forum_modules/sximporter/importer.py

   1 # -*- coding: utf-8 -*-\r
   2 \r
   3 from xml.dom import minidom\r
   4 from datetime import datetime\r
   5 import time\r
   6 import re\r
   7 from django.utils.translation import ugettext as _\r
   8 from django.template.defaultfilters import slugify\r
   9 from orm import orm\r
  10 \r
  11 def getText(el):\r
  12     rc = ""\r
  13     for node in el.childNodes:\r
  14         if node.nodeType == node.TEXT_NODE:\r
  15             rc = rc + node.data\r
  16     return rc.strip()\r
  17 \r
  18 msstrip = re.compile(r'^(.*)\.\d+')\r
  19 def readTime(ts):\r
  20     noms = msstrip.match(ts)\r
  21     if noms:\r
  22         ts = noms.group(1)\r
  23 \r
  24     return datetime(*time.strptime(ts, '%Y-%m-%dT%H:%M:%S')[0:6])\r
  25 \r
  26 def readEl(el):\r
  27     return dict([(n.tagName.lower(), getText(n)) for n in el.childNodes if n.nodeType == el.ELEMENT_NODE])\r
  28 \r
  29 def readTable(dump, name):\r
  30     return [readEl(e) for e in minidom.parseString(dump.read("%s.xml" % name)).getElementsByTagName('row')]\r
  31 \r
  32 class UnknownUser(object):\r
  33     counter = 0\r
  34     def __init__(self):\r
  35         UnknownUser.counter += 1\r
  36         self.number = UnknownUser.counter\r
  37 \r
  38     def __str__(self):\r
  39         return _("Unknown user %(number)d") % {'number': self.number}\r
  40 \r
  41     def __unicode__(self):\r
  42         return self.__str__()\r
  43 \r
  44     def encode(self, *args):\r
  45         return self.__str__()\r
  46 \r
  47 class IdMapper(dict):\r
  48     def __getitem__(self, key):\r
  49         key = int(key)\r
  50         return super(IdMapper, self).get(key, key)\r
  51 \r
  52     def __setitem__(self, key, value):\r
  53         super(IdMapper, self).__setitem__(int(key), int(value))\r
  54 \r
  55 openidre = re.compile('^https?\:\/\/')\r
  56 def userimport(dump, options):\r
  57     users = readTable(dump, "Users")\r
  58 \r
  59     user_by_name = {}\r
  60     uidmapper = IdMapper()\r
  61     merged_users = []\r
  62 \r
  63     owneruid = options.get('owneruid', None)\r
  64     #check for empty values\r
  65     if not owneruid:\r
  66         owneruid = None\r
  67 \r
  68     for sxu in users:\r
  69         create = True\r
  70 \r
  71         if sxu.get('id') == '-1':\r
  72             continue\r
  73 \r
  74         if int(sxu.get('id')) == int(owneruid):\r
  75             osqau = orm.User.objects.get(id=1)\r
  76             uidmapper[owneruid] = 1\r
  77             uidmapper[-1] = 1\r
  78             create = False\r
  79         else:\r
  80             username = sxu.get('displayname', sxu.get('displaynamecleaned', sxu.get('realname', UnknownUser())))\r
  81 \r
  82             if not isinstance(username, UnknownUser) and username in user_by_name:\r
  83                 if options.get('mergesimilar', False) and sxu.get('email', 'INVALID') == user_by_name[username].email:\r
  84                     osqau = user_by_name[username]\r
  85                     create = False\r
  86                     uidmapper[sxu.get('id')] = osqau.id\r
  87                 else:\r
  88                     inc = 1\r
  89                     while ("%s %d" % (username, inc)) in user_by_name:\r
  90                         inc += 1\r
  91 \r
  92                     username = "%s %d" % (username, inc)\r
  93 \r
  94         sxbadges = sxu.get('badgesummary', None)\r
  95         badges = {'1':'0','2':'0','3':'0'}\r
  96 \r
  97         if sxbadges:\r
  98             badges.update(dict([b.split('=') for b in sxbadges.split()]))\r
  99 \r
 100         if create:\r
 101             osqau = orm.User(\r
 102                 id           = sxu.get('id'),\r
 103                 username     = unicode(username),\r
 104                 password     = '!',\r
 105                 email        = sxu.get('email', ''),\r
 106                 is_superuser = sxu.get('usertypeid') == '5',\r
 107                 is_staff     = sxu.get('usertypeid') == '4',\r
 108                 is_active    = True,\r
 109                 date_joined  = readTime(sxu.get('creationdate')),\r
 110                 about         = sxu.get('aboutme', ''),\r
 111                 date_of_birth = sxu.get('birthday', None) and readTime(sxu['birthday']) or None,\r
 112                 email_isvalid = int(sxu.get('usertypeid')) > 2,\r
 113                 website       = sxu.get('websiteurl', ''),\r
 114                 reputation    = int(sxu.get('reputation')),\r
 115                 gold          = int(badges['1']),\r
 116                 silver        = int(badges['2']),\r
 117                 bronze        = int(badges['3']),\r
 118                 real_name     = sxu.get('realname', ''),\r
 119             )\r
 120 \r
 121             osqau.save()\r
 122 \r
 123             s = orm.SubscriptionSettings(user=osqau)\r
 124             s.save()\r
 125 \r
 126             user_by_name[osqau.username] = osqau\r
 127         else:\r
 128             new_about = sxu.get('aboutme', None)\r
 129             if new_about and osqau.about != new_about:\r
 130                 if osqau.about:\r
 131                     osqau.about = "%s\n|\n%s" % (osqau.about, new_about)\r
 132                 else:\r
 133                     osqau.about = new_about\r
 134 \r
 135             osqau.username = sxu.get('displayname', sxu.get('displaynamecleaned', sxu.get('realname', UnknownUser())))\r
 136             osqau.email = sxu.get('email', '')\r
 137             osqau.reputation += int(sxu.get('reputation'))\r
 138             osqau.gold += int(badges['1'])\r
 139             osqau.silver += int(badges['2'])\r
 140             osqau.bronze += int(badges['3'])\r
 141 \r
 142             merged_users.append(osqau.id)\r
 143             osqau.save()\r
 144 \r
 145 \r
 146         openid = sxu.get('openid', None)\r
 147         if openid and openidre.match(openid):\r
 148             assoc = orm.AuthKeyUserAssociation(user=osqau, key=openid, provider="openidurl")\r
 149             assoc.save()\r
 150 \r
 151     if uidmapper[-1] == -1:\r
 152         uidmapper[-1] = 1\r
 153 \r
 154     return (uidmapper, merged_users)\r
 155 \r
 156 def tagsimport(dump, uidmap):\r
 157     tags = readTable(dump, "Tags")\r
 158 \r
 159     tagmap = {}\r
 160 \r
 161     for sxtag in tags:\r
 162         otag = orm.Tag(\r
 163             id = int(sxtag['id']),\r
 164             name = sxtag['name'],\r
 165             used_count = int(sxtag['count']),\r
 166             created_by_id = uidmap[sxtag.get('userid', 1)],\r
 167         )\r
 168         otag.save()\r
 169 \r
 170         tagmap[otag.name] = otag\r
 171 \r
 172     return tagmap\r
 173 \r
 174 def postimport(dump, uidmap, tagmap):\r
 175     history = {}\r
 176     all = {}\r
 177 \r
 178     for h in readTable(dump, "PostHistory"):\r
 179         if not history.get(h.get('postid'), None):\r
 180             history[h.get('postid')] = []\r
 181 \r
 182         history[h.get('postid')].append(h)\r
 183 \r
 184     posts = readTable(dump, "Posts")\r
 185 \r
 186     for sxpost in posts:\r
 187         accepted = {}\r
 188 \r
 189         postclass = sxpost.get('posttypeid') == '1' and orm.Question or orm.Answer\r
 190 \r
 191         post = postclass(\r
 192             id = sxpost['id'],\r
 193             added_at = readTime(sxpost['creationdate']),\r
 194             body = sxpost['body'],\r
 195             score = sxpost.get('score', 0),\r
 196             vote_up_count = 0,\r
 197             vote_down_count = 0\r
 198         )\r
 199 \r
 200         if sxpost.get('deletiondate', None):\r
 201             post.deleted = True\r
 202             post.deleted_at = readTime(sxpost['deletiondate'])\r
 203             post.author_id = 1\r
 204         else:\r
 205             post.author_id = uidmap[sxpost['owneruserid']]\r
 206 \r
 207         if sxpost.get('lasteditoruserid', None):\r
 208             post.last_edited_by_id = uidmap[sxpost.get('lasteditoruserid')]\r
 209             post.last_edited_at = readTime(sxpost['lasteditdate'])\r
 210 \r
 211         if sxpost.get('communityowneddate', None):\r
 212             post.wiki = True\r
 213             post.wikified_at = readTime(sxpost['communityowneddate'])\r
 214 \r
 215         if sxpost.get('posttypeid') == '1': #question\r
 216             post.node_type = "question"\r
 217             post.title = sxpost['title']\r
 218 \r
 219             tagnames = sxpost['tags'].replace(u'ö', '-').replace(u'é', '').replace(u'à', '')\r
 220             post.tagnames = tagnames\r
 221 \r
 222             post.view_count = sxpost.get('viewcount', 0)\r
 223             post.favourite_count = sxpost.get('favoritecount', 0)\r
 224             post.answer_count = sxpost.get('answercount', 0)\r
 225 \r
 226             if sxpost.get('lastactivityuserid', None):\r
 227                 post.last_activity_by_id = uidmap[sxpost['lastactivityuserid']]\r
 228                 post.last_activity_at = readTime(sxpost['lastactivitydate'])\r
 229 \r
 230             if sxpost.get('closeddate', None):\r
 231                 post.closed = True\r
 232                 post.closed_by_id = 1\r
 233                 post.closed_at = datetime.now()\r
 234 \r
 235             if sxpost.get('acceptedanswerid', None):\r
 236                 accepted[int(sxpost.get('acceptedanswerid'))] = post\r
 237 \r
 238         else:\r
 239             post.node_type = "answer"\r
 240             post.parent_id = sxpost['parentid']\r
 241 \r
 242             if int(post.id) in accepted:\r
 243                 question = accepted[int(post.id)]\r
 244                 question.accepted_answer_id = post\r
 245                 question.save()\r
 246 \r
 247                 post.accepted = True\r
 248                 post.accepted_at = datetime.now()\r
 249                 post.accepted_by_id = question.author_id\r
 250 \r
 251         all[int(post.id)] = post\r
 252 \r
 253     return all\r
 254 \r
 255 def comment_import(dump, uidmap, posts):\r
 256     comments = readTable(dump, "PostComments")\r
 257     currid = max(posts.keys())\r
 258     mapping = {}\r
 259 \r
 260     for sxc in comments:\r
 261         currid += 1\r
 262         oc = orm.Node(\r
 263             id = currid,\r
 264             node_type = "comment",\r
 265             added_at = readTime(sxc['creationdate']),\r
 266             author_id = uidmap[sxc['userid']],\r
 267             body = sxc['text'],\r
 268             parent_id = sxc.get('postid'),\r
 269             vote_up_count = 0,\r
 270             vote_down_count = 0\r
 271         )\r
 272 \r
 273         if sxc.get('deletiondate', None):\r
 274             oc.deleted = True\r
 275             oc.deleted_at = readTime(sxc['deletiondate'])\r
 276             oc.deleted_by_id = uidmap[sxc['deletionuserid']]\r
 277             oc.author_id = uidmap[sxc['deletionuserid']]\r
 278         else:\r
 279             oc.author_id = uidmap[sxc['userid']]\r
 280 \r
 281 \r
 282         posts[oc.id] = oc\r
 283         mapping[int(sxc['id'])] = int(oc.id)\r
 284 \r
 285     return posts, mapping\r
 286 \r
 287 \r
 288 def save_posts(posts, tagmap):\r
 289     for post in posts.values():\r
 290         post.save()\r
 291 \r
 292         if post.node_type == "question":\r
 293             tags = filter(lambda t: t is not None, [tagmap.get(n, None) for n in post.tagnames.split()])\r
 294             post.tagnames = " ".join([t.name for t in tags]).strip()\r
 295             post.tags = tags\r
 296 \r
 297         create_and_activate_revision(post)\r
 298 \r
 299 \r
 300 def create_and_activate_revision(post):\r
 301     rev = orm.NodeRevision(\r
 302         author_id = post.author_id,\r
 303         body = post.body,\r
 304         node_id = post.id,\r
 305         revised_at = post.added_at,\r
 306         revision = 1,\r
 307         summary = 'Initial revision',\r
 308         tagnames = post.tagnames,\r
 309         title = post.title,\r
 310     )\r
 311 \r
 312     rev.save()\r
 313     post.active_revision_id = rev.id\r
 314     post.save()\r
 315 \r
 316 \r
 317 def post_vote_import(dump, uidmap, posts):\r
 318     votes = readTable(dump, "Posts2Votes")\r
 319 \r
 320     for sxv in votes:\r
 321         if sxv['votetypeid'] in ('2', '3'):\r
 322             ov = orm.Vote(\r
 323                 node_id = sxv['postid'],\r
 324                 user_id = uidmap[sxv['userid']],\r
 325                 voted_at = readTime(sxv['creationdate']),\r
 326                 vote = sxv['votetypeid'] == '2' and 1 or -1,\r
 327             )\r
 328 \r
 329             if sxv['votetypeid'] == '2':\r
 330                 posts[int(sxv['postid'])].vote_up_count += 1\r
 331             else:\r
 332                 posts[int(sxv['postid'])].vote_down_count += 1\r
 333 \r
 334             ov.save()\r
 335 \r
 336 def comment_vote_import(dump, uidmap, comments, posts):\r
 337     votes = readTable(dump, "Comments2Votes")\r
 338 \r
 339     for sxv in votes:\r
 340         if sxv['votetypeid'] in ('2', '3'):\r
 341             ov = orm.Vote(\r
 342                 node_id = comments[int(sxv['postcommentid'])],\r
 343                 user_id = uidmap[sxv['userid']],\r
 344                 voted_at = readTime(sxv['creationdate']),\r
 345                 vote = sxv['votetypeid'] == '2' and 1 or -1,\r
 346             )\r
 347 \r
 348             if sxv['votetypeid'] == '2':\r
 349                 posts[comments[int(sxv['postcommentid'])]].vote_up_count += 1\r
 350             else:\r
 351                 posts[comments[int(sxv['postcommentid'])]].vote_down_count += 1\r
 352 \r
 353             ov.save()\r
 354 \r
 355 \r
 356 \r
 357 def badges_import(dump, uidmap):\r
 358     node_ctype = orm['contenttypes.contenttype'].objects.get(name='node')\r
 359     obadges = dict([(b.slug, b) for b in orm.Badge.objects.all()])\r
 360     sxbadges = dict([(int(b['id']), b) for b in readTable(dump, "Badges")])\r
 361 \r
 362     sx_to_osqa = {}\r
 363 \r
 364     for id, sxb in sxbadges.items():\r
 365         slug = slugify(sxb['name'].replace('&', 'and'))\r
 366         if slug in obadges:\r
 367             sx_to_osqa[id] = obadges[slug]\r
 368         else:\r
 369             osqab = orm.Badge(\r
 370                 name = sxb['name'],\r
 371                 slug = slugify(sxb['name']),\r
 372                 description = sxb['description'],\r
 373                 multiple = sxb.get('single', 'false') == 'false',\r
 374                 awarded_count = 0,\r
 375                 type = sxb['class']                \r
 376             )\r
 377             osqab.save()\r
 378             sx_to_osqa[id] = osqab\r
 379 \r
 380     sxawards = readTable(dump, "Users2Badges")\r
 381     osqaawards = []\r
 382 \r
 383     for sxa in sxawards:\r
 384         badge = sx_to_osqa[int(sxa['badgeid'])]\r
 385         osqaa = orm.Award(\r
 386             user_id = uidmap[sxa['userid']],\r
 387             badge = badge,\r
 388             content_type = node_ctype,\r
 389             object_id = 1\r
 390         )\r
 391 \r
 392         osqaawards.append(osqaa)\r
 393         badge.awarded_count += 1\r
 394 \r
 395     for b in sx_to_osqa.values():\r
 396         b.save()\r
 397 \r
 398     for a in osqaawards:\r
 399         a.save()\r
 400 \r
 401 \r
 402 def reset_sequences():\r
 403     from south.db import db\r
 404     if db.backend_name == "postgres":\r
 405         db.start_transaction()\r
 406         db.execute_many(PG_SEQUENCE_RESETS)\r
 407         db.commit_transaction()\r
 408 \r
 409 def sximport(dump, options):\r
 410     uidmap, merged_users = userimport(dump, options)\r
 411     tagmap = tagsimport(dump, uidmap)\r
 412     posts = postimport(dump, uidmap, tagmap)\r
 413     posts, comments = comment_import(dump, uidmap, posts)\r
 414     save_posts(posts, tagmap)\r
 415     post_vote_import(dump, uidmap, posts)\r
 416     comment_vote_import(dump, uidmap, comments, posts)\r
 417     for post in posts.values():\r
 418         post.save()\r
 419     badges_import(dump, uidmap)\r
 420 \r
 421     from south.db import db\r
 422     db.commit_transaction()\r
 423 \r
 424     reset_sequences()\r
 425 \r
 426     \r
 427     \r
 428 PG_SEQUENCE_RESETS = """\r
 429 SELECT setval('"auth_user_groups_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "auth_user_groups";\r
 430 SELECT setval('"auth_user_user_permissions_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "auth_user_user_permissions";\r
 431 SELECT setval('"activity_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "activity";\r
 432 SELECT setval('"forum_subscriptionsettings_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_subscriptionsettings";\r
 433 SELECT setval('"forum_validationhash_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_validationhash";\r
 434 SELECT setval('"forum_authkeyuserassociation_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_authkeyuserassociation";\r
 435 SELECT setval('"tag_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "tag";\r
 436 SELECT setval('"forum_markedtag_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_markedtag";\r
 437 SELECT setval('"forum_node_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node";\r
 438 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";\r
 439 SELECT setval('"forum_noderevision_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_noderevision";\r
 440 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";\r
 441 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";\r
 442 SELECT setval('"favorite_question_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "favorite_question";\r
 443 SELECT setval('"forum_questionsubscription_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_questionsubscription";\r
 444 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";\r
 445 SELECT setval('"vote_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "vote";\r
 446 SELECT setval('"flagged_item_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "flagged_item";\r
 447 SELECT setval('"badge_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "badge";\r
 448 SELECT setval('"award_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "award";\r
 449 SELECT setval('"repute_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "repute";\r
 450 SELECT setval('"forum_node_tags_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_node_tags";\r
 451 SELECT setval('"forum_keyvalue_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_keyvalue";\r
 452 SELECT setval('"forum_openidnonce_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_openidnonce";\r
 453 SELECT setval('"forum_openidassociation_id_seq"', coalesce(max("id"), 1) + 2, max("id") IS NOT null) FROM "forum_openidassociation";\r
 454 """\r
 455 \r
 456 \r
 457     \r
 458