question.py

import collections
import logging
import operator
import regex as re

from copy import copy
from django.conf import settings as django_settings
from django.db import models
from django.db.models import Count, F, Q
from django.contrib.auth.models import User
from django.contrib.contenttypes.models import ContentType
from django.core import cache  # import cache, not from cache import cache, to be able to monkey-patch cache.cache in test cases
from django.core import exceptions as django_exceptions
from django.template.loader import get_template
from django.template import Context
from django.utils.translation import gettext as _
from django.utils.translation import get_language
from django.utils import timezone

import askbot
from askbot.conf import settings as askbot_settings
from askbot.models.tag import Tag, TagSynonym
from askbot.models.tag import get_tags_by_names
from askbot.models.tag import filter_accepted_tags, filter_suggested_tags
from askbot.models.tag import separate_unused_tags
from askbot.models.base import BaseQuerySetManager
from askbot.models.base import DraftContent, AnonymousContent
from askbot.models.user import Activity, Group, PERSONAL_GROUP_NAME_PREFIX
from askbot.models.fields import LanguageCodeField
from askbot import signals
from askbot import const
from askbot.utils.lists import LazyList
from askbot.utils.loading import load_plugin
from askbot.search import mysql
from askbot.utils import translation as translation_utils
from askbot.search.state_manager import DummySearchState

LOG = logging.getLogger(__name__)


def clean_tagnames(tagnames):
    """Cleans tagnames string so that the field fits the constraint of the
    database.
    TODO: remove this when the Thread.tagnames field is converted into
    text_field
    """
    tagnames = tagnames.strip().split()
    # see if the tagnames field fits into 125 bytes
    while True:
        encoded_tagnames = ' '.join(tagnames).encode('utf-8')
        length = len(encoded_tagnames)
        if length == 0:
            return ''
        elif length <= 125:
            return ' '.join(tagnames)
        else:
            tagnames.pop()


def default_title_renderer(thread):
    """renders thread title,
    can be overridden by setting
    ASKBOT_QUESTION_TITLE_RENDERER
    """
    if thread.is_private():
        attr = const.POST_STATUS['private']
    elif thread.closed:
        attr = const.POST_STATUS['closed']
    elif thread.deleted:
        attr = const.POST_STATUS['deleted']
    else:
        attr = None
    if attr is not None:
        return '%s %s' % (thread.title, str(attr))
    else:
        return thread.title


class ThreadQuerySet(models.query.QuerySet):

    def get_visible(self, user):
        """filters out threads not belonging to the user groups"""
        if user.is_authenticated:
            groups = user.get_groups()
        else:
            groups = [Group.objects.get_global_group()]
        return self.filter(groups__in=groups).distinct()

    def get_for_title_query(self, search_query):
        """returns threads matching title query
        TODO: possibly add tags
        TODO: implement full text search on relevant fields
        """

        if getattr(django_settings, 'ENABLE_HAYSTACK_SEARCH', False):
            from askbot.search.haystack.helpers import get_threads_from_query

            return self & get_threads_from_query(search_query)
        else:
            db_engine_name = askbot.get_database_engine_name()
            filter_parameters = {'deleted': False}
            if 'postgresql_psycopg2' in db_engine_name:
                from askbot.search import postgresql
                return postgresql.run_title_search(
                                        self, search_query
                                    ).filter(
                                        **filter_parameters
                                    ).order_by('-relevance')
            elif 'mysql' in db_engine_name and mysql.supports_full_text_search():
                filter_parameters['title__search'] = search_query
            else:
                filter_parameters['title__icontains'] = search_query

            if askbot.is_multilingual():
                filter_parameters['language_code'] = get_language()

            return self.filter(**filter_parameters)


class ThreadManager(BaseQuerySetManager):

    def get_queryset(self):
        return ThreadQuerySet(self.model)

    def get_tag_summary_from_threads(self, threads):
        """returns a humanized string containing up to
        five most frequently used
        unique tags coming from the ``threads``.
        Variable ``threads`` is an iterable of
        :class:`~askbot.models.Thread` model objects.

        This is not implemented yet as a query set method,
        because it is used on a list.
        """
        # TODO: In Python 2.6 there is collections.Counter() thing which would be very useful here
        # TODO: In Python 2.5 there is `defaultdict` which already would be an improvement
        tag_counts = dict()
        for thread in threads:
            for tag_name in thread.get_tag_names():
                if tag_name in tag_counts:
                    tag_counts[tag_name] += 1
                else:
                    tag_counts[tag_name] = 1
        tag_list = list(tag_counts.keys())
        tag_list.sort(key=lambda t: tag_counts[t], reverse=True)

        # note that double quote placement is important here
        if len(tag_list) == 0:
            return ''
        if len(tag_list) == 1:
            last_topic = '"'
        elif len(tag_list) <= 5:
            last_topic = _('" and "%s"') % tag_list.pop()
        else:
            tag_list = tag_list[:5]
            last_topic = _('" and more')

        return '"' + '", "'.join(tag_list) + str(last_topic)

    def create(self, *args, **kwargs):
        raise NotImplementedError

    def create_new(self, title, author, added_at, wiki, text, tagnames=None,
                   is_anonymous=False, is_private=False, group_id=None,
                   by_email=False, email_address=None, language=None,
                   ip_addr=None):
        """creates new thread"""
        # TODO: Some of this code will go to Post.objects.create_new

        language = language or get_language()
        tagnames = clean_tagnames(tagnames)

        thread = super(ThreadManager, self).create(
            title=title, tagnames=tagnames, last_activity_at=added_at,
            last_activity_by=author, language_code=language)

        # TODO: code below looks like ``Post.objects.create_new()``
        from askbot.models.post import Post
        question = Post(post_type='question', thread=thread, author=author,
                        added_at=added_at, wiki=wiki, is_anonymous=is_anonymous,
                        text=text, language_code=language)
        # html and summary fields are denormalized in .save() call
        if question.wiki:
            #DATED COMMENT
            # TODO: this is confusing - last_edited_at field
            # is used as an indicator whether question has been edited
            # but in principle, post creation should count as edit as well
            question.last_edited_by = question.author
            question.last_edited_at = added_at
            question.wikified_at = added_at

        # save question to have id for revision
        question.save()

        revision = question.add_revision(
            author=author,
            is_anonymous=is_anonymous,
            text=text,
            comment=str(const.POST_STATUS['default_version']),
            revised_at=added_at,
            by_email=by_email,
            email_address=email_address,
            ip_addr=ip_addr
        )

        # this is kind of bad, but we save assign privacy groups to posts and thread
        # this call is rather heavy, we should split into several functions
        parse_results = question.parse_and_save(author=author, is_private=is_private)

        # moderate inline html items (e.g. links, images)
        question.moderate_html()

        author_group = author.get_personal_group()
        thread.add_to_groups([author_group], visibility=ThreadToGroup.SHOW_PUBLISHED_RESPONSES)
        question.add_to_groups([author_group])

        if is_private or group_id:  # add groups to thread and question
            thread.make_private(author, group_id=group_id)
        else:
            thread.make_public()

        # INFO: Question has to be saved before update_tags() is called
        thread.update_tags(tagnames=tagnames,
                           user=author,
                           timestamp=added_at,
                           suppress_signals=[signals.tags_updated])

        # TODO: this is handled in signal because models for posts
        # are too spread out
        if revision.revision > 0:
            signals.post_updated.send(
                post=question, updated_by=author,
                newly_mentioned_users=parse_results['newly_mentioned_users'],
                timestamp=added_at, created=True, diff=parse_results['diff'],
                sender=question.__class__)

        return thread

    def get_for_query(self, search_query, qs=None):
        """returns a query set of questions,
        matching the full text query
        TODO: move to query set
        """
        if getattr(django_settings, 'ENABLE_HAYSTACK_SEARCH', False):
            from askbot.search.haystack.helpers import get_threads_from_query
            return get_threads_from_query(search_query)
        else:
            if not qs:
                qs = self.all()
    #        if getattr(settings, 'USE_SPHINX_SEARCH', False):
    #            matching_questions = Question.sphinx_search.query(search_query)
    #            question_ids = [q.id for q in matching_questions]
    #            return qs.filter(posts__post_type='question', posts__deleted=False, posts__self_question_id__in=question_ids)
            if askbot.get_database_engine_name().endswith('mysql') \
                and mysql.supports_full_text_search():
                return qs.filter(
                    models.Q(title__search=search_query) |
                    models.Q(tagnames__search=search_query) |
                    models.Q(posts__deleted=False, posts__text__search=search_query)
                )
            elif 'postgresql_psycopg2' in askbot.get_database_engine_name():
                from askbot.search import postgresql
                return postgresql.run_thread_search(qs, search_query)
            else:
                return qs.filter(
                    models.Q(title__icontains=search_query) |
                    models.Q(tagnames__icontains=search_query) |
                    models.Q(posts__deleted=False, posts__text__icontains=search_query)
                )

    # TODO: !! review, fix, and write tests for this
    def run_advanced_search(self, request_user, search_state):
        """
        all parameters are guaranteed to be clean
        however may not relate to database - in that case
        a relvant filter will be silently dropped

        """
        from askbot.conf import settings as askbot_settings  # Avoid circular import

        primary_filter = {
            'posts__post_type': 'question',
            'posts__deleted': False
        }

        lang_mode = askbot.get_lang_mode()
        if lang_mode == 'url-lang':
            primary_filter['language_code'] = get_language()
        elif lang_mode == 'user-lang':
            if request_user.is_authenticated:
                language_codes = request_user.get_languages()
            else:
                language_codes = list(dict(django_settings.LANGUAGES).keys())
            primary_filter['language_code__in'] = language_codes

        # TODO: add a possibility to see deleted questions
        qs = self.filter(**primary_filter)

        if askbot_settings.CONTENT_MODERATION_MODE == 'premoderation':
            if request_user.is_authenticated:
                qs = qs.filter(Q(approved=True) | Q(posts__author_id=request_user.pk))
            else:
                qs = qs.filter(approved=True)

        # if groups feature is enabled, filter out threads
        # that are private in groups to which current user does not belong
        if askbot_settings.GROUPS_ENABLED:
            # get group names
            qs = qs.get_visible(user=request_user)
        # run text search while excluding any modifier in the search string
        # like # tag [title: something] @user
        if search_state.stripped_query:
            qs = self.get_for_query(search_query=search_state.stripped_query, qs=qs)

        # we run other things after full text search, because
        # FTS may break the chain of the query set calls,
        # since it might go into an external asset, like Solr

        # search in titles, if necessary
        if search_state.query_title:
            qs = qs.filter(title__icontains=search_state.query_title)

        # search user names if @user is added to search string
        # or if user name exists in the search state
        if search_state.query_users:
            query_users = User.objects.filter(username__in=search_state.query_users)
            if query_users:
                # TODO: unify with search_state.author ?
                qs = qs.filter(posts__post_type='question',
                               posts__author__in=query_users)

        # unified tags - is list of tags taken from the tag selection
        # plus any tags added to the query string with #tag or [tag:something]
        # syntax.
        # run tag search in addition to these unified tags
        meta_data = {}
        tags = search_state.unified_tags()
        if len(tags) > 0:
            if askbot_settings.TAG_SEARCH_INPUT_ENABLED:
                # TODO: this may be gone or disabled per option
                # "tag_search_box_enabled"
                existing_tags = set()
                non_existing_tags = set()
                # we're using a one-by-one tag retreival, b/c
                # we want to take advantage of case-insensitive search indexes
                # in postgresql, plus it is most likely that there will be
                # only one or two search tags anyway
                for tag in tags:
                    try:
                        tag_record = Tag.objects.get(
                            name__iexact=tag, language_code=get_language())
                        existing_tags.add(tag_record.name)
                    except Tag.DoesNotExist:
                        non_existing_tags.add(tag)

                meta_data['non_existing_tags'] = list(non_existing_tags)
                tags = existing_tags
            else:
                meta_data['non_existing_tags'] = list()

            # AND across tags: fetch only threads tagged with every requested tag.
            # The natural ORM expression — one .filter(tags__name=tag) per tag —
            # produces a JOIN through the M2M for every tag, which scales
            # superlinearly and degrades to multi-second response times for
            # 6-8 tag queries even on small datasets. The subquery + HAVING
            # COUNT form is the relational idiom for set intersection.
            # Deduplicate: unified_tags() concatenates query_tags + tags,
            # so the same tag can appear twice; an inflated len() would
            # never match Count(distinct=True). Empty list means all
            # requested tags were unknown, in which case the search is
            # unfiltered by tags (the names are reported via meta_data).
            tags = list(set(tags))
            if tags:
                ThreadTagModel = self.model.tags.through
                matching_thread_ids = (
                    ThreadTagModel.objects
                    .filter(tag__name__in=tags)
                    .values('thread_id')
                    .annotate(matched=Count('tag_id', distinct=True))
                    .filter(matched=len(tags))
                    .values_list('thread_id', flat=True)
                )
                qs = qs.filter(id__in=matching_thread_ids)
        else:
            meta_data['non_existing_tags'] = list()

        if search_state.scope == 'unanswered':
            # Do not show closed questions in unanswered section
            qs = qs.filter(closed=False)
            if askbot_settings.UNANSWERED_QUESTION_MEANING == 'NO_ANSWERS':
                # TODO: this will introduce a problem if there are private answers
                # which are counted here
                # TODO: expand for different meanings of this
                qs = qs.filter(answer_count=0)
            elif askbot_settings.UNANSWERED_QUESTION_MEANING == 'NO_ACCEPTED_ANSWERS':
                qs = qs.filter(accepted_answer__isnull=True)
            elif askbot_settings.UNANSWERED_QUESTION_MEANING == 'NO_UPVOTED_ANSWERS':
                raise NotImplementedError()
            else:
                raise Exception('UNANSWERED_QUESTION_MEANING setting is wrong')

        elif search_state.scope == 'followed':
            followed_filter = models.Q(favorited_by=request_user)
            if 'followit' in django_settings.INSTALLED_APPS:
                followed_users = request_user.get_followed_users()
                followed_filter |= models.Q(posts__post_type__in=('question', 'answer'), posts__author__in=followed_users)

            # a special case: "personalized" main page only ==
            # if followed is the only available scope
            # if total number (regardless of users selections)
            # followed questions is < than a pagefull - we should mix in a list of
            # random questions
            if askbot_settings.ALL_SCOPE_ENABLED == askbot_settings.UNANSWERED_SCOPE_ENABLED == False:
                followed_question_count = qs.filter(followed_filter).distinct().count()
                if followed_question_count < 30:
                    # here we mix in anything
                    followed_filter |= models.Q(deleted=False)

            qs = qs.filter(followed_filter)

        # user contributed questions & answers
        if search_state.author:
            try:
                # TODO: maybe support selection by multiple authors
                u = User.objects.get(id=int(search_state.author))
            except User.DoesNotExist:
                meta_data['author_name'] = None
            else:
                qs = qs.filter(posts__post_type='question', posts__author=u,
                               posts__deleted=False)
                meta_data['author_name'] = u.username

        # get users tag filters
        if request_user and request_user.is_authenticated:
            # mark questions tagged with interesting tags
            # a kind of fancy annotation, would be nice to avoid it
            lang = get_language()
            interesting_tags = Tag.objects.filter(
                user_selections__user=request_user,
                user_selections__reason='good',
                language_code=lang)
            ignored_tags = Tag.objects.filter(
                user_selections__user=request_user,
                user_selections__reason='bad',
                language_code=lang)
            subscribed_tags = Tag.objects.none()
            if askbot_settings.SUBSCRIBED_TAG_SELECTOR_ENABLED:
                subscribed_tags = Tag.objects.filter(
                    user_selections__user=request_user,
                    user_selections__reason='subscribed',
                    language_code=lang)
                meta_data['subscribed_tag_names'] = [tag.name for tag in subscribed_tags]

            meta_data['interesting_tag_names'] = [tag.name for tag in interesting_tags]
            meta_data['ignored_tag_names'] = [tag.name for tag in ignored_tags]

            if request_user.display_tag_filter_strategy == const.INCLUDE_INTERESTING and (interesting_tags or request_user.has_interesting_wildcard_tags()):
                # filter by interesting tags only
                interesting_tag_filter = models.Q(tags__in=interesting_tags)
                if request_user.has_interesting_wildcard_tags():
                    interesting_wildcards = request_user.interesting_tags.split()
                    extra_interesting_tags = Tag.objects.get_by_wildcards(interesting_wildcards)
                    interesting_tag_filter |= models.Q(tags__in=extra_interesting_tags)
                qs = qs.filter(interesting_tag_filter)

            # get the list of interesting and ignored tags (interesting_tag_names, ignored_tag_names) = (None, None)
            if request_user.display_tag_filter_strategy == const.EXCLUDE_IGNORED and (ignored_tags or request_user.has_ignored_wildcard_tags()):
                # exclude ignored tags if the user wants to
                qs = qs.exclude(tags__in=ignored_tags)
                if request_user.has_ignored_wildcard_tags():
                    ignored_wildcards = request_user.ignored_tags.split()
                    extra_ignored_tags = Tag.objects.get_by_wildcards(ignored_wildcards)
                    qs = qs.exclude(tags__in=extra_ignored_tags)

            if request_user.display_tag_filter_strategy == const.INCLUDE_SUBSCRIBED \
                    and subscribed_tags:
                qs = qs.filter(tags__in=subscribed_tags)

            if askbot_settings.USE_WILDCARD_TAGS:
                meta_data['interesting_tag_names'].extend(request_user.interesting_tags.split())
                meta_data['ignored_tag_names'].extend(request_user.ignored_tags.split())

        QUESTION_ORDER_BY_MAP = {
            'age-desc': '-added_at',
            'age-asc': 'added_at',
            'activity-desc': '-last_activity_at',
            'activity-asc': 'last_activity_at',
            'answers-desc': '-answer_count',
            'answers-asc': 'answer_count',
            'votes-desc': '-points',
            'votes-asc': 'points',

            'relevance-desc': '-relevance', # special Postgresql-specific ordering, 'relevance' quaso-column is added by get_for_query()
        }

        orderby = QUESTION_ORDER_BY_MAP[search_state.sort]

        if not (getattr(django_settings, 'ENABLE_HAYSTACK_SEARCH', False) \
                and orderby == '-relevance'):
            # FIXME: this does not produces the very same results as postgres.
            qs = qs.extra(order_by=[orderby])
        # HACK: We add 'ordering_key' column as an alias and order by it, because when distict() is used,
        #       qs.extra(order_by=[orderby,]) is lost if only `orderby` column is from askbot_post!
        #       Removing distinct() from the queryset fixes the problem, but we have to use it here.
        # UPDATE: Apparently we don't need distinct, the query don't duplicate Thread rows!
        # qs = qs.extra(select={'ordering_key': orderby.lstrip('-')}, order_by=['-ordering_key' if orderby.startswith('-') else 'ordering_key'])
        # qs = qs.distinct()
        qs = qs.only(
            'id', 'title', 'view_count', 'answer_count', 'last_activity_at',
            'last_activity_by', 'closed', 'tagnames', 'accepted_answer'
        )
        return qs.distinct(), meta_data

    def precache_view_data_hack(self, threads):
        # TODO: Re-enable this when we have a good test cases to verify that it works properly.
        #
        #       E.g.: - make sure that not precaching give threads never increase # of db queries for the main page
        #             - make sure that it really works, i.e. stuff for non-cached threads is fetched properly
        # Precache data only for non-cached threads - only those will be rendered
        # threads = [thread for thread in threads if not thread.summary_html_cached()]

        thread_ids = [obj.id for obj in threads]
        from askbot.models.post import Post
        page_questions = Post.objects\
            .filter(post_type='question', thread__id__in=thread_ids)\
            .only('id', 'thread', 'points', 'is_anonymous',
                  'summary', 'post_type', 'deleted')
        page_question_map = {}
        for pq in page_questions:
            page_question_map[pq.thread_id] = pq
        for thread in threads:
            thread._question_cache = page_question_map[thread.id]

        last_activity_by_users = User.objects\
            .filter(id__in=[obj.last_activity_by_id for obj in threads])\
            .only('id', 'username', 'askbot_profile__country',
                  'askbot_profile__show_country')
        user_map = {}
        for la_user in last_activity_by_users:
            user_map[la_user.id] = la_user
        for thread in threads:
            thread._last_activity_by_cache = user_map[thread.last_activity_by_id]

    # TODO: this function is similar to get_response_receivers - profile this function against the other one
    def get_thread_contributors(self, thread_list):
        """Returns query set of Thread contributors"""
        # INFO: Evaluate this query to avoid subquery in the subsequent query below (At least MySQL can be awfully slow on subqueries)
        from askbot.models.post import Post
        u_id = list(
            Post.objects
                .filter(post_type__in=('question', 'answer'),
                        thread__in=thread_list)
                .values_list('author', flat=True)
                .distinct()
        )

        # TODO: this does not belong gere - here we select users with real faces
        # first and limit the number of users in the result for display
        # on the main page, we might also want to completely hide fake gravatars
        # and show only real images and the visitors - even if he does not have
        # a real image and try to prompt him/her to upload a picture
        from askbot.conf import settings as askbot_settings
        avatar_limit = askbot_settings.SIDEBAR_MAIN_AVATAR_LIMIT
        contributors = User.objects\
            .filter(id__in=u_id)\
            .order_by('askbot_profile__avatar_type')[:avatar_limit]
        return contributors

    def get_for_user(self, user):
        """returns threads where a given user had participated"""
        from askbot.models.post import PostRevision
        from askbot.models.post import Post
        post_ids = PostRevision.objects \
            .filter(author=user) \
            .values_list('post_id', flat=True) \
            .distinct()
        thread_ids = Post.objects \
            .filter(id__in=post_ids) \
            .values_list('thread_id', flat=True) \
            .distinct()
        return self.filter(id__in=thread_ids)


class ThreadToGroup(models.Model):
    """the "through" many-to-many relation between
    threads and groups - to distinguish full and "what's published"
    visibility of threads to various groups
    """
    SHOW_PUBLISHED_RESPONSES = 0
    SHOW_ALL_RESPONSES = 1
    VISIBILITY_CHOICES = (
        (SHOW_PUBLISHED_RESPONSES, 'show only published responses'),
        (SHOW_ALL_RESPONSES, 'show all responses')
    )
    thread = models.ForeignKey('Thread', on_delete=models.CASCADE)
    group = models.ForeignKey(Group, on_delete=models.CASCADE)
    visibility = models.SmallIntegerField(choices=VISIBILITY_CHOICES,
                                          default=SHOW_ALL_RESPONSES)

    class Meta:
        unique_together = ('thread', 'group')
        db_table = 'askbot_thread_groups'
        app_label = 'askbot'
        verbose_name = _("thread to group")
        verbose_name_plural = _("threads to groups")


class Thread(models.Model):
    title = models.CharField(max_length=300)

    tags = models.ManyToManyField('Tag', related_name='threads')
    groups = models.ManyToManyField(Group, through=ThreadToGroup,
                                    related_name='group_threads')

    # Denormalised data, transplanted from Question
    tagnames = models.CharField(max_length=125)
    view_count = models.PositiveIntegerField(default=0)
    favourite_count = models.PositiveIntegerField(default=0)
    answer_count = models.PositiveIntegerField(default=0)
    last_activity_at = models.DateTimeField(default=timezone.now)
    last_activity_by = models.ForeignKey(User, related_name='unused_last_active_in_threads', on_delete=models.CASCADE)
    language_code = LanguageCodeField()

    # TODO: these two are redundant (we used to have a "star" and "subscribe"
    # now merged into "followed")
    followed_by = models.ManyToManyField(User, related_name='followed_threads')
    favorited_by = models.ManyToManyField(User, through='FavoriteQuestion',
                                          related_name='unused_favorite_threads')

    closed = models.BooleanField(default=False)
    closed_by = models.ForeignKey(User, null=True, blank=True, on_delete=models.CASCADE)  # , related_name='closed_questions')
    closed_at = models.DateTimeField(null=True, blank=True)
    close_reason = models.SmallIntegerField(choices=const.CLOSE_REASONS,
                                            null=True,
                                            blank=True)
    close_reason_text = models.CharField(max_length=256, default='', blank=True)
    deleted = models.BooleanField(default=False, db_index=True)

    # denormalized data: the core approval of the posts is made
    # in the revisions. In the revisions there is more data about
    # approvals - by whom and when
    approved = models.BooleanField(default=True, db_index=True)

    accepted_answer = models.ForeignKey('Post', null=True, blank=True, related_name='+', on_delete=models.CASCADE)
    added_at = models.DateTimeField(auto_now_add=True)

    # db_column will be removed later
    points = models.IntegerField(default=0, db_column='score')

    objects = ThreadManager()

    class Meta:
        app_label = 'askbot'

    # property to support legacy themes in case there are.
    @property
    def score(self):
        return int(self.points)

    @score.setter
    def score(self, number):
        if number:
            self.points = int(number)

    def _question_post(self, refresh=False):
        if refresh and hasattr(self, '_question_cache'):
            delattr(self, '_question_cache')
        post = getattr(self, '_question_cache', None)
        if post:
            return post
        from askbot.models.post import Post
        self._question_cache = Post.objects.get(post_type='question', thread=self)
        return self._question_cache

    def apply_hinted_tags(self, hints=None, user=None, timestamp=None, silent=False):
        """match words in title and body with hints
        and apply some of the hints as tags,
        so that total number of tags in no more
        than the maximum allowed number of tags"""

        # 1) see how many tags we're missing,
        # if we don't need more we return
        existing_tags = self.get_tag_names()
        tags_count = len(existing_tags)
        if tags_count >= askbot_settings.MAX_TAGS_PER_POST:
            return

        # 2) get set of words from title and body
        post_text = self.title + ' ' + self._question_post().text
        post_text = post_text.lower()  # normalize
        post_words = set(post_text.split())

        # 3) get intersection set
        # normalize hints and tags and remember the originals
        orig_hints = dict()
        for hint in hints:
            orig_hints[hint.lower()] = hint

        norm_hints = list(orig_hints.keys())
        norm_tags = [v.lower() for v in existing_tags]

        common_words = (set(norm_hints) & post_words) - set(norm_tags)

        # 4) for each common word count occurances in corpus
        counts = dict()
        for word in common_words:
            counts[word] = sum([w.lower() == word.lower() for w in post_words])

        # 5) sort words by count
        sorted_words = sorted(
                        common_words,
                        key=lambda a: counts[a],
                        reverse=True)

        # 6) extract correct number of most frequently used tags
        need_tags = askbot_settings.MAX_TAGS_PER_POST - len(existing_tags)
        add_tags = sorted_words[0:need_tags]
        add_tags = [orig_hints[h] for h in add_tags]

        tagnames = ' '.join(existing_tags + add_tags)

        if askbot_settings.FORCE_LOWERCASE_TAGS:
            tagnames = tagnames.lower()

        self.retag(
            retagged_by=user,
            retagged_at=timestamp or timezone.now(),
            tagnames=' '.join(existing_tags + add_tags),
            silent=silent
        )

    def get_absolute_url(self):
        return self._question_post().get_absolute_url(thread=self)
        # question_id = self._question_post().id
        # return reverse('question', args = [question_id]) + slugify(self.title)

    def get_answer_count(self, user=None):
        """returns answer count depending on who the user is.
        When user groups are enabled and some answers are hidden,
        the answer count to show must be reflected accordingly"""
        if askbot_settings.GROUPS_ENABLED:
            return self.get_answers(user).count()
        return self.answer_count

    def get_oldest_answer_id(self, user=None):
        """give oldest visible answer id for the user"""
        answers = self.get_answers(user=user).order_by('added_at')
        if len(answers) > 0:
            return answers[0].id
        return None

    def get_answer_ids(self, user=None):
        """give the ids to all the answers for the user"""
        answers = self.get_answers(user=user)
        return [answer.id for answer in answers]

    def get_flag_counts_by_post_id(self, user):
        """Returns a dictionary post_id -> flag_count"""
        if user.is_anonymous:
            return {}

        from askbot.models import Post
        post_ids = Post.objects.filter(thread_id=self.pk).only('pk')
        flags = Activity.objects.filter(object_id__in=post_ids,
                                        content_type=ContentType.objects.get_for_model(Post),
                                        user_id=user.pk,
                                        activity_type=const.TYPE_ACTIVITY_MARK_OFFENSIVE)
        flags = flags.only('pk', 'object_id')
        result = collections.defaultdict(int)
        for flag in flags:
            result[flag.object_id] += 1
        return dict(result)

    def get_latest_revision(self, user=None):
        # TODO: add denormalized field to Thread model
        from askbot.models import Post, PostRevision
        posts_filter = {
            'thread': self,
            'post_type__in': ('question', 'answer'),
            'deleted': False
        }

        if user and user.is_authenticated and askbot_settings.GROUPS_ENABLED:
            # get post with groups shared with having at least
            # one of the user groups
            # of those posts return the latest revision
            posts_filter['groups__in'] = user.get_groups()

        posts = Post.objects.filter(**posts_filter)
        post_ids = list(posts.values_list('id', flat=True))

        revs = PostRevision.objects.filter(post__id__in=post_ids,
                                           revision__gt=0)
        try:
            return revs.order_by('-id')[0]
        except IndexError:
            return None

    def get_sharing_info(self, visitor=None):
        """returns a dictionary with abbreviated thread sharing info:
        * users - up to a certain number of users, excluding the visitor
        * groups - up to a certain number of groups
        * more_users_count - remaining count of shared-with users
        * more_groups_count - remaining count of shared-with groups
        """
        # "visitor" is implicit
        shared_users = self.get_users_shared_with(max_count=2,
                                                  exclude_user=visitor)
        groups = self.groups
        ugroups = groups.get_personal()
        ggroups = groups.exclude_personal()

        sharing_info = {
            'users': shared_users,
            'groups': self.get_groups_shared_with(max_count=3),
            'more_users_count': max(0, ugroups.count() - 3),
            'more_groups_count': max(0, ggroups.count() - 3)
        }
        return sharing_info

    def get_users_shared_with(self, max_count=None, exclude_user=None):
        """returns query set of users with whom
        this thread is shared
        """
        filter = (
            models.Q(thread=self, visibility=ThreadToGroup.SHOW_ALL_RESPONSES) &
            models.Q(group__name__startswith=PERSONAL_GROUP_NAME_PREFIX))

        if exclude_user:
            user_group = exclude_user.get_personal_group()
            filter = filter & ~models.Q(group_id=user_group.id)

        thread_groups = ThreadToGroup.objects.filter(filter)

        if max_count:
            thread_groups = thread_groups[:max_count]

        group_ids = thread_groups.values_list('group_id', flat=True)

        group_ids = list(group_ids)  # force query for MySQL

        from askbot.models import GroupMembership
        user_ids = GroupMembership.objects.filter(group__id__in=group_ids)\
            .values_list('user__id', flat=True)

        return User.objects.filter(id__in=user_ids)

    def get_groups_shared_with(self, max_count=None):
        """returns query set of groups with whom thread is shared"""
        thread_groups = ThreadToGroup.objects.filter(
            models.Q(thread=self, visibility=ThreadToGroup.SHOW_ALL_RESPONSES) &~
            models.Q(group__name__startswith=PERSONAL_GROUP_NAME_PREFIX))
        if max_count:
            thread_groups = thread_groups[:max_count]

        group_ids = thread_groups.values_list('group_id', flat=True)
        return Group.objects.filter(id__in=list(group_ids))  # force list 4 mysql

    def update_favorite_count(self):
        self.favourite_count = FavoriteQuestion.objects.filter(thread=self).count()
        self.save()

    def update_answer_count(self):
        self.answer_count = self.get_answers().count()
        self.save()

    def increase_view_count(self, increment=1):
        qset = Thread.objects.filter(id=self.id)
        qset.update(view_count=models.F('view_count') + increment)
        # get the new view_count back because other pieces of code relies on such behaviour
        self.view_count = qset.values('view_count')[0]['view_count']

        ####################################################################
        self.invalidate_cached_summary_html()
        if not getattr(django_settings, 'CELERY_TASK_ALWAYS_EAGER', False):
            self.update_summary_html()  # proactively regenerate thread summary html
        ####################################################################

    def set_closed_status(self, closed, closed_by, closed_at, close_reason_text):
        self.closed = closed
        self.closed_by = closed_by
        self.closed_at = closed_at
        self.close_reason_text = close_reason_text
        self.save()
        self.reset_cached_data()

    def set_tags_language_code(self, language_code=None):
        """sets language code to tags of this thread.
        If lang code of the tag does not coincide with that
        of thread, we replace the tag with the one of correct
        lang code. If necessary, tags are created and
        the used_counts are updated.
        """
        wrong_lang_tags = list()
        for tag in self.tags.all():
            if tag.language_code != language_code:
                wrong_lang_tags.append(tag)

        # remove wrong tags
        self.tags.remove(*wrong_lang_tags)
        # update used counts of the wrong tags
        wrong_lang_tag_names = list()
        for tag in wrong_lang_tags:
            wrong_lang_tag_names.append(tag.name)
            if tag.used_count > 0:
                tag.decrement_used_count()
                tag.save()

        # load existing tags and figure out which tags don't exist
        reused_tags, new_tagnames = get_tags_by_names(
            wrong_lang_tag_names, language_code=language_code)
        reused_tags.mark_undeleted()
        # tag moderation is in the call below
        created_tags = Tag.objects.create_in_bulk(
            language_code=self.language_code, tag_names=new_tagnames,
            user=self.last_activity_by, auto_approve=True)
        # add the tags
        added_tags = list(reused_tags) + list(created_tags)
        self.tags.add(*added_tags)
        # increment the used counts and save tags
        tag_ids = [tag.id for tag in added_tags]
        Tag.objects.filter(id__in=tag_ids).update(used_count=F('used_count')+1)

    def set_language_code(self, language_code=None):
        assert(language_code)

        # save language code on thread
        self.language_code = language_code
        self.save()

        # save language code on all posts
        # for some reason "update" fails in postgres - possibly b/c of the FTS
        for post in self.posts.all():
            post.language_code = language_code
            post.save()

        # update language of the reputes on the question
        question = self._question_post()
        from askbot.models import Repute
        reputes = Repute.objects.filter(question=question)
        reputes.update(language_code=language_code)

        # make sure that tags have correct language code
        self.set_tags_language_code(language_code)

    def set_accepted_answer(self, answer, actor, timestamp):
        if answer and answer.thread != self:
            raise ValueError("Answer doesn't belong to this thread")
        # TODO: in the future there may be >1 accepted answer
        self.accepted_answer = answer
        self.set_last_activity_info(timestamp, actor)
        self.save()
        answer.endorsed = True
        answer.endorsed_at = timestamp
        answer.endorsed_by = actor
        answer.save()

    def set_last_activity_info(self, last_activity_at, last_activity_by):
        self.last_activity_at = last_activity_at
        self.last_activity_by = last_activity_by

    def get_last_activity_info(self):
        post_ids = self.get_answers().values_list('id', flat=True)
        question = self._question_post()
        post_ids = list(post_ids)
        post_ids.append(question.id)
        from askbot.models import PostRevision
        revs = PostRevision.objects.filter(post__id__in=post_ids,
                                           revision__gt=0).order_by('-id')
        try:
            rev = revs[0]
            return rev.revised_at, rev.author
        except IndexError:
            return None, None

    def update_last_activity_info(self):
        timestamp, user = self.get_last_activity_info()

        if timestamp:
            self.set_last_activity_info(timestamp, user)
            self.save()

    def get_tag_names(self):
        "Creates a list of Tag names from the ``tagnames`` attribute."
        if self.tagnames.strip() == '':
            return list()
        else:
            return self.tagnames.split(' ')

    def get_title(self):
        title_renderer = load_plugin(
            'ASKBOT_QUESTION_TITLE_RENDERER',
            'askbot.models.question.default_title_renderer')
        return title_renderer(self)

    def get_answers_by_user(self, user):
        """regardless - deleted or not"""
        return self.posts.filter(post_type='answer', author=user,
                                 deleted=False)

    def has_answer_by_user(self, user):
        # use len to cache the queryset
        return len(self.get_answers_by_user(user)) > 0

    def has_moderator(self, user):
        """true if ``user`` is also a thread moderator"""
        if user.is_anonymous:
            return False
        return user.is_administrator_or_moderator()

    def tagname_meta_generator(self):
        return ','.join([str(tag) for tag in self.get_tag_names()])

    def all_answers(self):
        return self.posts.get_answers()

    def get_answers(self, user=None):
        """returns query set for answers to this question
        that may be shown to the given user
        """
        if user is None or user.is_anonymous:
            return self.posts.get_answers().filter(deleted=False)
        else:
            return self.posts.get_answers(
                                    user=user
                                ).filter(deleted=False)
            #    return self.posts.get_answers(user=user).filter(
            #                models.Q(deleted=False) \
            #                | models.Q(author=user) \
            #                | models.Q(deleted_by=user)
            #            )
            # we used to show deleted answers to admins,
            # users who deleted those answers and answer owners
            # but later decided to not show deleted answers at all
            # because it makes caching the post lists for thread easier
            # if user.is_administrator() or user.is_moderator():
            #    return self.posts.get_answers(user=user)
            # else:
            #    return self.posts.get_answers(user=user).filter(
            #                models.Q(deleted=False) \
            #                | models.Q(author=user) \
            #                | models.Q(deleted_by=user)
            #            )

    def invalidate_cached_summary_html(self):
        """Invalidates cached summary html in all activated languages"""
        langs = translation_utils.get_language_codes()
        keys = [self.get_summary_cache_key(v) for v in langs]
        cache.cache.delete_many(keys)

    def get_summary_cache_key(self, lang=None):
        lang = lang or get_language()
        return 'thread-question-summary-%d-%s' % (self.id, lang)

    def get_post_data_cache_key(self, sort_method=None, groups=None): #pylint: disable=missing-docstring
        key = f'thread-data-{self.id}-{sort_method}'
        if not groups:
            return key
        return key + '-' + '-'.join(sorted([group.id for group in groups]))

    def invalidate_cached_post_data(self):
        """needs to be called when anything notable
        changes in the post data - on votes, adding,
        deleting, editing content"""
        # we can call delete_many() here if using Django > 1.2
        sort_methods = [v[0] for v in const.ANSWER_SORT_METHODS]
        keys = [self.get_post_data_cache_key(v) for v in sort_methods]
        cache.cache.delete_many(keys)

    def reset_cached_data(self):
        self.clear_cached_data()
        self.update_summary_html()

    def clear_cached_data(self):
        self.invalidate_cached_post_data()
        self.invalidate_cached_summary_html()

    def get_public_posts(self):
        kwargs = {
            'deleted': False,
            'post_type__in': ('question', 'answer', 'comment'),
        }
        if askbot.is_multilingual():
            kwargs['language_code'] = self.language_code or get_language()
        return self.posts.filter(**kwargs)

    def get_personalized_post_data(self, post_data, user):
        """Returns `post_data` data structure,
        personalized for user"""

        def reverse_comments(post_data):
            question = post_data[0]
            answers = post_data[1]
            question.reverse_cached_comments()
            list(map(lambda v: v.reverse_cached_comments(), answers))
            return post_data

        def find_posts(posts, need_ids):
            """posts - is source list
            need_ids - set of post ids
            """
            found = {}
            for post in posts:
                if post.id in need_ids:
                    found[post.id] = post
                    need_ids.remove(post.id)
                    comments = post.get_cached_comments()
                    found.update(find_posts(comments, need_ids))
            return found

        def post_type_ord(p):
            """need to sort by post type"""
            if p.is_question():
                return 0
            elif p.is_answer():
                return 1
            return 2

        post_data = list(post_data)
        from askbot.models import PostRevision
        suggested_revs = PostRevision.objects.filter(author=user, revision=0)

        # users who don't have posts in moderation will incur
        # quick query and exit
        if not suggested_revs.count():
            return post_data

        # for the remaining users we'll try to patch the thread data
        # with the users suggested edits
        question = post_data[0]
        answers = post_data[1]
        post_to_author = post_data[2]

        all_posts = copy(answers)
        if question:
            all_posts.append(question)

        post_ids = [post.id for post in all_posts]

        suggested_revs = suggested_revs.filter(post_id__in=post_ids)
        suggested_post_ids = [rev.post_id for rev in suggested_revs]
        post_id_set = set(suggested_post_ids)

        # posts that we need to patch with user's suggested edits
        posts = find_posts(all_posts, post_id_set)
        rev_map = dict(list(zip(suggested_post_ids, suggested_revs)))

        # patch approved posts with users's revisions pending moderation
        for post_id, post in list(posts.items()):
            rev = rev_map[post_id]
            # patching work
            post.text = rev.text
            parse_data = post.parse_post_text()
            post.html = parse_data['html']
            post.summary = post.get_snippet()
            post_to_author[post_id] = rev.author_id
            post.set_runtime_needs_moderation()

        # patch post data with users' posts pending moderation
        pending_posts = list(self.posts.filter(approved=False, author=user))
        suggested_revs = PostRevision.objects.filter(post_id__in=[post.id for post in pending_posts])
        rev_map = {rev.post_id: rev for rev in suggested_revs}
        for post in sorted(pending_posts, key=post_type_ord):
            rev = rev_map[post.id]
            post.text = rev.text
            post.html = post.parse_post_text()['html']
            post.summary = post.get_snippet()
            post_to_author[post.id] = rev.author_id
            if post.is_comment():
                parents = find_posts(all_posts, set([post.parent_id]))
                parent = list(parents.values())[0]
                parent.add_cached_comment(post)
            if post.is_answer():
                answers.insert(0, post)
                all_posts.append(post)# add b/c there may be self-comments
            if post.is_question():
                post_data[0] = post
                all_posts.append(post)

        return post_data

    def get_post_data_for_question_view(self, user=None, sort_method=None):
        """loads post data for use in the question details view
        """
        post_data = self.get_cached_post_data(user=user, sort_method=sort_method)
        if user.is_anonymous:
            return post_data

        if not (askbot_settings.CONTENT_MODERATION_MODE == 'premoderation' and user.is_watched()):
            return post_data

        return self.get_personalized_post_data(post_data, user)

    def get_groups_for_get_post_data(self, user):
        """Returns groups necessary for `Thread.get_post_data`"""
        if not askbot_settings.GROUPS_ENABLED:
            return []

        if not user or user.is_anonymous:
            return [Group.objects.get_global_group(),]

        all_groups = set(user.get_groups())
        personal = set([user.get_personal_group()])
        return list(all_groups - personal)


    def get_cached_post_data(self, user=None, sort_method=None):
        """returns cached post data, as calculated by
        the method get_post_data()"""
        sort_method = sort_method or askbot_settings.DEFAULT_ANSWER_SORT_METHOD
        groups = self.get_groups_for_get_post_data(user)
        if groups:
            # A temporary hacky plug. 
            # We could cache by groups as well, but in that case invalidation
            # becomes an issue.
            return self.get_post_data(sort_method=sort_method, groups=groups)

        key = self.get_post_data_cache_key(sort_method, groups)
        post_data = cache.cache.get(key)
        if not post_data:
            post_data = self.get_post_data(sort_method=sort_method, groups=groups)
            cache.cache.set(key, post_data, const.LONG_TIME)
        return post_data

    def get_post_data(self, sort_method=None, groups=None):
        """
        returns a tuple of four values:
        * question
        * answers as list
        * list of post ids 
        * list of published post ids

        the returned posts are pre-stuffed with the comments
        the posts and the comments sorted in the correct order
        """
        sort_method = sort_method or askbot_settings.DEFAULT_ANSWER_SORT_METHOD

        if groups:
            posts = self.posts.filter(groups__in=groups, approved=True)
            if len(groups) > 1:
                # important for >1 group
                posts = posts.distinct()
        else:
            posts = self.posts.filter(approved=True)

        order_by_method = {
                        'latest':'-added_at',
                        'oldest':'added_at',
                        'votes':'-points'
                    }

        default_answer_sort_method = askbot_settings.DEFAULT_ANSWER_SORT_METHOD
        default_order_by_method = order_by_method[default_answer_sort_method]
        order_by = order_by_method.get(sort_method, default_order_by_method)
        # we add secondary sort method for the answers to make
        # discussion more coherent
        if order_by != default_order_by_method:
            order_by = (order_by, default_order_by_method)
        else:
            order_by = (order_by,)

        posts = posts.order_by(*order_by)
        # 1) collect question, answer and comment posts and list of post id's
        answers = list()
        post_map = dict()
        comment_map = dict()
        post_to_author = dict()
        question_post = None

        for post in posts:

            if post.post_type not in ('question', 'answer', 'comment'):
                continue

            # precache some revision data
            first_rev = post.get_earliest_revision()
            last_rev = post.get_latest_revision()
            first_rev.post = post
            last_rev.post = post

            # pass through only deleted question posts
            if post.deleted and post.post_type != 'question':
                continue

            if not post.is_approved():
                continue

            post_to_author[post.id] = post.author_id

            if post.post_type == 'answer':
                answers.append(post)
                post_map[post.id] = post

            if post.post_type == 'comment':
                if post.parent_id not in comment_map:
                    comment_map[post.parent_id] = list()
                comment_map[post.parent_id].append(post)

            if post.post_type == 'question':
                assert(question_post is None)
                post_map[post.id] = post
                question_post = post

        # 2) sort comments in the temporal order
        for comment_list in list(comment_map.values()):
            comment_list.sort(key=operator.attrgetter('added_at'))

        # 3) attach comments to question and the answers
        for post_id, comment_list in list(comment_map.items()):
            try:
                post_map[post_id].set_cached_comments(comment_list)
            except KeyError:
                pass  # comment to deleted answer - don't want it

        if askbot_settings.SHOW_ACCEPTED_ANSWER_FIRST:
            if self.has_accepted_answer() and not self.accepted_answer.deleted:
                # Put the accepted answer to front
                # the second check is for the case when accepted answer is deleted
                if self.accepted_answer_id in post_map:
                    accepted_answer = post_map[self.accepted_answer_id]
                    answers.remove(accepted_answer)
                    answers.insert(0, accepted_answer)

        # when there was the "private forum" feature,
        # this block below was under an if branch - if the
        # if question_post and not question_post.is_approved() and user != question_post.author:
        published_answer_ids = self.posts\
            .get_answers()\
            .filter(deleted=False)\
            .order_by(*order_by)\
            .values_list('id', flat=True)

        published_answer_ids = list(reversed(published_answer_ids))
        # now put those answers first
        answer_map = dict([(answer.id, answer) for answer in answers])
        for answer_id in published_answer_ids:
            # note that answer map may not contain answers publised
            # to the question enquirer, because current user may
            # not have access to that answer, so we use the .get() method
            answer = answer_map.get(answer_id, None)
            if answer:
                answers.remove(answer)
                answers.insert(0, answer)

        return (question_post, answers, post_to_author, published_answer_ids)

    def has_accepted_answer(self):
        return self.accepted_answer_id is not None

    def get_similarity(self, other_thread=None):
        """return number of tags in the other question
        that overlap with the current question (self)
        """
        my_tags = set(self.get_tag_names())
        others_tags = set(other_thread.get_tag_names())
        return len(my_tags & others_tags)

    def get_similar_threads(self):
        """
        Get 10 similar threads for given one.
        Threads with the individual tags will be added to list if above questions are not full.

        This function has a limitation that it will
        retrieve only 100 records then select 10 most similar
        from that list as querying entire database may
        be very expensive - this function will benefit from
        some sort of optimization
        """

        def get_data():
            # TODO: code in this function would be simpler if
            # we had question post id denormalized on the thread
            tags_list = self.get_tag_names()
            similar_threads = Thread.objects\
                .filter(tags__name__in=tags_list, language_code=self.language_code)\
                .exclude(id=self.id)\
                .exclude(posts__post_type='question', posts__deleted=True)\
                .distinct()[:100]
            similar_threads = list(similar_threads)

            for thread in similar_threads:
                thread.similarity = self.get_similarity(other_thread=thread)

            similar_threads.sort(key=operator.attrgetter('similarity'), reverse=True)
            similar_threads = similar_threads[:10]

            # Denormalize questions to speed up template rendering
            # TODO: just denormalize question_post_id on the thread!
            thread_map = dict([(thread.id, thread) for thread in similar_threads])
            from askbot.models.post import Post
            questions = Post.objects.get_questions()
            questions = questions.select_related('thread').filter(thread__in=similar_threads)
            for q in questions:
                thread_map[q.thread_id].question_denorm = q

            # Postprocess data for the final output
            result = list()
            for thread in similar_threads:
                question_post = getattr(thread, 'question_denorm', None)
                # unfortunately the if statement below is necessary due to
                # a possible bug
                # all this proves that it's wrong to reference threads by
                # the question post id in the question page urls!!!
                # this is a "legacy" problem inherited from the old models
                if question_post:
                    url = question_post.get_absolute_url()
                    title = thread.get_title()
                    result.append({'url': url, 'title': title})

            return result

        def get_cached_data():
            """similar thread data will expire
            with the default expiration delay
            """
            key = 'similar-threads-%s' % self.id
            data = cache.cache.get(key)
            if data is None:
                data = get_data()
                cache.cache.set(key, data)
            return data

        return LazyList(get_cached_data)

    def remove_author_anonymity(self):
        """removes anonymous flag from the question
        and all its revisions
        the function calls update method to make sure that
        signals are not called
        """
        # note: see note for the is_anonymous field
        # it is important that update method is called - not save,
        # because we do not want the signals to fire here
        thread_question = self._question_post()
        from askbot.models.post import Post
        Post.objects.filter(id=thread_question.id).update(is_anonymous=False)
        thread_question.revisions.all().update(is_anonymous=False)

    def is_followed_by(self, user=None):
        """True if thread is followed by user"""
        if user and user.is_authenticated:
            return self.followed_by.filter(id=user.id).count() > 0
        return False

    def is_moderated(self):
        """True, if tread has SHOW_PUBLISHED_RESPONSES
        group memberships"""
        return askbot_settings.GROUPS_ENABLED

    def add_child_posts_to_groups(self, groups):
        """adds questions and answers of the thread to
        given groups, comments are taken care of implicitly
        by the underlying ``Post`` methods
        """
        post_types = ('question', 'answer')
        posts = self.posts.filter(post_type__in=post_types)
        for post in posts:
            post.add_to_groups(groups)

    def remove_child_posts_from_groups(self, groups):
        """removes child posts from given groups"""
        post_ids = self.posts.all().values_list('id', flat=True)
        group_ids = [group.id for group in groups]
        from askbot.models.post import PostToGroup
        PostToGroup.objects\
            .filter(post__id__in=post_ids, tag__id__in=group_ids)\
            .delete()

    def add_to_groups(self, groups,
                      visibility=ThreadToGroup.SHOW_ALL_RESPONSES,
                      recursive=False):
        """adds thread to a list of groups
        ``groups`` argument may be any iterable of groups
        """
        for group in groups:
            # TODO: change to bulk create when django 1.3 goes out of use
            thread_group, created = ThreadToGroup.objects.get_or_create(
                                                    thread=self,
                                                    group=group
                                                )

            if thread_group.visibility != visibility:
                thread_group.visibility = visibility
                thread_group.save()

        if recursive:
            # comments are taken care of automatically
            self.add_child_posts_to_groups(groups)

    def remove_from_groups(self, groups, recursive=False):
        thread_groups = ThreadToGroup.objects\
            .filter(thread=self, group__in=groups)
        thread_groups.delete()
        if recursive:
            self.remove_child_posts_from_groups(groups)

    def make_public(self, recursive=False):
        """adds the global group to the thread"""
        groups = (Group.objects.get_global_group(),)
        self.add_to_groups(groups, recursive=recursive)
        if not recursive:
            self._question_post().make_public()

    def make_private(self, user, group_id=None):
        """adds thread to all user's groups, excluding
        the global, or to a group given by id.
        The add by ID now only works if user belongs to that group
        """
        if group_id:
            group = Group.objects.get(id=group_id)
            groups = [group]
            self.add_to_groups(groups)

            global_group = Group.objects.get_global_group()
            if group != global_group:
                self.remove_from_groups((global_group,))
        else:
            groups = user.get_groups(private=True)
            self.add_to_groups(groups)
            self.remove_from_groups((Group.objects.get_global_group(),))

        self._question_post().make_private(user, group_id)

        if len(groups) == 0:
            message = _('Sharing did not work, because group is unknown')
            user.message_set.create(message=message)

    def is_private(self):
        """true, if thread belongs to the global group"""
        if askbot_settings.GROUPS_ENABLED:
            group = Group.objects.get_global_group()
            return not self.groups.filter(id=group.id).exists()
        return False

    def remove_tags_by_names(self, tagnames):
        """removes tags from thread by names"""
        removed_tags = list()
        for tag in self.tags.all():
            if tag.name in tagnames:
                tag.decrement_used_count()
                removed_tags.append(tag)
        self.tags.remove(*removed_tags)
        return removed_tags

    def update_tags(self, tagnames=None, user=None,
                    timestamp=None, suppress_signals=None):
        """
        Updates Tag associations for a thread to match the given
        tagname string.
        When an added tag does not exist - it is created
        If tag moderation is on - new tags are placed on the queue

        Tag use counts are recalculated
        A signal tags updated is sent

        TagSynonym is used to replace tag names

        ``suppress_signals`` - list of signal objects to suppress.
        E.g. on thread creation we suppress ``signals.tags_updated``
        to avoid a spurious "QUESTION_RETAGGED" analytics event,
        while still allowing other signals (like badge awarding) to fire.

        *IMPORTANT*: self._question_post() has to
        exist when update_tags() is called!
        """
        if tagnames.strip() == '':
            return

        previous_tags = list(self.tags.filter(status=Tag.STATUS_ACCEPTED))

        ordered_updated_tagnames = [t for t in tagnames.strip().split(' ')]
        updated_tagnames_tmp = set(ordered_updated_tagnames)

        # apply TagSynonym
        updated_tagnames = set()
        for tag_name in updated_tagnames_tmp:
            try:
                tag_synonym = TagSynonym.objects\
                    .get(source_tag_name=tag_name,
                         language_code=self.language_code)
                updated_tagnames.add(tag_synonym.target_tag_name)
                tag_synonym.auto_rename_count += 1
                tag_synonym.save()
            except TagSynonym.DoesNotExist:
                updated_tagnames.add(tag_name)

        previous_tagnames = set([tag.name for tag in previous_tags])
        removed_tagnames = previous_tagnames - updated_tagnames

        # remove tags from the question's tags many2many relation
        # used_count values are decremented on all tags
        removed_tags = self.remove_tags_by_names(removed_tagnames)

        # modified tags go on to recounting their use
        # TODO - this can actually be done asynchronously - not so important
        modified_tags, _unused_tags = separate_unused_tags(removed_tags)

        modified_tags = removed_tags
        # add new tags to the relation
        added_tagnames = updated_tagnames - previous_tagnames

        if added_tagnames:
            # find reused tags
            reused_tags, new_tagnames = get_tags_by_names(
                added_tagnames, language_code=self.language_code)
            reused_tags.mark_undeleted()

            added_tags = list(reused_tags)
            # tag moderation is in the call below
            created_tags = Tag.objects\
                .create_in_bulk(language_code=self.language_code,
                                tag_names=new_tagnames, user=user)

            added_tags.extend(created_tags)
            # TODO: not nice that assignment of added_tags is way above
            self.tags.add(*added_tags)
            modified_tags.extend(added_tags)
        else:
            added_tags = Tag.objects.none()

        # Save denormalized tag names on thread. Preserve order from user input.
        accepted_added_tags = filter_accepted_tags(added_tags)
        added_tagnames = set([tag.name for tag in accepted_added_tags])
        final_tagnames = (previous_tagnames - removed_tagnames) | added_tagnames
        ordered_final_tagnames = list()
        for tagname in ordered_updated_tagnames:
            if tagname in final_tagnames:
                ordered_final_tagnames.append(tagname)

        self.tagnames = ' '.join(ordered_final_tagnames)

        # TODO: factor out - tell author about suggested tags
        suggested_tags = filter_suggested_tags(added_tags)
        if len(suggested_tags) > 0:
            # 1) notify author that the tag is going to be moderated
            # TODO: factor this out
            if len(suggested_tags) == 1:
                msg = _(
                    'Tag %s is new and will be submitted for the '
                    'moderators approval'
                ) % suggested_tags[0].name
            else:
                msg = _(
                    'Tags %s are new and will be submitted for the '
                    'moderators approval'
                ) % ', '.join([tag.name for tag in suggested_tags])
            user.message_set.create(message=msg)
            # 2) TODO: notify moderators about newly suggested tags
        # if there are any modified tags, update their use counts
        modified_tags = set(modified_tags)
        if modified_tags:
            Tag.objects.update_use_counts(modified_tags)
            if not suppress_signals:
                suppress_signals = []
            from askbot.models.badges import award_badges_signal
            if award_badges_signal not in suppress_signals:
                for tag in modified_tags:
                    award_badges_signal.send(None,
                                             event='update_tag',
                                             actor=user,
                                             context_object=tag,
                                             timestamp=timestamp)
            if signals.tags_updated not in suppress_signals:
                signals.tags_updated.send(None,
                                          thread=self,
                                          tags=modified_tags,
                                          user=user,
                                          timestamp=timestamp)
            return True

        return False

    def add_tag(self, user=None, timestamp=None, tag_name=None, silent=False):
        """adds one tag to thread"""
        tag_names = self.get_tag_names()
        if tag_name in tag_names:
            return
        tag_names.append(tag_name)

        self.retag(retagged_by=user, retagged_at=timestamp,
                   tagnames=' '.join(tag_names), silent=silent)

    def retag(self, retagged_by=None, retagged_at=None, tagnames=None, silent=False):
        """changes thread tags"""
        if None in (retagged_by, retagged_at, tagnames):
            raise Exception('arguments retagged_at, retagged_by and tagnames are required')

        tagnames = clean_tagnames(tagnames)
        self.tagnames = tagnames
        self.save()

        thread_question = self._question_post()

        # Update the Question itself
        if not silent:
            thread_question.last_edited_at = retagged_at
            # thread_question.thread.last_activity_at = retagged_at
            thread_question.last_edited_by = retagged_by
            # thread_question.thread.last_activity_by = retagged_by
            thread_question.save()

        # Update the Thread's tag associations
        self.update_tags(tagnames=tagnames, user=retagged_by, timestamp=retagged_at)

        # Create a new revision
        latest_revision = thread_question.get_latest_revision()

        from askbot.models.post import PostRevision
        PostRevision.objects.create(
            post=thread_question,
            title=latest_revision.title,
            author=retagged_by,
            revised_at=retagged_at,
            tagnames=tagnames,
            summary=str(const.POST_STATUS['retagged']),
            text=latest_revision.text
        )

    def has_favorite_by_user(self, user):
        if not user.is_authenticated:
            return False

        return FavoriteQuestion.objects.filter(thread=self, user=user).exists()

    def get_last_update_info(self):
        posts = list(self.posts.select_related('author', 'last_edited_by'))

        last_updated_at = posts[0].added_at
        last_updated_by = posts[0].author

        for post in posts:
            last_updated_at, last_updated_by = max((last_updated_at, last_updated_by), (post.added_at, post.author))
            if post.last_edited_at:
                last_updated_at, last_updated_by = max((last_updated_at, last_updated_by), (post.last_edited_at, post.last_edited_by))

        return last_updated_at, last_updated_by

    def get_summary_html(self, search_state=None, visitor=None):
        html = self.get_cached_summary_html(visitor) or self.update_summary_html(visitor)
        # TODO: this work may be pushed onto javascript we post-process tag names
        # in the snippet so that tag urls match the search state
        # use `<<<` and `>>>` because they cannot be confused with user input
        # - if user accidentialy types <<<tag-name>>> into question title or body,
        # then in html it'll become escaped like this: &lt;&lt;&lt;tag-name&gt;&gt;&gt;
        regex = re.compile(
            r'<<<(%s)>>>' % const.TAG_REGEX_BARE,
            re.UNICODE
        )

        if search_state is None:
            search_state = DummySearchState()

        while True:
            match = regex.search(html)
            if not match:
                break
            seq = match.group(0)  # e.g "<<<my-tag>>>"
            tag = match.group(1)  # e.g "my-tag"
            full_url = search_state.add_tag(tag).full_url()
            html = html.replace(seq, full_url)

        return html

    def get_cached_summary_html(self, visitor=None):
        # TODO: remove this plug by adding cached foreign user group
        # parameter to the key. Now with groups on caching is turned off
        # parameter visitor is there to get summary out by the user groups
        if askbot_settings.GROUPS_ENABLED:
            return None
        return cache.cache.get(self.get_summary_cache_key())

    def update_summary_html(self, visitor=None):
        # TODO: it is quite wrong that visitor is an argument here
        # because we do not include any visitor-related info in the cache key
        # ideally cache should be shareable between users, so straight up
        # using the user id for cache is wrong, we could use group
        # memberships, but in that case we'd need to be more careful with
        # cache invalidation
        context = {
            'thread': self,
            # fetch new question post to make sure we're up-to-date
            'question': self._question_post(refresh=True),
            'search_state': DummySearchState(),
            'visitor': visitor
        }
        from askbot.views.context import get_extra as get_extra_context
        context.update(get_extra_context('ASKBOT_QUESTION_SUMMARY_EXTRA_CONTEXT', None, context))
        template = get_template('questions/question_summary.html')
        html = template.render(Context(context))
        # INFO: Timeout is set to 30 days:
        # * timeout=0/None is not a reliable cross-backend way to set infinite timeout
        # * We probably don't need to pollute the cache with threads older than 30 days
        # * Additionally, Memcached treats timeouts > 30day as dates (https://code.djangoproject.com/browser/django/tags/releases/1.3/django/core/cache/backends/memcached.py#L36),
        #   which probably doesn't break anything but if we can stick to 30 days then let's stick to it
        cache.cache.set(self.get_summary_cache_key(), html,
                        timeout=const.LONG_TIME)
        return html

    def summary_html_cached(self):
        return self.get_summary_cache_key() in cache.cache


class QuestionView(models.Model):
    question = models.ForeignKey('Post', related_name='viewed', on_delete=models.CASCADE)
    who = models.ForeignKey(User, related_name='question_views', on_delete=models.CASCADE)
    when = models.DateTimeField()

    class Meta:
        app_label = 'askbot'


class FavoriteQuestion(models.Model):
    """A favorite Question of a User."""
    thread = models.ForeignKey(Thread, on_delete=models.CASCADE)
    user = models.ForeignKey(User, related_name='user_favorite_questions', on_delete=models.CASCADE)
    added_at = models.DateTimeField(default=timezone.now)

    class Meta:
        app_label = 'askbot'
        db_table = 'favorite_question'
        verbose_name = _("favorite question")
        verbose_name_plural = _("favorite questions")

    def __str__(self):
        return '[%s] favorited at %s' % (self.user, self.added_at)


class DraftQuestion(DraftContent):
    """Provides space to solve unpublished draft
    questions. Contents is used to populate the Ask form.
    """
    author = models.ForeignKey(User, on_delete=models.CASCADE)
    title = models.CharField(max_length=300, null=True)
    tagnames = models.CharField(max_length=125, null=True)

    class Meta:
        app_label = 'askbot'


class AnonymousQuestion(AnonymousContent):
    """question that was asked before logging in
    maybe the name is a little misleading, the user still
    may or may not want to stay anonymous after the question
    is published
    """
    title = models.CharField(max_length=300)
    tagnames = models.CharField(max_length=125)
    is_anonymous = models.BooleanField(default=False)

    def publish(self, user):
        added_at = timezone.now()
        # TODO: wrong - use User.post_question() instead
        try:
            user.assert_can_post_text(self.text)

        except django_exceptions.PermissionDenied:
            # delete previous draft questions (only one is allowed anyway)
            prev_drafts = DraftQuestion.objects.filter(author=user)
            prev_drafts.delete()
            # convert this question to draft
            DraftQuestion.objects.create(
                author=user, title=self.title, text=self.text,
                tagnames=self.tagnames)
        else:
            Thread.objects.create_new(
                title=self.title, added_at=added_at, author=user,
                wiki=self.wiki, is_anonymous=self.is_anonymous,
                tagnames=self.tagnames, text=self.text)
            DraftQuestion.objects.filter(author=user).delete()

        finally:
            self.delete()