marcj / topsort
High-Performance TopSort/Dependency resolving algorithm
Fund package maintenance!
marcj
Installs: 10 438 346
Dependents: 53
Suggesters: 0
Security: 0
Stars: 239
Watchers: 4
Forks: 18
Open Issues: 2
Requires
- php: >=7.3
Requires (Dev)
- codeclimate/php-test-reporter: dev-master
- phpunit/phpunit: ^9
- symfony/console: ~2.5 || ~3.0 || ~4.0
README
This library provides several implementations of a Topological Sort (topSort). In additional to the plain sorting algorithm it provides several implementations of a Grouped Topological Sort, means you can pass items with a type which will be grouped together in the sorting. With its implementation of using strings instead of arrays its over 20x faster than regular implementations.
What is it?
A topological sort is useful for determining dependency loading. It tells you which elements need to be proceeded first in order to fulfill all dependencies in the correct order.
Example usage: Unit of Work (relations), simple Package manager, Dependency Injection, ...
Examples:
$sorter = new StringSort(); $sorter->add('car1', ['owner1', 'brand1']); $sorter->add('brand1'); $sorter->add('brand2'); $sorter->add('owner1', ['brand1']); $sorter->add('owner2', ['brand2']); $result = $sorter->sort(); // output would be: [ 'brand1', 'owner1', 'car1', 'brand2', 'owner2' ]
Sometimes you want to group equal types together (imagine a UnitOfWork which wants to combine all elements from the same type to stored those in one batch):
$sorter = new GroupedStringSort(); $sorter->add('car1', 'car', ['owner1', 'brand1']); $sorter->add('brand1', 'brand'); $sorter->add('brand2', 'brand'); $sorter->add('owner1', 'user', ['brand1']); $sorter->add('owner2', 'user', ['brand2']); $result = $sorter->sort(); // output would be: [ 'brand2', 'brand1', 'owner2', 'owner1', 'car1' ] $groups = $sorter->getGroups(); [ {type: 'brand', level: 0, position: 0, length: 2}, {type: 'user', level: 1, position: 2, length: 2}, {type: 'car', level: 2, position: 4, length: 1}, ] //of course there may be several groups with the same type, if the dependency graphs makes this necessary. foreach ($groups as $group) { $firstItem = $result[$group->position]; $allItemsOfThisGroup = array_slice($result, $group->position, $group->length); }
You can only store strings as elements.
To sort PHP objects you can stored its hash instead. $sorter->add(spl_object_hash($obj1), [spl_object_hash($objt1Dep)])
.
Installation
Use composer package: marcj/topsort
{
"require": {
"marcj/topsort": "~0.1"
}
}
include 'vendor/autoload.php'; $sorter = new GroupedStringSort; $sorter->add(...); $result = $sorter->sort();
Implementations
tl;dr: Use FixedArraySort
for normal topSort or GroupedStringSort
for grouped topSort since its always the fastest
and has a good memory footprint.
ArraySort
This is the most basic, most inefficient implementation of topSort using plain php arrays.
FixedArraySort
This uses \SplFixedArray of php and is therefore much more memory friendly.
StringSort
This uses a string as storage and has therefore no array overhead. It's thus a bit faster and has almost equal memory footprint like FixedArraySort. Small drawback: You can not store element ids containing a null byte.
GroupedArraySort
This is the most basic, not so efficient implementation of grouped topSort using plain php arrays.
GroupedStringSort
This uses a string as storage and has therefore no array operations overhead. It's extremely faster and has better memory footprint than GroupedArraySort. Small drawback: You can not store element ids containing a null byte.
Benchmarks with PHP 7.0.9
Test data: 1/3 has two edges, 1/3 has one edge and 1/3 has no edges. Use the benchmark
command in ./bin/console
to play with it.