owner_report.py

import json
import time
import pandas as pd

def get_entry(member, role, project):
    parts = member.split(':')
    type = parts[0]
    email = parts[1]

    entry = {
        'project': project,
        'role': role,
        'type': type,
        'email': email
    }

    if project['parent']['type'] == 'folder':
        parent_id = project['parent']['id']
        folder = folders.get(parent_id)
        entry['project']['parent']['folder'] = folder['displayName']
        entry['project']['parent']['path'] = folder['path']

    return entry


projects_by_user = {}
entries = []
folders = {}

df = pd.DataFrame()

with open('folders.json', 'r') as jsonfile:
    folders = json.load(jsonfile)

with open('projects.json', 'r') as jsonfile:
    projects = json.load(jsonfile)

    for project_id in projects.keys():
        proj = projects[project_id]

        try:
            for binding in proj['iam_policy']['bindings']:
                for member in binding['members']:
                    entry = get_entry(member=member,
                                    role=binding['role'],
                                    project=proj['project'])

                    
                    entry_df = pd.json_normalize(entry)
                    entries.append(entry_df)
                    
                    #local_part = entry['email'].split('@')[0].lower()
                    #if not projects_by_user.get(local_part):
                    #    projects_by_user[local_part] = []

                    #projects_by_user[local_part].append(entry)
        except:
            pass

#print(json.dumps(projects_by_user))
df = df.append(other=entries)

# get rid of the .'s in the column names created by json_normalize
df.columns = df.columns.str.replace(r".", "_")
df.columns = df.columns.str.replace("-", "_")

# add the date of the audit so we can create a time series
df['audit_time'] = pd.Timestamp.now().isoformat()

# convert all field values to string type
df = df.astype(str)

# workaround for pandas v1.1.1, due to the fact that astype(str) will convert a np.nan to the literal string 'nan'....
# so we'll just flip it back to a none type....
df = df.replace(['nan'], [None])

#output to row delimited json
df.to_json(path_or_buf='owners_nldj.json',orient='records', lines=True, date_format='iso')