sfa/importer/dummyimporter.py

   1 #
   2 # Dummy importer
   3 #
   4 # requirements
   5 #
   6 # read the planetlab database and update the local registry database accordingly
   7 # so we update the following collections
   8 # . authorities                 (from pl sites)
   9 # . node                        (from pl nodes)
  10 # . users+keys                  (from pl persons and attached keys)
  11 #                       known limitation : *one* of the ssh keys is chosen at random here
  12 #                       xxx todo/check xxx at the very least, when a key is known to the registry
  13 #                       and is still current in plc
  14 #                       then we should definitely make sure to keep that one in sfa...
  15 # . slice+researchers           (from pl slices and attached users)
  16 #
  17
  18 import os
  19
  20 from sfa.util.config import Config
  21 from sfa.util.xrn import Xrn, get_leaf, get_authority, hrn_to_urn
  22
  23 from sfa.trust.gid import create_uuid
  24 from sfa.trust.certificate import convert_public_key, Keypair
  25
  26 from sfa.storage.alchemy import dbsession
  27 from sfa.storage.model import RegRecord, RegAuthority, RegSlice, RegNode, RegUser, RegKey
  28
  29 from sfa.dummy.dummyshell import DummyShell
  30 from sfa.dummy.dummyxrn import hostname_to_hrn, slicename_to_hrn, email_to_hrn, hrn_to_dummy_slicename
  31
  32 def _get_site_hrn(interface_hrn, site):
  33      hrn = ".".join([interface_hrn, site['name']])
  34      return hrn
  35
  36
  37 class DummyImporter:
  38
  39     def __init__ (self, auth_hierarchy, logger):
  40         self.auth_hierarchy = auth_hierarchy
  41         self.logger=logger
  42
  43     def add_options (self, parser):
  44         # we don't have any options for now
  45         pass
  46
  47     # hrn hash is initialized from current db
  48     # remember just-created records as we go
  49     # xxx might make sense to add a UNIQUE constraint in the db itself
  50     def remember_record_by_hrn (self, record):
  51         tuple = (record.type, record.hrn)
  52         if tuple in self.records_by_type_hrn:
  53             self.logger.warning ("DummyImporter.remember_record_by_hrn: duplicate (%s,%s)"%tuple)
  54             return
  55         self.records_by_type_hrn [ tuple ] = record
  56
  57     # ditto for pointer hash
  58     def remember_record_by_pointer (self, record):
  59         if record.pointer == -1:
  60             self.logger.warning ("DummyImporter.remember_record_by_pointer: pointer is void")
  61             return
  62         tuple = (record.type, record.pointer)
  63         if tuple in self.records_by_type_pointer:
  64             self.logger.warning ("DummyImporter.remember_record_by_pointer: duplicate (%s,%s)"%tuple)
  65             return
  66         self.records_by_type_pointer [ ( record.type, record.pointer,) ] = record
  67
  68     def remember_record (self, record):
  69         self.remember_record_by_hrn (record)
  70         self.remember_record_by_pointer (record)
  71
  72     def locate_by_type_hrn (self, type, hrn):
  73         return self.records_by_type_hrn.get ( (type, hrn), None)
  74
  75     def locate_by_type_pointer (self, type, pointer):
  76         return self.records_by_type_pointer.get ( (type, pointer), None)
  77
  78     # a convenience/helper function to see if a record is already known
  79     # a former, broken, attempt (in 2.1-9) had been made
  80     # to try and use 'pointer' as a first, most significant attempt
  81     # the idea being to preserve stuff as much as possible, and thus
  82     # to avoid creating a new gid in the case of a simple hrn rename
  83     # however this of course doesn't work as the gid depends on the hrn...
  84     #def locate (self, type, hrn=None, pointer=-1):
  85     #    if pointer!=-1:
  86     #        attempt = self.locate_by_type_pointer (type, pointer)
  87     #        if attempt : return attempt
  88     #    if hrn is not None:
  89     #        attempt = self.locate_by_type_hrn (type, hrn,)
  90     #        if attempt : return attempt
  91     #    return None
  92
  93     # this makes the run method a bit abtruse - out of the way
  94
  95     def run (self, options):
  96         config = Config ()
  97         interface_hrn = config.SFA_INTERFACE_HRN
  98         root_auth = config.SFA_REGISTRY_ROOT_AUTH
  99         shell = DummyShell (config)
 100
 101         ######## retrieve all existing SFA objects
 102         all_records = dbsession.query(RegRecord).all()
 103
 104         # create hash by (type,hrn)
 105         # we essentially use this to know if a given record is already known to SFA
 106         self.records_by_type_hrn = \
 107             dict ( [ ( (record.type, record.hrn) , record ) for record in all_records ] )
 108         # create hash by (type,pointer)
 109         self.records_by_type_pointer = \
 110             dict ( [ ( (record.type, record.pointer) , record ) for record in all_records
 111                      if record.pointer != -1] )
 112
 113         # initialize record.stale to True by default, then mark stale=False on the ones that are in use
 114         for record in all_records: record.stale=True
 115
 116         ######## retrieve Dummy TB data
 117         # Get all plc sites
 118         # retrieve only required stuf
 119         sites = [shell.GetTestbedInfo()]
 120         # create a hash of sites by login_base
 121 #        sites_by_login_base = dict ( [ ( site['login_base'], site ) for site in sites ] )
 122         # Get all dummy TB users
 123         users = shell.GetUsers()
 124         # create a hash of users by user_id
 125         users_by_id = dict ( [ ( user['user_id'], user) for user in users ] )
 126         # Get all dummy TB public keys
 127         keys = []
 128         for user in users:
 129             if 'keys' in user:
 130                 keys.extend(user['keys'])
 131         # create a dict user_id -> [ keys ]
 132         keys_by_person_id = {}
 133         for user in users:
 134              if 'keys' in user:
 135                  keys_by_person_id[user['user_id']] = user['keys']
 136         # Get all dummy TB nodes
 137         nodes = shell.GetNodes()
 138         # create hash by node_id
 139         nodes_by_id = dict ( [ ( node['node_id'], node, ) for node in nodes ] )
 140         # Get all dummy TB slices
 141         slices = shell.GetSlices()
 142         # create hash by slice_id
 143         slices_by_id = dict ( [ (slice['slice_id'], slice ) for slice in slices ] )
 144
 145
 146         # start importing
 147         for site in sites:
 148             site_hrn = _get_site_hrn(interface_hrn, site)
 149             # import if hrn is not in list of existing hrns or if the hrn exists
 150             # but its not a site record
 151             site_record=self.locate_by_type_hrn ('authority', site_hrn)
 152             if not site_record:
 153                 try:
 154                     urn = hrn_to_urn(site_hrn, 'authority')
 155                     if not self.auth_hierarchy.auth_exists(urn):
 156                         self.auth_hierarchy.create_auth(urn)
 157                     auth_info = self.auth_hierarchy.get_auth_info(urn)
 158                     site_record = RegAuthority(hrn=site_hrn, gid=auth_info.get_gid_object(),
 159                                                pointer= -1,
 160                                                authority=get_authority(site_hrn))
 161                     site_record.just_created()
 162                     dbsession.add(site_record)
 163                     dbsession.commit()
 164                     self.logger.info("DummyImporter: imported authority (site) : %s" % site_record)
 165                     self.remember_record (site_record)
 166                 except:
 167                     # if the site import fails then there is no point in trying to import the
 168                     # site's child records (node, slices, persons), so skip them.
 169                     self.logger.log_exc("DummyImporter: failed to import site. Skipping child records")
 170                     continue
 171             else:
 172                 # xxx update the record ...
 173                 pass
 174             site_record.stale=False
 175
 176             # import node records
 177             for node in nodes:
 178                 site_auth = get_authority(site_hrn)
 179                 site_name = site['name']
 180                 node_hrn =  hostname_to_hrn(site_auth, site_name, node['hostname'])
 181                 # xxx this sounds suspicious
 182                 if len(node_hrn) > 64: node_hrn = node_hrn[:64]
 183                 node_record = self.locate_by_type_hrn ( 'node', node_hrn )
 184                 if not node_record:
 185                     try:
 186                         pkey = Keypair(create=True)
 187                         urn = hrn_to_urn(node_hrn, 'node')
 188                         node_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 189                         node_record = RegNode (hrn=node_hrn, gid=node_gid,
 190                                                pointer =node['node_id'],
 191                                                authority=get_authority(node_hrn))
 192                         node_record.just_created()
 193                         dbsession.add(node_record)
 194                         dbsession.commit()
 195                         self.logger.info("DummyImporter: imported node: %s" % node_record)
 196                         self.remember_record (node_record)
 197                     except:
 198                         self.logger.log_exc("DummyImporter: failed to import node")
 199                 else:
 200                     # xxx update the record ...
 201                     pass
 202                 node_record.stale=False
 203
 204             site_pis=[]
 205             # import users
 206             for user in users:
 207                 user_hrn = email_to_hrn(site_hrn, user['email'])
 208                 # xxx suspicious again
 209                 if len(user_hrn) > 64: user_hrn = user_hrn[:64]
 210                 user_urn = hrn_to_urn(user_hrn, 'user')
 211
 212                 user_record = self.locate_by_type_hrn ( 'user', user_hrn)
 213
 214                 # return a tuple pubkey (a dummy TB key object) and pkey (a Keypair object)
 215
 216                 def init_user_key (user):
 217                     pubkey = None
 218                     pkey = None
 219                     if  user['keys']:
 220                         # randomly pick first key in set
 221                         for key in user['keys']:
 222                              pubkey = key
 223                              try:
 224                                 pkey = convert_public_key(pubkey)
 225                                 break
 226                              except:
 227                                 continue
 228                         if not pkey:
 229                             self.logger.warn('DummyImporter: unable to convert public key for %s' % user_hrn)
 230                             pkey = Keypair(create=True)
 231                     else:
 232                         # the user has no keys. Creating a random keypair for the user's gid
 233                         self.logger.warn("DummyImporter: user %s does not have a NITOS public key"%user_hrn)
 234                         pkey = Keypair(create=True)
 235                     return (pubkey, pkey)
 236
 237                 # new user
 238                 try:
 239                     if not user_record:
 240                         (pubkey,pkey) = init_user_key (user)
 241                         user_gid = self.auth_hierarchy.create_gid(user_urn, create_uuid(), pkey)
 242                         user_gid.set_email(user['email'])
 243                         user_record = RegUser (hrn=user_hrn, gid=user_gid,
 244                                                  pointer=user['user_id'],
 245                                                  authority=get_authority(user_hrn),
 246                                                  email=user['email'])
 247                         if pubkey:
 248                             user_record.reg_keys=[RegKey (pubkey)]
 249                         else:
 250                             self.logger.warning("No key found for user %s"%user_record)
 251                         user_record.just_created()
 252                         dbsession.add (user_record)
 253                         dbsession.commit()
 254                         self.logger.info("DummyImporter: imported person: %s" % user_record)
 255                         self.remember_record ( user_record )
 256
 257                     else:
 258                         # update the record ?
 259                         # if user's primary key has changed then we need to update the
 260                         # users gid by forcing an update here
 261                         sfa_keys = user_record.reg_keys
 262                         def key_in_list (key,sfa_keys):
 263                             for reg_key in sfa_keys:
 264                                 if reg_key.key==key: return True
 265                             return False
 266                         # is there a new key in Dummy TB ?
 267                         new_keys=False
 268                         for key in user['keys']:
 269                             if not key_in_list (key,sfa_keys):
 270                                 new_keys = True
 271                         if new_keys:
 272                             (pubkey,pkey) = init_user_key (user)
 273                             user_gid = self.auth_hierarchy.create_gid(user_urn, create_uuid(), pkey)
 274                             if not pubkey:
 275                                 user_record.reg_keys=[]
 276                             else:
 277                                 user_record.reg_keys=[ RegKey (pubkey)]
 278                             self.logger.info("DummyImporter: updated person: %s" % user_record)
 279                     user_record.email = user['email']
 280                     dbsession.commit()
 281                     user_record.stale=False
 282                 except:
 283                     self.logger.log_exc("DummyImporter: failed to import user %d %s"%(user['user_id'],user['email']))
 284
 285
 286             # import slices
 287             for slice in slices:
 288                 slice_hrn = slicename_to_hrn(site_hrn, slice['slice_name'])
 289                 slice_record = self.locate_by_type_hrn ('slice', slice_hrn)
 290                 if not slice_record:
 291                     try:
 292                         pkey = Keypair(create=True)
 293                         urn = hrn_to_urn(slice_hrn, 'slice')
 294                         slice_gid = self.auth_hierarchy.create_gid(urn, create_uuid(), pkey)
 295                         slice_record = RegSlice (hrn=slice_hrn, gid=slice_gid,
 296                                                  pointer=slice['slice_id'],
 297                                                  authority=get_authority(slice_hrn))
 298                         slice_record.just_created()
 299                         dbsession.add(slice_record)
 300                         dbsession.commit()
 301                         self.logger.info("DummyImporter: imported slice: %s" % slice_record)
 302                         self.remember_record ( slice_record )
 303                     except:
 304                         self.logger.log_exc("DummyImporter: failed to import slice")
 305                 else:
 306                     # xxx update the record ...
 307                     self.logger.warning ("Slice update not yet implemented")
 308                     pass
 309                 # record current users affiliated with the slice
 310                 slice_record.reg_researchers = \
 311                     [ self.locate_by_type_pointer ('user',user_id) for user_id in slice['user_ids'] ]
 312                 dbsession.commit()
 313                 slice_record.stale=False
 314
 315         ### remove stale records
 316         # special records must be preserved
 317         system_hrns = [interface_hrn, root_auth, interface_hrn + '.slicemanager']
 318         for record in all_records:
 319             if record.hrn in system_hrns:
 320                 record.stale=False
 321             if record.peer_authority:
 322                 record.stale=False
 323
 324         for record in all_records:
 325             try:        stale=record.stale
 326             except:
 327                 stale=True
 328                 self.logger.warning("stale not found with %s"%record)
 329             if stale:
 330                 self.logger.info("DummyImporter: deleting stale record: %s" % record)
 331                 dbsession.delete(record)
 332                 dbsession.commit()