forum_modules/sximporter/importer.py

   1 # -*- coding: utf-8 -*-
   2
   3 from datetime import datetime
   4 import time
   5 import re
   6 import os
   7 import gc
   8 from django.utils.translation import ugettext as _
   9
  10 from django.utils.encoding import force_unicode
  11
  12 try:
  13     from cPickle import loads, dumps
  14 except ImportError:
  15     from pickle import loads, dumps
  16
  17 from copy import deepcopy
  18 from base64 import b64encode, b64decode
  19 from zlib import compress, decompress
  20
  21 from xml.sax import make_parser
  22 from xml.sax.handler import ContentHandler
  23
  24 def create_orm():
  25     from django.conf import settings
  26     from south.orm import FakeORM
  27
  28     get_migration_number_re = re.compile(r'^((\d+)_.*)\.py$')
  29
  30     migrations_folder = os.path.join(settings.SITE_SRC_ROOT, 'forum/migrations')
  31
  32     highest_number = 0
  33     highest_file = None
  34
  35     for f in os.listdir(migrations_folder):
  36         if os.path.isfile(os.path.join(migrations_folder, f)):
  37             m = get_migration_number_re.match(f)
  38
  39             if m:
  40                 found = int(m.group(2))
  41
  42                 if found > highest_number:
  43                     highest_number = found
  44                     highest_file = m.group(1)
  45
  46     mod = __import__('forum.migrations.%s' % highest_file, globals(), locals(), ['forum.migrations'])
  47     return FakeORM(getattr(mod, 'Migration'), "forum")
  48
  49 orm = create_orm()
  50
  51 class SXTableHandler(ContentHandler):
  52     def __init__(self, fname, callback):
  53         self.in_row = False
  54         self.el_data = {}
  55         self.ch_data = ''
  56
  57         self.fname = fname.lower()
  58         self.callback = callback
  59
  60     def startElement(self, name, attrs):
  61         if name.lower() == self.fname:
  62             pass
  63         elif name.lower() == "row":
  64             self.in_row = True
  65
  66     def characters(self, ch):
  67         self.ch_data += ch
  68
  69     def endElement(self, name):
  70         if name.lower() == self.fname:
  71             pass
  72         elif name.lower() == "row":
  73             self.callback(self.el_data)
  74
  75             self.in_row = False
  76             del self.el_data
  77             self.el_data = {}
  78         elif self.in_row:
  79             self.el_data[name.lower()] = self.ch_data.strip()
  80             del self.ch_data
  81             self.ch_data = ''
  82
  83
  84 def readTable(path, name, callback):
  85     parser = make_parser()
  86     handler = SXTableHandler(name, callback)
  87     parser.setContentHandler(handler)
  88
  89     f = os.path.join(path, "%s.xml" % name)
  90     parser.parse(f)
  91
  92
  93 def dbsafe_encode(value):
  94     return force_unicode(b64encode(compress(dumps(deepcopy(value)))))
  95
  96 def getText(el):
  97     rc = ""
  98     for node in el.childNodes:
  99         if node.nodeType == node.TEXT_NODE:
 100             rc = rc + node.data
 101     return rc.strip()
 102
 103 msstrip = re.compile(r'^(.*)\.\d+')
 104 def readTime(ts):
 105     noms = msstrip.match(ts)
 106     if noms:
 107         ts = noms.group(1)
 108
 109     return datetime(*time.strptime(ts, '%Y-%m-%dT%H:%M:%S')[0:6])
 110
 111 #def readEl(el):
 112 #    return dict([(n.tagName.lower(), getText(n)) for n in el.childNodes if n.nodeType == el.ELEMENT_NODE])
 113
 114 #def readTable(dump, name):
 115 #    for e in minidom.parseString(dump.read("%s.xml" % name)).getElementsByTagName('row'):
 116 #        yield readEl(e)
 117 #return [readEl(e) for e in minidom.parseString(dump.read("%s.xml" % name)).getElementsByTagName('row')]
 118
 119 google_accounts_lookup = re.compile(r'^https?://www.google.com/accounts/')
 120 yahoo_accounts_lookup = re.compile(r'^https?://me.yahoo.com/a/')
 121
 122 openid_lookups = [
 123         re.compile(r'^https?://www.google.com/profiles/(?P<uname>\w+(\.\w+)*)/?$'),
 124         re.compile(r'^https?://me.yahoo.com/(?P<uname>\w+(\.\w+)*)/?$'),
 125         re.compile(r'^https?://openid.aol.com/(?P<uname>\w+(\.\w+)*)/?$'),
 126         re.compile(r'^https?://(?P<uname>\w+(\.\w+)*).myopenid.com/?$'),
 127         re.compile(r'^https?://flickr.com/(\w+/)*(?P<uname>\w+(\.\w+)*)/?$'),
 128         re.compile(r'^https?://technorati.com/people/technorati/(?P<uname>\w+(\.\w+)*)/?$'),
 129         re.compile(r'^https?://(?P<uname>\w+(\.\w+)*).wordpress.com/?$'),
 130         re.compile(r'^https?://(?P<uname>\w+(\.\w+)*).blogspot.com/?$'),
 131         re.compile(r'^https?://(?P<uname>\w+(\.\w+)*).livejournal.com/?$'),
 132         re.compile(r'^https?://claimid.com/(?P<uname>\w+(\.\w+)*)/?$'),
 133         re.compile(r'^https?://(?P<uname>\w+(\.\w+)*).pip.verisignlabs.com/?$'),
 134         re.compile(r'^https?://getopenid.com/(?P<uname>\w+(\.\w+)*)/?$'),
 135         re.compile(r'^https?://[\w\.]+/(\w+/)*(?P<uname>\w+(\.\w+)*)/?$'),
 136         re.compile(r'^https?://(?P<uname>[\w\.]+)/?$'),
 137         ]
 138
 139 def final_username_attempt(sxu):
 140     openid = sxu.get('openid', None)
 141
 142     if openid:
 143         if google_accounts_lookup.search(openid):
 144             return UnknownGoogleUser(sxu.get('id'))
 145         if yahoo_accounts_lookup.search(openid):
 146             return UnknownYahooUser(sxu.get('id'))
 147
 148         for lookup in openid_lookups:
 149             if lookup.search(openid):
 150                 return lookup.search(openid).group('uname')
 151
 152     return UnknownUser(sxu.get('id'))
 153
 154 class UnknownUser(object):
 155     def __init__(self, id):
 156         self._id = id
 157
 158     def __str__(self):
 159         return _("user-%(id)s") % {'id': self._id}
 160
 161     def __unicode__(self):
 162         return self.__str__()
 163
 164     def encode(self, *args):
 165         return self.__str__()
 166
 167 class UnknownGoogleUser(UnknownUser):
 168     def __str__(self):
 169         return _("user-%(id)s (google)") % {'id': self._id}
 170
 171 class UnknownYahooUser(UnknownUser):
 172     def __str__(self):
 173         return _("user-%(id)s (yahoo)") % {'id': self._id}
 174
 175
 176 class IdMapper(dict):
 177
 178     def __init__(self):
 179         self.default = 1
 180
 181     def __getitem__(self, key):
 182         key = int(key)
 183         return super(IdMapper, self).get(key, self.default)
 184
 185     def __setitem__(self, key, value):
 186         super(IdMapper, self).__setitem__(int(key), int(value))
 187
 188 class IdIncrementer():
 189     def __init__(self, initial):
 190         self.value = initial
 191
 192     def inc(self):
 193         self.value += 1
 194
 195 openidre = re.compile('^https?\:\/\/')
 196 def userimport(path, options):
 197
 198     usernames = []
 199     openids = set()
 200     uidmapper = IdMapper()
 201
 202     authenticated_user = options.get('authenticated_user', None)
 203     owneruid = options.get('owneruid', None)
 204     #check for empty values
 205     if not owneruid:
 206         owneruid = None
 207     else:
 208         owneruid = int(owneruid)
 209
 210     def callback(sxu):
 211         create = True
 212         set_mapper_defaults = False
 213
 214         if sxu.get('id') == '-1':
 215             return
 216         #print "\n".join(["%s : %s" % i for i in sxu.items()])
 217
 218         if (owneruid and (int(sxu.get('id')) == owneruid)) or (
 219             (not owneruid) and len(uidmapper)):
 220
 221             set_mapper_defaults = True
 222
 223             if authenticated_user:
 224                 osqau = orm.User.objects.get(id=authenticated_user.id)
 225
 226                 for assoc in orm.AuthKeyUserAssociation.objects.filter(user=osqau):
 227                     openids.add(assoc.key)
 228
 229                 uidmapper[owneruid] = osqau.id
 230                 create = False
 231
 232         sxbadges = sxu.get('badgesummary', None)
 233         badges = {'1':'0', '2':'0', '3':'0'}
 234
 235         if sxbadges:
 236             badges.update(dict([b.split('=') for b in sxbadges.split()]))
 237
 238         if create:
 239             username = unicode(sxu.get('displayname',
 240                                sxu.get('displaynamecleaned', sxu.get('realname', final_username_attempt(sxu)))))[:30]
 241
 242             if username in usernames:
 243             #if options.get('mergesimilar', False) and sxu.get('email', 'INVALID') == user_by_name[username].email:
 244             #    osqau = user_by_name[username]
 245             #    create = False
 246             #    uidmapper[sxu.get('id')] = osqau.id
 247             #else:
 248                 inc = 0
 249
 250                 while True:
 251                     inc += 1
 252                     totest = "%s %d" % (username[:29 - len(str(inc))], inc)
 253
 254                     if not totest in usernames:
 255                         username = totest
 256                         break
 257
 258             osqau = orm.User(
 259                     id           = sxu.get('id'),
 260                     username     = username,
 261                     password     = '!',
 262                     email        = sxu.get('email', ''),
 263                     is_superuser = sxu.get('usertypeid') == '5',
 264                     is_staff     = sxu.get('usertypeid') == '4',
 265                     is_active    = True,
 266                     date_joined  = readTime(sxu.get('creationdate')),
 267                     last_seen    = readTime(sxu.get('lastaccessdate')),
 268                     about         = sxu.get('aboutme', ''),
 269                     date_of_birth = sxu.get('birthday', None) and readTime(sxu['birthday']) or None,
 270                     email_isvalid = int(sxu.get('usertypeid')) > 2,
 271                     website       = sxu.get('websiteurl', ''),
 272                     reputation    = int(sxu.get('reputation')),
 273                     gold          = int(badges['1']),
 274                     silver        = int(badges['2']),
 275                     bronze        = int(badges['3']),
 276                     real_name     = sxu.get('realname', '')[:30],
 277                     location      = sxu.get('location', ''),
 278                     )
 279
 280             osqau.save()
 281
 282             user_joins = orm.Action(
 283                     action_type = "userjoins",
 284                     action_date = osqau.date_joined,
 285                     user = osqau
 286                     )
 287             user_joins.save()
 288
 289             rep = orm.ActionRepute(
 290                     value = 1,
 291                     user = osqau,
 292                     date = osqau.date_joined,
 293                     action = user_joins
 294                     )
 295             rep.save()
 296
 297             try:
 298                 orm.SubscriptionSettings.objects.get(user=osqau)
 299             except:
 300                 s = orm.SubscriptionSettings(user=osqau)
 301                 s.save()
 302
 303             uidmapper[osqau.id] = osqau.id
 304         else:
 305             new_about = sxu.get('aboutme', None)
 306             if new_about and osqau.about != new_about:
 307                 if osqau.about:
 308                     osqau.about = "%s\n|\n%s" % (osqau.about, new_about)
 309                 else:
 310                     osqau.about = new_about
 311
 312             osqau.username = sxu.get('displayname',
 313                                      sxu.get('displaynamecleaned', sxu.get('realname', final_username_attempt(sxu))))
 314             osqau.email = sxu.get('email', '')
 315             osqau.reputation += int(sxu.get('reputation'))
 316             osqau.gold += int(badges['1'])
 317             osqau.silver += int(badges['2'])
 318             osqau.bronze += int(badges['3'])
 319
 320             osqau.date_joined = readTime(sxu.get('creationdate'))
 321             osqau.website = sxu.get('websiteurl', '')
 322             osqau.date_of_birth = sxu.get('birthday', None) and readTime(sxu['birthday']) or None
 323             osqau.location = sxu.get('location', '')
 324             osqau.real_name = sxu.get('realname', '')
 325
 326             #merged_users.append(osqau.id)
 327             osqau.save()
 328
 329         if set_mapper_defaults:
 330             uidmapper[-1] = osqau.id
 331             uidmapper.default = osqau.id
 332
 333         usernames.append(osqau.username)
 334
 335         openid = sxu.get('openid', None)
 336         if openid and openidre.match(openid) and (not openid in openids):
 337             assoc = orm.AuthKeyUserAssociation(user=osqau, key=openid, provider="openidurl")
 338             assoc.save()
 339             openids.add(openid)
 340
 341         openidalt = sxu.get('openidalt', None)
 342         if openidalt and openidre.match(openidalt) and (not openidalt in openids):
 343             assoc = orm.AuthKeyUserAssociation(user=osqau, key=openidalt, provider="openidurl")
 344             assoc.save()
 345             openids.add(openidalt)
 346
 347     readTable(path, "Users", callback)
 348
 349     #if uidmapper[-1] == -1:
 350     #    uidmapper[-1] = 1
 351
 352     return uidmapper
 353
 354 def tagsimport(dump, uidmap):
 355
 356     tagmap = {}
 357
 358     def callback(sxtag):
 359         otag = orm.Tag(
 360                 id = int(sxtag['id']),
 361                 name = sxtag['name'],
 362                 used_count = int(sxtag['count']),
 363                 created_by_id = uidmap[sxtag.get('userid', 1)],
 364                 )
 365         otag.save()
 366
 367         tagmap[otag.name] = otag
 368
 369     readTable(dump, "Tags", callback)
 370
 371     return tagmap
 372
 373 def add_post_state(name, post, action):
 374     if not "(%s)" % name in post.state_string:
 375         post.state_string = "%s(%s)" % (post.state_string, name)
 376         post.save()
 377
 378     try:
 379         state = orm.NodeState.objects.get(node=post, state_type=name)
 380         state.action = action
 381         state.save()
 382     except:
 383         state = orm.NodeState(node=post, state_type=name, action=action)
 384         state.save()
 385
 386 def remove_post_state(name, post):
 387     if "(%s)" % name in post.state_string:
 388         try:
 389             state = orm.NodeState.objects.get(state_type=name, post=post)
 390             state.delete()
 391         except:
 392             pass
 393     post.state_string = "".join("(%s)" % s for s in re.findall('\w+', post.state_string) if s != name)
 394
 395 def postimport(dump, uidmap, tagmap):
 396     all = []
 397
 398     def callback(sxpost):
 399         nodetype = (sxpost.get('posttypeid') == '1') and "nodetype" or "answer"
 400
 401         post = orm.Node(
 402                 node_type = nodetype,
 403                 id = sxpost['id'],
 404                 added_at = readTime(sxpost['creationdate']),
 405                 body = sxpost['body'],
 406                 score = sxpost.get('score', 0),
 407                 author_id = sxpost.get('deletiondate', None) and 1 or uidmap[sxpost.get('owneruserid', 1)]
 408                 )
 409
 410         post.save()
 411
 412         create_action = orm.Action(
 413                 action_type = (nodetype == "nodetype") and "ask" or "answer",
 414                 user_id = post.author_id,
 415                 node = post,
 416                 action_date = post.added_at
 417                 )
 418
 419         create_action.save()
 420
 421         if sxpost.get('lasteditoruserid', None):
 422             revise_action = orm.Action(
 423                     action_type = "revise",
 424                     user_id = uidmap[sxpost.get('lasteditoruserid')],
 425                     node = post,
 426                     action_date = readTime(sxpost['lasteditdate']),
 427                     )
 428
 429             revise_action.save()
 430             post.last_edited = revise_action
 431
 432         if sxpost.get('communityowneddate', None):
 433             wikify_action = orm.Action(
 434                     action_type = "wikify",
 435                     user_id = 1,
 436                     node = post,
 437                     action_date = readTime(sxpost['communityowneddate'])
 438                     )
 439
 440             wikify_action.save()
 441             add_post_state("wiki", post, wikify_action)
 442
 443         if sxpost.get('lastactivityuserid', None):
 444             post.last_activity_by_id = uidmap[sxpost['lastactivityuserid']]
 445             post.last_activity_at = readTime(sxpost['lastactivitydate'])
 446
 447         if sxpost.get('posttypeid') == '1': #question
 448             post.node_type = "question"
 449             post.title = sxpost['title']
 450
 451             tagnames = sxpost['tags'].replace(u'ö', '-').replace(u'é', '').replace(u'à', '')
 452             post.tagnames = tagnames
 453
 454             post.extra_count = sxpost.get('viewcount', 0)
 455
 456             add_tags_to_post(post, tagmap)
 457
 458         else:
 459             post.parent_id = sxpost['parentid']
 460
 461         post.save()
 462
 463         all.append(int(post.id))
 464         create_and_activate_revision(post)
 465
 466         del post
 467
 468     readTable(dump, "Posts", callback)
 469
 470     return all
 471
 472 def comment_import(dump, uidmap, posts):
 473     currid = IdIncrementer(max(posts))
 474     mapping = {}
 475
 476     def callback(sxc):
 477         currid.inc()
 478         oc = orm.Node(
 479                 id = currid.value,
 480                 node_type = "comment",
 481                 added_at = readTime(sxc['creationdate']),
 482                 author_id = uidmap[sxc.get('userid', 1)],
 483                 body = sxc['text'],
 484                 parent_id = sxc.get('postid'),
 485                 )
 486
 487         if sxc.get('deletiondate', None):
 488             delete_action = orm.Action(
 489                     action_type = "delete",
 490                     user_id = uidmap[sxc['deletionuserid']],
 491                     action_date = readTime(sxc['deletiondate'])
 492                     )
 493
 494             oc.author_id = uidmap[sxc['deletionuserid']]
 495             oc.save()
 496
 497             delete_action.node = oc
 498             delete_action.save()
 499
 500             add_post_state("deleted", oc, delete_action)
 501         else:
 502             oc.author_id = uidmap[sxc.get('userid', 1)]
 503             oc.save()
 504
 505         create_action = orm.Action(
 506                 action_type = "comment",
 507                 user_id = oc.author_id,
 508                 node = oc,
 509                 action_date = oc.added_at
 510                 )
 511
 512         create_and_activate_revision(oc)
 513
 514         create_action.save()
 515         oc.save()
 516
 517         posts.append(int(oc.id))
 518         mapping[int(sxc['id'])] = int(oc.id)
 519
 520     readTable(dump, "PostComments", callback)
 521     return posts, mapping
 522
 523
 524 def add_tags_to_post(post, tagmap):
 525     tags = [tag for tag in [tagmap.get(name.strip()) for name in post.tagnames.split(u' ') if name] if tag]
 526     post.tagnames = " ".join([t.name for t in tags]).strip()
 527     post.tags = tags
 528
 529
 530 def create_and_activate_revision(post):
 531     rev = orm.NodeRevision(
 532             author_id = post.author_id,
 533             body = post.body,
 534             node_id = post.id,
 535             revised_at = post.added_at,
 536             revision = 1,
 537             summary = 'Initial revision',
 538             tagnames = post.tagnames,
 539             title = post.title,
 540             )
 541
 542     rev.save()
 543     post.active_revision_id = rev.id
 544     post.save()
 545
 546 def post_vote_import(dump, uidmap, posts):
 547     close_reasons = {}
 548
 549     def close_callback(r):
 550         close_reasons[r['id']] = r['name']
 551
 552     readTable(dump, "CloseReasons", close_callback)
 553
 554     user2vote = []
 555
 556     def callback(sxv):
 557         action = orm.Action(
 558                 user_id=uidmap[sxv['userid']],
 559                 action_date = readTime(sxv['creationdate']),
 560                 )
 561
 562         if not int(sxv['postid']) in posts: return
 563         node = orm.Node.objects.get(id=sxv['postid'])
 564         action.node = node
 565
 566         if sxv['votetypeid'] == '1':
 567             answer = node
 568             question = orm.Node.objects.get(id=answer.parent_id)
 569
 570             action.action_type = "acceptanswer"
 571             action.save()
 572
 573             answer.marked = True
 574
 575             question.extra_ref_id = answer.id
 576
 577             answer.save()
 578             question.save()
 579
 580         elif sxv['votetypeid'] in ('2', '3'):
 581             if not (action.node.id, action.user_id) in user2vote:
 582                 user2vote.append((action.node.id, action.user_id))
 583
 584                 action.action_type = (sxv['votetypeid'] == '2') and "voteup" or "votedown"
 585                 action.save()
 586
 587                 ov = orm.Vote(
 588                         node_id = action.node.id,
 589                         user_id = action.user_id,
 590                         voted_at = action.action_date,
 591                         value = sxv['votetypeid'] == '2' and 1 or -1,
 592                         action = action
 593                         )
 594                 ov.save()
 595             else:
 596                 action.action_type = "unknown"
 597                 action.save()
 598
 599         elif sxv['votetypeid'] in ('4', '12', '13'):
 600             action.action_type = "flag"
 601             action.save()
 602
 603             of = orm.Flag(
 604                     node = action.node,
 605                     user_id = action.user_id,
 606                     flagged_at = action.action_date,
 607                     reason = '',
 608                     action = action
 609                     )
 610
 611             of.save()
 612
 613         elif sxv['votetypeid'] == '5':
 614             action.action_type = "favorite"
 615             action.save()
 616
 617         elif sxv['votetypeid'] == '6':
 618             action.action_type = "close"
 619             action.extra = dbsafe_encode(close_reasons[sxv['comment']])
 620             action.save()
 621
 622             node.marked = True
 623             node.save()
 624
 625         elif sxv['votetypeid'] == '7':
 626             action.action_type = "unknown"
 627             action.save()
 628
 629             node.marked = False
 630             node.save()
 631
 632             remove_post_state("closed", node)
 633
 634         elif sxv['votetypeid'] == '10':
 635             action.action_type = "delete"
 636             action.save()
 637
 638         elif sxv['votetypeid'] == '11':
 639             action.action_type = "unknown"
 640             action.save()
 641
 642             remove_post_state("deleted", node)
 643
 644         else:
 645             action.action_type = "unknown"
 646             action.save()
 647
 648         if sxv.get('targetrepchange', None):
 649             rep = orm.ActionRepute(
 650                     action = action,
 651                     date = action.action_date,
 652                     user_id = uidmap[sxv['targetuserid']],
 653                     value = int(sxv['targetrepchange'])
 654                     )
 655
 656             rep.save()
 657
 658         if sxv.get('voterrepchange', None):
 659             rep = orm.ActionRepute(
 660                     action = action,
 661                     date = action.action_date,
 662                     user_id = uidmap[sxv['userid']],
 663                     value = int(sxv['voterrepchange'])
 664                     )
 665
 666             rep.save()
 667
 668         if action.action_type in ("acceptanswer", "delete", "close"):
 669             state = {"acceptanswer": "accepted", "delete": "deleted", "close": "closed"}[action.action_type]
 670             add_post_state(state, node, action)
 671
 672     readTable(dump, "Posts2Votes", callback)
 673
 674
 675 def comment_vote_import(dump, uidmap, comments):
 676     user2vote = []
 677     comments2score = {}
 678
 679     def callback(sxv):
 680         if sxv['votetypeid'] == "2":
 681             comment_id = comments[int(sxv['postcommentid'])]
 682             user_id = uidmap[sxv['userid']]
 683
 684             if not (comment_id, user_id) in user2vote:
 685                 user2vote.append((comment_id, user_id))
 686
 687                 action = orm.Action(
 688                         action_type = "voteupcomment",
 689                         user_id = user_id,
 690                         action_date = readTime(sxv['creationdate']),
 691                         node_id = comment_id
 692                         )
 693                 action.save()
 694
 695                 ov = orm.Vote(
 696                         node_id = comment_id,
 697                         user_id = user_id,
 698                         voted_at = action.action_date,
 699                         value = 1,
 700                         action = action
 701                         )
 702
 703                 ov.save()
 704
 705                 if not comment_id in comments2score:
 706                     comments2score[comment_id] = 1
 707                 else:
 708                     comments2score[comment_id] += 1
 709
 710     readTable(dump, "Comments2Votes", callback)
 711
 712     for cid, score in comments2score.items():
 713         orm.Node.objects.filter(id=cid).update(score=score)
 714
 715
 716 def badges_import(dump, uidmap, post_list):
 717
 718     sxbadges = {}
 719
 720     def sxcallback(b):
 721         sxbadges[int(b['id'])] = b
 722
 723     readTable(dump, "Badges", sxcallback)
 724
 725     obadges = dict([(b.cls, b) for b in orm.Badge.objects.all()])
 726     user_badge_count = {}
 727
 728     sx_to_osqa = {}
 729
 730     for id, sxb in sxbadges.items():
 731         cls = "".join(sxb['name'].replace('&', 'And').split(' '))
 732
 733         if cls in obadges:
 734             sx_to_osqa[id] = obadges[cls]
 735         else:
 736             osqab = orm.Badge(
 737                     cls = cls,
 738                     awarded_count = 0,
 739                     type = sxb['class']
 740                     )
 741             osqab.save()
 742             sx_to_osqa[id] = osqab
 743
 744     osqaawards = []
 745
 746     def callback(sxa):
 747         badge = sx_to_osqa[int(sxa['badgeid'])]
 748
 749         user_id = uidmap[sxa['userid']]
 750         if not user_badge_count.get(user_id, None):
 751             user_badge_count[user_id] = 0
 752
 753         action = orm.Action(
 754                 action_type = "award",
 755                 user_id = user_id,
 756                 action_date = readTime(sxa['date'])
 757                 )
 758
 759         action.save()
 760
 761         osqaa = orm.Award(
 762                 user_id = uidmap[sxa['userid']],
 763                 badge = badge,
 764                 node_id = post_list[user_badge_count[user_id]],
 765                 awarded_at = action.action_date,
 766                 action = action
 767                 )
 768
 769         osqaa.save()
 770         badge.awarded_count += 1
 771
 772         user_badge_count[user_id] += 1
 773
 774     readTable(dump, "Users2Badges", callback)
 775
 776     for badge in obadges.values():
 777         badge.save()
 778
 779 def save_setting(k, v):
 780     try:
 781         kv = orm.KeyValue.objects.get(key=k)
 782         kv.value = v
 783     except:
 784         kv = orm.KeyValue(key = k, value = v)
 785
 786     kv.save()
 787
 788
 789 def pages_import(dump, currid):
 790     currid = IdIncrementer(currid)
 791     registry = {}
 792
 793     def callback(sxp):
 794         currid.inc()
 795         page = orm.Node(
 796                 id = currid.value,
 797                 node_type = "page",
 798                 title = sxp['name'],
 799                 body = b64decode(sxp['value']),
 800                 extra = dbsafe_encode({
 801                 'path': sxp['url'][1:],
 802                 'mimetype': sxp['contenttype'],
 803                 'template': (sxp['usemaster'] == "true") and "default" or "none",
 804                 'render': "html",
 805                 'sidebar': "",
 806                 'sidebar_wrap': True,
 807                 'sidebar_render': "html",
 808                 'comments': False
 809                 }),
 810                 author_id = 1
 811                 )
 812
 813         create_and_activate_revision(page)
 814
 815         page.save()
 816         registry[sxp['url'][1:]] = page.id
 817
 818         create_action = orm.Action(
 819                 action_type = "newpage",
 820                 user_id = page.author_id,
 821                 node = page
 822                 )
 823
 824         create_action.save()
 825
 826         if sxp['active'] == "true" and sxp['contenttype'] == "text/html":
 827             pub_action = orm.Action(
 828                     action_type = "publish",
 829                     user_id = page.author_id,
 830                     node = page
 831                     )
 832
 833             pub_action.save()
 834             add_post_state("published", page, pub_action)
 835
 836     readTable(dump, "FlatPages", callback)
 837
 838     save_setting('STATIC_PAGE_REGISTRY', dbsafe_encode(registry))
 839
 840 sx2osqa_set_map = {
 841 u'theme.html.name': 'APP_TITLE',
 842 u'theme.html.footer': 'CUSTOM_FOOTER',
 843 u'theme.html.sidebar': 'SIDEBAR_UPPER_TEXT',
 844 u'theme.html.sidebar-low': 'SIDEBAR_LOWER_TEXT',
 845 u'theme.html.welcome': 'APP_INTRO',
 846 u'theme.html.head': 'CUSTOM_HEAD',
 847 u'theme.html.header': 'CUSTOM_HEADER',
 848 u'theme.css': 'CUSTOM_CSS',
 849 }
 850
 851 html_codes = (
 852 ('&amp;', '&'),
 853 ('&lt;', '<'),
 854 ('&gt;', '>'),
 855 ('&quot;', '"'),
 856 ('&#39;', "'"),
 857 )
 858
 859 def html_decode(html):
 860     html = force_unicode(html)
 861
 862     for args in html_codes:
 863         html = html.replace(*args)
 864
 865     return html
 866
 867
 868 def static_import(dump):
 869     sx_unknown = {}
 870
 871     def callback(set):
 872         if unicode(set['name']) in sx2osqa_set_map:
 873             save_setting(sx2osqa_set_map[set['name']], dbsafe_encode(html_decode(set['value'])))
 874         else:
 875             sx_unknown[set['name']] = html_decode(set['value'])
 876
 877     readTable(dump, "ThemeTextResources", callback)
 878
 879     save_setting('SXIMPORT_UNKNOWN_SETS', dbsafe_encode(sx_unknown))
 880
 881 def disable_triggers():
 882     from south.db import db
 883     if db.backend_name == "postgres":
 884         db.execute_many(PG_DISABLE_TRIGGERS)
 885         db.commit_transaction()
 886         db.start_transaction()
 887
 888 def enable_triggers():
 889     from south.db import db
 890     if db.backend_name == "postgres":
 891         db.start_transaction()
 892         db.execute_many(PG_ENABLE_TRIGGERS)
 893         db.commit_transaction()
 894
 895 def reset_sequences():
 896     from south.db import db
 897     if db.backend_name == "postgres":
 898         db.start_transaction()
 899         db.execute_many(PG_SEQUENCE_RESETS)
 900         db.commit_transaction()
 901
 902 def reindex_fts():
 903     from south.db import db
 904     if db.backend_name == "postgres":
 905         db.start_transaction()
 906         db.execute_many("UPDATE forum_noderevision set id = id WHERE TRUE;")
 907         db.commit_transaction()
 908
 909
 910 def sximport(dump, options):
 911     try:
 912         disable_triggers()
 913         triggers_disabled = True
 914     except:
 915         triggers_disabled = False
 916
 917     uidmap = userimport(dump, options)
 918     tagmap = tagsimport(dump, uidmap)
 919     gc.collect()
 920
 921     posts = postimport(dump, uidmap, tagmap)
 922     gc.collect()
 923
 924     posts, comments = comment_import(dump, uidmap, posts)
 925     gc.collect()
 926
 927     post_vote_import(dump, uidmap, posts)
 928     gc.collect()
 929
 930     comment_vote_import(dump, uidmap, comments)
 931     gc.collect()
 932
 933     badges_import(dump, uidmap, posts)
 934
 935     pages_import(dump, max(posts))
 936     static_import(dump)
 937     gc.collect()
 938
 939     from south.db import db
 940     db.commit_transaction()
 941
 942     reset_sequences()
 943
 944     if triggers_disabled:
 945         enable_triggers()
 946         reindex_fts()
 947
 948
 949 PG_DISABLE_TRIGGERS = """
 950 ALTER table auth_user DISABLE TRIGGER ALL;
 951 ALTER table auth_user_groups DISABLE TRIGGER ALL;
 952 ALTER table auth_user_user_permissions DISABLE TRIGGER ALL;
 953 ALTER table forum_keyvalue DISABLE TRIGGER ALL;
 954 ALTER table forum_action DISABLE TRIGGER ALL;
 955 ALTER table forum_actionrepute DISABLE TRIGGER ALL;
 956 ALTER table forum_subscriptionsettings DISABLE TRIGGER ALL;
 957 ALTER table forum_validationhash DISABLE TRIGGER ALL;
 958 ALTER table forum_authkeyuserassociation DISABLE TRIGGER ALL;
 959 ALTER table forum_tag DISABLE TRIGGER ALL;
 960 ALTER table forum_markedtag DISABLE TRIGGER ALL;
 961 ALTER table forum_node DISABLE TRIGGER ALL;
 962 ALTER table forum_nodestate DISABLE TRIGGER ALL;
 963 ALTER table forum_node_tags DISABLE TRIGGER ALL;
 964 ALTER table forum_noderevision DISABLE TRIGGER ALL;
 965 ALTER table forum_node_tags DISABLE TRIGGER ALL;
 966 ALTER table forum_questionsubscription DISABLE TRIGGER ALL;
 967 ALTER table forum_vote DISABLE TRIGGER ALL;
 968 ALTER table forum_flag DISABLE TRIGGER ALL;
 969 ALTER table forum_badge DISABLE TRIGGER ALL;
 970 ALTER table forum_award DISABLE TRIGGER ALL;
 971 ALTER table forum_openidnonce DISABLE TRIGGER ALL;
 972 ALTER table forum_openidassociation DISABLE TRIGGER ALL;
 973 """
 974
 975 PG_ENABLE_TRIGGERS = """
 976 ALTER table auth_user ENABLE TRIGGER ALL;
 977 ALTER table auth_user_groups ENABLE TRIGGER ALL;
 978 ALTER table auth_user_user_permissions ENABLE TRIGGER ALL;
 979 ALTER table forum_keyvalue ENABLE TRIGGER ALL;
 980 ALTER table forum_action ENABLE TRIGGER ALL;
 981 ALTER table forum_actionrepute ENABLE TRIGGER ALL;
 982 ALTER table forum_subscriptionsettings ENABLE TRIGGER ALL;
 983 ALTER table forum_validationhash ENABLE TRIGGER ALL;
 984 ALTER table forum_authkeyuserassociation ENABLE TRIGGER ALL;
 985 ALTER table forum_tag ENABLE TRIGGER ALL;
 986 ALTER table forum_markedtag ENABLE TRIGGER ALL;
 987 ALTER table forum_node ENABLE TRIGGER ALL;
 988 ALTER table forum_nodestate ENABLE TRIGGER ALL;
 989 ALTER table forum_node_tags ENABLE TRIGGER ALL;
 990 ALTER table forum_noderevision ENABLE TRIGGER ALL;
 991 ALTER table forum_node_tags ENABLE TRIGGER ALL;
 992 ALTER table forum_questionsubscription ENABLE TRIGGER ALL;
 993 ALTER table forum_vote ENABLE TRIGGER ALL;
 994 ALTER table forum_flag ENABLE TRIGGER ALL;
 995 ALTER table forum_badge ENABLE TRIGGER ALL;
 996 ALTER table forum_award ENABLE TRIGGER ALL;
 997 ALTER table forum_openidnonce ENABLE TRIGGER ALL;
 998 ALTER table forum_openidassociation ENABLE TRIGGER ALL;
 999 """
1000
1001 PG_SEQUENCE_RESETS = """
1002 SELECT setval('"auth_user_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "auth_user";
1003 SELECT setval('"auth_user_groups_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "auth_user_groups";
1004 SELECT setval('"auth_user_user_permissions_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "auth_user_user_permissions";
1005 SELECT setval('"forum_keyvalue_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_keyvalue";
1006 SELECT setval('"forum_action_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_action";
1007 SELECT setval('"forum_actionrepute_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_actionrepute";
1008 SELECT setval('"forum_subscriptionsettings_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_subscriptionsettings";
1009 SELECT setval('"forum_validationhash_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_validationhash";
1010 SELECT setval('"forum_authkeyuserassociation_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_authkeyuserassociation";
1011 SELECT setval('"forum_tag_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_tag";
1012 SELECT setval('"forum_markedtag_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_markedtag";
1013 SELECT setval('"forum_node_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node";
1014 SELECT setval('"forum_nodestate_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_nodestate";
1015 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";
1016 SELECT setval('"forum_noderevision_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_noderevision";
1017 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";
1018 SELECT setval('"forum_questionsubscription_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_questionsubscription";
1019 SELECT setval('"forum_vote_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_vote";
1020 SELECT setval('"forum_flag_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_flag";
1021 SELECT setval('"forum_badge_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_badge";
1022 SELECT setval('"forum_award_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_award";
1023 SELECT setval('"forum_openidnonce_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_openidnonce";
1024 SELECT setval('"forum_openidassociation_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_openidassociation";
1025 """
1026
1027
1028
1029