get_object_list_from_bucket using boto3

boto3_list.py

import json
import re
from io import BytesIO, StringIO
from typing import List

import requests
import bs4
import boto3

s3 = boto3.client('s3')
prefix = '/asdf/123'
REG = re.compile('^([\w]+/){3}data.json')
res_lis = []


def get_datajson_files(response) -> List[str]:
    if not response['IsTruncated']:
        return []

    token = response['NextContinuationToken']
    top_level_datajson_list = [obj['Key'] for obj in response['Contents'] if REG.search(obj['Key'])]
    print(top_level_datajson_list)

    if len(top_level_datajson_list) >= 1:
        res_lis.extend(top_level_datajson_list)

    return get_datajson_files(s3.list_objects_v2(Bucket='bucket.asdf.zxcv', StartAfter=prefix, ContinuationToken=token))

objs = s3.list_objects_v2(Bucket='cdn.zipview.kr', StartAfter=prefix)
    # get_datajson_files(objs)