A simple wrapper API to make usage of Java collections, lambdas and CompletableFuture more simple, concise and enjoyable.
Java
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
src
.gitignore
.travis.yml
LICENSE
README.md
pom.xml

README.md

LambdaOmega

A simple wrapper API to make usage of Java collections, lambdas and CompletableFuture more simple, concise and enjoyable.

Build Status codecov.io

What’s in the box?

LambdaOmega consists of only a few classes. For brevity reasons, most of their names consist of a single letter.

  • L (“List”) is a wrapper for List.
  • S (“Set”) is a wrapper for Set.
  • M (“Map”) is a wrapper for Map.
  • R (“Range”) is a wrapper for an IntStream range.
  • V2 (“Vector 2D”) represents a 2D vector (= a 2-ary tuple). It can be converted into a Map.Entry.
  • F (“Function”) is a wrapper for functional interfaces (lambda expressions) which also provides helper methods to convert functions.
  • C (“Collection”) is the base class for L, S and M and provides additional helper methods to convert collections.
  • U (“Utils”) provides additional miscellaneous helper methods.
  • Promise is a wrapper and a drop-in replacement for CompletableFuture, providing several simplifications and fixes for the API. It can be used independently of all the other classes. It’s discussed in a separate section.

Why you should use it

  • Although Java’s Collections / lambdas are everywhere, you feel like their API is really cumbersome.
  • You feel even more so if you know and love Groovy or JavaScript / lodash.
  • It perfectly fits unit tests where fluid, maintainable code is key.

Other benefits:

  • Small footprint (JAR < 130KB), no other dependencies.
  • Thoroughly tested (coverage >= 90%).
  • Human-readable documentation (here and in the API docs).
  • Free & Open source (New BSD license).

How to use it

Use JitPack to add its dependency to your Maven project:

<dependency>
    <groupId>com.github.codebulb</groupId>
    <artifactId>LambdaOmega</artifactId>
    <version>0.3</version>
</dependency>
...
<repository>
    <id>jitpack.io</id>
    <url>https://jitpack.io</url>
</repository>

Replace the version by the tag / commit hash of your choice or -SNAPSHOT to get the newest SNAPSHOT.

Not using Maven? You can download the JAR directly from JitPack’s servers.

Visit JitPack’s docs for more information.

Getting started with Collections

The heart of LambdaOmega are the wrapper classes L, S and M which wrap around vanilla Java Collection List, Set or Map, respectively (decorator pattern) to provide a more concise and more enjoyable API to the underlying collection. You can wrap an L around everything which can be turned into a List: a List, a Stream, varargs:

import static ch.codebulb.lambdaomega.L.*;

L<Integer> myL = l(0, 1, 2);

The wrapper function invokes the L constructor and returns the L. You can’t invoke that constructor explicitly, always use the convenience wrapper function instead.

It’s best practice to statically import the l() method so you can write just l() instead of L.l(). The same goes for the other collections, e.g. S.s():

S<Integer> myS = s(0, 1, 2);

In the most simple case, you just want to return the underlying Collection:

List<Integer> mylist = l(0, 1, 2).l;

There’s a shorthand method for this simple case:

List<Integer> list = list(0, 1, 2);

Note that l(Collection) returns a one-element L with the collection provided being the one element. To create an L containing all the elements of the collection, use

L<Integer> flatList = L(list(0, 1, 2));

Some wrapper methods return another wrapper (e.g. of type L) or this; these are “intermediate” functions. You can chain another function call after them (builder pattern). Other wrapper methods return the underlying Java collection (e.g. of type List) or any other type; these are “terminal” functions. The wrapper API adheres to a simple method naming schema:

  • methods with names starting with an UpperCase letter are intermediate functions
  • methods with names consisting of a single letter are intermediate functions as well. These are typically shorthand aliases for another intermediate function.
  • methods with names starting with a lowerCase letter are terminal functions

There are no exceptions to these rules. (Although they don’t apply to the special Promise class.)

With these rules in mind, we can start working with a L, e.g. add a new element:

List<Integer> outcome = l(0, 1, 2).add(3);

This is a terminal function. Using the equivalent intermediate function, we can chain method calls:

L<Integer> modifiedList = l(0, 1, 2).Add(3).Add(4, 5);

There’s a shorthand for the Add() method:

L<Integer> modifiedList2 = l(0, 1, 2).a(3).a(4, 5);

When constructing a Map, you make intense use of method chaining:

Map<String, Integer> map = m("a", 0).Insert("b", 1).i("c", 2).m;

(we cover Map’s insert() method in a bit.)

There’s a two-arg constructor for empty Maps of an explicit key / value type. If the “key” class is omitted, it’s assumed to be String:

M<String, Integer> emptyM = m(Integer.class);

Collection methods

None of the LambdaOmega collections implements Java’s Collection or Map interface, for two reasons:

  • We don’t want to inherit this bloated, flawed API.
  • It encourages us to use LambdaOmega collections locally only and unwrap them into Java collections for external method calls.

However, L, S and M implement a set of methods closely inspired by Collection and Map; most methods are actually the same, others are augmented to e.g. accept varargs or to provide a more reasonable return type.

Some of these methods exist both as intermediate and terminal function. For the most common methods, a shorthand one-letter variant exists.

We’ve already met the add / Add / a method on L and S. There’s also addAll / AddAll / A; all of these also works with varargs:

List<Integer> zeroToSix = l(0, 1, 2).A(list(3, 4)).addAll(l(5, 6));

For M, there’s an add method variant named insert / Insert / i / insertAll / InsertAll / I which will invoke put(), but only after a check preventing you from inserting the same key twice; otherwise, an IndexAlreadyPresentException is thrown and the map is not modified. This method comes in handy e.g. in unit tests when you explicitly build a map and you want to make sure that you don’t accidentally insert the same key twice. Note that this check costs significantly more performance than just performing put().

m("a", 0).i("b", 1).I(m("c", 2).i("d", 3), m("c", 9)); // exception because of "c"!

For L or M, you can set a single element with set / Set / s / put / Put / p or set all elements included in a Map data structure with setAll / SetAll / S / putAll / PutAll / P.

m("a", 0).p("b", 1).p("c", 2).P(m("d", 3).i("e", 4));

For L or M, you can get a single element or multiple elements with get / Get / g:

String a = l("a", "b", "c").g(0);
L<String> aAndC = l("a", "b", "c").g(0, 2);

For L or S, you can remove a single element or a collection with remove / Remove / r / removeAll / RemoveAll / R:

L<String> aAndB = l("a", "b", "c", "d", "e").r("c").R(list("d", "e"));

For M, you can remove a single element or a collection with deleteKey / DeleteKey / d / deleteAllKeys / DeleteAllKeys / D or deleteValue / DeleteValue / deleteAllValues / DeleteAllValues (it’s called “delete” instead of “remove” to avoid naming clashes for the one-letter abbreviation):

M<String, Integer> bAndC = m("a", 0).i("b", 1).i("c", 2).d("a");

Also, you can convert an L / S / M into almost any collection with a corresponding to…(…) method:

Set<Integer> set = l(0, 1, 2).toSet();

These conversion methods internally use the C.to…(…) static helper methods. You can call them directly to convert collection without the need to create intermediate L / S / M instances.

There are a lot of additional methods for L, S and M. For more information, visit the API docs.

A List is a Map and a Map is a List

Now comes the fun part. Because the LambdaOmega API lives independently of vanilla Java Collection / Map API, it features its own API which is more simple, consistent and powerful at the same time, whilst keeping it as close to the original Java APIs as possible.

Most importantly, a List is also a Map from int to T, and a Map is also a List of entries. More precisely, both L and M implement the SequentialI (Collection-like access) interface as well as the IndexedI (Map-like access) interface. Note that similar to its plain Java Set counterpart, S only implements the SequentialI interface. Thus, you can call L methods such as remove(M.E) on a M:

m("a", 0).i("b", 1).i("c", 2).remove(e("b", 1));

and M methods such as insert(Integer, T) on a L:

L<String> abcd = l("a", "b").I(m(2, "c").i(3, "d"));

Finally, indexed access using get() is actually provided by the I interface which is implemented by L, M, and F, thus you can use “one interface to rule them all”:

I<Integer, String> indexed = m(0, "a").i(1, "b");
String b = indexed.g(1);
indexed = l("a", "b");
b = indexed.g(1);
indexed = f((Integer it) -> it == 0 ? "a" : "b");
b = indexed.g(1);

Collection methods with Lambdas

One of the main reasons to use LambdaOmega are the simplifications to use collections with lambda expressions in functional programming. When compared with vanilla Java lambda use, the API is much more simple and concise. Many of these modifications are inspired by the equivalent Groovy semantics.

Most importantly, functions are invoked on the collection object itself, without the need to create an intermediate stream.

To execute a forEach loop:

l("a", "b", "c").forEach(it -> println(it));

(U.println() is a shorthand for System.out.println().)

Keep in mind that a L is also a Map, thus there’s a variant of the forEach loop which takes both a key and value argument whereby in a L, the key is the index (this method is thus equivalent to a Groovy forEachWithIndex() loop).

l("a", "b", "c").forEach((index, it) -> println(index + ": " + it));

Again note how intermediary functions are marked by a name starting with an UpperCase letter. Here we invoke two consecutive map operations:

List<Integer> list246 = l(0, 1, 2).Map(it -> it + 1).map(it -> it * 2);

Some functions incorporate syntax changes (when compared to their vanilla Java couterpart) to facilitate their usage. With mapEntries(), you can map list elements to a Map. Note that the lambda returns a M.E Map Entry which is much simpler than having two separate return values as in the corresponding vanilla Java method.

Map<String, Integer> stringsToSize = l("a", "ab", "abc").mapEntries(it -> e(it, it.length()));

(e() is a shorthand to create a M.E with key, value.)

find() returns null or a result instead of an Option:

Integer negative = l(1, 2, -1, 3).find(it -> it < 0);

There are sort operations which work with consecutive sort key mappers:

l("John Smith", "Bob Miller", "Tim Miller", "Anna Smith").sortAscBy(it -> it.split(" ")[1], it -> it.split(" ")[0]);

There are also additional functional operations such as flatten(), flattenDeep(), join(), partition(), and more.

List<Integer> list123456 = l(0, 1, l(2, 3, l(4)), list(5, 6)).<Integer> flattenDeep();

There’s also a WithDefault() method which allows you for any I to register a function the return value of which is returned if a get() access would return null or get out of bounds (as inspired by Groovy). You can use this e.g. to easily create a Map of Lists:

M<String, L> withDefault = m(L.class).WithDefault(it -> l());
withDefault.g("a").a(1);
println(withDefault); // prints M{a=L[1]}

Take a look at the API docs to see all available functional operations.

Functions

The F class serves as a simple wrapper around any kind of FunctionalInterface, providing a unified API to any kind of function:

int two = f((Integer it) -> it + 1).call(1); // Function
int three = f((Integer x, Integer y) -> x + y).call(1, 2); // BiFunction
boolean yes = f((Integer it) -> it > 0).call(1); // Predicate

It also features some useful transformation methods for Java’s FunctionalInterface implementations. For instance, it may be useful to combine a 2-ary <K, V> function into a 1-ary E<K, V> function or vice versa to write more concise code:

F.function((Integer k, Integer v) -> k - v).apply(e(3, 2))
F.biFunction((E<Integer, Integer> it) -> it.k - it.v).apply(3, 2)

Moreover, LambdaOmega features two additional interfaces which act as a convenient multi-interface to Java’s FunctionalInterfaces:

  • FunctionalI combines all 1-ary function interfaces (except for UnaryOperator which is mutually incompatible with Function)
  • BiFunctionalI combines all 2-ary function interfaces (except for BinaryOperator which is mutually incompatible with BiFunction)

This allows you to use “one interface to rule them all”, instead of being forced to choose from 29 / 12 mutually incompatible interfaces when defining a lambda function:

private static boolean javaTestMethod(Function<Integer, Boolean> function) {
    return function.apply(SYSTEM_PROPERTY);
}

Predicate<Integer> javaPredicate = it -> it > 0;
// different functions have different APIs
boolean oneIsPositive = javaPredicate.test(1);
// primitive functions are incompatible with object functions
javaTestMethod(javaPredicate); // COMPILE ERROR
Function<Integer, Boolean> javaPredicateAsFunction = it -> javaPredicate.test(it);
boolean testOutput = javaTestMethod(javaPredicateAsFunction);

Use only one interface with LambdaOmega:

FunctionalI<Integer, Boolean> predicate = it -> it > 0;
// every FunctionalI has the method "call"
boolean twoIsPositive = predicate.call(2);
// can be used with any FunctionalInterface API
testOutput = javaTestMethod(predicate);

Having only one common functional interface has these benefits:

  • #call(…) as a single common interface method
  • plug the interface everywhere a lambda function is expected (with above constraints).

You can use the F class to explicitly create a FunctionalI / BiFunctionalI. For every F, the respective FunctionalI / BiFunctionalI is available:

// vanilla Java function casting not supported because of primitives
Function<Integer, Integer> primitivePred1 = (int x) -> x + 1; // COMPILE ERROR
FunctionalI<Integer, Boolean> primitivePred2 = (int x) -> x > 1; // COMPILE ERROR
FunctionalI<Integer, Boolean> primitivePred = f((int x) -> x > 1).f;

For the most important FunctionalInterfaces, there’s a shorthand:

FunctionalI<Integer, Boolean> pred = F((int x) -> x > 1);

That way, any lambda function can be turned into a FunctionalI / BiFunctionalI for global compatibility with all FunctionalInterfaces (again, with above constraints):

javaTestMethod(F((int it) -> it > 0));

Ranges

The R class represents an int range. It’s basically syntactic sugar to create ranges using the Java 8 stream API.

Create a range and convert it into a List like this:

List<Integer> list012 = r(0).to(3).list;

Support for range access has been added to L:

List<String> listBcd = l("a", "b", "c", "d", "e").get(r(1).to(3));

2D vectors

The V2 class represents a 2-dimensional vector = a 2-ary tuple:

V2<String, Integer> vector = v("a", 0);

By simply adding another value pair, the whole structure turns into a L:

V2.LV2<String, Integer> vector2 = v("a", 0).a("a", 1);

Promises

This library also features a drop-in-replacement for CompletableFuture, simplifying and fixing its partially flawed API. I went into some more details in the accompanying blog post.

Things to consider and Best practices

  • Performance: LambdaOmega doesn’t make use of reflection. It really just wraps Collection API method calls and should perform similarly to vanilla API calls.
    • However, intermediate operations on LambdaOmega collections will always create a new Collection much unlike vanilla Java collections which just operate on a stream: l(0, 1, 2).Map(it -> it + 1).map(it -> it * 2); // Creates a new Collection on #l(…), #Map(…) and #map(…) This may decrease performance for big collections. Thus, even though LambdaOmega should perform OK in everyday situations, keep in mind that it is not and will never be built primarily for speed.
  • LambdaOmega is not built as a replacement for any of the Java core classes / interfaces. It’s hence best practice to use LambdaOmega collections and other abstractions locally only for the coder’s convenience and to not expose them in a public API to reduce dependency on this library.
  • However, LambdaOmega fits perfectly where you make intense use of Java’s collection API and keeping your code clean and concise is key. It’s thus especially useful for e.g. JUnit test code which typically involves lots of collection boilerplate code.

Project status and future plans

LambdaOmega is RELEASED, feedback is welcome.

I will use the issue tracker to plan features for future releases. Feel free to use it to submit feature requests or bug reports, or make your own contribution to the project.

This is a private project I’ve started for my own pleasure and usage and to learn more about Java 8’s collection API, and I have no plans for (commercial) support. I decided to publish it open sourced so that everyone who is interested is free to use it at his own discretion.

Version history

  • V. 0.3:
  • S enhancements: Fully implement S (Set).
  • V. 0.2:
  • F enhancements: Support for all FunctionalInterfaces and more transformations in the F class.
  • V. 0.1: First release

For more information

Please visit the accompanying blog post to learn more about why I created this library or check out the API docs: