github_org_control: Update to the latest version (#9322)

This commit is contained in:
Alexander Zhogov authored and GitHub committed 2021-12-20 21:04:28 +03:00
1 parent 46c9755b6e
commit c10a27f601
15 files changed
+672 -499

No files matched your search

Whitespace-only changes.
+139
View File
@@ -0,0 +1,139 @@
# Copyright (C) 2018-2021 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
"""
Check GitHub organization and invite members
"""
# pylint: disable=fixme,no-member,too-many-locals
import sys
from pathlib import Path
from argparse import ArgumentParser
sys.path.append(str(Path(__file__).resolve().parents[1]))
from github_org_control.configs import Config
from github_org_control.github_api import GithubOrgApi, get_dev_emails, print_users
from github_org_control.ldap_api import LdapApi, print_user_info, InfoLevel
def remove_members(gh_api, cfg_emails, org_emails, dev_emails, org_emails_no_in_ldap):
"""Checks and remove members"""
print(
f"\n{'=' * 10} Check accounts below and remove from the GitHub organization or "
f"configuration {'=' * 10}"
)
cfg_emails_no_in_org = sorted(cfg_emails.difference(org_emails))
print(
f"\nCfg developer emails - absent in GitHub organization {len(cfg_emails_no_in_org)}:",
"; ".join(cfg_emails_no_in_org),
)
non_member_ignored_logins = set(Config().IGNORE_LOGINS).difference(
set(gh_api.org_members_by_login.keys())
)
print(
f"\nIgnored logins - absent in GitHub organization {len(non_member_ignored_logins)}:\n",
"\n".join(non_member_ignored_logins),
)
org_emails_no_in_dev = sorted(org_emails.difference(dev_emails))
print(
f"\nOrg member emails - absent in cfg and LDAP PDLs {len(org_emails_no_in_dev)}:",
"; ".join(org_emails_no_in_dev),
)
print(
f"\nOrg member emails - absent in LDAP at all {len(org_emails_no_in_ldap)}:",
"; ".join(sorted(org_emails_no_in_ldap)),
)
print("\nOrg members - no real name:")
members_to_fix_name = sorted(gh_api.members_to_fix_name, key=lambda member: member.email)
print_users(members_to_fix_name)
print(
"\nOrg member emails - no real name:",
"; ".join([member.email.lower() for member in members_to_fix_name]),
)
print("\nOrg members - no Intel emails:")
print_users(gh_api.members_to_remove)
gh_api.remove_users(org_emails_no_in_ldap | gh_api.members_to_remove)
def main():
"""The main entry point function"""
arg_parser = ArgumentParser()
arg_parser.add_argument(
"--cfg-file",
metavar="PATH",
default=Config.default_cfg_path,
help=f"Path to json configuration file, e.g. {Config.default_cfg_path}",
)
arg_parser.add_argument("--teams", action="store_true", help="Check GitHub teams")
arg_parser.add_argument("--no-ldap", action="store_true", help="Don't use LDAP info")
args, unknown_args = arg_parser.parse_known_args()
Config(args.cfg_file, unknown_args)
gh_api = GithubOrgApi()
if args.teams:
gh_api.get_org_teams()
return
cfg_emails = get_dev_emails()
print(f"\nCfg developer emails {len(cfg_emails)}:", "; ".join(sorted(cfg_emails)))
dev_emails = set()
dev_emails.update(cfg_emails)
if not args.no_ldap:
ldap_api = LdapApi()
ldap_emails = ldap_api.get_user_emails()
dev_emails.update(ldap_emails)
print(f"\nLDAP developer emails {len(ldap_emails)}:", "; ".join(sorted(ldap_emails)))
cfg_emails_no_in_ldap = ldap_api.get_absent_emails(cfg_emails)
print(
f"\nCfg developer emails - absent in LDAP at all {len(cfg_emails_no_in_ldap)}:",
"; ".join(sorted(cfg_emails_no_in_ldap)),
)
cfg_ldap_inters = cfg_emails.intersection(ldap_emails)
print(
f"\nCfg developer emails - present in LDAP developers {len(cfg_ldap_inters)}:",
"; ".join(sorted(cfg_ldap_inters)),
)
org_emails = gh_api.get_org_emails()
print(f"\nOrg emails {len(org_emails)}:", "; ".join(sorted(org_emails)))
org_emails_no_in_ldap = set()
if not args.no_ldap:
org_ldap_diff = org_emails.difference(ldap_emails)
print(
f"\nOrg member emails - absent in LDAP developers {len(org_ldap_diff)}:",
"; ".join(sorted(org_ldap_diff)),
)
for email in org_ldap_diff:
user_info = ldap_api.get_user_info_by_email(email)
if user_info:
print_user_info(user_info, InfoLevel.PDL)
else:
org_emails_no_in_ldap.add(email)
org_pendig_invitation_emails = gh_api.get_org_invitation_emails()
invite_emails = dev_emails.difference(org_emails).difference(org_pendig_invitation_emails)
print(f"\nInvite emails {len(invite_emails)}:", "; ".join(sorted(invite_emails)))
valid_github_users = gh_api.get_valid_github_users(invite_emails)
gh_api.invite_users(valid_github_users)
remove_members(gh_api, cfg_emails, org_emails, dev_emails, org_emails_no_in_ldap)
if __name__ == "__main__":
main()
+261
View File
@@ -0,0 +1,261 @@
# Copyright (C) 2018-2021 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
"""
Check GitHub PRs and set labels by type and categories, e.g. 'ExternalPR', 'category: ci'
"""
# pylint: disable=fixme,no-member
import re
import sys
import datetime
from enum import Enum
from pathlib import Path
from argparse import ArgumentParser
sys.path.append(str(Path(__file__).resolve().parents[1]))
from github_org_control import github_api
from github_org_control.configs import Config
class PrType(Enum):
"""Constants for type of GitHub pull request by author membership"""
EXTERNAL = "ExternalPR"
INTEL = "ExternalIntelPR"
ORG = "OpenvinoPR"
BAD = "BadPR"
def get_pr_labels(pull):
"""Gets PR labels as set"""
pr_lables = set()
for label in pull.labels:
pr_lables.add(label.name)
return pr_lables
def set_pr_labels(pull, labels):
"""Sets new PR labels (all previously set labels are removed)"""
if not labels or Config().DRY_RUN:
return
print("Set PR labels:", labels)
# set_labels() should accept list but fails with empty "AssertionError:"
pull.set_labels(labels)
def add_pr_labels(pull, labels):
"""Adds PR labels"""
if not labels or Config().DRY_RUN:
return
print("Add PR labels:", labels)
for label in labels:
pull.add_to_labels(label)
def get_pr_type_by_labels(pull):
"""Gets PR type using labels"""
pr_lables = get_pr_labels(pull)
pr_types = set(type.value for type in PrType)
pr_types_labels = pr_lables & pr_types
if not pr_types_labels:
return None
if len(pr_types_labels) > 1:
print(f"Duplicated labels: {pr_types_labels}")
return PrType.BAD
return PrType(PrType(pr_types_labels.pop()))
def get_label_by_team_name_re(team_name):
"""Generates label by PR reviwer team name using regular expressions"""
if "admins" in team_name:
return "category: ci"
re_compile_label = re.compile(rf"{Config().GITHUB_REPO}-(.+)-maintainers")
re_label = re_compile_label.match(team_name)
if re_label:
return f"category: {re_label.group(1).strip()}"
return None
def get_label_by_team_name_map(team_name):
"""Generates label by PR reviwer team name using config map"""
return Config().TEAM_TO_LABEL.get(team_name)
def get_category_labels(pull):
"""Gets list of category labels by all PR reviwer teams"""
labels = []
pr_lables = get_pr_labels(pull)
for reviewer_team in pull.get_review_requests()[1]:
reviewer_label = get_label_by_team_name_map(reviewer_team.name)
if reviewer_label and reviewer_label not in pr_lables:
labels.append(reviewer_label)
return labels
def get_pr_info_str(pull):
"""Gets info about PR using a few workarounds"""
pr_title = pull.title.encode("ASCII", "ignore").decode()
# Workaround for PyGithub issue: https://github.com/PyGithub/PyGithub/issues/512
pr_created_at = pull.created_at.replace(tzinfo=datetime.timezone.utc).astimezone()
return (
f"PR: {pull.number} - {pr_title} - Created: {pr_created_at} - "
f"Labels: {get_pr_labels(pull)} - Type: {get_pr_type_by_labels(pull)}"
)
def update_labels(gh_api, pull, non_org_intel_pr_users, non_org_pr_users):
"""Checks and updates labels"""
print("Check and update labels:")
pr_type_by_labels = get_pr_type_by_labels(pull)
add_labels = []
# Checks PR source type
if gh_api.is_org_user(pull.user):
print(" - Org user")
elif github_api.is_intel_email(pull.user.email) or github_api.is_intel_company(
pull.user.company
):
print(" - Non org user with Intel email or company")
non_org_intel_pr_users.add(pull.user)
if pr_type_by_labels is not PrType.INTEL:
print(f'NO "{PrType.INTEL.value}" label: ', end="")
github_api.print_users(pull.user)
add_labels.append(PrType.INTEL.value)
elif github_api.is_user_ignored(pull.user):
print(" - IGNORED non org user with NO Intel email or company")
else:
print(" - Non org user with NO Intel email or company")
non_org_pr_users.add(pull.user)
if pr_type_by_labels is not PrType.EXTERNAL:
print(f'NO "{PrType.EXTERNAL.value}" label: ', end="")
github_api.print_users(pull.user)
add_labels.append(PrType.EXTERNAL.value)
add_labels += get_category_labels(pull)
add_pr_labels(pull, add_labels)
def get_wrong_commits(pull):
"""Returns commits with incorrect user and email"""
pr_author_email = (pull.user.email or "").lower()
print("GitHub PR author email:", pr_author_email)
print("Check commits:")
wrong_commits = set()
for commit in pull.get_commits():
# import pprint; pprint.pprint(commit.raw_data)
print("Commit SHA:", commit.sha)
# Use raw data because commit author can be non GitHub user
commit_author_email = (commit.raw_data["commit"]["author"]["email"] or "").lower()
commit_committer_email = (commit.raw_data["commit"]["committer"]["email"] or "").lower()
print(" Commit author email:", commit_author_email)
print(" Commit committer email:", commit_committer_email)
if not github_api.is_valid_user(commit.author):
print(
" ERROR: User with the commit author email is absent in GitHub:",
commit.raw_data["commit"]["author"]["name"],
)
wrong_commits.add(commit.sha)
if not github_api.is_valid_user(commit.committer):
print(
" ERROR: User with the commit committer email is absent in GitHub:",
commit.raw_data["commit"]["committer"]["name"],
)
wrong_commits.add(commit.sha)
if not commit.raw_data["commit"]["verification"]["verified"]:
print(
" WARNING: The commit is not verified. Reason:",
commit.raw_data["commit"]["verification"]["reason"],
)
if pr_author_email != commit_author_email or pr_author_email != commit_committer_email:
print(" WARNING: Commit emails and GitHub PR author public email are differnt")
return wrong_commits
def main():
"""The main entry point function"""
arg_parser = ArgumentParser()
arg_parser.add_argument(
"--cfg-file",
metavar="PATH",
default=Config.default_cfg_path,
help=f"Path to json configuration file, e.g. {Config.default_cfg_path}",
)
arg_parser.add_argument(
"--pr", metavar="NUMBER", help="Get GitHub pull request with the number"
)
arg_parser.add_argument(
"--pr-state",
default="open",
choices=["open", "closed"],
help="Set GitHub pull request state",
)
arg_parser.add_argument(
"--newer", metavar="MINUTES", help="Get newly created GitHub pull request only"
)
arg_parser.add_argument(
"--check-commits",
action="store_true",
help="Check and compare git commit email with GitHub account email",
)
args, unknown_args = arg_parser.parse_known_args()
Config(args.cfg_file, unknown_args)
gh_api = github_api.GithubOrgApi()
if args.pr:
pulls = [gh_api.repo.get_pull(int(args.pr))]
else:
pulls = gh_api.repo.get_pulls(state=args.pr_state)
print(f"\nPRs count ({args.pr_state}):", pulls.totalCount)
if args.newer:
pr_created_after = (
datetime.datetime.now() - datetime.timedelta(minutes=int(args.newer))
).astimezone()
print("Checking PRs created after:", pr_created_after)
non_org_intel_pr_users = set()
non_org_pr_users = set()
wrong_pulls = {}
for pull in pulls:
pr_created_at = pull.created_at.replace(tzinfo=datetime.timezone.utc).astimezone()
if args.newer and pr_created_at <= pr_created_after:
print(f"\nIGNORE: {get_pr_info_str(pull)}")
continue
print(f"\n{get_pr_info_str(pull)}")
if args.check_commits:
wrong_commits = get_wrong_commits(pull)
if wrong_commits:
wrong_pulls[pull.number] = wrong_commits
else:
update_labels(gh_api, pull, non_org_intel_pr_users, non_org_pr_users)
if wrong_pulls:
for pull_number, wrong_commits in wrong_pulls.items():
print(
f"\nERROR: Remove or replace wrong commits in the PR {pull_number}:\n ",
"\n ".join(wrong_commits),
)
print(
"\nAbout commit signature verification:\n ",
"https://docs.github.com/en/github/authenticating-to-github/"
"managing-commit-signature-verification/about-commit-signature-verification",
)
sys.exit(1)
if non_org_intel_pr_users:
print("\nNon org user with Intel email or company:")
github_api.print_users(non_org_intel_pr_users)
if non_org_pr_users:
print("\nNon org user with NO Intel email or company:")
github_api.print_users(non_org_pr_users)
if __name__ == "__main__":
main()
+45
View File
@@ -0,0 +1,45 @@
{
"GITHUB_TOKEN": "<Put token here or set as arg or as env variable>",
"GITHUB_ORGANIZATION": "openvinotoolkit",
"GITHUB_REPO": "openvino",
"IGNORE_LOGINS": [
"openvino-ci",
"openvino-pushbot",
"lab-nerval",
"lab-nerval-onnx-ci",
"onnx-watchdog-agent",
"workbench-ci-bot",
"openvino-pot-ci"
],
"MAX_MEMBERS_TO_REMOVE": 15,
"EMAILS_FILE_PATH": "dev_emails-test.txt",
"PROXIES": {
"HTTP_PROXY": null,
"HTTPS_PROXY": null,
"NO_PROXY": "localhost,127.0.0.1,.intel.com"
},
"DRY_RUN": false,
"TEAM_TO_LABEL": {
"openvino-admins": "category: CI",
"openvino-maintainers": "category: IE common",
"openvino-docs-maintainers": "category: docs",
"openvino-ie-maintainers": "category: IE common",
"openvino-ie-cpu-maintainers": "category: CPU",
"openvino-ie-gna-maintainers": "category: GNA",
"openvino-ie-gpu-maintainers": "category: GPU",
"openvino-ie-lpt-maintainers": "category: LP transformations",
"openvino-ie-multi-maintainers": "category: MULTI",
"openvino-ie-python-api-maintainers": "category: python api",
"openvino-ie-template-maintainers": "category: TEMPLATE",
"openvino-ie-tests-maintainers": "category: IE Tests",
"openvino-ie-vpu-maintainers": "category: VPU",
"openvino-mo-maintainers": "category: MO",
"openvino-ngraph-maintainers": "category: nGraph",
"openvino-scripts-maintainers": "category: build",
"openvino-tests-maintainers": "category: IE Tests",
"openvino-tools-maintainers": "category: tools",
"openvino-pot-maintainers": "category: POT",
"openvino-configuration-mgmt": "category: dependency_changes",
"openvino-samples-maintainers": "category: samples"
}
}
+120
View File
@@ -0,0 +1,120 @@
# Copyright (C) 2018-2021 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
"""
Configurations management
"""
# pylint: disable=fixme,broad-except
import os
import sys
import ast
import json
from pathlib import Path
if sys.hexversion < 0x3060000:
raise Exception("Python version must be >= 3.6")
class ConfigException(Exception):
"""Base configuration exception"""
class Config:
"""Configuration wrapper"""
_instance = None
_properties = None
default_cfg_path = Path(__file__).resolve().parent / "config.json"
def __new__(cls, *_args, **_kwargs):
if not Config._instance:
Config._instance = super(Config, cls).__new__(cls)
return Config._instance
def __init__(self, file_path=None, cli_args=None):
"""
:param file_path: Path to json configuration file
:type file_path: String
:param args: List of argparse arguments with patterns: 'name=value' or 'name'
:type args: list
"""
if Config._properties:
return
self._file_path = file_path or Config.default_cfg_path
self._cli_args = cli_args or []
self._json_cfg = {}
self._args = {}
self._load_cfg()
self._parse_cli_args()
Config._properties = {}
for name, value in self._json_cfg.items():
if hasattr(self, name):
raise ConfigException(f"Duplicating prosperity: {name}")
property_value = self._args.get(name) or os.getenv(name)
if property_value:
# Try to set prosperity_value as Python literal structures, e.g. DRY_RUN=False
try:
property_value = ast.literal_eval(property_value)
except Exception:
pass
if not isinstance(property_value, type(value)):
raise ConfigException(f"Python type of {name} parameter must be {type(value)}")
else:
property_value = value
Config._properties[name] = property_value
self.set_proxy()
def __getattr__(self, attr_name):
if attr_name in self._properties:
return self._properties.get(attr_name)
raise AttributeError(f"'{self.__class__.__name__}' object has no attribute '{attr_name}'")
def _load_cfg(self):
"""Load the json configuration file"""
try:
with open(self._file_path, encoding="utf-8") as conf:
self._json_cfg = json.load(conf)
except Exception as exc:
raise ConfigException("Failed to load configuration from:", self._file_path) from exc
def _parse_cli_args(self):
"""Parse argparse arguments with patterns: 'name=value' or 'name'"""
for cli_arg in self._cli_args:
arg = cli_arg.split("=")
if arg[0] not in self._json_cfg:
raise ConfigException(f"Unsupported argument: {arg}")
self._args[arg[0]] = True if len(arg) == 1 else "=".join(arg[1:])
@property
def properties(self):
"""Get all properties as Dict"""
return self._properties
def set_proxy(self):
"""Set proxies"""
for proxy_name, url in self._properties["PROXIES"].items():
if url is not None:
print(f"Set proxy: {proxy_name}={url}")
os.environ[proxy_name] = url
def _test():
"""Test and debug"""
print("Config.default_cfg_path:", Config.default_cfg_path)
cfg = Config(cli_args=["DRY_RUN", 'PROXIES={"NO_PROXY": "localhost"}'])
print("Config.properties:", cfg.properties)
print("cfg.PROXIES:", cfg.PROXIES)
if __name__ == "__main__":
_test()
@@ -0,0 +1,9 @@
# good comment
Last_name, First_name <first_name.last_name@intel.com>
first_name.last_name@intel.com
openvino_pushbot@intel.com
# Wrong emails
foo@foo.com
foo1 foo2
foo1 foo2@intel.com
+384
View File
@@ -0,0 +1,384 @@
# Copyright (C) 2018-2021 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
"""
GitHub API for controlling organization
"""
# pylint: disable=fixme,no-member
import re
import sys
import time
import typing
from pathlib import Path
from github import Github, GithubException, RateLimitExceededException, IncompletableObject
from github.PaginatedList import PaginatedList
sys.path.append(str(Path(__file__).resolve().parents[1]))
from github_org_control.configs import Config
class GithubApiException(Exception):
"""Base GitHub API exception"""
def is_valid_user(user):
"""Checks that user is valid github.Github object"""
try:
return user and user.login
except IncompletableObject:
return False
def is_user_ignored(user):
"""Checks that user should be ignored"""
if is_valid_user(user) and user.login.lower() not in Config().IGNORE_LOGINS:
return False
return True
def is_valid_name(name):
"""Checks that GitHub user's name is valid"""
return name and len(name) >= 3 and " " in name
def is_intel_email(email):
"""Checks that email is valid Intel email"""
return email and len(email) > 10 and " " not in email and email.lower().endswith("@intel.com")
def is_intel_company(company):
"""Checks that company contains intel"""
return company and "intel" in company.lower()
def is_valid_intel_user(user):
"""Checks that user is valid GitHub and Intel user"""
try:
return is_valid_user(user) and is_valid_name(user.name) and is_intel_email(user.email)
except IncompletableObject:
return False
def print_users(users):
"""Print list of users in different formats: list, set, PaginatedList"""
if isinstance(users, (list, set, PaginatedList)):
users_count = users.totalCount if isinstance(users, PaginatedList) else len(users)
print(f"GitHub users {users_count} (login - name - company - email - valid):")
else:
users = [users]
for user in users:
if not is_valid_user(user):
print("WRONG GitHub user: ???")
continue
try:
name = user.name
except IncompletableObject:
name = "???"
try:
company = user.company
except IncompletableObject:
company = "???"
try:
email = user.email
except IncompletableObject:
email = "???"
valid_check = "OK" if is_valid_intel_user(user) else "FIX"
if not is_intel_email(email):
valid_check += " email"
if not is_valid_name(name):
valid_check += " name"
print(f'{user.login} - "{name}" - "{company}" - {email} - {valid_check}')
def get_dev_emails():
"""
Read a file with developer emails. Supported email formats
first_name.last_name@intel.com
Import from Outlook: Last_name, First_name <first_name.last_name@intel.com>
"""
re_email = re.compile(r".+<(.+)>")
emails = set()
cfg = Config()
with open(cfg.properties["EMAILS_FILE_PATH"]) as file_obj:
for line in file_obj:
line = line.strip().lower()
if not line or line.startswith("#"):
continue
re_outlook_email = re_email.match(line)
if re_outlook_email:
line = re_outlook_email.group(1).strip()
if not is_intel_email(line):
print(f'Wrong email in {cfg.properties["EMAILS_FILE_PATH"]}: {line}')
continue
emails.add(line)
return emails
class GithubOrgApi:
"""Common API for GitHub organization"""
def __init__(self):
self._cfg = Config()
self.github = Github(self._cfg.GITHUB_TOKEN)
self.github_org = self.github.get_organization(self._cfg.GITHUB_ORGANIZATION)
self.repo = self.github.get_repo(f"{self._cfg.GITHUB_ORGANIZATION}/{self._cfg.GITHUB_REPO}")
self.github_users_by_email = {}
self.org_members_by_login = {}
self.members_to_remove = set()
self.members_to_fix_name = set()
def is_org_user(self, user):
"""Checks that user is a member of GitHub organization"""
if is_valid_user(user):
# user.get_organization_membership(self.github_org) doesn't work with org members
# permissions, GITHUB_TOKEN must be org owner now
return self.github_org.has_in_members(user)
return False
def get_org_emails(self):
"""Gets and prints emails of all GitHub organization members"""
org_members = self.github_org.get_members()
org_emails = set()
print(f"\nOrg members {org_members.totalCount} (login - name - company - email - valid):")
for org_member in org_members:
self.org_members_by_login[org_member.login.lower()] = org_member
print_users(org_member)
if is_intel_email(org_member.email):
email = org_member.email.lower()
org_emails.add(email)
self.github_users_by_email[email] = org_member
if not is_valid_name(org_member.name):
self.members_to_fix_name.add(org_member)
elif not is_user_ignored(org_member):
self.members_to_remove.add(org_member)
print("\nOrg members - no Intel emails:")
print_users(self.members_to_remove)
print("\nOrg members - no real name:")
print_users(self.members_to_fix_name)
print(
"\nOrg member emails - no real name:",
"; ".join([member.email.lower() for member in self.members_to_fix_name]),
)
return org_emails
def get_org_invitation_emails(self):
"""Gets GitHub organization teams prints info"""
org_invitations = self.github_org.invitations()
org_invitation_emails = set()
print(
f"\nOrg invitations {org_invitations.totalCount} "
"(login - name - company - email - valid):"
)
for org_invitation in org_invitations:
print_users(org_invitation)
if is_user_ignored(org_invitation):
continue
if is_intel_email(org_invitation.email):
org_invitation_emails.add(org_invitation.email.lower())
else:
print("Strange org invitation:", org_invitation)
print(
f"\nOrg invitation emails {len(org_invitation_emails)}:",
"; ".join(org_invitation_emails),
)
return org_invitation_emails
def get_org_teams(self):
"""Gets GitHub organization teams prints info"""
teams = []
org_teams = self.github_org.get_teams()
print("\nOrg teams count:", org_teams.totalCount)
for team in org_teams:
teams.append(team.name)
print(f"\nTeam: {team.name} - parent: {team.parent}")
repos = team.get_repos()
print("Repos:")
for repo in repos:
print(f" {repo.name} -", team.get_repo_permission(repo))
team_maintainers = team.get_members(role="maintainer")
team_maintainer_logins = set()
for maintainer in team_maintainers:
team_maintainer_logins.add(maintainer.login)
team_members = team.get_members(role="member")
team_member_logins = set()
for member in team_members:
team_member_logins.add(member.login)
members = team.get_members(role="all")
member_emails = []
print("Members (role - login - name - company - email - valid):")
for user in members:
if user.login in team_maintainer_logins:
print(" Maintainer - ", end="")
elif user.login in team_member_logins:
print(" Member - ", end="")
else:
# It is not possible to check child teams members
print(" ??? - ", end="")
print_users(user)
if is_intel_email(user.email) and not is_user_ignored(user):
member_emails.append(user.email.lower())
print(f"Intel emails {len(member_emails)}:", "; ".join(member_emails))
return teams
def get_github_user_by_email(self, email):
"""Gets GitHub user by email"""
if email in self.github_users_by_email:
return self.github_users_by_email.get(email)
def search_users():
paginated_users = self.github.search_users(f"{email} in:email")
# Minimize the GitHub Rate Limit
users = []
for user in paginated_users:
users.append(user)
if len(users) == 1:
return users[0]
if len(users) == 0:
return None
raise GithubApiException(
f"ERROR: Found {len(users)} GitHub accounts with the same email {email}"
)
try:
user = search_users()
except RateLimitExceededException:
print("WARNING: RateLimitExceededException")
time.sleep(30)
user = search_users()
self.github_users_by_email[email] = user
return user
def get_valid_github_users(self, emails):
"""Gets valid GitHub users by email and prints status"""
valid_users = set()
wrong_emails = set()
no_account_emails = set()
no_account_names = set()
print(f"\nGitHub users from {len(emails)} invite emails (email - status):")
for email in emails:
if not is_intel_email(email):
print(f"{email} - Non Intel email")
wrong_emails.add(email)
continue
# You can make up to 30 requests per minute; https://developer.github.com/v3/search/
time.sleep(2)
user = self.get_github_user_by_email(email)
if not user:
print(f"{email} - No valid GitHub account")
no_account_emails.add(email)
continue
if user.email and user.email.lower() == email:
if is_valid_name(user.name):
print(f"{email} - OK")
valid_users.add(user)
else:
print(f"{email} - No valid name in GitHub account: ", end="")
print_users(user)
no_account_names.add(email)
else:
print(f"{email} - Non public or wrong email in GitHub account: ", end="")
print_users(user)
no_account_emails.add(email)
print("\nValid users:")
print_users(valid_users)
print(f"\nWrong emails {len(wrong_emails)}:", "; ".join(wrong_emails))
print(
f"\nIntel emails - No valid GitHub account {len(no_account_emails)}:",
"; ".join(no_account_emails),
)
print(
f"\nIntel emails - No valid name in GitHub account {len(no_account_names)}:",
"; ".join(no_account_names),
)
return valid_users
def invite_users(self, users):
"""Invites users to GitHub organization and prints status"""
if not isinstance(users, typing.Iterable):
users = [users]
print(f"\nInvite {len(users)} users:")
for user in users:
if isinstance(user, str):
print(f"Email: {user}")
self.github_org.invite_user(email=user)
else:
print(f'{user.login} - "{user.name}" - {user.email} - ', end="")
try:
if is_user_ignored(user):
print("Ignored")
continue
if self._cfg.DRY_RUN:
print("Dry run")
continue
self.github_org.invite_user(user=user)
print("OK")
except GithubException as exc:
print(f'FAIL: {exc.data["errors"][0]["message"]}')
def remove_users(self, users):
"""Removes users from GitHub organization"""
if not isinstance(users, typing.Iterable):
users = [users]
print(f"\nRemove {len(users)} users:")
dry_run = self._cfg.DRY_RUN
if not dry_run and len(users) > self._cfg.MAX_MEMBERS_TO_REMOVE:
print(
"WARNING: Review is required for removing members more than "
f"{self._cfg.MAX_MEMBERS_TO_REMOVE}"
)
# TODO: Add notification
dry_run = True
for user in users:
member = self.get_github_user_by_email(user) if isinstance(user, str) else user
print(f'{member.login} - "{member.name}" - {member.email} - ', end="")
try:
if is_user_ignored(member):
print("Ignored")
continue
if dry_run:
print("Dry run")
continue
self.github_org.remove_from_membership(member)
print("OK")
except GithubException as exc:
print(f'FAIL: {exc.data["errors"][0]["message"]}')
def _test():
"""Test and debug"""
Config(cli_args=["DRY_RUN=True"])
dev_emails = get_dev_emails()
print("dev_emails:", dev_emails)
gh_api = GithubOrgApi()
gh_api.get_org_emails()
if __name__ == "__main__":
_test()
+247
View File
@@ -0,0 +1,247 @@
# Copyright (C) 2018-2021 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
"""
Gets info about users and groups via LDAP
"""
# pylint: disable=fixme,no-member
import sys
from enum import Enum
from pathlib import Path
from ldap3 import Server, Connection, ALL, SUBTREE
sys.path.append(str(Path(__file__).resolve().parents[1]))
from github_org_control.configs import Config
class LdapApiException(Exception):
"""Base LDAP API exception"""
class InfoLevel(Enum):
"""Constants for printing user info from LDAP"""
PDL = "PDL" # Public Distribution List (group of e-mail addresses)
FULL = "Full"
def print_user_info(info, info_level=None):
"""Pretty-print of a user info data structure (dict). info_level is the InfoLevel Enum"""
if not info or not info.get("mail"):
raise LdapApiException("ERROR: No info or absent mail")
def get_membership():
if info_level == InfoLevel.PDL:
membership_info = " PDLs:"
elif info_level == InfoLevel.FULL:
membership_info = " memberOf :"
else:
return ""
# Grouping groups by purpose
if info_level == InfoLevel.PDL:
sort_key = lambda i: i.split(",", 1)[0].lower()
else:
sort_key = lambda i: i.split(",", 1)[1] + i.split(",", 1)[0].lower()
for item in sorted(info["memberOf"], key=sort_key):
if info_level == InfoLevel.PDL and "OU=Delegated" not in item:
continue
membership_info += f"\n {item}"
return membership_info
try:
text_info = (
f'\n{info["cn"]} <{info["mail"]}>; {info["sAMAccountName"]}; {info["employeeID"]}'
f'\n Org group: {info["intelSuperGroupDescr"]} ({info["intelSuperGroupShortName"]}) /'
f' {info["intelGroupDescr"]} ({info["intelGroupShortName"]}) /'
f' {info["intelDivisionDescr"]} ({info["intelDivisionShortName"]}) /'
f' {info["intelOrgUnitDescr"]}'
f'\n Manager: {info.get("manager")}'
f'\n Location: {info["intelRegionCode"]} / {info["co"]} / {info["intelSiteCode"]} /'
f' {info["intelBldgCode"]} ({info.get("intelSiteName")}) /'
f' {info["physicalDeliveryOfficeName"]}'
f'\n Other: {info["employeeType"]} | {info["intelExportCountryGroup"]} |'
f' {info["whenCreated"]} | {info["intelCostCenterDescr"]} | {info["jobDescription"]}'
)
except Exception as exc:
raise LdapApiException(
f'ERROR: Failed to get info about "{info["mail"]}". '
f"Exception occurred:\n{repr(exc)}"
) from exc
print(text_info)
membership = get_membership()
if info_level == InfoLevel.PDL and membership:
print(membership)
elif info_level == InfoLevel.FULL:
for key in sorted(info):
if isinstance(info[key], list):
if key == "memberOf":
print(membership)
else:
print(f" {key} :")
for item in info[key]:
print(" ", item)
else:
print(f" {key} : {info[key]}")
class LdapApi:
"""LDAP API for getting user info and emails"""
_binary_blobs = ["thumbnailPhoto", "msExchUMSpokenName", "msExchBlockedSendersHash"]
_check_existing = [
"intelExportCountryGroup",
"physicalDeliveryOfficeName",
"intelSuperGroupShortName",
"intelGroupShortName",
"intelDivisionShortName",
]
null = "<null>"
def __init__(self):
self._cfg = Config()
self.server = Server(self._cfg.LDAP_SERVER, get_info=ALL)
self.connection = Connection(
self.server, user=self._cfg.LDAP_USER, password=self._cfg.LDAP_PASSWORD, auto_bind=True
)
self.connection.bind()
def get_user_emails(self, groups=None):
"""Gets emails of LDAP groups and sub-groups"""
print("\nGet emails from LDAP groups:")
processed_ldap_members = {}
def process_group_members(member, parent_group):
if member in processed_ldap_members:
processed_ldap_members[member]["parent_groups"].append(parent_group)
print(
"\nWARNING: Ignore LDAP member to avoid duplication and recursive cycling "
f"of PDLs: {member}\n "
f'email: {processed_ldap_members[member].get("email")}\n parent_groups:'
)
for group in processed_ldap_members[member].get("parent_groups", []):
print(7 * " ", group)
return
processed_ldap_members[member] = {"email": None, "parent_groups": [parent_group]}
# AD moves terminated users to the boneyard OU in case the user returns,
# so it can be reactivated with little effort.
# After 30 days it is removed and the unix personality becomes unlinked.
if "OU=Boneyard" in member:
return
self.connection.search(
member, r"(objectClass=*)", SUBTREE, attributes=["cn", "member", "mail"]
)
# print(self.connection.entries)
if not self.connection.response:
raise LdapApiException(f"ERROR: empty response. LDAP member: {member}")
# Check that the member is worker.
# The response can contain several items, but the first item is valid only
if "OU=Workers" in member:
if self.connection.response[0]["attributes"]["mail"]:
processed_ldap_members[member]["email"] = self.connection.response[0][
"attributes"
]["mail"].lower()
return
raise LdapApiException(
f"ERROR: no mail. LDAP worker: {member}\n" f"{self.connection.entries}"
)
if len(self.connection.response) > 1:
raise LdapApiException(
f"ERROR: multiple responses for {member}: "
f"{len(self.connection.response)}\n"
f"{self.connection.entries}"
)
if self.connection.response[0]["attributes"]["member"]:
for group_member in self.connection.response[0]["attributes"]["member"]:
process_group_members(group_member, member)
else:
print(f"\nERROR: no members in LDAP group: {member}\n{self.connection.entries}")
for group in groups or self._cfg.LDAP_PDLs:
print("\nProcess ROOT LDAP group:", group)
process_group_members(group, "ROOT")
return {
member.get("email") for member in processed_ldap_members.values() if member.get("email")
}
def _get_user_info(self, query):
"""Gets user info from LDAP as dict matching key and values pairs from query"""
query_filter = "".join(f"({key}={value})" for key, value in query.items())
for domain in self._cfg.LDAP_DOMAINS:
search_base = f"OU=Workers,DC={domain},DC=corp,DC=intel,DC=com"
self.connection.search(
search_base,
f"(&(objectcategory=person)(objectclass=user)(intelflags=1){query_filter})",
SUBTREE,
attributes=["*"],
)
if self.connection.response:
if len(self.connection.response) > 1:
raise LdapApiException(
f"ERROR: multiple responses for {query_filter}: "
f"{len(self.connection.response)}\n"
f"{self.connection.entries}"
)
info = self.connection.response[0]["attributes"]
# remove long binary blobs
for blob in LdapApi._binary_blobs:
info[blob] = b""
for key in LdapApi._check_existing:
if not info.get(key):
info[key] = LdapApi.null
return info
return {}
def get_user_info_by_idsid(self, idsid):
"""Gets user info from LDAP as dict using account name for searching"""
return self._get_user_info({"sAMAccountName": idsid})
def get_user_info_by_name(self, name):
"""Gets user info from LDAP as dict using common name for searching"""
return self._get_user_info({"cn": name})
def get_user_info_by_email(self, email):
"""Gets user info from LDAP as dict using emails for searching"""
return self._get_user_info({"mail": email})
def get_absent_emails(self, emails):
"""Checks users by email in LDAP and returns absent emails"""
absent_emails = set()
for email in emails:
if not self.get_user_info_by_email(email):
absent_emails.add(email)
return absent_emails
def _test():
"""Test and debug"""
ldap = LdapApi()
emails = ldap.get_user_emails()
print(f'\nLDAP emails count: {len(emails)}\n{"; ".join(emails)}')
emails = ["foo@intel.com"]
for email in emails:
info = ldap.get_user_info_by_email(email)
if info:
print_user_info(info, InfoLevel.PDL)
else:
print(f"\n{email} - not found")
if __name__ == "__main__":
_test()
@@ -0,0 +1 @@
pylint==2.11.1
@@ -0,0 +1,2 @@
PyGithub==1.55
ldap3==2.7