tree maintained by sonars

A nested set is a smart way to implement an ordered tree that allows for fast, non-recursive queries. For example, you can fetch all descendants of a node in a single query, no matter how deep the tree. The drawback is that insertions/moves/deletes require complex SQL, but that is handled behind the curtains by this package!

Nested sets are appropriate for ordered trees (e.g. menus, commercial categories) and big trees that must be queried efficiently (e.g. threaded posts).

See the wikipedia entry for nested sets for more info. Also, this is a good introductory tutorial: http://www.evanpetersen.com/item/nested-sets.html

The theory behind, a TL;DR version

An easy way to visualize how a nested set works is to think of a parent entity surrounding all of its children, and its parent surrounding it, etc. So this tree:

root
  |_ Child 1
    |_ Child 1.1
    |_ Child 1.2
  |_ Child 2
    |_ Child 2.1
    |_ Child 2.2

Could be visualized like this:

 ___________________________________________________________________
|  Root                                                             |
|    ____________________________    ____________________________   |
|   |  Child 1                  |   |  Child 2                  |   |
|   |   __________   _________  |   |   __________   _________  |   |
|   |  |  C 1.1  |  |  C 1.2 |  |   |  |  C 2.1  |  |  C 2.2 |  |   |
1   2  3_________4  5________6  7   8  9_________10 11_______12 13  14
|   |___________________________|   |___________________________|   |
|___________________________________________________________________|

The numbers represent the left and right boundaries. The table then might look like this:

id | parent_id | lft  | rgt  | depth | data
 1 |           |    1 |   14 |     0 | root
 2 |         1 |    2 |    7 |     1 | Child 1
 3 |         2 |    3 |    4 |     2 | Child 1.1
 4 |         2 |    5 |    6 |     2 | Child 1.2
 5 |         1 |    8 |   13 |     1 | Child 2
 6 |         5 |    9 |   10 |     2 | Child 2.1
 7 |         5 |   11 |   12 |     2 | Child 2.2

To get all children of a parent node, you

SELECT * WHERE lft IS BETWEEN parent.lft AND parent.rgt

To get the number of children, it's

(right - left - 1)/2

To get a node and all its ancestors going back to the root, you

SELECT * WHERE node.lft IS BETWEEN lft AND rgt

As you can see, queries that would be recursive and prohibitively slow on ordinary trees are suddenly quite fast. Nifty, isn't it?

Installation

Tree works with Laravel 5 onwards. You can add it to your composer.json file with:

"sonars/tree": "~1.1"

Run composer install to install it.

As with most Laravel 5 packages you'll then need to register the Tree service provider. To do that, head over your config/app.php file and add the following line into the providers array:

'Tree\Providers\TreeServiceProvider',

Getting started

After the package is correctly installed the easiest way to get started is to run the provided generator:

php artisan tree:install MODEL

Replace model by the class name you plan to use for your Nested Set model.

The generator will install a migration and a model file into your application configured to work with the Nested Set behaviour provided by Tree. You SHOULD take a look at those files, as each of them describes how they can be customized.

Next, you would probably run artisan migrate to apply the migration.

Model configuration

In order to work with Tree, you must ensure that your model class extends Tree\Node.

This is the easiest it can get:

class Category extends Tree\Node {

}

This is a slightly more complex example where we have the column names customized:

class Dictionary extends Tree\Node {

  protected $table = 'dictionary';

  // 'parent_id' column name
  protected $parentColumn = 'parent_id';

  // 'lft' column name
  protected $leftColumn = 'lidx';

  // 'rgt' column name
  protected $rightColumn = 'ridx';

  // 'depth' column name
  protected $depthColumn = 'nesting';

  // guard attributes from mass-assignment
  protected $guarded = array('id', 'parent_id', 'lidx', 'ridx', 'nesting');

}

Remember that, obviously, the column names must match those in the database table.

Migration configuration

You must ensure that the database table that supports your Tree models has the following columns:

parent_id: a reference to the parent (int)
lft: left index bound (int)
rgt: right index bound (int)
depth: depth or nesting level (int)

Here is a sample migration file:

class Category extends Migration {

  public function up() {
    Schema::create('categories', function(Blueprint $table) {
      $table->increments('id');

      $table->integer('parent_id')->nullable();
      $table->integer('lft')->nullable();
      $table->integer('rgt')->nullable();
      $table->integer('depth')->nullable();

      $table->string('name', 255);

      $table->timestamps();
    });
  }

  public function down() {
    Schema::drop('categories');
  }

}

You may freely modify the column names, provided you change them both in the migration and the model.

Usage

After you've configured your model and run the migration, you are now ready to use Tree with your model. Below are some examples.

Creating a root node
Inserting nodes
Deleting nodes
Getting the nesting level of a node
Moving nodes around
Asking questions to your nodes
Relations
Root and Leaf scopes
Accessing the ancestry/descendancy chain
Limiting levels of children returned
Custom sorting column
Dumping the hierarchy tree
Model events: moving and moved
Scope support
Validation
Tree rebuilding
Soft deletes
Seeding/Mass assignment
Misc/Utility functions

Creating a root node

By default, all nodes are created as roots:

$root = Category::create(['name' => 'Root category']);

Alternatively, you may find yourself in the need of converting an existing node into a root node:

$node->makeRoot();

You may also nullify it's parent_id column to accomplish the same behaviour:

// This works the same as makeRoot()
$node->parent_id = null;
$node->save();

Inserting nodes

// Directly with a relation
$child1 = $root->children()->create(['name' => 'Child 1']);

// with the `makeChildOf` method
$child2 = Category::create(['name' => 'Child 2']);
$child2->makeChildOf($root);

Deleting nodes

$child1->delete();

Descendants of deleted nodes will also be deleted and all the lft and rgt bound will be recalculated. Pleases note that, for now, deleting and deleted model events for the descendants will not be fired.

Getting the nesting level of a node

The getLevel() method will return current nesting level, or depth, of a node.

$node->getLevel() // 0 when root

Moving nodes around

Tree provides several methods for moving nodes around:

moveLeft(): Find the left sibling and move to the left of it.
moveRight(): Find the right sibling and move to the right of it.
moveToLeftOf($otherNode): Move to the node to the left of ...
moveToRightOf($otherNode): Move to the node to the right of ...
makeNextSiblingOf($otherNode): Alias for moveToRightOf.
makeSiblingOf($otherNode): Alias for makeNextSiblingOf.
makePreviousSiblingOf($otherNode): Alias for moveToLeftOf.
makeChildOf($otherNode): Make the node a child of ...
makeFirstChildOf($otherNode): Make the node the first child of ...
makeLastChildOf($otherNode): Alias for makeChildOf.
makeRoot(): Make current node a root node.

For example:

$root = Creatures::create(['name' => 'The Root of All Evil']);

$dragons = Creatures::create(['name' => 'Here Be Dragons']);
$dragons->makeChildOf($root);

$monsters = new Creatures(['name' => 'Horrible Monsters']);
$monsters->save();

$monsters->makeSiblingOf($dragons);

$demons = Creatures::where('name', '=', 'demons')->first();
$demons->moveToLeftOf($dragons);

Asking questions to your nodes

You can ask some questions to your Tree nodes:

isRoot(): Returns true if this is a root node.
isLeaf(): Returns true if this is a leaf node (end of a branch).
isChild(): Returns true if this is a child node.
isDescendantOf($other): Returns true if node is a descendant of the other.
isSelfOrDescendantOf($other): Returns true if node is self or a descendant.
isAncestorOf($other): Returns true if node is an ancestor of the other.
isSelfOrAncestorOf($other): Returns true if node is self or an ancestor.
equals($node): current node instance equals the other.
insideSubtree($node): Checks whether the given node is inside the subtree defined by the left and right indices.
inSameScope($node): Returns true if the given node is in the same scope as the current one. That is, if every column in the scoped property has the same value in both nodes.

Using the nodes from the previous example:

$demons->isRoot(); // => false

$demons->isDescendantOf($root) // => true

Relations

Tree provides two self-referential Eloquent relations for your nodes: parent and children.

$parent = $node->parent()->get();

$children = $node->children()->get();

Root and Leaf scopes

Tree provides some very basic query scopes for accessing the root and leaf nodes:

// Query scope which targets all root nodes
Category::roots()

// All leaf nodes (nodes at the end of a branch)
Category:allLeaves()

You may also be interested in only the first root:

$firstRootNode = Category::root();

Accessing the ancestry/descendancy chain

There are several methods which Tree offers to access the ancestry/descendancy chain of a node in the Nested Set tree. The main thing to keep in mind is that they are provided in two ways:

First as query scopes, returning an Illuminate\Database\Eloquent\Builder instance to continue to query further. To get actual results from these, remember to call get() or first().

ancestorsAndSelf(): Targets all the ancestor chain nodes including the current one.
ancestors(): Query the ancestor chain nodes excluding the current one.
siblingsAndSelf(): Instance scope which targets all children of the parent, including self.
siblings(): Instance scope targeting all children of the parent, except self.
leaves(): Instance scope targeting all of its nested children which do not have children.
descendantsAndSelf(): Scope targeting itself and all of its nested children.
descendants(): Set of all children & nested children.
immediateDescendants(): Set of all children nodes (non-recursive).

Second, as methods which return actual Tree\Node instances (inside a Collection object where appropiate):

getRoot(): Returns the root node starting at the current node.
getAncestorsAndSelf(): Retrieve all of the ancestor chain including the current node.
getAncestorsAndSelfWithoutRoot(): All ancestors (including the current node) except the root node.
getAncestors(): Get all of the ancestor chain from the database excluding the current node.
getAncestorsWithoutRoot(): All ancestors except the current node and the root node.
getSiblingsAndSelf(): Get all children of the parent, including self.
getSiblings(): Return all children of the parent, except self.
getLeaves(): Return all of its nested children which do not have children.
getDescendantsAndSelf(): Retrieve all nested children and self.
getDescendants(): Retrieve all of its children & nested children.
getImmediateDescendants(): Retrieve all of its children nodes (non-recursive).

Here's a simple example for iterating a node's descendants (provided a name attribute is available):

$node = Category::where('name', '=', 'Books')->first();

foreach($node->getDescendantsAndSelf() as $descendant) {
  echo "{$descendant->name}";
}

Limiting the levels of children returned

In some situations where the hierarchy depth is huge it might be desirable to limit the number of levels of children returned (depth). You can do this in Tree by using the limitDepth query scope.

The following snippet will get the current node's descendants up to a maximum of 5 depth levels below it:

$node->descendants()->limitDepth(5)->get();

Similarly, you can limit the descendancy levels with both the getDescendants and getDescendantsAndSelf methods by supplying the desired depth limit as the first argument:

// This will work without depth limiting
// 1. As usual
$node->getDescendants();
// 2. Selecting only some attributes
$other->getDescendants(array('id', 'parent_id', 'name'));
...
// With depth limiting
// 1. A maximum of 5 levels of children will be returned
$node->getDescendants(5);
// 2. A max. of 5 levels of children will be returned selecting only some attrs
$other->getDescendants(5, array('id', 'parent_id', 'name'));

Custom sorting column

By default in Tree all results are returned sorted by the lft index column value for consistency.

If you wish to change this default behaviour you need to specify in your model the name of the column you wish to use to sort your results like this:

protected $orderColumn = 'name';

Dumping the hierarchy tree

Tree extends the default Eloquent\Collection class and provides the toHierarchy method to it which returns a nested collection representing the queried tree.

Retrieving a complete tree hierarchy into a regular Collection object with its children properly nested is as simple as:

$tree = Category::where('name', '=', 'Books')->first()->getDescendantsAndSelf()->toHierarchy();

Model events: `moving` and `moved`

Tree models fire the following events: moving and moved every time a node is moved around the Nested Set tree. This allows you to hook into those points in the node movement process. As with normal Eloquent model events, if false is returned from the moving event, the movement operation will be cancelled.

The recommended way to hook into those events is by using the model's boot method:

class Category extends Tree\Node {

  public static function boot() {
    parent::boot();

    static::moving(function($node) {
      // Before moving the node this function will be called.
    });

    static::moved(function($node) {
      // After the move operation is processed this function will be
      // called.
    });
  }

}

Scope support

Tree provides a simple method to provide Nested Set "scoping" which restricts what we consider part of a nested set tree. This should allow for multiple nested set trees in the same database table.

To make use of the scoping funcionality you may override the scoped model attribute in your subclass. This attribute should contain an array of the column names (database fields) which shall be used to restrict Nested Set queries:

class Category extends Tree\Node {
  ...
  protected $scoped = array('company_id');
  ...
}

In the previous example, company_id effectively restricts (or "scopes") a Nested Set tree. So, for each value of that field we may be able to construct a full different tree.

$root1 = Category::create(['name' => 'R1', 'company_id' => 1]);
$root2 = Category::create(['name' => 'R2', 'company_id' => 2]);

$child1 = Category::create(['name' => 'C1', 'company_id' => 1]);
$child2 = Category::create(['name' => 'C2', 'company_id' => 2]);

$child1->makeChildOf($root1);
$child2->makeChildOf($root2);

$root1->children()->get(); // <- returns $child1
$root2->children()->get(); // <- returns $child2

All methods which ask or traverse the Nested Set tree will use the scoped attribute (if provided).

Please note that, for now, moving nodes between scopes is not supported.

Validation

The ::isValidNestedSet() static method allows you to check if your underlying tree structure is correct. It mainly checks for these 3 things:

Check that the bound indexes lft, rgt are not null, rgt values greater than lft and within the bounds of the parent node (if set).
That there are no duplicates for the lft and rgt column values.
As the first check does not actually check root nodes, see if each root has the lft and rgt indexes within the bounds of its children.

All of the checks are scope aware and will check each scope separately if needed.

Example usage, given a Category node class:

Category::isValidNestedSet()
=> true

Tree rebuilding

Tree supports for complete tree-structure rebuilding (or reindexing) via the ::rebuild() static method.

This method will re-index all your lft, rgt and depth column values, inspecting your tree only from the parent <-> children relation standpoint. Which means that you only need a correctly filled parent_id column and Tree will try its best to recompute the rest.

This can prove quite useful when something has gone horribly wrong with the index values or it may come quite handy when converting from another implementation (which would probably have a parent_id column).

This operation is also scope aware and will rebuild all of the scopes separately if they are defined.

Simple example usage, given a Category node class:

Category::rebuild()

Valid trees (per the isValidNestedSet method) will not get rebuilt. To force the index rebuilding process simply call the rebuild method with true as the first parameter:

Category::rebuild(true);

Soft deletes

Tree comes with limited support for soft-delete operations. What I mean by limited is that the testing is still limited and the soft delete functionality is changing in the upcoming 4.2 version of the framework, so use this feature wisely.

For now, you may consider a safe restore() operation to be one of:

Restoring a leaf node
Restoring a whole sub-tree in which the parent is not soft-deleted

Seeding/Mass-assignment

Because Nested Set structures usually involve a number of method calls to build a hierarchy structure (which result in several database queries), Tree provides two convenient methods which will map the supplied array of node attributes and create a hierarchy tree from them:

buildTree($nodeList): (static method) Maps the supplied array of node attributes into the database.
makeTree($nodeList): (instance method) Maps the supplied array of node attributes into the database using the current node instance as the parent for the provided subtree.

Both methods will create new nodes when the primary key is not supplied, update or create if it is, and delete all nodes which are not present in the affecting scope. Understand that the affecting scope for the buildTree static method is the whole nested set tree and for the makeTree instance method are all of the current node's descendants.

For example, imagine we wanted to map the following category hierarchy into our database:

TV & Home Theater
Tablets & E-Readers
Computers
- Laptops
  - PC Laptops
  - Macbooks (Air/Pro)
- Desktops
- Monitors
Cell Phones

This could be easily accomplished with the following code:

$categories = [
  ['id' => 1, 'name' => 'TV & Home Theather'],
  ['id' => 2, 'name' => 'Tablets & E-Readers'],
  ['id' => 3, 'name' => 'Computers', 'children' => [
    ['id' => 4, 'name' => 'Laptops', 'children' => [
      ['id' => 5, 'name' => 'PC Laptops'],
      ['id' => 6, 'name' => 'Macbooks (Air/Pro)']
    ]],
    ['id' => 7, 'name' => 'Desktops'],
    ['id' => 8, 'name' => 'Monitors']
  ]],
  ['id' => 9, 'name' => 'Cell Phones']
];

Category::buildTree($categories) // => true

After that, we may just update the hierarchy as needed:

$categories = [
  ['id' => 1, 'name' => 'TV & Home Theather'],
  ['id' => 2, 'name' => 'Tablets & E-Readers'],
  ['id' => 3, 'name' => 'Computers', 'children' => [
    ['id' => 4, 'name' => 'Laptops', 'children' => [
      ['id' => 5, 'name' => 'PC Laptops'],
      ['id' => 6, 'name' => 'Macbooks (Air/Pro)']
    ]],
    ['id' => 7, 'name' => 'Desktops', 'children' => [
      // These will be created
      ['name' => 'Towers Only'],
      ['name' => 'Desktop Packages'],
      ['name' => 'All-in-One Computers'],
      ['name' => 'Gaming Desktops']
    ]]
    // This one, as it's not present, will be deleted
    // ['id' => 8, 'name' => 'Monitors'],
  ]],
  ['id' => 9, 'name' => 'Cell Phones']
];

Category::buildTree($categories); // => true

The makeTree instance method works in a similar fashion. The only difference is that it will only perform operations on the descendants of the calling node instance.

So now imagine we already have the following hierarchy in the database:

Electronics
Health Fitness & Beaty
Small Appliances
Major Appliances

If we execute the following code:

$children = [
  ['name' => 'TV & Home Theather'],
  ['name' => 'Tablets & E-Readers'],
  ['name' => 'Computers', 'children' => [
    ['name' => 'Laptops', 'children' => [
      ['name' => 'PC Laptops'],
      ['name' => 'Macbooks (Air/Pro)']
    ]],
    ['name' => 'Desktops'],
    ['name' => 'Monitors']
  ]],
  ['name' => 'Cell Phones']
];

$electronics = Category::where('name', '=', 'Electronics')->first();
$electronics->makeTree($children); // => true

Would result in:

Electronics
- TV & Home Theater
- Tablets & E-Readers
- Computers
  - Laptops
    - PC Laptops
    - Macbooks (Air/Pro)
  - Desktops
  - Monitors
- Cell Phones
Health Fitness & Beaty
Small Appliances
Major Appliances

Updating and deleting nodes from the subtree works the same way.

Misc/Utility functions

Node extraction query scopes

Tree provides some query scopes which may be used to extract (remove) selected nodes from the current results set.

withoutNode(node): Extracts the specified node from the current results set.
withoutSelf(): Extracts itself from the current results set.
withoutRoot(): Extracts the current root node from the results set.

$node = Category::where('name', '=', 'Some category I do not want to see.')->first();

$root = Category::where('name', '=', 'Old boooks')->first();
var_dump($root->descendantsAndSelf()->withoutNode($node)->get());
... // <- This result set will not contain $node

Get a nested list of column values

The ::getNestedList() static method returns a key-value pair array indicating a node's depth. Useful for silling select elements, etc.

It expects the column name to return, and optionally: the column to use for array keys (will use id if none supplied) and/or a separator:

public static function getNestedList($column, $key = null, $seperator = ' ');

An example use case:

$nestedList = Category::getNestedList('name');
// $nestedList will contain an array like the following:
// array(
//   1 => 'Root 1',
//   2 => ' Child 1',
//   3 => ' Child 2',
//   4 => '  Child 2.1',
//   5 => ' Child 3',
//   6 => 'Root 2'
// );

Further information

You may find additional information, usage examples and/or frequently asked questions about Tree in the wiki.

Feel free to browse the wiki after finishing this README:

https://github.com/etrepat/baum/wiki

Contributing

Thinking of contributing? Maybe you've found some nasty bug? That's great news!

Fork & clone the project: git clone git@github.com:your-username/baum.git.
Run the tests and make sure that they pass with your setup: phpunit.
Create your bugfix/feature branch and code away your changes. Add tests for your changes.
Make sure all the tests still pass: phpunit.
Push to your fork and submit new a pull request.

Please see the CONTRIBUTING.md file for extended guidelines and/or recommendations.

License

Tree is licensed under the terms of the MIT License (See LICENSE file for details).

Coded by Estanislau Trepat (etrepat). I'm also @etrepat on twitter.