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