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

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