forum_modules/sximporter/importer.py

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