ioredis-tree
v1.0.1
Published
A robust tree structure implementation for Redis
Downloads
52
Readme
ioredis-tree
A robust tree structure implementation for Redis
Install
npm install ioredis-tree
Usage
var Redis = require('ioredis');
var redis = require('ioredis-tree')(new Redis());
redis.tinsert('files', 'parent', 'node');
NOTE
Starting from v1.0.0, ioredis-tree assumes that nodes containing char ':' are the root of the tree for performance reasons.
API
TINSERT key parent node
Insert node
to parent
. If parent
does not exist, a new tree with root of parent
is created.
Example
redis.tinsert('mytree', '1', '2');
Creates:
+-----+
| 1 |
+----+-----+
|
+--+--+
| 2 |
+-----+
redis.tinsert('mytree', '1', '4');
Creates:
+-----+
| 1 |
+----+-----+----+
| |
+--+--+ +--+--+
| 2 | | 4 |
+-----+ +-----+
TINSERT
accepts one of the three optional options to specified where to insert the node into:
INDEX
: Insert the node to the specified index. Index starts with0
, and it can also be negative numbers indicating offsets starting at the end of the list. For example,-1
is the last element of the list,-2
the penultimate, and so on. If the index is out of the range, the node will insert into the tail (when positive) or head (when negative).BEFORE
: Insert the node before the specified node. Insert to the head when the specified node is not found.AFTER
: Insert the node after the specified node. Insert to the tail when the specified node is not found.
Continue with our example:
redis.tinsert('mytree', '1', '3', { before: '4' });
// Or:
// redis.tinsert('mytree', '1', '3', { after: '2' });
// redis.tinsert('mytree', '1', '3', { index: 1 });
// redis.tinsert('mytree', '1', '3', { index: -2 });
Creates:
+-----+
| 1 |
+----+--+--+----+
| | |
+--+--+ +--+--+ +--+--+
| 2 | | 3 | | 4 |
+-----+ +-----+ +-----+
redis.tinsert('mytree', '2', '5', { index: 1000 });
Creates:
+-----+
| 1 |
+----+--+--+----+
| | |
+--+--+ +--+--+ +--+--+
| 2 | | 3 | | 4 |
+--+--+ +-----+ +-----+
|
+--+--+
| 5 |
+-----+
A node can have multiple parents, say we insert 4
into 5
:
redis.tinsert('mytree', '5', '4');
Creates:
+-----+
| 1 |
+----+--+--+----+
| | |
+--+--+ +--+--+ +--+--+
| 2 | | 3 | | 4 |
+--+--+ +-----+ +-----+
|
+--+--+
| 5 |
+--+--+
|
+--+--+
| 4 |
+-----+
It's not allowed to move a node into its posterity, which will lead an error:
redis.tinsert('mytree', '3', '1');
// ERR parent node cannot be the posterity of new node
TPARENTS key node
Get the parents of the node. Returns an empty array when doesn't have parent.
redis.tparents('mytree', '5'); // ['2']
redis.tparents('mytree', '1'); // []
redis.tparents('mytree', '4'); // ['5', '1']
redis.tparents('non-exists tree', '1'); // []
redis.tparents('mytree', 'non-exists node'); // []
The order of parents is random.
TPATH key from to
Get the path between from
and to
. The depth of from
must lower than to
. Return null
if from
isn't a ancestor of to
.
redis.tpath('mytree', '1', '5'); // ['2']
redis.tpath('mytree', '1', '3'); // []
redis.tpath('mytree', '1', '7'); // null
If there's more than one path between the two nodes, the shorter path will be returned:
redis.tpath('mytree', '1', '4'); // []
TCHILDREN key node
Get the children of the node. Each node has at least two properties:
node
: The node itself.hasChild
:true
orfalse
, whether the node has at least one child.
If the hasChild
is true
, there will be an additional children
property, which is an array containing the children of the node.
redis.tchildren('mytree', '1');
// [
// { node: '2', hasChild: true, children: [{ node: '5', hasChild: false }] },
// { node: '3', hasChild: false },
// { node: '4', hasChild: false }
// ]
redis.tchildren('mytree', '5'); // []
redis.tchildren('non-exists tree', '1'); // []
redis.tchildren('mytree', 'non-exists node'); // []
TCHILDREN
accepts a LEVEL
option to specified how many levels of children to fetch:
redis.tchildren('mytree', '1', { level: 1 });
// [
// { node: '2', hasChild: true },
// { node: '3', hasChild: false },
// { node: '4', hasChild: false }
// ]
Notice that although node '2' has a child (its hasChild
is true
), it doesn't has the children
property since we are only insterested in the first level children by specifying { level: 1 }
.
TREM key parent count node
Remove the reference of a node from a parent.
redis.trem('mytree', '5', 0, '4');
Creates:
+-----+
| 1 |
+----+--+--+----+
| | |
+--+--+ +--+--+ +--+--+
| 2 | | 3 | | 4 |
+--+--+ +-----+ +-----+
|
+--+--+
| 5 |
+-----+
The count
argument influences the operation in the following ways:
- count > 0: Remove nodes moving from head to tail.
- count < 0: Remove nodes moving from tail to head.
- count = 0: Remove all nodes.
TREM
returns the remaining nodes in the parent.
TMREM key node
Remove the node from all parents. Use not
option to exclude a parent.
redis.tmrem('mytree', '2', { not: '3' });
TDESTROY key node
Destroy a node recursively and remove all references of it. Returns the count of nodes being deleted.
redis.tdestroy('mytree', '2'); // returns 2, since "2" and "5" are deleted.
Creates:
+-----+
| 1 |
+----+--+--+----+
| |
+--+--+ +--+--+
| 3 | | 4 |
+-----+ +-----+
TMOVECHILDREN key sourceNode targetNode [APPEND|PREPEND]
Move all the children of the sourceNode
to targetNode
. By default the new nodes will be
appended to the target node, if PREPEND
option is passed, the new nodes will be prepended
to the target node.
redis.tmovechildren('mytree', 'source', 'target', 'PREPEND');
TEXISTS key node
Returns if node exists.
redis.texists('mytree', '2'); // 0
redis.texists('mytree', '1'); // 1
TRENAME key node name
Rename a node.
redis.trename('mytree', '2', '5');
Creates:
+-----+
| 1 |
+----+--+--+----+
| |
+--+--+ +--+--+
| 5 | | 4 |
+-----+ +-----+
TPRUNE key node
Prune a tree to remove its children from parents which don't belong to the node.
Given the following two trees:
+-----+
| 1 |
+----+--+--+----+
| |
+--+--+ +--+--+
| 5 | | 4 |
+-----+ +-----+
+-----+
| 6 |
+----+--+--+
|
+--+--+
| 5 |
+-----+
redis.tprune('mytree', '1');
Creates:
+-----+
| 1 |
+----+--+--+----+
| |
+--+--+ +--+--+
| 5 | | 4 |
+-----+ +-----+
+-----+
| 6 |
+--+--+
Cluster Compatibility
This module supports Redis Cluster by ensuring all nodes that are belong to a tree have a same slot.
Performance
Benefiting from the high performance of Redis, modifying a tree is very fast. For instance, getting all children of a tree with the level of 100 recursively in a iMac 5k costs 4ms.