skip to navigation
skip to content

cypher_kernel 0.2.1

A Cypher kernel for Jupyter

Latest Version: 0.2.3

# Cypher Kernel

This is a small Jupyter kernel wrapping the Cypher language and Neo4j [](


## Why? Do I need a Cypher language kernel?

I started working on this kernel as I am teaching on Cypher and Neo4j as part of a database course on Copenhagen Business Academy.

Usually, my lecture notes are in Jupyter notebooks and I use the Jupter extension [RISE]( to make slideshows with executable code.

In the classes on graph databases, I want to focus on the query language Cypher. That is, I do not want to have Cypher embedded in Python such as with [icypher]( -a `%cypher` magic sending queries with the help of `py2neo`- or in [plain Python notebooks](

## How does it look like?

In case you do not have a Python environment with Jupyter installed on your system, you can see in the following how the rendered notebooks -not the presentations- could look like, see:

* or

## Installation

To install the `cypher_kernel` from PyPi:

pip install cypher_kernel
python -m cypher_kernel.install

To work on this code directly, you may want to:

git clone
cd cypher_kernel
python install
pip install .
python -m cypher_kernel.install

## Configuration

To configure a Neo4j user, password, and the address of the Neo4j REST API, you can specify the values in the configuration file `cypher_config.yml`. Normally, this file is located under `~/.jupyter/`. In case the file is not existent the following default configuration is used:

user: 'neo4j'
pwd: 'neo4j'
host: 'localhost:7474'
connect_result_nodes: False
cmd_timeout: null

## Using the Cypher Kernel

**Notebook**: The *New* menu in the notebook should show an option for an `Cypher` notebook.

**Console frontends**: To use it with the console frontends, add `--kernel cypher` to their command line arguments.

## Neo4j for Presentations

To get quickly started -under the assumption you have Docker installed- start up a Neo4j DBMS instance with:

docker run --rm --publish=7474:7474 --publish=7687:7687 neo4j
docker run --rm --publish=7474:7474 --publish=7687:7687 --env=NEO4J_dbms_memory_pagecache_size=4G neo4j

* Navigate with your browser to http://localhost:7474
* Login with `neo4j` as username and password respectively
* Change the password to a new one. **OBS** Do not forget to add this password to the `cypher_config.yml`, see above.
* Now, create a new Cypher notebook.

See more on configuring the Neo4j container

### What? I have Docker but no `pip` and other Python stuff?!

Likely the easiest way to get started, have:

* A Linux/OS X (Windows should work but I cannot test it at the moment...)
* An installation of Anaconda (with Python 3.6), see Download and install it according to their documentation, see
* A Docker installation, see Alternatively, a native installation of Neo4j, see
* The `cypher_kernel`:
pip install cypher_kernel
python -m cypher_kernel.install
* The Jupyter Notebook server up and running:
jupyter notebook

That should be it...

File Type Py Version Uploaded on Size
cypher_kernel-0.2.1.tar.gz (md5) Source 2018-03-13 2MB