import json

import numpy as np

import pickle

from scipy import sparse

from sknetwork.data import Bunch, from_adjacency_list

def load_json_dataset(inpath: str, filename: str):

"""Load JSON data.

Parameters

----------

inpath: str

Path to data

filename: str

File name"""

with open(f'{inpath}/{filename}.json') as f:

data = json.load(f)

return data

if __name__ == '__main__':

inpath = 'data'

filename = 'londonGraph'

dataset = Bunch()

meta = Bunch()

nodelabel2idx = dict()

# Load data

data = load_json_dataset(inpath, filename)

# Metadata

meta.name = data.get('descriptorName')

meta.description = ''

meta.source = ''

meta.date = 'February 2023'

# Attributes information

names = []

names_col = np.asarray(data.get('attributesName'))

vertices = data.get('vertices')

biadjacency_dense = np.zeros((len(vertices), len(names_col)))

print(f'Initial number of vertices: {len(vertices)}')

for i, v in enumerate(vertices):

nodelabel2idx[v.get('vertexId')] = i

biadjacency_dense[i, :] = np.asarray(v.get('descriptorsValues'))

names.append(v.get('vertexId'))

biadjacency = sparse.csr_matrix(biadjacency_dense)

print(f'Biadjacency shape: {biadjacency.shape}')

# Vertex information

edges = data.get('edges')

adj_list = {}

for i, e in enumerate(edges):

adj_list[nodelabel2idx.get(e.get('vertexId'))] = [nodelabel2idx.get(x) for x in e.get('connected_vertices')]

adjacency = from_adjacency_list(adj_list, directed=False, reindex=False)

print(f'Adjacency: {adjacency.shape}, #edges: {adjacency.nnz}')

dataset.adjacency = adjacency

dataset.names = np.asarray(names)

dataset.biadjacency = biadjacency

dataset.names_col = names_col

dataset.meta = meta

# Save dataset

with open(f'{inpath}/{filename}', 'bw') as f:

pickle.dump(dataset, f)