 loc) · 57 KB
Raw

About Oracle R2DBC

The Oracle R2DBC Driver is a Java library that supports reactive programming with Oracle Database.

Oracle R2DBC implements the R2DBC Service Provider Interface (SPI) as specified by the Reactive Relational Database Connectivity (R2DBC) project. The R2DBC SPI exposes Reactive Streams as an abstraction for remote database operations. Reactive Streams is a well defined standard for asynchronous, non-blocking, and back-pressured communication. This standard allows an R2DBC driver to interoperate with other reactive libraries and frameworks, such as Spring, Project Reactor, RxJava, and Akka Streams.

Learn More About R2DBC:

R2DBC Project Home Page

R2DBC Javadocs v1.0.0.RELEASE

R2DBC Specification v1.0.0.RELEASE

Learn More About Reactive Streams:

Reactive Streams Project Home Page

Reactive Streams Javadocs v1.0.3

Reactive Streams Specification v1.0.3

About This Version

The 1.3.0 release Oracle R2DBC implements version 1.0.0.RELEASE of the R2DBC SPI.

New features in this release:

Updated dependencies:

  • Updated Oracle JDBC from 21.11.0.0 to 23.6.0.24.10
  • Updated Project Reactor from 3.5.11 to 3.6.11

Installation

Oracle R2DBC can be obtained from Maven Central.

<dependency>
  <groupId>com.oracle.database.r2dbc</groupId>
  <artifactId>oracle-r2dbc</artifactId>
  <version>1.3.0</version>
</dependency>

Oracle R2DBC can also be built from source using Maven: mvn clean install -DskipTests=true

If -DskipTests=true is omitted from the command above, then it will execute end-to-end tests which connect to an Oracle Database. Tests read the connection configuration from src/test/resources/config.properties.

Oracle R2DBC is compatible with JDK 11 (or newer), and has the following runtime dependencies:

  • R2DBC SPI 1.0.0.RELEASE
  • Reactive Streams 1.0.3
  • Project Reactor 3.6.11
  • Oracle JDBC 23.6.0.24.10 for JDK 11 (ojdbc11.jar)

The Oracle R2DBC Driver has been verified with Oracle Database versions 18, 19, 21, and 23.

Integration with Spring and Other Libraries

Oracle R2DBC can only interoperate with libraries that support the 1.0.0.RELEASE version of the R2DBC SPI. When using libraries like Spring and r2dbc-pool, be sure to use a version which supports the 1.0.0.RELEASE of the SPI.

Oracle R2DBC depends on the JDK 11 build of Oracle JDBC 23.6.0.24.10. Other libraries may depend on a different version of Oracle JDBC, and this version may be incompatible. To resolve incompatibilities, it may be necessary to explicitly declare the dependency in your project, ie:

<dependency>
    <groupId>com.oracle.database.jdbc</groupId>
    <artifactId>ojdbc11</artifactId>
    <version>23.6.0.24.10</version>
</dependency>

Basic Code Examples

The following method returns an Oracle R2DBC ConnectionFactory

  static ConnectionFactory getConnectionFactory() {
    String user = getUser();
    char[] password = getPassword();
    try {
      return ConnectionFactories.get(
        ConnectionFactoryOptions.builder()
          .option(ConnectionFactoryOptions.DRIVER, "oracle")
          .option(ConnectionFactoryOptions.HOST, "db.host.example.com")
          .option(ConnectionFactoryOptions.PORT, 1521)
          .option(ConnectionFactoryOptions.DATABASE, "db.service.name")
          .option(ConnectionFactoryOptions.USER, user)
          .option(ConnectionFactoryOptions.PASSWORD, CharBuffer.wrap(password))
          .build());
    }
    finally {
      Arrays.fill(password, (char)0);
    }
  }

The following method uses Project Reactor's Flux to open a connection, execute a SQL query, and then close the connection:

Flux.usingWhen(
  getConnectionFactory().create(),
  connection ->
    Flux.from(connection.createStatement(
      "SELECT 'Hello, Oracle' FROM sys.dual")
      .execute())
      .flatMap(result ->
        result.map(row -> row.get(0, String.class))),
  Connection::close)
  .doOnNext(System.out::println)
  .doOnError(Throwable::printStackTrace)
  .subscribe();

When executed, the code above will asynchronously print the result of the SQL query.

The next example uses a named parameter marker, :locale_name, in the SQL command:

Flux.usingWhen(
  getConnectionFactory().create(),
  connection ->
    Flux.from(connection.createStatement(
      "SELECT greeting FROM locale WHERE locale_name = :locale_name")
      .bind("locale_name", "France")
      .execute())
      .flatMap(result ->
        result.map(row ->
          String.format("%s, Oracle", row.get("greeting", String.class)))),
  Connection::close)
  .doOnNext(System.out::println)
  .doOnError(Throwable::printStackTrace)
  .subscribe();

Like the previous example, executing the code above will asynchronously print a greeting message. "France" is set as the bind value for locale_name, so the query should return a greeting like "Bonjour" when row.get("greeting") is called.

Additional code examples appear throughout this document, and can also be found here.

Help

For help programming with Oracle R2DBC, ask questions on Stack Overflow tagged with [oracle] and [r2dbc]. The development team monitors Stack Overflow regularly.

Issues may be opened as described in our contribution guide.

Contributing

This project welcomes contributions from the community. Before submitting a pull request, please review our contribution guide.

Security

Please consult the security guide for our responsible security vulnerability disclosure process.

License

Copyright (c) 2021, 2024 Oracle and/or its affiliates.

This software is dual-licensed to you under the Universal Permissive License (UPL) 1.0 as shown at https://oss.oracle.com/licenses/upl or Apache License 2.0 as shown at http://www.apache.org/licenses/LICENSE-2.0. You may choose either license.

Documentation

This document specifies the behavior of the R2DBC SPI as implemented for the Oracle Database. This SPI implementation is referred to as the "Oracle R2DBC Driver" or "Oracle R2DBC" throughout the remainder of this document.

The Oracle R2DBC Driver implements behavior specified by the R2DBC 1.0.0.RELEASE Specification and Javadoc

Publisher objects created by Oracle R2DBC implement behavior specified by the Reactive Streams 1.0.3 Specification and Javadoc

The R2DBC and Reactive Streams specifications include requirements that are optional for a compliant implementation. Oracle R2DBC's implementation of these optional are specified in this document. This document also specifies additional functionality that is supported by Oracle R2DBC, but is not part of the R2DBC 1.0.0 Specification.

Connection Creation

The Oracle R2DBC Driver is identified by the name "oracle". The driver implements a ConnectionFactoryProvider located by an R2DBC URL identifing "oracle" as a driver, or by a DRIVER ConnectionFactoryOption with the value of "oracle".

Support for Standard R2DBC Options

The following standard ConnectionFactoryOptions are supported by Oracle R2DBC:

  • DRIVER
  • HOST
  • PORT
  • DATABASE
    • The database option is interpreted as the service name of an Oracle Database instance. System Identifiers (SID) are not recognized.
  • USER
  • PASSWORD
  • SSL
  • CONNECT_TIMEOUT
  • STATEMENT_TIMEOUT.
  • PROTOCOL
    • Accepted protocol values are "tcps", "ldap", and "ldaps"

Support for Extended R2DBC Options

Oracle R2DBC extends the standard set of R2DBC options to offer functionality that is specific to Oracle Database and the Oracle JDBC Driver. Extended options are declared in the OracleR2dbcOptions class.

Support for Supplier and Publisher as Option Values

Most options can have a value provided by a Supplier or Publisher.

Oracle R2DBC requests the value of an Option from a Supplier or Publisher each time the Publisher returned by ConnectionFactory.create() creates a new Connection. Each Connection can then be configured with values that change over time, such as a password which is periodically rotated.

If a Supplier provides the value of an Option, then Oracle R2DBC requests the value by invoking Supplier.get(). If get() returns null, then no value is configured for the Option. If get() throws a RuntimeException, then it is set as the initial cause of an R2dbcException emitted by the Publisher returned by ConnectionFactory.create(). The Supplier must have a thread safe get() method, as multiple subscribers may request connections concurrently.

If a Publisher provides the value of an Option, then Oracle R2DBC requests the value by subscribing to the Publisher and signalling demand. The first value emitted to onNext will be used as the value of the Option. If the Publisher emits onComplete before onNext, then no value is configured for the Option. If the Publisher emits onError before onNext, then the Throwable is set as the initial cause of an R2dbcException emitted by the Publisher returned by ConnectionFactory.create().

The following example configures the PASSWORD option with a Supplier:

  void configurePassword(ConnectionFactoryOptions.Builder optionsBuilder) {
  
    // Cast the PASSWORD option
    Option<Supplier<CharSequence>> suppliedOption = OracleR2dbcOptions.supplied(PASSWORD);
    
    // Supply a password
    Supplier<CharSequence> supplier = () -> getPassword();
    
    // Configure the builder
    optionsBuilder.option(suppliedOption, supplier); 
  }

A more concise example configures TLS_WALLET_PASSWORD as a Publisher

  void configurePassword(ConnectionFactoryOptions.Builder optionsBuilder) {
    optionsBuilder.option(
      OracleR2dbcOptions.published(TLS_WALLET_PASSWORD),
      Mono.fromSupplier(() -> getWalletPassword()));
  }

These examples use the supplied(Option) and published(Option) methods declared by oracle.r2dbc.OracleR2dbcOptions. These methods cast an Option<T> to Option<Supplier<T>> and Option<Publisher<T>>, respectively. It is necessary to cast the generic type of the Option when calling ConnectionFactoryOptions.Builder.option(Option<T>, T) in order for the call to compile and not throw a ClassCastException at runtime. It is not strictly required that supplied(Option) or published(Option) be used to cast the Option. These methods are only meant to offer code readability and convenience.

Note that the following code would compile, but fails at runtime with a ClassCastException:

  void configurePassword(ConnectionFactoryOptions.Builder optionsBuilder) {
    Publisher<CharSequence> publisher = Mono.fromSupplier(() -> getPassword());
    // Doesn't work. Throws ClassCastException at runtime:
    optionsBuilder.option(PASSWORD, PASSWORD.cast(publisher));
  }

To avoid a ClassCastException, the generic type of an Option must match the actual type of the value passed to ConnectionFactoryOptions.Builder.option(Option<T>, T).

For a small set of options, providing values with a Supplier or Publisher is not supported:

  • DRIVER
  • PROTOCOL

Providing values for these options would not be interoperable with io.r2dbc.spi.ConnectionFactories and r2dbc-pool.

Normally, Oracle R2DBC will not retain references to Option values after ConnectionFactories.create(ConnectionFactoryOptions) returns. However, if the value of at least one Option is provided by a Supplier or Publisher, then Oracle R2DBC will retain a reference to all Option values until the ConnectionFactory.create() Publisher emits a Connection or error. This is important to keep in mind when Option values may be mutated. In particular, a password may only be cleared from memory after the create() Publisher emits a Connection or error.

Configuring an Oracle Net Descriptor

The oracle.r2dbc.OracleR2dbcOptions.DESCRIPTOR option may be used to configure an Oracle Net Descriptor of the form (DESCRIPTION=...). If this option is used to configure a descriptor, then it is invalid to specify any other option that conflicts with information in the descriptor. Conflicting options include HOST, PORT, DATABASE, and SSL. These options all conflict with information that appears in a descriptor.

The DESCRIPTOR option has the name oracle.r2dbc.descriptor. This name can be used to configure a descriptor in the query section of an R2DBC URL:

r2dbc:oracle://?oracle.r2dbc.descriptor=(DESCRIPTION=...)

The DESCRIPTOR constant may also be used to configure a descriptor programmatically:

ConnectionFactoryOptions.builder()
  .option(OracleR2dbcOptions.DESCRIPTOR, "(DESCRIPTION=...)")

The DESCRIPTOR option may be set to an aliased entry of a tnsnames.ora file. Use the TNS_ADMIN option to specify the directory where tnsnames.ora is located:

r2dbc:oracle://?oracle.r2dbc.descriptor=myAlias&TNS_ADMIN=/path/to/tnsnames/

Configuring an LDAP URL

Use ldap or ldaps as the URL protocol to have an Oracle Net Descriptor retrieved from an LDAP server:

r2dbc:oracle:ldap://ldap.example.com:7777/sales,cn=OracleContext,dc=com
r2dbc:oracle:ldaps://ldap.example.com:7778/sales,cn=OracleContext,dc=com

Use a space separated list of LDAP URIs for fail over and load balancing:

r2dbc:oracle:ldap://ldap1.example.com:7777/sales,cn=OracleContext,dc=com%20ldap://ldap2.example.com:7777/sales,cn=OracleContext,dc=com%20ldap://ldap3.example.com:7777/sales,cn=OracleContext,dc=com

Space characters in a URL must be percent encoded as %20

An LDAP server request will block a thread for network I/O when Oracle R2DBC creates a new connection.

Configuring a java.util.concurrent.Executor

The oracle.r2dbc.OracleR2dbcOptions.EXECUTOR option configures a java.util.concurrent.Executor for executing asynchronous callbacks. The EXECUTOR option may be used to configure an Executor programmatically:

ConnectionFactoryOptions.builder()
  .option(OracleR2dbcOptions.EXECUTOR, getExecutor())

There is no way to configure an executor with a URL query parameter

If this option is not configured, then the common java.util.concurrent.ForkJoinPool is used as a default.

Configuring Oracle JDBC Connection Properties

A subset of Oracle JDBC's connection properties are defined as Option constants in the OracleR2dbcOptions class. If an Oracle JDBC property is not defined as an Option, in most cases it can instead be configured by a connection properties file or a JVM system property instead. Pull requests to add missing options are also a welcome addition.

When a connection property is defined in OracleR2dbcOptions, it may be configured as an R2DBC URL parameter. For example, the following URL configures the oracle.net.wallet_location connection property:

r2dbcs:oracle://db.host.example.com:1522/db.service.name?oracle.net.wallet_location=/path/to/wallet/

And, the OracleR2dbcOptions constants can be used in programmatic configuration:

 ConnectionFactoryOptions.builder()
  .option(OracleR2dbcOptions.TLS_WALLET_LOCATION, "/path/to/wallet")

All Oracle JDBC connection properties defined in OracleR2dbcOptions are listed in the next sections.

TLS/SSL Connection Properties
Database Tracing Connection Properties
Oracle Net Encryption Connection Properties
Kerberos Connection Properties
LDAP Connection Properties
Miscellaneous Connection Properties

Thread Safety

Oracle R2DBC's ConnectionFactory and ConnectionFactoryProvider are the only classes that have a thread safe implementation. All other classes implemented by Oracle R2DBC are not thread safe. For instance, it is not safe for multiple threads to concurrently access a single instance of Result.

It is recommended to use a Reactive Streams library such as Project Reactor or RxJava to manage the consumption of non-thread safe objects

While it is not safe for multiple threads to concurrently access the same object, it is safe from them to do so with different objects from the same Connection. For example, two threads can concurrently subscribe to two Statement objects from the same Connection. When this happens, the two statements are executed in a "pipeline". Pipelining will be covered in the next section.

Pipelining

Pipelining allows Oracle R2DBC to send a call without having to wait for a previous call to complete. If all requirements are met, then pipelining will be activated by concurrently subscribing to publishers from the same connection. For example, the following code concurrently subscribes to two statements:

Flux.merge(
  connection.createStatement(
    "INSERT INTO example (id, value) VALUES (0, 'X')")
    .execute(),
  connection.createStatement(
    "INSERT INTO example (id, value) VALUES (1, 'Y')")
    .execute())

When the Publisher returned by merge is subscribed to, both INSERTs are immediately sent to the database. The network traffic can be visualized as:

TIME | ORACLE R2DBC     | NETWORK | ORACLE DATABASE
-----+------------------+---------+-----------------
   0 | Send INSERT-X    | ------> | WAITING
   0 | Send INSERT-Y    | ------> | WAITING
   1 | WAITING          | <------ | Send Result-X
   1 | WAITING          | <------ | Send Result-Y
   2 | Receive Result-X |         | WAITING
   2 | Receive Result-Y |         | WAITING

In this visual, 1 unit of TIME is required to transfer data over the network. The TIME column is only measuring network latency. It does not include computational time spent on executing the INSERTs.

The key takeaway from this visual is that the INSERTs are sent and received concurrently, rather than sequentially. Both INSERTs are sent at TIME=0, and both are received at TIME=1. And, the results are both sent at TIME=1, and are received at TIME=2.

Recall that TIME is not measuring computational time. If each action by Oracle R2DBC and Oracle Database requires 0.1 units of computational TIME, then we can say:

INSERTs are sent at TIME=0.1 and TIME=0.2, and are received at TIME=1.1 and TIME=1.2. And, the results are sent at TIME=1.2 and TIME=1.3, and are received at TIME=2.2 and TIME=2.3.

This is a bit more complicated to think about, but it is important to keep in mind. All database calls will require at least some computational time.

Below is another visual of the network traffic, but in this case the INSERTs are sent and received without pipelining:

TIME | ORACLE R2DBC     | NETWORK | ORACLE DATABASE
-----+------------------+---------+-----------------
   0 | Send INSERT-X    | ------> | WAITING
   1 | WAITING          | <------ | Send Result-X
   2 | Receive Result-X |         | WAITING
   2 | Send INSERT-Y    | ------> | WAITING
   3 | WAITING          | <------ | Send Result-Y
   4 | Receive Result-Y |         | WAITING

This visual shows a sequential process of sending and receiving. It can be compared to the concurrent process seen in the previous visual. In both cases, Oracle R2DBC and Oracle Database have the same number of WAITING actions. These actions are waiting for network transfers. And in both cases, each network transfer requires 1 unit of TIME.

So if network latency is the same, and the number of WAITING actions are the same (,and the computational times are the same), then how are these INSERTs completing in less TIME with pipelining? The answer is that pipelining allowed the network transfer times to be waited for concurrently.

In the first visual, with pipelining, the database waits for both INSERT-X and INSERT-Y at TIME=0. Compare that to the second visual, without pipelining, where the database waits for INSERT-X at TIME=0, and then waits again for INSERT-Y at TIME=2. That's 1 additional unit of TIME when compared to pipelining. The other additional unit of TIME happens on the Oracle R2DBC side. Without pipelining, it waits for Result-X at TIME=1, and then waits again for Result-Y at TIME=3. With pipelining, it waits for both results concurrently at TIME=1.

Requirements for Pipelining

There are some requirements which must be met in order to use pipelining. As explained in the previous section, the availability of pipelining can have a significant impact on performance. Users should review the requirements listed in this section when developing applications that rely on this performance gain.

In terms of functional behavior, the availability of pipelining will have no impact: With or without it, the same database calls are going be executed. Users who are not relying on pipelining performance do not necessarily need to review the requirements listed in this section. Oracle JDBC is designed to automatically check for these requirements, and it will fallback to using sequential network transfers if any requirement is not met.

Product Versions

Pipelining is only available with Oracle Database version 23.4 or newer. It also requires an Oracle JDBC version of 23.4 or newer, but this is already a transitive dependency of Oracle R2DBC.

Out Of Band Breaks

Pipelining requires out-of-band (OOB) breaks (ie: TCP urgent data) for cancelling statement execution. The Oracle JDBC Driver automatically checks if OOB is available, and will disable pipelining if it is not. The availability of OOB may depend on the operating system where Oracle R2DBC is running. Notably, OOB is not available on Mac OS (or at least not available in the way which Oracle JDBC needs it to be for sending TCP urgent data to Oracle Database).

For experimentation only, Mac OS users can choose to by-pass the OOB requirement by setting a JVM system property:

-Doracle.jdbc.disablePipeline=false

Bypassing the OOB requirement on Mac OS will result in non-functional implementations of Connection.setStatementTimeout(Duration), and Subscription.cancel() for a Subscription from Statement.execute().

Reactive Streams

Every method implemented by Oracle R2DBC that returns a Publisher has a JavaDoc which specifies the Publisher's behavior with regard to deferred execution and support for multiple Subscribers.

Oracle R2DBC's implementation of Publishers that emit one or zero items will typically defer execution until a Subscriber subscribes, support multiple Subscribers, and cache the result of a database call (the same result of the same call is emitted to each Subscriber).

Oracle R2DBC's implementation of Publishers that emit multiple items will typically defer execution until a Subscriber signals demand, and not support multiple subscribers.

Errors and Warnings

Oracle R2DBC creates R2dbcExceptions having the same ORA-XXXXX error codes used by Oracle Database and Oracle JDBC. The Database Error Messages document provides a reference for all ORA-XXXXX error codes.

Warning messages from Oracle Database are emitted as oracle.r2dbc.OracleR2dbcWarning segments. These segments may be consumed using Result.flatMap(Function):

result.flatMap(segment -> {
  if (segment instanceof OracleR2dbcWarning) {
    logWarning(((OracleR2dbcWarning)segment).getMessage());
    return emptyPublisher();
  }
  else if (segment instanceof Result.Message){
    ... handle an error ...
  }
  else {
    ... handle other segment types ...
  }
})

Unlike the errors of standard Result.Message segments, if a warning is not consumed by flatMap, then it will be silently discarded when a Result is consumed using the map or getRowsUpdated methods.

Transactions

Oracle R2DBC uses READ COMMITTED as the default transaction isolation level.

Oracle R2DBC also supports the SERIALIZABLE isolation level. If SERIALIZABLE isolation is configured, then the oracle.r2dbc.OracleR2dbcOptions.ENABLE_QUERY_RESULT_CACHE option must also be configured as false to avoid phantom reads.

READ COMMITTED and SERIALIZABLE are the only isolation levels supported by Oracle Database

Oracle Database does not support a lock wait timeout that is configurable within the scope of a transaction or session. Oracle R2DBC implements SPI methods that configure a lock wait timeout to throw UnsupportedOperationException.

Statements

Oracle R2DBC supports SQL execution with the Statement SPI.

Parameter Markers

A SQL command passed to Connection.createStatement(String) may include named parameter markers, unnamed parameter markers, or both.

Unnamed parameter markers may appear in SQL as a question mark (?):

connection.createStatement(
  "SELECT value FROM example WHERE id=?")
  .bind(0, 99)

The bind method must be called with a zero-based index to set the value of an unnamed parameter.

Named parameter markers may appear in SQL as a colon character (:) followed by an alpha-numeric name:

connection.createStatement(
  "SELECT value FROM example WHERE id=:id")
  .bind("id", 99)

The bind method may be called with a String valued name, or with zero-based index, to set the value of a named parameter. Parameter names are case-sensitive.

Batch Execution

The Statement.add() method may be used execute a DML command multiple times with a batch of different bind values. Oracle Database only supports batch execution for DML type SQL commands (INSERT/UPDATE/DELETE). Attempting to execute a SELECT query with a batch of bind values will result in an error.

Returning Generated Values

The Statement.returnGeneratedValues(String...) method may be called to return generated values from basic forms of INSERT and UPDATE statements.

If an empty set of column names is passed to returnGeneratedValues, the Statement will return the ROWID of each row affected by an INSERT or UPDATE.

Programmers are advised not to use the ROWID as if it were a primary key. The ROWID of a row change, or be reassigned to a different row. See https://asktom.oracle.com/pls/apex/asktom.search?tag=is-it-safe-to-use-rowid-to-locate-a-row for more information.

Returning generated values is only supported for INSERT and UPDATE commands in which a RETURNING INTO clause would be valid. For example, if a table is declared as:

CREATE TABLE example (
  id NUMBER PRIMARY KEY GENERATED ALWAYS AS IDENTITY,
  value VARCAHR(100))

Returning generated values is supported for the following statement:

connection.createStatement(
  "INSERT INTO example(value) VALUES (:value)")
  .bind("value", "x")
  .returningGeneratedValues("id")

This statement is supported because the INSERT could be written to include a RETURNING INTO clause:

INSERT INTO example(value) VALUES (:value) RETURING id INTO :id

As a counter example, returning generated values is not supported for the following statement:

connection.createStatement(
  "INSERT INTO example (value) SELECT 'y' FROM sys.dual")
  .returningGeneratedValues("id")

This statement is not supported because it can not be written to include a RETURNING INTO clause.

The Oracle Database SQL Language Reference specifies the INSERT and UPDATE commands for which a RETURNING INTO clause is supported.

For the INSERT syntax, see: https://docs.oracle.com/en/database/oracle/oracle-database/23/sqlrf/INSERT.html

For the UPDATE syntax, see: https://docs.oracle.com/en/database/oracle/oracle-database/23/sqlrf/UPDATE.html

Procedural Calls

The SQL string passed to Connection.createStatement(String) may execute a PL/SQL call:

connection.createStatement("BEGIN sayHello(:name_in, :greeting_out); END;")

OUT parameters are registered by invoking Statement.bind(int, Object) or Statement.bind(String, Object) with an instance of io.r2dbc.spi.Parameter implementing the io.r2dbc.spi.Parameter.Out marker interface:

statement.bind("greeting_out", Parameters.out(R2dbcType.VARCHAR))

Likewise, an IN OUT parameter would be registered by invoking Statement.bind(int, Object) or Statement.bind(String, Object) with an instance of io.r2dbc.spi.Parameter implementing both the io.r2dbc.spi.Parameter.Out and io.r2dbc.spi.Parameter.In marker interfaces.

OUT parameters are consumed by invoking Result.map(Function):

result.map(outParameters -> outParameters.get("greeting_out", String.class))

If a procedural call returns multiple results, the publisher returned by Statement.execute() emits one Result for each cursor returned by DBMS_SQL.RETURN_RESULT in the procedure. The order in which each Result is emitted corresponds to the order in which the procedure returns each cursor.

If a procedure returns cursors, and also has out parameters, then the Result for the out parameters is emitted last, after the Result for each cursor.

Type Mappings

Oracle R2DBC supports type mappings between Java and SQL for non-standard data types of Oracle Database.

Oracle SQL Type Java Type
JSON javax.json.JsonObject or oracle.sql.json.OracleJsonObject
DATE java.time.LocalDateTime
INTERVAL DAY TO SECOND java.time.Duration
INTERVAL YEAR TO MONTH java.time.Period
SYS_REFCURSOR io.r2dbc.spi.Result
VECTOR double[], float[], byte[], boolean[] or oracle.sql.VECTOR

Unlike the standard SQL type named "DATE", the Oracle Database type named "DATE" stores values for year, month, day, hour, minute, and second. The standard SQL type only stores year, month, and day. LocalDateTime objects are able to store the same values as a DATE in Oracle Database.

BLOB, CLOB, and NCLOB

Oracle R2DBC allows large objects (LOBs) to be read and written as a reactive stream, or as a fully materialized value.

Prefetched LOB Data

When a SQL query returns a LOB column, only a portion of the LOB's content is received in the response from Oracle Database. The portion received in the SQL query response is referred to as "prefetched data". Any content remaining after the prefetched portion must be fetched with additional database calls.

For example, if a SQL query returns a LOB that is 100MB in size, then the response might prefetch only the first 1MB of the LOB's content. Additional database calls would be required to fetch the remaining 99MB of content.

By default, Oracle R2DBC attempts to prefetch the entire content of a LOB. Oracle R2DBC will request up to 1GB of prefetched data from Oracle Database when executing a SQL query.

Materialzed Type Mapping

The Row.get(...) method allows LOB values to be mapped into materialized types like ByteBuffer and String. If the entire LOB has been prefetched, then Row.get(...) can return a ByteBuffer/String without any additional database calls. However, if the LOB value is larger than the prefetch size, then Row.get(...) must execute a blocking database call to fetch the remainder of that value.

Streamed Type Mapping

In a system that consumes very large LOBs, a very large amount of memory will be consumed if the entire LOB is prefetched. When a LOB is too large to be prefetched entirely, a smaller prefetch size can be configured using the oracle.jdbc.defaultLobPrefetchSize option, and the LOB can be consumed as a stream. By mapping LOB columns to Blob or Clob objects, the content can be consumed as a reactive stream.

ARRAY

Oracle Database supports ARRAY as a user defined type only. A CREATE TYPE command is used to define an ARRAY type:

CREATE TYPE MY_ARRAY AS ARRAY(8) OF NUMBER

Oracle R2DBC defines oracle.r2dbc.OracleR2dbcType.ArrayType as a Type for representing user defined ARRAY types. A Parameter with a type of ArrayType must be used when binding array values to a Statement.

Publisher<Result> arrayBindExample(Connection connection) {
  Statement statement =
    connection.createStatement("INSERT INTO example VALUES (:array_bind)");

  // Use the name defined for an ARRAY type:
  // CREATE TYPE MY_ARRAY AS ARRAY(8) OF NUMBER
  ArrayType arrayType = OracleR2dbcTypes.arrayType("MY_ARRAY");
  Integer[] arrayValues = {1, 2, 3};
  statement.bind("arrayBind", Parameters.in(arrayType, arrayValues));

  return statement.execute();
}

A Parameter with a type of ArrayType must also be used when binding OUT parameters of a PL/SQL call.

Publisher<Result> arrayOutBindExample(Connection connection) {
  Statement statement =
    connection.createStatement("BEGIN; exampleCall(:array_bind); END;");

  // Use the name defined for an ARRAY type:
  // CREATE TYPE MY_ARRAY AS ARRAY(8) OF NUMBER
  ArrayType arrayType = OracleR2dbcTypes.arrayType("MY_ARRAY");
  statement.bind("arrayBind", Parameters.out(arrayType));

  return statement.execute();
}

ARRAY values may be consumed from a Row or OutParameter as a Java array. The element type of the Java array may be any Java type that is supported as a mapping for the SQL type of the ARRAY. For instance, if the ARRAY type is NUMBER, then a Integer[] mapping is supported:

Publisher<Integer[]> arrayMapExample(Result result) {
  return result.map(readable -> readable.get("arrayValue", Integer[].class));
}

OBJECT

Oracle Database supports OBJECT as a user defined type. A CREATE TYPE command is used to define an OBJECT type:

CREATE TYPE PET AS OBJECT(
  name VARCHAR(128),
  species VARCHAR(128),
  weight NUMBER,
  birthday DATE)

Oracle R2DBC defines oracle.r2dbc.OracleR2dbcType.ObjectType as a Type for representing user defined OBJECT types. A Parameter with a type of ObjectType may be used to bind OBJECT values to a Statement.

Use an Object[] to bind the attribute values of an OBJECT by index:

Publisher<Result> objectArrayBindExample(Connection connection) {
  Statement statement =
    connection.createStatement("INSERT INTO petTable VALUES (:petObject)");

  // Bind the attributes of the PET OBJECT defined above
  ObjectType objectType = OracleR2dbcTypes.objectType("PET");
  Object[] attributeValues = {
    "Derby",
    "Dog",
    22.8,
    LocalDate.of(2015, 11, 07)
  };
  statement.bind("petObject", Parameters.in(objectType, attributeValues));

  return statement.execute();
}

Use a Map<String,Object> to bind the attribute values of an OBJECT by name:

Publisher<Result> objectMapBindExample(Connection connection) {
  Statement statement =
    connection.createStatement("INSERT INTO petTable VALUES (:petObject)");

  // Bind the attributes of the PET OBJECT defined above
  ObjectType objectType = OracleR2dbcTypes.objectType("PET");
  Map<String,Object> attributeValues = Map.of(
    "name", "Derby",
    "species", "Dog",
    "weight", 22.8,
    "birthday", LocalDate.of(2015, 11, 07));
  statement.bind("petObject", Parameters.in(objectType, attributeValues));

  return statement.execute();
}

A Parameter with a type of ObjectType must be used when binding OUT parameters of OBJECT types for a PL/SQL call:

Publisher<Result> objectOutBindExample(Connection connection) {
  Statement statement =
    connection.createStatement("BEGIN; getPet(:petObject); END;");

  ObjectType objectType = OracleR2dbcTypes.objectType("PET");
  statement.bind("petObject", Parameters.out(objectType));

  return statement.execute();
}

OBJECT values may be consumed from a Row or OutParameter as an oracle.r2dbc.OracleR2dbcObject. The OracleR2dbcObject interface is a subtype of io.r2dbc.spi.Readable. Attribute values may be accessed using the standard get methods of Readable. The get methods of OracleR2dbcObject support all SQL to Java type mappings defined by the R2DBC Specification:

Publisher<Pet> objectMapExample(Result result) {
  return result.map(row -> {
    OracleR2dbcObject oracleObject = row.get(0, OracleR2dbcObject.class); 
    return new Pet(
      oracleObject.get("name", String.class),
      oracleObject.get("species", String.class),
      oracleObject.get("weight", Float.class),
      oracleObject.get("birthday", LocalDate.class));
  });
}

Instances of OracleR2dbcObject may be passed directly to Statement bind methods:

Publisher<Result> objectBindExample(
  OracleR2dbcObject oracleObject, Connection connection) {
  Statement statement =
    connection.createStatement("INSERT INTO petTable VALUES (:petObject)");
  
  statement.bind("petObject", oracleObject);

  return statement.execute();
}

Attribute metadata is exposed by the getMetadata method of OracleR2dbcObject:

void printObjectMetadata(OracleR2dbcObject oracleObject) {
  OracleR2dbcObjectMetadata metadata = oracleObject.getMetadata();
  OracleR2dbcTypes.ObjectType objectType = metadata.getObjectType();
  
  System.out.println("Object Type: " + objectType);
  metadata.getAttributeMetadatas()
    .stream()
    .forEach(attributeMetadata -> {
      System.out.println("\tAttribute Name: " + attributeMetadata.getName()));
      System.out.println("\tAttribute Type: " + attributeMetadata.getType()));
    });
}

REF Cursor

Use the oracle.r2dbc.OracleR2dbcTypes.REF_CURSOR type to bind SYS_REFCURSOR out parameters:

Publisher<Result> executeProcedure(Connection connection) {
  return connection.createStatement(
    "BEGIN example_procedure(:cursor_parameter); END;")
  .bind("cursor_parameter", Parameters.out(OracleR2dbcTypes.REF_CURSOR))
  .execute()
}

A SYS_REFCURSOR out parameter can be mapped to an io.r2dbc.spi.Result:

Publisher<Result> mapOutParametersResult(Result outParametersResult) {
  return outParametersResult.map(outParameters ->
    outParameters.get("cursor_parameter", Result.class));
}

The rows of a SYS_REFCURSOR may be consumed from the Result it is mapped to:

Publisher<ExampleObject> mapRefCursorRows(Result refCursorResult) {
  return refCursorResult.map(row ->
    new ExampleObject(
      row.get("id_column", Long.class),
      row.get("value_column", String.class)));
}

VECTOR

The default mapping for the VECTOR data type is the oracle.sql.VECTOR class. Instances of this class may be passed to Statement.bind(int/String, Object):

void bindVector(Statement statement, float[] floatArray) throws SQLException {
  final VECTOR vector;
  try {
    vector = VECTOR.ofFloat32Values(floatArray);
  }
  catch (SQLException sqlException) {
    throw new IllegalArgumentException(sqlException);
  }
  statement.bind("vector", vector);
}

The oracle.sql.VECTOR class defines factory methods that convert a double[], float[], long[], int[], short[], byte[], or boolean[] into a VECTOR of a specific dimension type:

  • ofFloat64Values
  • ofFloat32Values
  • ofInt8Values
  • ofBinaryValues (only supports boolean[] and byte[])

These factory methods may perform a lossy conversion, such as when converting a double[] into a VECTOR of FLOAT32 dimensions. The JavaDocs of these methods specify which conversions are lossy.

The OracleR2dbcTypes.VECTOR type descriptor can be used to register an OUT or IN/OUT parameter:

void registerOutVector(Statement statement) {
  Parameter outVector = Parameters.out(OracleR2dbcTypes.VECTOR);
  statement.bind("out_vector", outVector);
}

The OracleR2dbcTypes.VECTOR type descriptor can also be used as an alternative to oracle.sql.VECTOR for binding a double[], float[], byte[], or boolean[]. Arrays of these types are respectively converted into a VECTOR of FLOAT64, FLOAT32, INT8, or BINARY dimensions.

void bindVector(Statement statement, float[] floatArray) {
  Parameter inVector = Parameters.in(OracleR2dbcTypes.VECTOR, floatArray);
  statement.bind("in_vector", inVector);
}

Note that passing arrays directly into Statement.bind(int/String, Object) will NOT create a VECTOR bind. The R2DBC Specification already defines ARRAY as the default mapping for Java arrays, not VECTOR.

A VECTOR column or OUT parameter is converted to oracle.sql.VECTOR by default. But a VECTOR column or OUT parameter having FLOAT64, FLOAT32, or INT8 dimensions may also be converted to double[], float[], long[], int[], short[], byte[], or boolean[]. A VECTOR of BINARY dimensions may only be converted to boolean[] or byte[]. These array classes may be passed to the get methods of a Readable:

float[] getVector(io.r2dbc.Readable readable) {
  return readable.get("vector", float[].class);
}

Secure Programming Guidelines

The following security related guidelines should be adhered to when programming with the Oracle R2DBC Driver.

Defend Against SQL Injection Attacks

  • Always specify the parameters of a SQL command using the bind methods of io.r2dbc.spi.Statement.
    • Do not use String concatenation to specify parameters of a SQL command.
    • Do not use format Strings to specify parameters of a SQL command.

Protect Passwords

  • Do not hard code passwords in your source code.
  • Avoid hard coding passwords in the R2DBC URL.
    • When handling URL strings in code, be aware that a clear text password may appear in the user info section.
  • Use a sensitive io.r2dbc.spi.Option to specify passwords.
    • If possible, specify the Option's value as an instance of java.nio.CharBuffer or java.lang.StringBuffer and clear the contents immediately after ConnectionFactories.get(ConnectionFactoryOptions) has returned. Oracle R2DBC's implementation of ConnectionFactory does not retain a reference to the clear text password.

Protect Network Communications

  • Use SSL/TLS if possible. Use any of the following methods to enable SSL/TLS:
    • Specify the boolean value of true for io.r2dbc.spi.ConnectionFactoryOptions.SSL
    • Specify "r2dbcs:" as the R2DBC URL schema.
    • Specify "ssl=true" in the query section of the R2DBC URL.
  • Use Option.sensitiveValueOf(String) when creating an Option that specifies a password.
    • Option.sensitiveValueOf(OracleConnection.CONNECTION_PROPERTY_WALLET_PASSWORD)
      • An SSO wallet does not require a password.
    • Option.sensitiveValueOf(OracleConnection.CONNECTION_PROPERTY_THIN_JAVAX_NET_SSL_KEYSTOREPASSWORD)
    • Option.sensitiveValueOf(OracleConnection.CONNECTION_PROPERTY_THIN_JAVAX_NET_SSL_TRUSTSTOREPASSWORD)

Defend Against Denial-of-Service Attacks

  • Use a connection pool and configure a maximum size to limit the number of database sessions created by ConnectionFactory.create()
  • Enforce a maximum batch size to limit invocations of Statement.add() or Batch.add(String).
  • Enforce a maximum fetch size to limit values supplied to Statement.fetchSize(int).
  • Enforce a maximum buffer size to limit memory usage when reading Blob and Clob objects.