如何从 psycopg2 查询结果有效地创建 networkx 图？

Question

I'm building a networkx graph using data retrieved from a Postgres DB using psycopg2. I can create the graph just fine, but I'm wondering if there is a more efficient and/or Pythonic way to do it.

Current code:

DG = nx.DiGraph()
cur.execute(edgeQuery)
for row in cur:
    self.DG.add_edge(
        row[0],            # fnode
        row[1],            # tnode
        weight=row[3],     # cost
        name=row[4]
    )

The networkx documentation indicates multiple edges can be created at once by feeding in an ebunch . I could create an ebunch by iterating the cursor results, but that wouldn't be any more efficient than my current solution. I feel like there's got to be a more efficient way to translate my cursor results into networkx edges. Maybe something like zip? I'd also like to identify a pythonic way to do it for ease of future maintenance (and to satisfy my own curiosity).

Answer 1

Assuming I understand the data-structure of your "cur" correctly: You should be able to convert it to an array and then slice for a compact notation. For example:

import numpy as np
import networkx as nx

graph = nx.DiGraph()
x= np.array([[1,2,0.5,"o"],[3,4,0.2,"a"]])
graph.add_edges_from(x[:,:2],weight = x[:,2], name = x[:,3])

the add_edges_from() also saves you the for-loop "trouble" :) Like this, the edges are now:

print(graph.edges)

[('1', '2'), ('3', '4')]

I hope this is what you're looking for.

Answer 2

There is a much simpler way using Pandas.

import pandas as pd
import networkx as nx
import psycopg2

*** define psql variables***
con = psycopg2.connect(host=pg_host, port=pg_port, dbname=dbase, user=user, password=pw,options=f"-c search_path={schema}")

sql = "select a,b from nodetable"


df = pd.read_sql(sql,con)

G = nx.Graph()
G = nx.from_pandas_edgelist(df, 'a', 'b')