from math import dist
from os import write
import glob
import os.path
import csv
from unicodedata import name
from google.auth.transport.requests import Request
from google.oauth2.credentials import Credentials
from google_auth_oauthlib.flow import InstalledAppFlow
from googleapiclient.discovery import build
from googleapiclient.errors import HttpError
./result.csv 削除
file_names = glob.glob(‘./result.csv’)
for file_name in file_names:
os.remove(file_name)
If modifying these scopes, delete the file token.json.
SCOPES = [‘https://www.googleapis.com/auth/admin.directory.group.readonly’]
def main():
“””Shows basic usage of the Drive v3 API.
Prints the names and ids of the first 10 files the user has access to.
“””
creds = None
# The file token.json stores the user’s access and refresh tokens, and is
# created automatically when the authorization flow completes for the first
# time.
#token.jsonが実在すればTRUE
if os.path.exists('token.json'):
#Jsonから資格情報を作成
creds = Credentials.from_authorized_user_file('token.json', SCOPES)
# If there are no (valid) credentials available, let the user log in.
if not creds or not creds.valid:
if creds and creds.expired and creds.refresh_token:
creds.refresh(Request())
else:
flow = InstalledAppFlow.from_client_secrets_file(
'client-key_sd-inventory-check.json', SCOPES)
creds = flow.run_local_server(port=0)
# Save the credentials for the next run
with open('token.json', 'w') as token:
token.write(creds.to_json())
try:
#create service
service = build('admin', 'directory_v1', credentials=creds)
cnt = 0
group_name = list()
group_email = list()
#グループ名を全部取得
results = service.groups().list(domain='staging.gaba.co.jp').execute()
results = results.get('groups', [])
while (cnt < len(results)):
groups = results[cnt]
#特定のグループのみリストに入れる
if 'infra' in groups.get('name') or 'ML_sendastest2' in groups.get('name') :
group_name.append(groups.get('name'))
group_email.append(groups.get('email'))
cnt = cnt + 1
#group_emailに入れた内容をキーに読み込んでいく
i = 0
while (i < len(group_email)):
result = service.members().list(groupKey=group_email[i]).execute()
result = result.get('members')
#writing
labels = ['group_name','email','role','type','status','kind', 'etag', 'id']
with open('./result.csv', 'a' ,newline='') as f:
writer = csv.DictWriter(f, fieldnames=labels )
if i == 0:
writer.writeheader()
for elem in result:
elem['group_name']=group_name[i]
writer.writerow(elem)
i=i+1
except HttpError as error:
# TODO(developer) - Handle errors from drive API.
print(f'An error occurred: {error}')
if name == ‘main‘:
main()
コメント