def gmail_initial_sync(crispin_client, db_session, log, folder_name, shared_state, local_uids, uid_download_stack, msg_create_fn): remote_uid_count = len(set(crispin_client.all_uids())) remote_g_metadata, sync_info = get_g_metadata( crispin_client, db_session, log, folder_name, local_uids, shared_state['syncmanager_lock']) sync_type, update_uid_count = sync_info remote_uids = sorted(remote_g_metadata.keys(), key=int) log.info(remote_uid_count=len(remote_uids)) if folder_name == crispin_client.folder_names()['all']: log.info(local_uid_count=len(local_uids)) with shared_state['syncmanager_lock']: log.debug('gmail_initial_sync grabbed syncmanager_lock') deleted_uids = remove_deleted_uids( crispin_client.account_id, db_session, log, folder_name, local_uids, remote_uids) delete_uid_count = len(deleted_uids) local_uids = set(local_uids) - deleted_uids unknown_uids = set(remote_uids) - local_uids # Persist the num(messages) to sync (any type of sync: download, # update or delete) before we start. # Note that num_local_deleted, num_local_updated ARE the numbers to # delete/update too since we make those changes rightaway before we start # downloading messages. update_uid_counts(db_session, log, crispin_client.account_id, folder_name, remote_uid_count=remote_uid_count, download_uid_count=len(unknown_uids), update_uid_count=update_uid_count, delete_uid_count=delete_uid_count, sync_type=sync_type) if folder_name == crispin_client.folder_names()['inbox']: # We don't do an initial dedupe for Inbox because we do thread # expansion, which means even if we have a given msgid downloaded, we # miiight not have the whole thread. This means that restarts cause # duplicate work, but hopefully these folders aren't too huge. message_download_stack = LifoQueue() flags = crispin_client.flags(unknown_uids) for uid in unknown_uids: if uid in flags: message_download_stack.put( GMessage(uid, remote_g_metadata[uid], flags[uid].flags, flags[uid].labels)) new_uid_poller = spawn(check_new_g_thrids, crispin_client.account_id, crispin_client.PROVIDER, folder_name, log, message_download_stack, shared_state['poll_frequency'], shared_state['syncmanager_lock']) download_queued_threads(crispin_client, db_session, log, folder_name, message_download_stack, shared_state['syncmanager_lock']) elif folder_name in uid_download_folders(crispin_client): full_download = deduplicate_message_download( crispin_client, db_session, log, shared_state['syncmanager_lock'], remote_g_metadata, unknown_uids) add_uids_to_stack(full_download, uid_download_stack) new_uid_poller = spawn(check_new_uids, crispin_client.account_id, crispin_client.PROVIDER, folder_name, log, uid_download_stack, shared_state['poll_frequency'], shared_state['syncmanager_lock']) download_queued_uids(crispin_client, db_session, log, folder_name, uid_download_stack, len(local_uids), len(unknown_uids), shared_state['syncmanager_lock'], gmail_download_and_commit_uids, msg_create_fn) else: raise MailsyncError( 'Unknown Gmail sync folder: {}'.format(folder_name)) # Complete X-GM-MSGID mapping is no longer needed after initial sync. rm_cache(remote_g_metadata_cache_file(crispin_client.account_id, folder_name)) new_uid_poller.kill()
def gmail_initial_sync(crispin_client, log, folder_name, shared_state, local_uids, uid_download_stack, msg_create_fn): remote_uid_count = len(set(crispin_client.all_uids())) remote_g_metadata, update_uid_count = get_g_metadata( crispin_client, log, folder_name, local_uids, shared_state['syncmanager_lock']) remote_uids = sorted(remote_g_metadata.keys(), key=int) log.info(remote_uid_count=len(remote_uids)) if folder_name == crispin_client.folder_names()['all']: log.info(local_uid_count=len(local_uids)) with shared_state['syncmanager_lock']: log.debug('gmail_initial_sync grabbed syncmanager_lock') with session_scope(ignore_soft_deletes=False) as db_session: deleted_uids = remove_deleted_uids( crispin_client.account_id, db_session, log, folder_name, local_uids, remote_uids) delete_uid_count = len(deleted_uids) local_uids = set(local_uids) - deleted_uids unknown_uids = set(remote_uids) - local_uids # Persist the num(messages) to sync (any type of sync: download, # update or delete) before we start. # Note that num_local_deleted, num_local_updated ARE the numbers to # delete/update too since we make those changes rightaway before we # start downloading messages. update_uid_counts(db_session, log, crispin_client.account_id, folder_name, remote_uid_count=remote_uid_count, download_uid_count=len(unknown_uids), update_uid_count=update_uid_count, delete_uid_count=delete_uid_count) if folder_name == crispin_client.folder_names()['inbox']: # We don't do an initial dedupe for Inbox because we do thread # expansion, which means even if we have a given msgid downloaded, we # miiight not have the whole thread. This means that restarts cause # duplicate work, but hopefully these folders aren't too huge. message_download_stack = LifoQueue() flags = crispin_client.flags(unknown_uids) for uid in unknown_uids: if uid in flags: message_download_stack.put( GMessage(uid, remote_g_metadata[uid], flags[uid].flags, flags[uid].labels)) new_uid_poller = spawn(check_new_g_thrids, crispin_client.account_id, crispin_client.PROVIDER, folder_name, log, message_download_stack, shared_state['poll_frequency'], shared_state['syncmanager_lock']) download_queued_threads(crispin_client, log, folder_name, message_download_stack, shared_state['syncmanager_lock']) elif folder_name in uid_download_folders(crispin_client): full_download = deduplicate_message_download( crispin_client, log, shared_state['syncmanager_lock'], remote_g_metadata, unknown_uids) add_uids_to_stack(full_download, uid_download_stack) new_uid_poller = spawn(check_new_uids, crispin_client.account_id, folder_name, log, uid_download_stack, shared_state['poll_frequency'], shared_state['syncmanager_lock']) download_queued_uids(crispin_client, log, folder_name, uid_download_stack, len(local_uids), len(unknown_uids), shared_state['syncmanager_lock'], gmail_download_and_commit_uids, msg_create_fn) else: raise MailsyncError( 'Unknown Gmail sync folder: {}'.format(folder_name)) # Complete X-GM-MSGID mapping is no longer needed after initial sync. rm_cache(remote_g_metadata_cache_file(crispin_client.account_id, folder_name)) new_uid_poller.kill()
def check_new_g_thrids(account_id, provider, folder_name, log, message_download_stack, poll_frequency, syncmanager_lock): """ Check for new X-GM-THRIDs and add them to the download stack. We do this by comparing local UID lists to remote UID lists, maintaining the invariant that (stack uids)+(local uids) == (remote uids). We also remove local messages that have disappeared from the remote, since it's totally probable that users will be archiving mail as the initial sync goes on. We grab a new IMAP connection from the pool for this to isolate its actions from whatever the main greenlet may be doing. Runs until killed. (Intended to be run in a greenlet.) """ with connection_pool(account_id).get() as crispin_client: with session_scope(ignore_soft_deletes=False) as db_session: crispin_client.select_folder(folder_name, uidvalidity_cb( db_session, crispin_client.account_id)) while True: log.info('Checking for new/deleted messages during initial sync.') remote_uids = set(crispin_client.all_uids()) # We lock this section to make sure no messages are being modified # in the database while we make sure the queue is in a good state. with syncmanager_lock: log.debug('check_new_g_thrids acquired syncmanager_lock') with session_scope(ignore_soft_deletes=False) as db_session: local_uids = set(account.all_uids(account_id, db_session, folder_name)) stack_uids = {gm.uid for gm in message_download_stack.queue} local_with_pending_uids = local_uids | stack_uids deleted_uids = remove_deleted_uids( account_id, db_session, log, folder_name, local_uids, remote_uids) log.info(deleted_uid_count=len(deleted_uids)) # filter out messages that have disappeared on the remote side new_message_download_stack = [gm for gm in message_download_stack.queue if gm.uid in remote_uids] # add in any new uids from the remote new_uids = [uid for uid in remote_uids if uid not in local_with_pending_uids] flags = crispin_client.flags(new_uids) g_metadata = crispin_client.g_metadata(new_uids) log.info('adding new messages to download queue', count=min(len(flags), len(g_metadata))) for new_uid in new_uids: # could have disappeared from the folder in the meantime if new_uid in flags and new_uid in g_metadata: new_message_download_stack.append( GMessage(new_uid, g_metadata[new_uid], flags[new_uid].flags, flags[new_uid].labels)) message_download_stack.queue = sorted( new_message_download_stack, key=lambda m: m.uid) log.info('idling', timeout=poll_frequency) crispin_client.conn.idle() crispin_client.conn.idle_check(timeout=poll_frequency) crispin_client.conn.idle_done() log.info('IDLE detected changes or timeout reached')
def check_new_g_thrids(account_id, provider_name, folder_name, log, message_download_stack, poll_frequency, syncmanager_lock): """ Check for new X-GM-THRIDs and add them to the download stack. We do this by comparing local UID lists to remote UID lists, maintaining the invariant that (stack uids)+(local uids) == (remote uids). We also remove local messages that have disappeared from the remote, since it's totally probable that users will be archiving mail as the initial sync goes on. We grab a new IMAP connection from the pool for this to isolate its actions from whatever the main greenlet may be doing. Runs until killed. (Intended to be run in a greenlet.) """ with connection_pool(account_id).get() as crispin_client: crispin_client.select_folder(folder_name, uidvalidity_cb(crispin_client.account_id)) while True: log.info('Checking for new/deleted messages during initial sync.') remote_uids = set(crispin_client.all_uids()) # We lock this section to make sure no messages are being modified # in the database while we make sure the queue is in a good state. with syncmanager_lock: log.debug('check_new_g_thrids acquired syncmanager_lock') with session_scope(ignore_soft_deletes=False) as db_session: local_uids = set(account.all_uids(account_id, db_session, folder_name)) stack_uids = {gm.uid for gm in message_download_stack.queue} local_with_pending_uids = local_uids | stack_uids deleted_uids = remove_deleted_uids( account_id, db_session, log, folder_name, local_uids, remote_uids) log.info(deleted_uid_count=len(deleted_uids)) # filter out messages that have disappeared on the remote side new_message_download_stack = [gm for gm in message_download_stack.queue if gm.uid in remote_uids] # add in any new uids from the remote new_uids = [uid for uid in remote_uids if uid not in local_with_pending_uids] flags = crispin_client.flags(new_uids) g_metadata = crispin_client.g_metadata(new_uids) log.info('adding new messages to download queue', count=min(len(flags), len(g_metadata))) for new_uid in new_uids: # could have disappeared from the folder in the meantime if new_uid in flags and new_uid in g_metadata: new_message_download_stack.append( GMessage(new_uid, g_metadata[new_uid], flags[new_uid].flags, flags[new_uid].labels)) message_download_stack.queue = sorted( new_message_download_stack, key=lambda m: m.uid) with session_scope(ignore_soft_deletes=False) as db_session: update_uid_counts( db_session, log, crispin_client.account_id, folder_name, remote_uid_count=len(remote_uids), download_uid_count=message_download_stack.qsize(), delete_uid_count=len(deleted_uids)) log.info('idling', timeout=poll_frequency) crispin_client.conn.idle() crispin_client.conn.idle_check(timeout=poll_frequency) crispin_client.conn.idle_done() log.info('IDLE detected changes or timeout reached')