chiark - git - matthewv - gooswapper/blob - gooswapper.py

   1 #!/usr/bin/env python3
   2
   3 # Copyright (C) 2018 Genome Research Limited
   4 #
   5 # Author: Matthew Vernon <mv3@sanger.ac.uk>
   6 #
   7 # This program is free software: you can redistribute it and/or modify
   8 # it under the terms of the GNU Affero General Public License as published
   9 # by the Free Software Foundation, either version 3 of the License, or
  10 # (at your option) any later version.
  11 #
  12 # This program is distributed in the hope that it will be useful,
  13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15 # GNU Affero General Public License for more details.
  16 #
  17 # You should have received a copy of the GNU Affero General Public License
  18 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  19
  20 import sys
  21 import getpass
  22 import os
  23 import os.path
  24 import pickle
  25 import collections
  26 import argparse
  27 import time
  28 import logging
  29 logger = logging.getLogger('gooswapper')
  30 logger.setLevel(logging.INFO)
  31 consolelog = logging.StreamHandler()
  32 consolelog.setLevel(logging.INFO)
  33 logformatter = logging.Formatter('%(name)s - %(levelname)s - %(message)s')
  34 consolelog.setFormatter(logformatter)
  35 logger.addHandler(consolelog)
  36 #We can't use this, because that way all the libraries' logs spam us
  37 #logging.basicConfig(level=logging.INFO)
  38
  39 #Exchange-related library
  40 sys.path.append("/upstreams/exchangelib")
  41 import exchangelib
  42
  43 #Google calendar-api libraries
  44 import httplib2
  45 import apiclient.discovery
  46 import oauth2client
  47 import oauth2client.file
  48 import oauth2client.client
  49 import googleapiclient.errors
  50
  51 #Not sure what the distribution approach here is...
  52 gcal_client_id = '805127902516-ptbbtgpq9o8pjr6r3k6hsm60j589o85u.apps.googleusercontent.com'
  53 gcal_client_secret = '8hpdxV3MauorryTDoZ1YK8JO'
  54
  55 #scope URL for r/w calendar access
  56 scope = 'https://www.googleapis.com/auth/calendar'
  57 #flow object, for doing OAuth2.0 stuff
  58 flow = oauth2client.client.OAuth2WebServerFlow(gcal_client_id,
  59                                                gcal_client_secret,
  60                                                scope)
  61
  62 gsdir = os.path.expanduser("~/.gooswapper")
  63 gcal_authpath = gsdir + "/.gooswap_gcal_creds.dat"
  64
  65 cachepath=None
  66
  67 exchange_credential = None
  68
  69 CachedExEvent=collections.namedtuple('CachedExEvent',
  70                                      ['changekey','gcal_link'])
  71
  72 class ex_gcal_link(exchangelib.ExtendedProperty):
  73     distinguished_property_set_id = 'PublicStrings'
  74     property_name = "google calendar event id"
  75     property_type = 'String'
  76
  77 try:
  78     exchangelib.CalendarItem.get_field_by_fieldname('gcal_link')
  79 except ValueError:
  80     exchangelib.CalendarItem.register('gcal_link',ex_gcal_link)
  81
  82 #useful if you want to replay an event
  83 def drop_from_ex_cache(itemid):
  84     with open(cachepath,"rb") as f:
  85         cache = pickle.load(f)
  86     cache.pop(itemid)
  87     with open(cachepath,"wb") as f:
  88         pickle.dump(cache,f)
  89
  90 def get_ex_event_by_itemid(calendar,itemid):
  91     return calendar.get(item_id=itemid)
  92
  93 def get_ex_event_by_id_and_changekey(acct,itemid,changekey):
  94     l=list(acct.fetch([(itemid,changekey)]))
  95     return list(acct.fetch([(itemid,changekey)]))[0]
  96
  97 def get_gcal_event_by_eventid(gcal_acct,eventId,gcal_id="primary"):
  98     return gcal_acct.events().get(calendarId=gcal_id,eventId=eventId).execute()
  99
 100 def get_gcal_recur_instance(gcal_acct,gcal_master,start,gcal_id="primary"):
 101     if gcal_master is None:
 102         logger.warning("Cannot get recurrences from event with null gcal id")
 103         return None
 104     ans = gcal_acct.events().instances(calendarId=gcal_id,
 105                                        eventId=gcal_master,
 106                                        originalStart=start.isoformat(),
 107                                        showDeleted=True).execute()
 108     if len(ans['items']) != 1:
 109         logger.error("Searching for recurrance instance returned %d events" % \
 110                      len(ans['items']))
 111         return None
 112     return ans['items'][0]
 113
 114 def get_ex_cred(username="SANGER\mv3",password=None):
 115     if password is None:
 116         password = getpass.getpass(prompt="Password for user %s: " % username)
 117     return exchangelib.ServiceAccount(username,password)
 118
 119 def ex_login(username,emailaddr,ad_cache_path=None):
 120     global exchange_credential
 121     autodiscover = True
 122     if exchange_credential is None:
 123         exchange_credential = get_ex_cred(username)
 124     if ad_cache_path is not None:
 125         try:
 126             with open(ad_cache_path,"rb") as f:
 127                 url,auth_type = pickle.load(f)
 128                 autodiscover = False
 129         except FileNotFoundError:
 130             pass
 131
 132     if autodiscover:
 133         ex_ac = exchangelib.Account(emailaddr,
 134                                     credentials = exchange_credential,
 135                                     autodiscover = autodiscover)
 136         if ad_cache_path is not None:
 137             cache=(ex_ac.protocol.service_endpoint,
 138                    ex_ac.protocol.auth_type)
 139             with open(ad_cache_path,"wb") as f:
 140                 pickle.dump(cache,f)
 141     else:
 142         ex_conf = exchangelib.Configuration(service_endpoint=url,
 143                                             credentials=exchange_credential,
 144                                             auth_type=auth_type)
 145         ex_ac = exchangelib.Account(emailaddr,
 146                                     config=ex_conf,
 147                                     autodiscover=False,
 148                                     access_type=exchangelib.DELEGATE)
 149
 150     return ex_ac
 151
 152 def get_ex_events(calendar):
 153     ans={}
 154     for event in calendar.all().only('changekey','item_id','gcal_link'):
 155         if event.item_id in ans:
 156             logger.warning("Event item_id %s was duplicated!" % event.item_id)
 157         ans[event.item_id] = CachedExEvent(event.changekey,event.gcal_link)
 158     logger.info("%d events found" % len(ans))
 159     return ans
 160
 161 def ex_event_changes(old,new):
 162     olds = set(old.keys())
 163     news = set(new.keys())
 164     added = list(news-olds)
 165     deleted = list(olds-news)
 166     changed = []
 167     #intersection - i.e. common to both sets
 168     for event in olds & news:
 169         if old[event].changekey != new[event].changekey:
 170             changed.append(event)
 171     logger.info("%d events updated, %d added, %d deleted" % (len(changed),
 172                                                               len(added),
 173                                                               len(deleted)))
 174     return added, deleted, changed
 175
 176 #exchangelib gives us days in recurrence patterns as integers,
 177 #RFC5545 wants SU,MO,TU,WE,TH,FR,SA
 178 #it has a utility function to convert to Monday, Tuesday, ...
 179 def rr_daystr_from_int(i):
 180     return exchangelib.recurrence._weekday_to_str(i).upper()[:2]
 181
 182 #for monthly patterns, we want the week (or -1 for last) combined with each
 183 #day specified
 184 def rr_daystr_monthly(p):
 185     if p.week_number == 5:
 186         wn = "-1"
 187     else:
 188         wn = str(p.week_number)
 189     return ",".join([wn + rr_daystr_from_int(x) for x in p.weekdays])
 190
 191 def rrule_from_ex(event,gcal_tz):
 192     if event.type != "RecurringMaster":
 193         logger.error("Cannot make recurrence from not-recurring event")
 194         return None
 195     if event.recurrence is None:
 196         logger.error("Empty recurrence structure")
 197         return None
 198     if isinstance(event.recurrence.pattern,
 199                   exchangelib.recurrence.DailyPattern):
 200         rr = "RRULE:FREQ=DAILY;INTERVAL=%d" % event.recurrence.pattern.interval
 201     elif isinstance(event.recurrence.pattern,
 202                     exchangelib.recurrence.WeeklyPattern):
 203         rr = "RRULE:FREQ=WEEKLY;INTERVAL=%d;BYDAY=%s;WKST=%s" % \
 204                           (event.recurrence.pattern.interval,
 205                            ",".join([rr_daystr_from_int(x) for x in event.recurrence.pattern.weekdays]),
 206                            rr_daystr_from_int(event.recurrence.pattern.first_day_of_week) )
 207     elif isinstance(event.recurrence.pattern,
 208                     exchangelib.recurrence.RelativeMonthlyPattern):
 209         rr = "RRULE:FREQ=MONTHLY;INTERVAL=%d;BYDAY=%s" % \
 210                      (event.recurrence.pattern.interval,
 211                       rr_daystr_monthly(event.recurrence.pattern))
 212     elif isinstance(event.recurrence.pattern,
 213                     exchangelib.recurrence.AbsoluteMonthlyPattern):
 214         rr = "RRULE:FREQ=MONTHLY;INTERVAL=%d;BYMONTHDAY=%d" % \
 215                           (event.recurrence.pattern.interval,
 216                            event.recurrence.pattern.day_of_month)
 217     elif isinstance(event.recurrence.pattern,
 218                     exchangelib.recurrence.AbsoluteYearlyPattern):
 219         rr = "RRULE:FREQ=YEARLY;BYMONTH=%d;BYMONTHDAY=%d" % \
 220                           (event.recurrence.pattern.month,
 221                            event.recurrence.pattern.day_of_month)
 222     elif isinstance(event.recurrence.pattern,
 223                     exchangelib.recurrence.RelativeYearlyPattern):
 224         rr = "RRULE:FREQ=YEARLY;BYMONTH=%d;BYDAY=%s" % \
 225                           (event.recurrence.pattern.month,
 226                            rr_daystr_monthly(event.recurrence.pattern))
 227     else:
 228         logger.error("Recurrence %s not supported" % event.recurrence)
 229         return None
 230     if isinstance(event.recurrence.boundary,
 231                   exchangelib.recurrence.EndDatePattern):
 232         rr += ";UNTIL={0:%Y}{0:%m}{0:%d}".format(event.recurrence.boundary.end)
 233     elif isinstance(event.recurrence.boundary,
 234                     exchangelib.recurrence.NoEndPattern):
 235         pass #no end date to set
 236     else:
 237         logger.error("Recurrence %s not supported" % event.recurrence)
 238         return None
 239     return [rr]
 240
 241 def modify_recurring(ex_acct,gcal_acct,gcal_tz,
 242                      events,master,gcal_id="primary"):
 243     if master.modified_occurrences is not None:
 244         for mod in master.modified_occurrences:
 245             instance = get_gcal_recur_instance(gcal_acct,master.gcal_link,
 246                                                mod.original_start,gcal_id)
 247             if instance is None: #give up after first failure
 248                 return
 249             mod_event = get_ex_event_by_itemid(ex_acct.calendar,mod.item_id)
 250             gevent = build_gcal_event_from_ex(mod_event,gcal_tz)
 251             gevent = gcal_acct.events().update(calendarId=gcal_id,
 252                                                eventId=instance.get('id'),
 253                                                body=gevent,
 254                                                sendUpdates="none").execute()
 255             mod_event.gcal_link = gevent.get("id")
 256             mod_event.save(update_fields=["gcal_link"])
 257     if master.deleted_occurrences is not None:
 258         for d in master.deleted_occurrences:
 259             instance = get_gcal_recur_instance(gcal_acct,master.gcal_link,
 260                                                d.start,gcal_id)
 261             if instance is None: #give up after any failure
 262                 return
 263             if instance["status"] != "cancelled":
 264                 instance["status"]="cancelled"
 265                 gcal_acct.events().update(calendarId=gcal_id,
 266                                           eventId=instance.get('id'),
 267                                           body=instance,
 268                                           sendUpdates="none").execute()
 269
 270 def build_gcal_event_from_ex(event,gcal_tz):
 271     gevent={}
 272     gevent["summary"]=event.subject
 273     #Use the event's timezones if possible, otherwise fall back to the
 274     #target calendar timezone
 275     if event._start_timezone is not None:
 276         stz = event._start_timezone
 277     else:
 278         stz = gcal_tz
 279     if event._end_timezone is not None:
 280         etz = event._end_timezone
 281     else:
 282         etz = gcal_tz
 283     if event.is_all_day:
 284         gevent["end"]={"date": str(event.end.astimezone(etz).date())}
 285         gevent["start"]={"date": str(event.start.astimezone(stz).date())}
 286     else:
 287         gevent["end"]={"dateTime": event.end.astimezone(etz).isoformat(),
 288                        "timeZone": str(etz)}
 289         gevent["start"]={"dateTime": event.start.astimezone(stz).isoformat(),
 290                          "timeZone": str(stz)}
 291     if event.text_body is not None and event.text_body.strip() != '':
 292         gevent["description"] = event.text_body
 293     if event.location is not None:
 294         gevent["location"] = event.location
 295     gevent["extendedProperties"]={"shared": {"ex_id": event.item_id}}
 296     return gevent
 297
 298 def add_ex_to_gcal(ex_acct,
 299                    gcal_acct,gcal_tz,events,
 300                    added,
 301                    gcal_id="primary"):
 302     for ev_id in added:
 303         event = get_ex_event_by_itemid(ex_acct.calendar,ev_id)
 304         gevent = build_gcal_event_from_ex(event,gcal_tz)
 305         if event.type=="RecurringMaster":
 306             rr = rrule_from_ex(event,gcal_tz)
 307             if rr is not None:
 308                 gevent["recurrence"] = rr
 309                 print(gevent)
 310             else:
 311                 logger.warning("Unable to set recurrence for %s" % event.item_id)
 312                 continue #don't make the gcal event
 313         gevent = gcal_acct.events().insert(calendarId=gcal_id,
 314                                            body=gevent).execute()
 315         event.gcal_link = gevent.get("id")
 316         event.save(update_fields=["gcal_link"])
 317         if event.type=="RecurringMaster" and (event.deleted_occurrences or \
 318                                               event.modified_occurrences):
 319             modify_recurring(ex_acct,gcal_acct,gcal_tz,
 320                              events,event,gcal_id)
 321             #changekey is updated by the above
 322             event.refresh()
 323         events[event.item_id] = events[event.item_id]._replace(changekey=event.changekey,gcal_link=event.gcal_link)
 324
 325 def del_ex_to_gcal(ex_acct, gcal_acct, events, deleted, gcal_id="primary"):
 326     for ev_id in deleted:
 327         if events[ev_id].gcal_link is not None:
 328             gevent = get_gcal_event_by_eventid(gcal_acct,
 329                                                events[ev_id].gcal_link,
 330                                                gcal_id)
 331             if gevent["status"] != "cancelled":
 332                 gcal_acct.events().delete(calendarId=gcal_id,
 333                                           eventId=events[ev_id].gcal_link,
 334                                           sendUpdates="none").execute()
 335
 336 def update_ex_to_gcal(ex_acct,
 337                       gcal_acct,gcal_tz,
 338                       events,changed,
 339                       gcal_id="primary"):
 340     for ev_id in changed:
 341         event = get_ex_event_by_itemid(ex_acct.calendar,ev_id)
 342         if event.gcal_link is None:
 343             logger.warning("Cannot apply update where event has no gcal link")
 344             continue
 345         gevent = build_gcal_event_from_ex(event,gcal_tz)
 346         if event.type=="RecurringMaster":
 347             rr = rrule_from_ex(event,gcal_tz)
 348             if rr is not None:
 349                 gevent["recurrence"] = rr
 350                 if event.deleted_occurrences or \
 351                    event.modified_occurrences:
 352                     modify_recurring(ex_acct,gcal_acct,gcal_tz,
 353                                      events,event,gcal_id)
 354                     event.refresh() #changekey is updated by the above
 355                     events[event.item_id] = events[event.item_id]._replace(changekey=event.changekey,gcal_link=event.gcal_link)
 356             else:
 357                 logger.warning("Unable to set recurrence for %s" % event.item_id)
 358                 continue #don't make the gcal event
 359         try: #may fail if we don't own the event
 360             gevent = gcal_acct.events().update(calendarId=gcal_id,
 361                                                eventId=event.gcal_link,
 362                                                body=gevent,
 363                                                sendUpdates="none").execute()
 364         except googleapiclient.errors.HttpError as err:
 365             if err.resp.status == 403:
 366                 pass
 367
 368 def match_ex_to_gcal(ex_acct,gcal_acct,gcal_tz,events,gcal_id="primary",ignore_link=True):
 369     recur = 0
 370     matched = 0
 371     skipped = 0
 372     toadd = []
 373     for ev_id in events:
 374         event = get_ex_event_by_itemid(ex_acct.calendar,ev_id)
 375         if event.gcal_link is not None and ignore_link is False:
 376             skipped += 1
 377             continue
 378         missing=True
 379         matches = gcal_acct.events().list(calendarId=gcal_id,
 380                                           timeMin=event.start.isoformat(),
 381                                           timeMax=event.end.isoformat()).execute()
 382         for ge in matches['items']:
 383             if ( ge.get("summary") is None and event.subject is None ) or \
 384                ( ge.get("summary") is not None and event.subject is not None \
 385                  and ge['summary'].strip()==event.subject.strip()):
 386                 logger.info("Matching '%s' starting at %s" % (event.subject,
 387                                                               event.start.isoformat()))
 388                 event.gcal_link = ge['id']
 389                 event.save(update_fields=["gcal_link"])
 390                 events[event.item_id] = events[event.item_id]._replace(changekey=event.changekey,gcal_link=event.gcal_link)
 391                 gevent = {}
 392                 gevent["start"] = ge["start"]
 393                 gevent["end"] = ge["end"]
 394                 gevent["extendedProperties"]={"shared": {"ex_id": event.item_id}}
 395                 try:
 396                     gcal_acct.events().update(calendarId=gcal_id,
 397                                               eventId=event.gcal_link,
 398                                               body=gevent,
 399                                               sendUpdates="none").execute()
 400                 #this may fail if we don't own the event
 401                 except googleapiclient.errors.HttpError as err:
 402                     if err.resp.status == 403:
 403                         pass
 404                 matched += 1
 405                 missing = False
 406                 break
 407         if missing == True:
 408             toadd.append(ev_id)
 409     logger.info("Matched %d events, skipped %d with existing link, and %d recurring ones" % (matched,skipped,recur))
 410     return toadd
 411
 412 def get_gcal_cred(args):
 413     #each such file can only store a single credential
 414     storage = oauth2client.file.Storage(gcal_authpath)
 415     gcal_credential = storage.get()
 416     #if no credential found, or they're invalid (e.g. expired),
 417     #then get a new one; pass --noauth_local_webserver on the command line
 418     #if you don't want it to spawn a browser
 419     if gcal_credential is None or gcal_credential.invalid:
 420         gcal_credential = oauth2client.tools.run_flow(flow,
 421                                                       storage,
 422                                                       args)
 423     return gcal_credential
 424
 425 def gcal_login(args):
 426     gcal_credential = get_gcal_cred(args)
 427     # Object to handle http requests; could add proxy details
 428     http = httplib2.Http()
 429     http = gcal_credential.authorize(http)
 430     return apiclient.discovery.build('calendar', 'v3', http=http)
 431
 432 def get_gcal_timezone(gcal_account,calendarid="primary"):
 433     gcal = gcal_account.calendars().get(calendarId=calendarid).execute()
 434     return exchangelib.EWSTimeZone.timezone(gcal['timeZone'])
 435
 436 def main():
 437     ap=argparse.ArgumentParser(description="Gooswapper calendar sync",
 438                                parents=[oauth2client.tools.argparser])
 439     ap.add_argument("exchuser",help="Exchange user e.g. 'SANGER\mv3'")
 440     ap.add_argument("exchemail",
 441                     help="Exchange calendar email e.g. ISGGroup@sanger.ac.uk")
 442     ap.add_argument("-g","--gcalid",help="google Calendar ID")
 443     ap.add_argument("-l","--loop",help="keep running indefinitely",
 444                     action="store_true")
 445     args = ap.parse_args()
 446     if args.gcalid is None:
 447         gcal_id = "primary"
 448     else:
 449         gcal_id = args.gcalid
 450
 451     #Make our config dir if it doesn't exist
 452     if not os.path.exists(gsdir):
 453         os.mkdir(gsdir,0o700)
 454     #Cache file is specific to the Exchange calendar
 455     global cachepath
 456     cachepath = gsdir + "/.cache-%s" % \
 457                 (args.exchemail.replace('@','_').replace('/','_'))
 458
 459     #log in to the accounts
 460     ex_account = ex_login(args.exchuser,args.exchemail,
 461                           gsdir+"/.gooswapper_exch_conf.dat")
 462     gcal_account = gcal_login(args)
 463     gcal_tz = get_gcal_timezone(gcal_account,gcal_id)
 464
 465     #Main loop (broken at the end if login is false)
 466     while True:
 467         try:
 468             with open(cachepath,"rb") as f:
 469                 cache = pickle.load(f)
 470         except FileNotFoundError:
 471             cache = None
 472
 473         current = get_ex_events(ex_account.calendar)
 474
 475         if cache is not None:
 476             added,deleted,changed = ex_event_changes(cache,current)
 477             add_ex_to_gcal(ex_account,gcal_account,gcal_tz,current,
 478                            added,gcal_id)
 479             #delete op needs the "cache" set, as that has the link ids in
 480             #for events that are now deleted
 481             del_ex_to_gcal(ex_account,gcal_account,cache,deleted,gcal_id)
 482             update_ex_to_gcal(ex_account,gcal_account,gcal_tz,current,
 483                               changed,gcal_id)
 484         else:
 485             toadd = match_ex_to_gcal(ex_account,gcal_account,gcal_tz,current,
 486                                      gcal_id)
 487             add_ex_to_gcal(ex_account,gcal_account,gcal_tz,current,
 488                            toadd,gcal_id)
 489
 490         with open(cachepath,"wb") as f:
 491             pickle.dump(current,f)
 492
 493         #If not looping, break here (after 1 run)
 494         if args.loop==False:
 495             break
 496         #otherwise, wait 10 minutes, then go round again
 497         time.sleep(600)
 498
 499 if __name__ == "__main__":
 500     main()
 501
 502