Provides utils to deal with Java Streams. Especially add missing features like takeWhile
and batch
.
This is one of the missing features of Java 8 streaming API.
Continue a stream only until all elements matches a specific predicate.
Reference: <T> Stream<T> StreamUtils.takeWhile(Stream<T> input, Predicate<T> predicate)
Example:
// Prints out random numbers to stdout until 66 was generated
// 66 will not appear in the stdout.
Random random = new Random();
Stream<Integer> stream = Stream.generate(() -> random.nextInt(100));
StreamUtils.takeWhile(stream, number -> number != 66)
.forEach(System.out::println);
This is one of the missing features of Java 8 streaming API.
Generate a stream and decide by your own when it should be done.
The regular Streaming API could only produce endless streams.
A solution could be our Stream takeWhile
but our generate
is sometimes a smarter solution.
Reference: <T> Stream<T> StreamUtils.generate(Stream<T> input, Predicate<T> predicate)
Example:
// Prints out random numbers to stdout until 66 was generated
// 66 will not appear in the stdout.
Random random = new Random();
StreamUtils.generate(() -> {
int candidate = random.nextInt(100);
return candidate == 66 ? Value.end() : Value.valueOf(candidate);
}).forEach(System.out::println);
This is one of the missing features of Java 8 streaming API.
Slice a stream with an unknown length in batches with a fixed size.
Perfect for example to query a batch of elements from a database to do with all of their IDs another batch query to the database to enrich them with more information.
Reference: <T> Stream<T> StreamUtils.batch(Stream<T> input, int batchSize)
Example:
// Prints out random numbers to stdout in ten blocks until 66 was generated
// 66 will not appear in the stdout.
Random random = new Random();
Stream<Integer> stream = StreamUtils.generate(() -> {
int candidate = random.nextInt(100);
return candidate == 66 ? Value.end() : Value.valueOf(candidate);
});
StreamUtils.batch(stream, 10)
.forEach(System.out::println);
How to integrate a batch back again into a single stream?
Stream<Integer> sourceStream = ...;
Stream<List<Integer>> batchedStream = StreamUtils.batch(sourceStream, 10);
Stream<Integer> enrichedStream = batchedStream.flatMap(batch -> {
// Do something with the batch...
...
// Flat map all elements again into a single element stream...
return batch.stream();
});
enrichedStream.forEach(System.out::println);
This is one of the missing features of JDBC API.
Simply converts a java.sql.ResultSet
into a java.util.stream.Stream
of ResultSets.
For every element in this stream ResultSet.next()
will be called
... which means every element in this stream represents a row.
Reference: Stream<ResultSet> JdbcUtils.toStream(ResultSet resultSet)
Example:
// Prints of every row the first column as string to stdout.
[..]
Stream<ResultSet> stream = JdbcUtils.toStream(resultSet);
stream
.forEach(row -> System.out.println(row.getString(1)));
Is similar to the toStream
version of above but also includes already mapping functionality
which also handles transparent java.sql.SQLException
s which means: You are not always required
to implement annoying try {..} catch {..}
blocks for you mappings.
Reference: <T> Stream<T> JdbcUtils.toStream(ResultSet resultSet, SqlFunction<ResultSet, T> mapper)
Example:
// Prints of every row the first column as string to stdout.
[..]
Stream<String> stream = JdbcUtils.toStream(resultSet, row -> row.getString(1));
stream
.forEach(System.out::println));
You can directly register our repository if you want always the latest version. The central can be versions behind.
<repositories>
<repository>
<id>central</id>
<url>https://repo.maven.apache.org/maven2</url>
</repository>
<repository>
<id>echocat</id>
<url>https://packages.echocat.org/maven</url>
</repository>
</repositories>
repositories {
mavenCentral()
maven {
url "https://packages.echocat.org/maven"
}
}
Find your desired version you want to install (usually the latest one) by looking it up in our repository or directly at the Maven Central.
<dependency>
<groupId>org.echocat.java-stream-utils</groupId>
<artifactId>java-stream-utils</artifactId>
<version><!-- THE VERSION --></version>
</dependency>
compile 'org.echocat.java-stream-utils:java-stream-utils:<THE VERSION>'
The MySQL JDBC driver by default stores at first the whole result of the database in the memory before
it will be processed. This delivers in many cases with small result sets a better performance instead
streaming it one by one. But this is very memory consuming and slow for large result sets - and also
dangerous - it may cause OutOfMemoryError
s.
You can do the following thing to enable the streaming results (works for MySQL):
Statement statement = connection.createStatement(ResultSet.TYPE_FORWARD_ONLY, ResultSet.CONCUR_READ_ONLY)
statement.setFetchSize(Integer.MIN_VALUE);
java-stream-utils is an open source project of echocat. So if you want to make this project even better, you can contribute to this project on Github by fork us.
If you commit code to this project you have to accept that this code will be released under the license of this project.
See LICENSE file.