Download the PHP package longitude-one/neo4j-php-client without Composer

Neo4j PHP Client

This Neo4j PHP Client is a fork of the GraphAware Neo4j PHP Client. Since this package was not maintained since 2017 and was archived to a readonly repository, LongitudeOne decided to fork it and to create a really new package. A simple fork was a problem because issues and pull requests were still centralized on graphaware/neo4j-php-client.

Introduction

Originally created by GraphAware, Neo4j-PHP-Client is the most advanced and flexible Neo4j Client for PHP. LongitudeOne is now maintaining this project.

What is Neo4j?

Neo4j is a transactional, open-source graph database. A graph database manages data in a connected data structure, capable of representing any kind of data in a very accessible way. Information is stored in nodes and relationships connecting them, both of which can have arbitrary properties. To learn more visit What is a Graph Database?

Key features

• Supports multiple connections
• Support for Bolt binary protocol
• Built-in and automatic support for Neo4j Enterprise HA Master-Slave Mode with auto slaves fallback

Neo4j Version Support

This project cannot be used with the version 4.0+.

Neo4j Feature Support

Requirements

As PHP7.2 and less are no more maintained, neo4j-php-client is now only tested with PHP7.3+ .

• PHP >= 7.3
• ext-bcmath
• ext-json
• ext-mbstring
• A Neo4j database (minimum version 3.0, maximum version 4.0 excluded)

Getting Help

To get help, you should:

• Ask a question on StackOverflow
• For bugs, please feel free to create a new issue on GitHub

Implementations

• Symfony Framework Bundle

Installation and basic usage

Installation

Add the library to your composer dependencies :

Require the composer autoloader, configure your connection by providing a connection alias and your connection settings :

You're now ready to connect to your database.

NB: The build method will process configuration settings and return you a Client instance.

Basic Usage

Sending a Cypher Query

Sending a Cypher Query with parameters

Reading a Result

A Record object contains the values of one record from your Cypher query :

Cypher statements and Stacks

Ideally, you would stack your statements and issue them all at once in order to improve performance.

You can create Cypher statement stacks that act as a Bag and run this stack with the client, example :

Tagging your Cypher statements

Sometimes, you may want to retrieve a specific result from a Stack, an easy way to do this is to tag your Cypher statements.

The tag is passed via the 3rd argument of the run or push methods :

Working with Result sets

Basics

The run method returns you a single Result object. Other methods where you can expect multiple results returns a ResultCollection object which is Traversable.

The Result object contains the records and the summary of the statement, the following methods are available in the API :

Summary

The ResultSummary contains the Statement, the Statistics and the QueryPlan if available :

Record Values

Each record contains one row of values returned by the Cypher query :

The client takes care of the hydration of Graph objects to PHP Objects, so it is for Node, Relationship and Path :

Node

• labels() : returns an array of labels (string)
• identity() : returns the internal ID of the node
• values() : returns the properties of the node (array)
• value($key) : returns the value for the given property key
• hasValue($key) : returns whether or not the nodes has a property with the given key
• keys() : returns you an array representing the keys of the node properties
• hasLabel($label) : returns whether or not the node has the given label (boolean)

Relationship

• type() : returns the relationship type
• identity() : returns the internal ID of the relationship
• values() : returns the properties of the relationship (array)
• value($key) : returns the value for the given property key
• hasValue($key) : returns whether or not the relationship has a property with the given key
• keys() : returns you an array representing the keys of the relationship properties
• startNodeIdentity : returns the start node id
• endNodeIdentity : returns the end node id

Path

• start() : returns the start node of the path
• end() : returns the end node of the path
• length() : returns the length of the path
• nodes() : returns all the nodes in the path
• relationships : returns all the relationships in the path

Handling Results (from v3 to v4)

There are 3 main concepts around this topic :

1. a Result
2. a Record
3. a RecordValue

Let's take a look at a query we do in the browser containing multiple possibilities of types :

Result

A Result is a collection of Record objects, every row you see in the browser is a Record and contains Record Values.

• In blue the Result
• In orange a Record
• In green a RecordValue

Record

In contrary to the previous versions of the client, there is no more automatic merging of all the records into one big record, so you will need to iterate all the records from the Result :

Record Value

Every record contains a collection of Record Values, which are identified by a key, the key is the identifier you give in the RETURN clause of the Cypher query. In our example, a Record will contain the following keys :

• addr
• n
• ids

In order to access the value, you make use of the get() method on the Record object :

The type of the value is depending of what you return from Neo4j, in our case the following values will be returned :

• a string for the addr value
• a Node for the n value
• an array of integers for the ids value

Meaning that :

Node, Relationship and Path objects have then further methods, so if you know that the node returned by the identifier n has a countries property on it, you can access it like this :

The Record object contains three methods for IDE friendlyness, namely :

This does not offer something extra, just that the docblocks hint the IDE for autocompletion.

Extra: ResultCollection

When you use Stack objects for sending multiple statements at once, it will return you a ResultCollection object containing a collection of Results. So you need to iterate the results before accessing the records.

Working with Transactions

The Client provides a Transaction object that ease how you would work with transactions.

Creating a Transaction

At this stage, nothing has been sent to the server yet (the statement BEGIN has not been sent), this permits to stack queries or Stack objects before commiting them.

Stack a query

Again, until now nothing has been sent.

Run a query in a Transaction

Sometimes you want to get an immediate result of a statement inside the transaction, this can be done with the run method :

If the transaction has not yet begun, the BEGIN of the transaction will be done automatically.

You can also push or run Stacks

Commit and Rollback

if you have queued statements in your transaction (those added with the push methods) and you have finish your job, you can commit the transaction and receive the results :

After a commit, you will not be able to push or run statements in this transaction.

Working with multiple connections

Generally speaking, you would better use HAProxy for running Neo4j in a cluster environment. However sometimes it makes sense to have full control to which instance you send your statements.

Let's assume a environment with 3 neo4j nodes :

By default, the $client->run() command will send your Cypher statements to the first registered connection in the list.

You can specify to which connection to send the statement by specifying its alias as 4th argument to the run parameter :

The client is also aware of the manually configured master connection, so sending your writes can be easier with :

Helper Methods

Returns an array of Label objects.

Event Dispatching

3 types of events are dispatched during the run methods :

• PreRunEvent : before the statement or stack is run.
• PostRunEvent : after the statement or stack is run.
• FailureEvent : in case of failure, you can disable the client to throw an exception with this event.

Registering listeners

Example :

The event dispatcher is available via the client with the $client->getEventDispatcher methods.

Settings

Timeout (deprecated)

You can configure a global timeout for the connections :

The timeout by default is 5 seconds.

This feature is deprecated and will be removed in version 5. See Http client settings below.

TLS

You can enable TLS encryption for the Bolt Protocol by passing a Configuration instance when building the connection, here is a simple example :

HTTP client settings

We use HTTPlug to give you full control of the HTTP client. Version 4 of the Neo4jClient comes with Guzzle6 by default to preserve backward compatibility. Version 5 will give you the option to choose whatever client you want. Read more about HTTPlug in their documentation.

To configure your client you may add it to Configuration. Below is an example using php-http/curl-client.

License

The library is released under the MIT License, refer to the LICENSE file bundled with this package.