NebulaGraph Python Client

Getting Started

Note: Ensure you are using the correct version, refer to the Capability Matrix for how the Python client version corresponds to the NebulaGraph Database version.

Accessing NebulaGraph

• Get Started Notebook - A Jupyter Notebook to get started with NebulaGraph Python client, with latest features and examples.

• For first-time trying out Python client, go through Quick Example: Connecting to GraphD Using Graph Client.

• If your Graph Application is a Web Service dedicated to one Graph Space, go with Singleton of Session Pool, check Using the Session Pool: A Guide.

• If you're building Graph Analysis Tools(Scan instead of Query), you may want to use the Storage Client to scan vertices and edges, see Quick Example: Using Storage Client to Scan Vertices and Edges.

• For parameterized query, see Example: Server-Side Evaluated Parameters.

Handling Query Results

• On how to form a query result into a Pandas DataFrame, see Example: Fetching Query Results into a Pandas DataFrame.

• On how to render/visualize the query result, see Example: Extracting Edge and Vertex Lists from Query Results, it demonstrates how to extract lists of edges and vertices from any query result by utilizing the ResultSet.dict_for_vis() method.

• On how to get rows of dict/JSON structure with primitive types, see Example: Retrieve Primitive Typed Results.

Jupyter Notebook Integration

If you are about to access NebulaGraph within Jupyter Notebook, you may want to use the NebulaGraph Jupyter Extension, which provides a more interactive way to access NebulaGraph. See also this on Google Colab: NebulaGraph on Google Colab.

Obtaining nebula3-python

Method 1: Installation via pip

# for v3.x
pip install nebula3-python==$version
# for v2.x
pip install nebula2-python==$version

Method 2: Installation via source

Click to expand

• Clone from GitHub

git clone https://github.com/vesoft-inc/nebula-python.git
cd nebula-python

• Install from source

For python version >= 3.7.0

pip install .

For python version >= 3.6.2, < 3.7.0

python3 setup.py install

Quick Example: Connecting to GraphD Using Graph Client

from nebula3.gclient.net import ConnectionPool
from nebula3.Config import Config

# define a config
config = Config()
config.max_connection_pool_size = 10
# init connection pool
connection_pool = ConnectionPool()
# if the given servers are ok, return true, else return false
ok = connection_pool.init([('127.0.0.1', 9669)], config)

# option 1 control the connection release yourself
# get session from the pool
session = connection_pool.get_session('root', 'nebula')

# select space
session.execute('USE basketballplayer')

# show tags
result = session.execute('SHOW TAGS')
print(result)

# release session
session.release()

# option 2 with session_context, session will be released automatically
with connection_pool.session_context('root', 'nebula') as session:
    session.execute('USE basketballplayer')
    result = session.execute('SHOW TAGS')
    print(result)

# close the pool
connection_pool.close()

Using the Session Pool: A Guide

The session pool is a collection of sessions that are managed by the pool. It is designed to improve the efficiency of session management and to reduce the overhead of session creation and destruction.

Session Pool comes with the following assumptions:

1. A space must already exist in the database prior to the initialization of the session pool.
2. Each session pool is associated with a single user and a single space to ensure consistent access control for the user. For instance, a user may possess different access permissions across various spaces. To execute queries in multiple spaces, consider utilizing several session pools.
3. Whenever sessionPool.execute() is invoked, the session executes the query within the space specified in the session pool configuration.
4. It is imperative to avoid executing commands through the session pool that would alter passwords or remove users.

For more details, see SessionPoolExample.py.

Example: Server-Side Evaluated Parameters

To enable parameterization of the query, refer to the following example:

Note: Not all tokens of a query can be parameterized. You can quickly verify it via iPython or Nebula-Console in an interactive way.

params = {
    "p1": 3,
    "p2": True,
    "p3": "Bob",
    "ids": ["player100", "player101"], # second query
}

resp = client.execute_py(
    "RETURN abs($p1)+3 AS col1, (toBoolean($p2) and false) AS col2, toLower($p3)+1 AS col3",
    params,
)
resp = client.execute_py(
    "MATCH (v) WHERE id(v) in $ids RETURN id(v) AS vertex_id",
    params,
)

For further information, consult Params.py.

Example: Extracting Edge and Vertex Lists from Query Results

For graph visualization purposes, the following code snippet demonstrates how to effortlessly extract lists of edges and vertices from any query result by utilizing the ResultSet.dict_for_vis() method.

result = session.execute(
    'GET SUBGRAPH WITH PROP 2 STEPS FROM "player101" YIELD VERTICES AS nodes, EDGES AS relationships;')

data_for_vis = result.dict_for_vis()

Then, we could pass the data_for_vis to a front-end visualization library such as vis.js, d3.js or Apache ECharts. There is an example of Apache ECharts in exapmple/apache_echarts.html.

The dict/JSON structure with dict_for_vis() is as follows:

Click to expand

{
    'nodes': [
        {
            'id': 'player100',
            'labels': ['player'],
            'props': {
                'name': 'Tim Duncan',
                'age': '42',
                'id': 'player100'
            }
        },
        {
            'id': 'player101',
            'labels': ['player'],
            'props': {
                'age': '36',
                'name': 'Tony Parker',
                'id': 'player101'
            }
        }
    ],
    'edges': [
        {
            'src': 'player100',
            'dst': 'player101',
            'name': 'follow',
            'props': {
                'degree': '95'
            }
        }
    ],
    'nodes_dict': {
        'player100': {
            'id': 'player100',
            'labels': ['player'],
            'props': {
                'name': 'Tim Duncan',
                'age': '42',
                'id': 'player100'
            }
        },
        'player101': {
            'id': 'player101',
            'labels': ['player'],
            'props': {
                'age': '36',
                'name': 'Tony Parker',
                'id': 'player101'
            }
        }
    },
    'edges_dict': {
        ('player100', 'player101', 0, 'follow'): {
            'src': 'player100',
            'dst': 'player101',
            'name': 'follow',
            'props': {
                'degree': '95'
            }
        }
    },
    'nodes_count': 2,
    'edges_count': 1
}

Example: Retrieve Primitive Typed Results

The executed result is typed as ResultSet, and you can inspect its structure using dir().

For each data cell in the ResultSet, you can use .cast() to retrieve raw wrapped data (with sugar) such as a Vertex (Node), Edge (Relationship), Path, Value (Int, Float, etc.). Alternatively, you can use .cast_primitive() to obtain values in primitive types like dict, int, or float, depending on your needs.

For more details, refer to FromResp.py.

Additionally, ResultSet.as_primitive() provides a convenient method to convert the result set into a list of dictionaries (similar to JSONL format) containing primitive values for each row.

result = session.execute('<your query>')

result_dict = result.as_primitive()
print(result_dict)

Example: Fetching Query Results into a Pandas DataFrame

For nebula3-python>=3.6.0:

Assuming you have pandas installed, you can use the following code to fetch query results into a pandas DataFrame:

pip3 install pandas

result = session.execute('<your query>')
df = result.as_data_frame()

For `nebula3-python<3.6.0`:

from nebula3.gclient.net import ConnectionPool
from nebula3.Config import Config
import pandas as pd
from typing import Dict
from nebula3.data.ResultSet import ResultSet

def result_to_df(result: ResultSet) -> pd.DataFrame:
    """
    build list for each column, and transform to dataframe
    """
    assert result.is_succeeded()
    columns = result.keys()
    d: Dict[str, list] = {}
    for col_num in range(result.col_size()):
        col_name = columns[col_num]
        col_list = result.column_values(col_name)
        d[col_name] = [x.cast() for x in col_list]
    return pd.DataFrame(d)

# define a config
config = Config()

# init connection pool
connection_pool = ConnectionPool()

# if the given servers are ok, return true, else return false
ok = connection_pool.init([('127.0.0.1', 9669)], config)

# option 2 with session_context, session will be released automatically
with connection_pool.session_context('root', 'nebula') as session:
    session.execute('USE <your graph space>')
    result = session.execute('<your query>')
    df = result_to_df(result)
    print(df)

# close the pool
connection_pool.close()

Quick Example: Using Storage Client to Scan Vertices and Edges

Storage Client enables you to scan vertices and edges from the storage service instead of the graph service w/ nGQL/Cypher. This is useful when you need to scan a large amount of data.

Click to expand

You should make sure the scan client can connect to the address of storage which see from SHOW HOSTS

from nebula3.mclient import MetaCache, HostAddr
from nebula3.sclient.GraphStorageClient import GraphStorageClient

# the metad servers's address
meta_cache = MetaCache([('172.28.1.1', 9559),
                        ('172.28.1.2', 9559),
                        ('172.28.1.3', 9559)],
                       50000)

# option 1 metad usually discover the storage address automatically
graph_storage_client = GraphStorageClient(meta_cache)

# option 2 manually specify the storage address
storage_addrs = [HostAddr(host='172.28.1.4', port=9779),
                 HostAddr(host='172.28.1.5', port=9779),
                 HostAddr(host='172.28.1.6', port=9779)]
graph_storage_client = GraphStorageClient(meta_cache, storage_addrs)

resp = graph_storage_client.scan_vertex(
        space_name='ScanSpace',
        tag_name='person')
while resp.has_next():
    result = resp.next()
    for vertex_data in result:
        print(vertex_data)

resp = graph_storage_client.scan_edge(
    space_name='ScanSpace',
    edge_name='friend')
while resp.has_next():
    result = resp.next()
    for edge_data in result:
        print(edge_data)

See ScanVertexEdgeExample.py for more details.

Compatibility Matrix

Nebula-Python Version	Compatible NebulaGraph Versions	Notes
3.8.2	3.x	Highly recommended. Latest release for NebulaGraph 3.x series.
master	master	Includes recent changes. Not yet released.
3.0.0 ~ 3.5.1	3.x	Compatible with any released version within the NebulaGraph 3.x series.
2.6.0	2.6.0, 2.6.1
2.5.0	2.5.0
2.0.0	2.0.0, 2.0.1
1.0	1.x

Directory Structure Overview

.
└──nebula-python
    │
    ├── nebula3                               // client source code
    │   ├── fbthrift                          // the RPC code generated from thrift protocol
    │   ├── common
    │   ├── data
    │   ├── graph
    │   ├── meta
    │   ├── net                               // the net code for graph client
    │   ├── storage                           // the storage client code
    │   ├── Config.py                         // the pool config
    │   └── Exception.py                      // the exceptions
    │
    ├── examples
    │   ├── FormatResp.py                     // the format response example
    │   ├── SessionPoolExample.py             // the session pool example
    │   ├── GraphClientMultiThreadExample.py  // the multi thread example
    │   ├── GraphClientSimpleExample.py       // the simple example
    │   └── ScanVertexEdgeExample.py          // the scan vertex and edge example(storage client)
    │
    ├── tests                                 // the test code
    │
    ├── setup.py                              // used to install or package
    │
    └── README.md                             // the introduction of nebula3-python

Contribute to Nebula-Python

Click to expand

To contribute, start by forking the repository. Next, clone your forked repository to your local machine. Remember to substitute {username} with your actual GitHub username in the URL below:

git clone https://github.com/{username}/nebula-python.git
cd nebula-python

For package management, we utilize PDM. Please begin by installing it:

pipx install pdm

Visit the PDM documentation for alternative installation methods.

Install the package and all dev dependencies:

pdm install

Make sure the Nebula server in running, then run the tests with pytest:

pdm test

Using the default formatter with black.

Please run pdm fmt to format python code before submitting.

See How to contribute for the general process of contributing to Nebula projects.