graphPaname/app/preprocessing.py

21 lines
431 B
Python

from json import load
from pandas import json_normalize, DataFrame
from constants import files, columns
def open_json(dataset):
with open(files[dataset]) as f:
json = load(f)
return json
def create_dataframe(dataset):
json = open_json(dataset)
df = json_normalize(
data=json["records"],
record_path=["fields"],
meta=columns[dataset],
errors="ignore",
)
return df