Zack Williams | 7f87d3e | 2020-06-05 12:36:39 -0700 | [diff] [blame^] | 1 | #!/usr/bin/env python3 |
| 2 | |
| 3 | # SPDX-FileCopyrightText: © 2020 Open Networking Foundation <support@opennetworking.org> |
| 4 | # SPDX-License-Identifier: Apache-2.0 |
| 5 | |
| 6 | from __future__ import absolute_import |
| 7 | |
| 8 | import argparse |
| 9 | import json |
| 10 | import logging |
| 11 | import urllib.request |
| 12 | import urllib.parse |
| 13 | import base64 |
| 14 | |
| 15 | # create shared logger |
| 16 | logging.basicConfig() |
| 17 | logger = logging.getLogger("cg") |
| 18 | |
| 19 | # global dict of jsonpath expressions -> compiled jsonpath parsers, as |
| 20 | # reparsing expressions in each loop results in 100x longer execution time |
| 21 | gjpaths = {} |
| 22 | |
| 23 | # gerrit headers is cached |
| 24 | gerrit_headers = [] |
| 25 | |
| 26 | # gerrit_groups is a cache of groups, to prevent multiple lookups |
| 27 | gerrit_groups = {} |
| 28 | |
| 29 | |
| 30 | def parse_crowdgroups_args(): |
| 31 | """ |
| 32 | parse CLI arguments |
| 33 | """ |
| 34 | |
| 35 | parser = argparse.ArgumentParser(description="Crowd-Gerrit group sync") |
| 36 | |
| 37 | # Positional args |
| 38 | parser.add_argument( |
| 39 | "settings", type=argparse.FileType("r"), help="JSON settings file", |
| 40 | ) |
| 41 | |
| 42 | parser.add_argument( |
| 43 | "--debug", action="store_true", help="Print additional debugging information" |
| 44 | ) |
| 45 | |
| 46 | return parser.parse_args() |
| 47 | |
| 48 | |
| 49 | def basic_auth_header(username, password): |
| 50 | """ |
| 51 | returns a tuple containing a http basic auth header |
| 52 | """ |
| 53 | creds_str = "%s:%s" % (username, password) |
| 54 | creds_b64 = base64.standard_b64encode(creds_str.encode("utf-8")) |
| 55 | |
| 56 | return ("Authorization", "Basic %s" % creds_b64.decode("utf-8")) |
| 57 | |
| 58 | |
| 59 | def json_api_get(url, headers, data=None, trim_prefix=False, allow_failure=False): |
| 60 | """ |
| 61 | Call JSON API endpoint, return data as a dict |
| 62 | """ |
| 63 | |
| 64 | # if data included, encode it as JSON |
| 65 | if data: |
| 66 | data_enc = str(json.dumps(data)).encode("utf-8") |
| 67 | |
| 68 | request = urllib.request.Request(url, data=data_enc, method="POST") |
| 69 | request.add_header("Content-Type", "application/json; charset=UTF-8") |
| 70 | else: |
| 71 | request = urllib.request.Request(url) |
| 72 | |
| 73 | # add headers tuples |
| 74 | for header in headers: |
| 75 | request.add_header(*header) |
| 76 | |
| 77 | try: |
| 78 | response = urllib.request.urlopen(request) |
| 79 | except urllib.error.HTTPError: |
| 80 | # asking for data that doesn't exist results in a 404, just return nothing |
| 81 | if allow_failure: |
| 82 | return None |
| 83 | logger.exception("Server encountered an HTTPError at URL: '%s'", url) |
| 84 | except urllib.error.URLError: |
| 85 | logger.exception("An URLError occurred at URL: '%s'", url) |
| 86 | else: |
| 87 | # docs: https://docs.python.org/3/library/json.html |
| 88 | jsondata = response.read() |
| 89 | logger.debug("API response: %s", jsondata) |
| 90 | |
| 91 | # optionally remove the gerrit "magic prefix" - docs: |
| 92 | # https://gerrit-review.googlesource.com/Documentation/rest-api.html#output |
| 93 | if trim_prefix: |
| 94 | jsondata = jsondata.lstrip(b")]}'\n") |
| 95 | |
| 96 | try: |
| 97 | data = json.loads(jsondata) |
| 98 | except json.decoder.JSONDecodeError: |
| 99 | # allow return of no data |
| 100 | if allow_failure: |
| 101 | return None |
| 102 | logger.exception("Unable to decode JSON") |
| 103 | else: |
| 104 | logger.debug("JSON decoded: %s", data) |
| 105 | |
| 106 | return data |
| 107 | |
| 108 | |
| 109 | def crowd_users_list(crowd_settings, groupname): |
| 110 | """ |
| 111 | Returns dict of email addresses mapped to usernames from Crowd, given a |
| 112 | group name |
| 113 | """ |
| 114 | |
| 115 | url = "%s/rest/usermanagement/1/group/user/direct?groupname=%s&expand=user" % ( |
| 116 | crowd_settings["url"], |
| 117 | groupname, |
| 118 | ) |
| 119 | |
| 120 | crowd_headers = [ |
| 121 | basic_auth_header(crowd_settings["username"], crowd_settings["password"]), |
| 122 | ("Accept", "application/json"), |
| 123 | ] |
| 124 | |
| 125 | cout = json_api_get(url, crowd_headers) |
| 126 | |
| 127 | userlist = {} |
| 128 | |
| 129 | for user in cout["users"]: |
| 130 | userlist[user["email"]] = user["name"] |
| 131 | |
| 132 | return userlist |
| 133 | |
| 134 | |
| 135 | def gerrit_group_id_by_name(gerrit_settings, groupname): |
| 136 | """ |
| 137 | Returns a gerrit group ID given the name of a group |
| 138 | Note - this is the "id" not the "group_id" in the dict. |
| 139 | """ |
| 140 | |
| 141 | # read from cache if found |
| 142 | if groupname not in gerrit_groups: |
| 143 | |
| 144 | gurl = "%s/groups/?m=%s" % (gerrit_settings["url"], groupname) |
| 145 | gout = json_api_get(gurl, gerrit_headers, None, True) |
| 146 | |
| 147 | gerrit_groups[groupname] = gout[groupname]["id"] |
| 148 | |
| 149 | logger.debug( |
| 150 | "Group: '%s' has group-id: '%s'", groupname, gerrit_groups[groupname] |
| 151 | ) |
| 152 | |
| 153 | return gerrit_groups[groupname] |
| 154 | |
| 155 | |
| 156 | def gerrit_users_list(gerrit_settings, groupname): |
| 157 | """ |
| 158 | Returns dict of lists of usernames and emails from Gerrit, given a group name |
| 159 | """ |
| 160 | |
| 161 | gid = gerrit_group_id_by_name(gerrit_settings, groupname) |
| 162 | |
| 163 | gurl = "%s/groups/%s/members" % (gerrit_settings["url"], gid) |
| 164 | gout = json_api_get(gurl, gerrit_headers, None, True) |
| 165 | |
| 166 | userlist = {} |
| 167 | |
| 168 | for user in gout: |
| 169 | userlist[user["email"]] = { |
| 170 | "_account_id": user.get("_account_id"), |
| 171 | "username": user.get("username", None), |
| 172 | } |
| 173 | |
| 174 | return userlist |
| 175 | |
| 176 | |
| 177 | def gerrit_check_crowd_id(gerrit_settings, aid, crowdusername): |
| 178 | """ |
| 179 | checks if a gerrit account has an external CrowdID identity string |
| 180 | """ |
| 181 | |
| 182 | gurl = "%s/accounts/%s/external_ids" % (gerrit_settings["url"], aid) |
| 183 | gout = json_api_get(gurl, gerrit_headers, None, True) |
| 184 | |
| 185 | crowd_id_string = ( |
| 186 | "https://crowd.opennetworking.org/openidserver/users/%s" % crowdusername |
| 187 | ) |
| 188 | |
| 189 | for identity in gout: |
| 190 | if identity["identity"] == crowd_id_string: |
| 191 | return aid |
| 192 | |
| 193 | return None |
| 194 | |
| 195 | |
| 196 | def gerrit_find_account_id(gerrit_settings, email, crowdusername): |
| 197 | """ |
| 198 | Returns a gerrit account ID given the email address of a user and the |
| 199 | corresponding crowd username, or None if no account exists. |
| 200 | """ |
| 201 | |
| 202 | gurl = "%s/accounts/?q=email:%s" % (gerrit_settings["url"], email) |
| 203 | gout = json_api_get(gurl, gerrit_headers, None, True, True) |
| 204 | |
| 205 | # check for users with multiple email addresses |
| 206 | if len(gout) > 1: |
| 207 | logger.warning("Email: '%s' register for more than one Gerrit account", email) |
| 208 | return None |
| 209 | |
| 210 | # may not have an account, return None if no account |
| 211 | if len(gout) == 0: |
| 212 | logger.warning("Email: '%s' doesn't have a Gerrit account", email) |
| 213 | return None |
| 214 | |
| 215 | aid_dict = gout[0] |
| 216 | logger.debug("Email: '%s' has account-id: '%s'", email, aid_dict["_account_id"]) |
| 217 | |
| 218 | return aid_dict["_account_id"] |
| 219 | |
| 220 | |
| 221 | def gerrit_lookup_account_ids(gerrit_settings, users): |
| 222 | """ |
| 223 | Returns a list of account-ids given a list of user tuples |
| 224 | """ |
| 225 | |
| 226 | account_ids = [] |
| 227 | |
| 228 | for user in users: |
| 229 | aid = gerrit_find_account_id(gerrit_settings, *user) |
| 230 | |
| 231 | if aid: |
| 232 | account_ids.append(aid) |
| 233 | |
| 234 | return account_ids |
| 235 | |
| 236 | |
| 237 | def gerrit_add_accounts_to_group(gerrit_settings, groupname, account_ids): |
| 238 | """ |
| 239 | Given a group name and a list of account-ids, add accounts from group |
| 240 | """ |
| 241 | |
| 242 | gid = gerrit_group_id_by_name(gerrit_settings, groupname) |
| 243 | |
| 244 | # https://gerrit-review.googlesource.com/Documentation/rest-api-groups.html#_add_group_members |
| 245 | members = {"members": account_ids} |
| 246 | |
| 247 | gurl = "%s/groups/%s/members.add" % (gerrit_settings["url"], gid) |
| 248 | gout = json_api_get(gurl, gerrit_headers, members, True) |
| 249 | |
| 250 | logger.debug("output of adding accounts to group: %s", gout) |
| 251 | |
| 252 | |
| 253 | def gerrit_remove_accounts_from_group(gerrit_settings, groupname, account_ids): |
| 254 | """ |
| 255 | Given a group name and a list of account-ids, remove accounts from group |
| 256 | """ |
| 257 | |
| 258 | gid = gerrit_group_id_by_name(gerrit_settings, groupname) |
| 259 | |
| 260 | # https://gerrit-review.googlesource.com/Documentation/rest-api-groups.html#_add_group_members |
| 261 | members = {"members": account_ids} |
| 262 | |
| 263 | gurl = "%s/groups/%s/members.delete" % (gerrit_settings["url"], gid) |
| 264 | gout = json_api_get(gurl, gerrit_headers, members, True, True) |
| 265 | |
| 266 | logger.debug("output of removing accounts to group: %s", gout) |
| 267 | |
| 268 | |
| 269 | def sync_group(settings, groupname): |
| 270 | """ |
| 271 | sync a group from Crowd to Gerrit |
| 272 | """ |
| 273 | |
| 274 | cusers = crowd_users_list(settings["crowd"], groupname) |
| 275 | |
| 276 | gusers = gerrit_users_list(settings["gerrit"], groupname) |
| 277 | |
| 278 | logger.info("Crowd users: %s" % cusers) |
| 279 | logger.info("Gerrit users: %s" % gusers) |
| 280 | |
| 281 | # lists of users to add or remove from gerrit |
| 282 | add_g = [] |
| 283 | rem_g = list(gusers.keys()) # default to removing all users |
| 284 | |
| 285 | # go through list of crowd users by email |
| 286 | for cemail, cuserid in cusers.items(): |
| 287 | # if a user is found, remove from rem_g list |
| 288 | if cemail in gusers: |
| 289 | rem_g.remove(cemail) |
| 290 | else: |
| 291 | add_g.append((cemail, cuserid)) |
| 292 | |
| 293 | logger.info("Users to add: %s" % add_g) |
| 294 | logger.info("Users to remove: %s" % rem_g) |
| 295 | |
| 296 | add_aids = gerrit_lookup_account_ids(settings["gerrit"], add_g) |
| 297 | |
| 298 | logger.info("Account-ids to add: %s" % add_aids) |
| 299 | |
| 300 | gerrit_aids = [gusers[e]["_account_id"] for e in gusers.keys()] |
| 301 | |
| 302 | logger.info("All Gerrit account-ids: %s" % gerrit_aids) |
| 303 | |
| 304 | aa_nodupes = [a for a in add_aids if a not in gerrit_aids] |
| 305 | |
| 306 | logger.info("Duplicates removed: %s" % aa_nodupes) |
| 307 | |
| 308 | if aa_nodupes: |
| 309 | gerrit_add_accounts_to_group(settings["gerrit"], groupname, aa_nodupes) |
| 310 | |
| 311 | remove_aids = [gusers[e]["_account_id"] for e in rem_g] |
| 312 | |
| 313 | # user with a different primary email could be in both add and remove lists |
| 314 | rem_filtered = [a for a in remove_aids if a not in add_aids] |
| 315 | |
| 316 | logger.info("Account-ids to remove: %s" % rem_filtered) |
| 317 | |
| 318 | if rem_filtered: |
| 319 | gerrit_remove_accounts_from_group(settings["gerrit"], groupname, rem_filtered) |
| 320 | |
| 321 | |
| 322 | # main function that calls other functions |
| 323 | if __name__ == "__main__": |
| 324 | |
| 325 | args = parse_crowdgroups_args() |
| 326 | |
| 327 | # only print log messages if debugging |
| 328 | if args.debug: |
| 329 | logger.setLevel(logging.DEBUG) |
| 330 | else: |
| 331 | logger.setLevel(logging.CRITICAL) |
| 332 | |
| 333 | # load settings from JSON file |
| 334 | settings = json.loads(args.settings.read()) |
| 335 | |
| 336 | # global, so this isn't run multiple times |
| 337 | gerrit_headers = [ |
| 338 | basic_auth_header( |
| 339 | settings["gerrit"]["username"], settings["gerrit"]["password"] |
| 340 | ), |
| 341 | ] |
| 342 | |
| 343 | # sync each group |
| 344 | for groupname in settings["groups"]: |
| 345 | sync_group(settings, groupname) |