graphs-for-js

Some JavaScript and TypeScript implementation of a graph data structure. Features: - Insert and remove nodes. - Connect and disconnect nodes. - Algorithms for graph structures.

Usage no npm install needed!

<script type="module">
  import graphsForJs from 'https://cdn.skypack.dev/graphs-for-js';
</script>

README

graphs-for-js NPM version Build Status Coverage Status

Implementations of graph data structures for JavaScript and TypeScript

Features:

  • Insert and remove nodes.
  • Create edges by connecting and disconnecting nodes.
  • Algorithms for graph data structures.

Table of contents generated with markdown-toc

Installation

$ npm install --save graphs-for-js

Graph Class Usage

Import the Graph class to initialize a graph.

The library supports 4 types of graphs:

  • Weighted, Directed graphs
    • Edges go in one-direction and can be assigned a value.
  • Unweighted, Directed graphs
    • Edges go in one-direction and cannot be assigned a value.
  • Weighted, Undirected graphs
    • Edges are bidirectional and can be assigned value.
  • Unweighted, Undirected graphs
    • Edges are bidirectional and cannot be assigned a value.

For each of the above graph types, there are also readonly, immutable versions.

Import the GraphBuilder

// With require()
const {Graph} = require('graphs-for-js')


// With import syntax
import {Graph} from 'graphs-for-js'

Key Function

Because JavaScript does not natively hash/map objects and their contents to unique values as object keys, the GraphBuilder accepts a function for mapping a node value to a string key.

If a Key Function is not given, then the default behaviors on node values are the following:

  • Primitive Types
    • This includes number, string, boolean, symbol, null, and undefined.
    • All primitives are converted to their string equivalents.
    • Caution: Floats are subject to float-precision issues.
  • Objects
    • A string representation of the object is used as the key, for which the string contains properties including keys and values.
    • For example, the object {a: 32, b: 23} is mapped as '{a:32,b:23}'
    • If an object value is circular, then it is mapped as the value of toString(), normally '[object Object]'

JavaScript initialization

const weightedGraph = new Graph()
  .keyFn(i => `${i}`).directed.weighted()

const unweightedGraph = new Graph()
  .noKey().directed.unweighted()

TypeScript initialization

Use the type parameters to specify the type of the nodes and of the edge values (if weighted).

const weightedGraph = new Graph<string, number>()
                          .noKey().directed.weighted()

const unweightedGraph = new Graph<string>()
                          .noKey().directed.unweighted()

You can also initiate a readonly graph which cannot be modified.

const weightedGraph = new Graph<number, number>()
                          .noKey().readonly.directed
                          .weighted([[1, 2, 5], [2, 3, 6]])
                          // Specify edges and implicitly the nodes

const unweightedGraph = new Graph<number>()
                          .noKey().readonly.directed
                          .unweighted([], [2, 3, 4, 5])  
                          // No edges, followed by an array of extra nodes.

Using the Graph

For the following examples, assume the nodes of graph are numbers.

The examples show the differences among each type of graph when necessary.

Insert nodes

graph.insert(0)  // Insert a single node
graph.insert(1, 2, 3) // Insert multiple nodes
const result = graph.insert(...[4,5,6,7]) // Use spread syntax for array inputs

console.log(result) // The number of nodes inserted

Removing nodes

graph.remove(0)  // Remove a single node
graph.remove(1, 2, 3) // Removes multiple nodes
const result = graph.remove(...[4,5,6,7]) // Use spread syntax for array inputs

console.log(result) // The number of nodes removed

Number of nodes

graph.count()

Forming edges

unweightedGraph.connect(1, 2) // Creates an edge from node 1 to node 2

weightedGraph.connect(1, 2, 0.5) // Creates an edge from node 1 to node 2 with weight 0.5

Removing edges

unweightedGraph.disconnect(1, 2) // Removes the edge from node 1 to node 2

weightedGraph.disconnect(1, 2) // Removes the edge from node 1 to node 2
weightedGraph.disconnect(1, 2, 0.5) // Removes the edge from node 1 to node 2 only if the weight of that edge is 0.5

undirectedGraph.connect(1, 2)
undirectedGraph.disconnect(2, 1) // Will remove the edge from node 1 to node 2 in a unweighted graph.

Get all of the nodes and edges in the graph

graph.nodes()  // Returns an array of node values
graph.edges()  // Returns an array of edges

/*
type Edge<V, E> = {
  source: V,
  target: V,
  value: E,
  undirected: boolean
}
*/

Incoming and Outgoing edges

graph.outgoingEdgesOf(2)  // Returns an array of edges whose source nodes are node 2
graph.incomingEdgesOf(2)  // Returns an array of edges whose target nodes are node 2

Degree of Edge

graph.degreeOf(2) // Degree of node 2
graph.inDegreeOf(2) // In-Degree of node 2
graph.outDegreeOf(2) // Out-Degree of node 2

Existence Methods

graph.contains(1)   // Contains node 1
graph.contains(1, 2, 3)  // Contains all three nodes



unweightedGraph.hasEdge(1, 2)  // Has edge from node 1 to node 2

weightedGraph.hasEdge(1, 2) // Has edge from node 1 to node 2
weightedGraph.hasEdge(1, 2, 0.5) // Returns true if there is an edge from node 1 to node 2 AND edge value is 0.5

undirectedGraph.hasEdge(x, y) === undirectedGraph.hasEdge(y, x)  // true

Edge Value

weightedGraph.weightOf(1, 4) 
// Returns the value of the edge from nodes 1 to 4, if it exists.
// If it does not exist, it returns undefined.

GraphUtil Usage

The GraphUtil module contains some helper functions/algorithms that can be used on the graphs.

  • hasCycle(graph)

    • Returns true if there is a cycle in the graph, false otherwise

  • findShortestPath(graph, start, end)

    • Finds the shortest path from the node start to the node end. Returns an object with the fields path and pathLength. If there exists a path, then path is an array of nodes in that path in order from start to end, and pathLength is the length of that path, i.e. the number of edges. If a path is not found, then path is an empty array, and pathLength is -1.

  • clone(graph)

    • Creates a new instance of a graph that contains all nodes and edges in the given graph. The type of graph returned is the same type of graph given, e.g. if an undirected, unweighted graph is given, then the cloned graph will also be undirected and unweighted.

  • topologicalSort(graph)

    • Topologically sorts the graph. Returns undefined if the given graph is not a DAG. Otherwise, it returns an array of nodes in topologically sorted order.

  • toAdjacencyMatrix(graph)

    • Converts the given graph into an adjacency matrix. Returns an object with 5 values:
    • type Result<V, E> = {
  // matrix[i][j] is true if there is an edge from node i to node j. Otherwise, it is false
  matrix: boolean[][]

  // valueMatrix[i][j] returns the value/weight on the edge from node i to j.
  // If there is no value or the edge does not exist, it is undefined.
  valueMatrix: (E | undefined)[][]
  
  // For each node n, nodeToIndex maps toKeyFn(n) to its index on the adjacency matrix
  nodeToIndex: Record<string, number>
  
  // Maps the index number on the adjacency matrix to the actual node value.
  indexToNode: V[]

  // An array of pairs, the node value and its index on the adjacency matrix.
  nodeIndexPairs: {node: V, index: number}[]
}

  • functional utility functions:

    • subset(graph, nodes)
      • Returns a new subgraph instance that contains a subset of its original nodes, where each node in that subset is in nodes.
      • The return type of subset is the same as the return type for clone.
    • mapNodes(graph, callback, newKeyFn?)
      • Creates a new graph instance whose nodes are the results of calling the given callback function on each node in the given graph.
      • The key function of the new graph is the given newKeyFn or the default key function if not given.
      • If 2 or more nodes result in the same key value (because of the callback function or the new key function), then those nodes are merged into one node, and each edge in those nodes are connected the newly merged node. If there was edge between two of those nodes, then the merged node will have a self loop.
    • mapEdges(graph, callback)
      • Creates a new graph instance whose edge values are the results of calling the given callback function on each edge value in the given graph.

  • serialize utility functions:

    • stringify(graph)
      • Creates a string using the nodes and edges of the graph.
    • parse(json)
      • Creates a graph using a serialized representation of the graph.

Examples

const {Graph, GraphUtil} = require('graphs-for-js')

const graph = new Graph().noKey().directed.unweighted()

// Returns true if there exists a cycle in `graph`. Otherwise false
GraphUtil.hasCycle(graph)

/*
  Finds the shortest path from the start node to the end node.
  Where V is the type of the node
  return type: {
    path: Array<V>  // The nodes in the discovered path. If no path is found, then this array is empty.
    pathLength: number // The number of edges in the discovered path. If no path is found, then this is -1.
  }
*/
GraphUtil.findShortestPath(graph, 1, 2)

Contributing

Feel free to contribute by adding changes to the graph implementation or by writing implementations for graph algorithms in GraphUtil! You can also add suggestions or issues here

If you'd like to make changes to the graph implementation or for GraphUtil:

  1. Fork this repository

  2. Create a feature branch

  3. Make changes 🛠

  4. Make a PR to merge into this repo

License

ISC © Anthony Yang