NewsBlur/apps/social/models.py

import os
import urllib.parse
import datetime
import time
import zlib
import hashlib
import redis
import re
import mongoengine as mongo
import random
import requests
import html.parser
import tweepy
from collections import defaultdict
from bs4 import BeautifulSoup
from mongoengine.queryset import Q
from django.conf import settings
from django.contrib.auth.models import User
from django.contrib.sites.models import Site
from django.urls import reverse
from django.template.loader import render_to_string
from django.template.defaultfilters import slugify
from django.core.mail import EmailMultiAlternatives
from apps.reader.models import UserSubscription, RUserStory
from apps.analyzer.models import MClassifierFeed, MClassifierAuthor, MClassifierTag, MClassifierTitle
from apps.analyzer.models import apply_classifier_titles, apply_classifier_feeds, apply_classifier_authors, apply_classifier_tags
from apps.rss_feeds.models import Feed, MStory
from apps.rss_feeds.text_importer import TextImporter
from apps.rss_feeds.page_importer import PageImporter
from apps.profile.models import Profile, MSentEmail
from vendor import facebook
from vendor import pynliner
from utils import log as logging
from utils import json_functions as json
from utils.feed_functions import relative_timesince, chunks
from utils.story_functions import truncate_chars, strip_tags, linkify, image_size
from utils.image_functions import ImageOps
from utils.scrubber import SelectiveScriptScrubber
from utils import s3_utils
from io import BytesIO

try:
    from apps.social.spam import detect_spammers
except ImportError:
    # logging.debug(" ---> ~SN~FRCouldn't find ~SBspam.py~SN.")
    pass

RECOMMENDATIONS_LIMIT = 5
IGNORE_IMAGE_SOURCES = [
    "http://feeds.feedburner.com"
]

class MRequestInvite(mongo.Document):
    email            = mongo.EmailField()
    request_date     = mongo.DateTimeField(default=datetime.datetime.now)
    invite_sent      = mongo.BooleanField(default=False)
    invite_sent_date = mongo.DateTimeField()

    meta = {
        'collection': 'social_invites',
        'allow_inheritance': False,
    }
    
    def __str__(self):
        return "%s%s" % (self.email, '*' if self.invite_sent else '')
    
    @classmethod
    def blast(cls):
        invites = cls.objects.filter(email_sent=None)
        print(' ---> Found %s invites...' % invites.count())
        
        for invite in invites:
            try:
                invite.send_email()
            except:
                print(' ***> Could not send invite to: %s. Deleting.' % invite.username)
                invite.delete()
        
    def send_email(self):
        user = User.objects.filter(username__iexact=self.username)
        if not user:
            user = User.objects.filter(email__iexact=self.username)
        if user:
            user = user[0]
            email = user.email or self.username
        else:
            user = {
                'username': self.username,
                'profile': {
                    'autologin_url': '/',
                }
            }
            email = self.username
        params = {
            'user': user,
        }
        text    = render_to_string('mail/email_social_beta.txt', params)
        html    = render_to_string('mail/email_social_beta.xhtml', params)
        subject = "Psst, you're in..."
        msg     = EmailMultiAlternatives(subject, text, 
                                         from_email='NewsBlur <%s>' % settings.HELLO_EMAIL,
                                         to=['<%s>' % (email)])
        msg.attach_alternative(html, "text/html")
        msg.send()
        
        self.email_sent = True
        self.save()
                
        logging.debug(" ---> ~BB~FM~SBSending email for social beta: %s" % self.username)


class MSocialProfile(mongo.Document):
    user_id              = mongo.IntField(unique=True)
    username             = mongo.StringField(max_length=30, unique=True)
    email                = mongo.StringField()
    bio                  = mongo.StringField(max_length=160)
    blurblog_title       = mongo.StringField(max_length=256)
    custom_bgcolor       = mongo.StringField(max_length=50)
    custom_css           = mongo.StringField()
    photo_url            = mongo.StringField()
    photo_service        = mongo.StringField()
    location             = mongo.StringField(max_length=40)
    website              = mongo.StringField(max_length=200)
    bb_permalink_direct  = mongo.BooleanField()
    subscription_count   = mongo.IntField(default=0)
    shared_stories_count = mongo.IntField(default=0)
    following_count      = mongo.IntField(default=0)
    follower_count       = mongo.IntField(default=0)
    following_user_ids   = mongo.ListField(mongo.IntField())
    follower_user_ids    = mongo.ListField(mongo.IntField())
    unfollowed_user_ids  = mongo.ListField(mongo.IntField())
    requested_follow_user_ids = mongo.ListField(mongo.IntField())
    popular_publishers   = mongo.StringField()
    stories_last_month   = mongo.IntField(default=0)
    average_stories_per_month = mongo.IntField(default=0)
    story_count_history  = mongo.ListField()
    story_days_history   = mongo.DictField()
    story_hours_history  = mongo.DictField()
    story_email_history  = mongo.ListField()
    feed_classifier_counts = mongo.DictField()
    favicon_color        = mongo.StringField(max_length=6)
    protected            = mongo.BooleanField()
    private              = mongo.BooleanField()
    
    meta = {
        'collection': 'social_profile',
        'indexes': ['user_id', 'following_user_ids', 'follower_user_ids', 'unfollowed_user_ids', 'requested_follow_user_ids'],
        'allow_inheritance': False,
    }
    
    def __str__(self):
        return "%s [%s] following %s/%s, shared %s" % (self.username, self.user_id, 
                                  self.following_count, self.follower_count, self.shared_stories_count)
    
    @classmethod
    def get_user(cls, user_id):
        try:
            profile = cls.objects.get(user_id=user_id)
        except cls.DoesNotExist:
            profile = cls.objects.create(user_id=user_id)
            profile.save()

        return profile
        
    def save(self, *args, **kwargs):
        if not self.username:
            self.import_user_fields()
        if not self.subscription_count:
            self.count_follows(skip_save=True)
        if self.bio and len(self.bio) > MSocialProfile.bio.max_length:
            self.bio = self.bio[:80]
        if self.bio:
            self.bio = strip_tags(self.bio)
        if self.website:
            self.website = strip_tags(self.website)
        if self.location:
            self.location = strip_tags(self.location)
        if self.custom_css:
            self.custom_css = strip_tags(self.custom_css)
            
        super(MSocialProfile, self).save(*args, **kwargs)
        if self.user_id not in self.following_user_ids:
            self.follow_user(self.user_id, force=True)
            self.count_follows()
        
        return self
            
    @property
    def blurblog_url(self):
        return "http://%s.%s/" % (
            self.username_slug,
            Site.objects.get_current().domain.replace('www.', ''))
    
    @property
    def blurblog_rss(self):
        return "%s%s" % (self.blurblog_url, reverse('shared-stories-rss-feed', 
                                                    kwargs={'user_id': self.user_id, 
                                                            'username': self.username_slug}))

    def find_stories(self, query, offset=0, limit=25):
        stories_db = MSharedStory.objects(
            Q(user_id=self.user_id) &
            (Q(story_title__icontains=query) |
             Q(story_author_name__icontains=query) |
             Q(story_tags__icontains=query))
        ).order_by('-shared_date')[offset:offset+limit]
        stories = Feed.format_stories(stories_db)
        
        return stories

    def recommended_users(self):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        following_key = "F:%s:F" % (self.user_id)
        social_follow_key = "FF:%s:F" % (self.user_id)
        profile_user_ids = []
        
        # Find potential twitter/fb friends
        services = MSocialServices.get_user(self.user_id)
        facebook_user_ids = [u.user_id for u in 
                            MSocialServices.objects.filter(facebook_uid__in=services.facebook_friend_ids).only('user_id')]
        twitter_user_ids = [u.user_id for u in 
                            MSocialServices.objects.filter(twitter_uid__in=services.twitter_friend_ids).only('user_id')]
        social_user_ids = facebook_user_ids + twitter_user_ids
        # Find users not currently followed by this user
        r.delete(social_follow_key)
        nonfriend_user_ids = []
        if social_user_ids:
            r.sadd(social_follow_key, *social_user_ids)
            nonfriend_user_ids = r.sdiff(social_follow_key, following_key)
            profile_user_ids = [int(f) for f in nonfriend_user_ids]
            r.delete(social_follow_key)
        
        # Not enough? Grab popular users.
        if len(nonfriend_user_ids) < RECOMMENDATIONS_LIMIT:
            homepage_user = User.objects.get(username='popular')
            suggested_users_list = r.sdiff("F:%s:F" % homepage_user.pk, following_key)
            suggested_users_list = [int(f) for f in suggested_users_list]
            suggested_user_ids = []
            slots_left = min(len(suggested_users_list), RECOMMENDATIONS_LIMIT - len(nonfriend_user_ids))
            for slot in range(slots_left):
                suggested_user_ids.append(random.choice(suggested_users_list))
            profile_user_ids.extend(suggested_user_ids)
        
        # Sort by shared story count
        profiles = MSocialProfile.profiles(profile_user_ids).order_by('-shared_stories_count')[:RECOMMENDATIONS_LIMIT]

        return profiles
    
    @property
    def username_slug(self):
        return slugify(self.username)
        
    def count_stories(self):
        # Popular Publishers
        self.save_popular_publishers()
        
    def save_popular_publishers(self, feed_publishers=None):
        if not feed_publishers:
            publishers = defaultdict(int)
            for story in MSharedStory.objects(user_id=self.user_id).only('story_feed_id')[:500]:
                publishers[story.story_feed_id] += 1
            feed_titles = dict((f.id, f.feed_title) 
                               for f in Feed.objects.filter(pk__in=list(publishers.keys())).only('id', 'feed_title'))
            feed_publishers = sorted([{'id': k, 'feed_title': feed_titles[k], 'story_count': v} 
                                      for k, v in list(publishers.items())
                                      if k in feed_titles],
                                     key=lambda f: f['story_count'],
                                     reverse=True)[:20]

        popular_publishers = json.encode(feed_publishers)
        if len(popular_publishers) < 1023:
            self.popular_publishers = popular_publishers
            self.save()
            return

        if len(popular_publishers) > 1:
            self.save_popular_publishers(feed_publishers=feed_publishers[:-1])
        
    @classmethod
    def profile(cls, user_id, include_follows=True):
        profile = cls.get_user(user_id)
        return profile.canonical(include_follows=True)
        
    @classmethod
    def profiles(cls, user_ids):
        profiles = cls.objects.filter(user_id__in=user_ids)
        return profiles

    @classmethod
    def profile_feeds(cls, user_ids):
        profiles = cls.objects.filter(user_id__in=user_ids)
        profiles = dict((p.user_id, p.feed()) for p in profiles)
        return profiles
        
    @classmethod
    def sync_all_redis(cls):
        for profile in cls.objects.all():
            profile.sync_redis(force=True)
    
    def sync_redis(self, force=False):
        self.following_user_ids = list(set(self.following_user_ids))
        self.save()
        
        for user_id in self.following_user_ids:
            self.follow_user(user_id, force=force)
        
        self.follow_user(self.user_id, force=force)
    
    @property
    def title(self):
        return self.blurblog_title if self.blurblog_title else self.username + "'s blurblog"
        
    def feed(self):
        params = self.canonical(compact=True)
        params.update({
            'feed_title': self.title,
            'page_url': reverse('load-social-page', kwargs={'user_id': self.user_id, 'username': self.username_slug}),
            'shared_stories_count': self.shared_stories_count,
        })
        return params
        
    def page(self):
        params = self.canonical(include_follows=True)
        params.update({
            'feed_title': self.title,
            'custom_css': self.custom_css,
        })
        return params
    
    @property
    def profile_photo_url(self):
        if self.photo_url:
            return self.photo_url
        return settings.MEDIA_URL + 'img/reader/default_profile_photo.png'
    
    @property
    def large_photo_url(self):
        photo_url = self.email_photo_url
        if 'graph.facebook.com' in photo_url:
            return photo_url + '?type=large'
        elif 'twimg' in photo_url:
            return photo_url.replace('_normal', '')
        elif '/avatars/' in photo_url:
            return photo_url.replace('thumbnail_', 'large_')
        return photo_url
            
    @property
    def email_photo_url(self):
        if self.photo_url:
            if self.photo_url.startswith('//'):
                self.photo_url = 'https:' + self.photo_url
            return self.photo_url
        domain = Site.objects.get_current().domain
        return 'https://' + domain + settings.MEDIA_URL + 'img/reader/default_profile_photo.png'
        
    def canonical(self, compact=False, include_follows=False, common_follows_with_user=None,
                  include_settings=False, include_following_user=None):
        domain = Site.objects.get_current().domain
        params = {
            'id': 'social:%s' % self.user_id,
            'user_id': self.user_id,
            'username': self.username,
            'photo_url': self.email_photo_url,
            'large_photo_url': self.large_photo_url,
            'location': self.location,
            'num_subscribers': self.follower_count,
            'feed_title': self.title,
            'feed_address': "http://%s%s" % (domain, reverse('shared-stories-rss-feed', 
                                    kwargs={'user_id': self.user_id, 'username': self.username_slug})),
            'feed_link': self.blurblog_url,
            'protected': self.protected,
            'private': self.private,
        }
        if not compact:
            params.update({
                'large_photo_url': self.large_photo_url,
                'bio': self.bio,
                'website': self.website,
                'shared_stories_count': self.shared_stories_count,
                'following_count': self.following_count,
                'follower_count': self.follower_count,
                'popular_publishers': json.decode(self.popular_publishers),
                'stories_last_month': self.stories_last_month,
                'average_stories_per_month': self.average_stories_per_month,
            })
        if include_settings:
            params.update({
                'custom_css': self.custom_css,
                'custom_bgcolor': self.custom_bgcolor,
                'bb_permalink_direct': self.bb_permalink_direct,
            })
        if include_follows:
            params.update({
                'photo_service': self.photo_service,
                'following_user_ids': self.following_user_ids_without_self[:48],
                'follower_user_ids': self.follower_user_ids_without_self[:48],
            })
        if common_follows_with_user:
            FOLLOWERS_LIMIT = 128
            with_user = MSocialProfile.get_user(common_follows_with_user)
            followers_youknow, followers_everybody = with_user.common_follows(self.user_id, direction='followers')
            following_youknow, following_everybody = with_user.common_follows(self.user_id, direction='following')
            params['followers_youknow'] = followers_youknow[:FOLLOWERS_LIMIT]
            params['followers_everybody'] = followers_everybody[:FOLLOWERS_LIMIT]
            params['following_youknow'] = following_youknow[:FOLLOWERS_LIMIT]
            params['following_everybody'] = following_everybody[:FOLLOWERS_LIMIT]
            params['requested_follow'] = common_follows_with_user in self.requested_follow_user_ids
        if include_following_user or common_follows_with_user:
            if not include_following_user:
                include_following_user = common_follows_with_user
            if include_following_user != self.user_id:
                params['followed_by_you'] = bool(self.is_followed_by_user(include_following_user))
                params['following_you'] = self.is_following_user(include_following_user)

        return params
    
    @property
    def following_user_ids_without_self(self):
        if self.user_id in self.following_user_ids:
            return [u for u in self.following_user_ids if u != self.user_id]
        return self.following_user_ids
        
    @property
    def follower_user_ids_without_self(self):
        if self.user_id in self.follower_user_ids:
            return [u for u in self.follower_user_ids if u != self.user_id]
        return self.follower_user_ids
        
    def import_user_fields(self, skip_save=False):
        user = User.objects.get(pk=self.user_id)
        self.username = user.username
        self.email = user.email

    def count_follows(self, skip_save=False):
        self.subscription_count = UserSubscription.objects.filter(user__pk=self.user_id).count()
        self.shared_stories_count = MSharedStory.objects.filter(user_id=self.user_id).count()
        self.following_count = len(self.following_user_ids_without_self)
        self.follower_count = len(self.follower_user_ids_without_self)
        if not skip_save:
            self.save()
        
    def follow_user(self, user_id, check_unfollowed=False, force=False):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        if check_unfollowed and user_id in self.unfollowed_user_ids:
            return
            
        if self.user_id == user_id:
            followee = self
        else:
            followee = MSocialProfile.get_user(user_id)
        
        logging.debug(" ---> ~FB~SB%s~SN (%s) following %s" % (self.username, self.user_id, user_id))
        
        if not followee.protected or force:
            if user_id not in self.following_user_ids:
                self.following_user_ids.append(user_id)
            elif not force:
                return
            
        if user_id in self.unfollowed_user_ids:
            self.unfollowed_user_ids.remove(user_id)
        self.count_follows()
        self.save()
        
        if followee.protected and user_id != self.user_id and not force:
            if self.user_id not in followee.requested_follow_user_ids:
                followee.requested_follow_user_ids.append(self.user_id)
                MFollowRequest.add(self.user_id, user_id)
        elif self.user_id not in followee.follower_user_ids:
            followee.follower_user_ids.append(self.user_id)
        followee.count_follows()
        followee.save()

        if followee.protected and user_id != self.user_id and not force:
            from apps.social.tasks import EmailFollowRequest
            EmailFollowRequest.apply_async(kwargs=dict(follower_user_id=self.user_id,
                                                       followee_user_id=user_id),
                                           countdown=settings.SECONDS_TO_DELAY_CELERY_EMAILS)
            return
            
        following_key = "F:%s:F" % (self.user_id)
        r.sadd(following_key, user_id)
        follower_key = "F:%s:f" % (user_id)
        r.sadd(follower_key, self.user_id)

        if user_id != self.user_id:
            MInteraction.new_follow(follower_user_id=self.user_id, followee_user_id=user_id)
            MActivity.new_follow(follower_user_id=self.user_id, followee_user_id=user_id)
        
        params = dict(user_id=self.user_id, subscription_user_id=user_id)
        try:
            socialsub = MSocialSubscription.objects.get(**params)
        except MSocialSubscription.DoesNotExist:
            socialsub = MSocialSubscription.objects.create(**params)
        socialsub.needs_unread_recalc = True
        socialsub.save()
        
        MFollowRequest.remove(self.user_id, user_id)
        
        if not force:
            from apps.social.tasks import EmailNewFollower
            EmailNewFollower.apply_async(kwargs=dict(follower_user_id=self.user_id,
                                                     followee_user_id=user_id),
                                         countdown=settings.SECONDS_TO_DELAY_CELERY_EMAILS)
        
        return socialsub
    
    def is_following_user(self, user_id):
        # XXX TODO: Outsource to redis
        return user_id in self.following_user_ids
    
    def is_followed_by_user(self, user_id):
        # XXX TODO: Outsource to redis
        return user_id in self.follower_user_ids
        
    def unfollow_user(self, user_id):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        if not isinstance(user_id, int):
            user_id = int(user_id)
        
        if user_id == self.user_id:
            # Only unfollow other people, not yourself.
            return

        if user_id in self.following_user_ids:
            self.following_user_ids.remove(user_id)
        if user_id not in self.unfollowed_user_ids:
            self.unfollowed_user_ids.append(user_id)
        self.count_follows()
        self.save()
        
        followee = MSocialProfile.get_user(user_id)
        if self.user_id in followee.follower_user_ids:
            followee.follower_user_ids.remove(self.user_id)
            followee.count_follows()
            followee.save()
        if self.user_id in followee.requested_follow_user_ids:
            followee.requested_follow_user_ids.remove(self.user_id)
            followee.count_follows()
            followee.save()
            MFollowRequest.remove(self.user_id, user_id)
        
        following_key = "F:%s:F" % (self.user_id)
        r.srem(following_key, user_id)
        follower_key = "F:%s:f" % (user_id)
        r.srem(follower_key, self.user_id)
        
        try:
            MSocialSubscription.objects.get(user_id=self.user_id, subscription_user_id=user_id).delete()
        except MSocialSubscription.DoesNotExist:
            return False
    
    def common_follows(self, user_id, direction='followers'):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        my_followers    = "F:%s:%s" % (self.user_id, 'F' if direction == 'followers' else 'F')
        their_followers = "F:%s:%s" % (user_id, 'f' if direction == 'followers' else 'F')
        follows_inter   = r.sinter(their_followers, my_followers)
        follows_diff    = r.sdiff(their_followers, my_followers)
        follows_inter   = [int(f) for f in follows_inter]
        follows_diff    = [int(f) for f in follows_diff]
        
        if user_id in follows_inter:
            follows_inter.remove(user_id)
        if user_id in follows_diff:
            follows_diff.remove(user_id)
        
        return follows_inter, follows_diff
    
    def send_email_for_new_follower(self, follower_user_id):
        user = User.objects.get(pk=self.user_id)
        if follower_user_id not in self.follower_user_ids:
            logging.user(user, "~FMNo longer being followed by %s" % follower_user_id)
            return
        if not user.email:
            logging.user(user, "~FMNo email to send to, skipping.")
            return
        elif not user.profile.send_emails:
            logging.user(user, "~FMDisabled emails, skipping.")
            return
        if self.user_id == follower_user_id:
            return
        
        emails_sent = MSentEmail.objects.filter(receiver_user_id=user.pk,
                                                sending_user_id=follower_user_id,
                                                email_type='new_follower')
        day_ago = datetime.datetime.now() - datetime.timedelta(days=1)
        for email in emails_sent:
            if email.date_sent > day_ago:
                logging.user(user, "~SK~FMNot sending new follower email, already sent before. NBD.")
                return
        
        follower_profile = MSocialProfile.get_user(follower_user_id)
        common_followers, _ = self.common_follows(follower_user_id, direction='followers')
        common_followings, _ = self.common_follows(follower_user_id, direction='following')
        if self.user_id in common_followers:
            common_followers.remove(self.user_id)
        if self.user_id in common_followings:
            common_followings.remove(self.user_id)
        common_followers = MSocialProfile.profiles(common_followers)
        common_followings = MSocialProfile.profiles(common_followings)
        
        data = {
            'user': user,
            'follower_profile': follower_profile,
            'common_followers': common_followers,
            'common_followings': common_followings,
        }
        
        text    = render_to_string('mail/email_new_follower.txt', data)
        html    = render_to_string('mail/email_new_follower.xhtml', data)
        subject = "%s is now following your Blurblog on NewsBlur!" % follower_profile.username
        msg     = EmailMultiAlternatives(subject, text, 
                                         from_email='NewsBlur <%s>' % settings.HELLO_EMAIL,
                                         to=['%s <%s>' % (user.username, user.email)])
        msg.attach_alternative(html, "text/html")
        msg.send()
        
        MSentEmail.record(receiver_user_id=user.pk, sending_user_id=follower_user_id,
                          email_type='new_follower')
                
        logging.user(user, "~BB~FM~SBSending email for new follower: %s" % follower_profile.username)

    def send_email_for_follow_request(self, follower_user_id):
        user = User.objects.get(pk=self.user_id)
        if follower_user_id not in self.requested_follow_user_ids:
            logging.user(user, "~FMNo longer being followed by %s" % follower_user_id)
            return
        if not user.email:
            logging.user(user, "~FMNo email to send to, skipping.")
            return
        elif not user.profile.send_emails:
            logging.user(user, "~FMDisabled emails, skipping.")
            return
        if self.user_id == follower_user_id:
            return
        
        emails_sent = MSentEmail.objects.filter(receiver_user_id=user.pk,
                                                sending_user_id=follower_user_id,
                                                email_type='follow_request')
        day_ago = datetime.datetime.now() - datetime.timedelta(days=1)
        for email in emails_sent:
            if email.date_sent > day_ago:
                logging.user(user, "~SK~FMNot sending follow request email, already sent before. NBD.")
                return
        
        follower_profile = MSocialProfile.get_user(follower_user_id)
        common_followers, _ = self.common_follows(follower_user_id, direction='followers')
        common_followings, _ = self.common_follows(follower_user_id, direction='following')
        if self.user_id in common_followers:
            common_followers.remove(self.user_id)
        if self.user_id in common_followings:
            common_followings.remove(self.user_id)
        common_followers = MSocialProfile.profiles(common_followers)
        common_followings = MSocialProfile.profiles(common_followings)
        
        data = {
            'user': user,
            'follower_profile': follower_profile,
            'common_followers': common_followers,
            'common_followings': common_followings,
        }
        
        text    = render_to_string('mail/email_follow_request.txt', data)
        html    = render_to_string('mail/email_follow_request.xhtml', data)
        subject = "%s has requested to follow your Blurblog on NewsBlur" % follower_profile.username
        msg     = EmailMultiAlternatives(subject, text, 
                                         from_email='NewsBlur <%s>' % settings.HELLO_EMAIL,
                                         to=['%s <%s>' % (user.username, user.email)])
        msg.attach_alternative(html, "text/html")
        msg.send()
        
        MSentEmail.record(receiver_user_id=user.pk, sending_user_id=follower_user_id,
                          email_type='follow_request')
                
        logging.user(user, "~BB~FM~SBSending email for follow request: %s" % follower_profile.username)
            
    def save_feed_story_history_statistics(self):
        """
        Fills in missing months between earlier occurances and now.
        
        Save format: [('YYYY-MM, #), ...]
        Example output: [(2010-12, 123), (2011-01, 146)]
        """
        now = datetime.datetime.utcnow()
        min_year = now.year
        total = 0
        month_count = 0

        # Count stories, aggregate by year and month. Map Reduce!
        map_f = """
            function() {
                var date = (this.shared_date.getFullYear()) + "-" + (this.shared_date.getMonth()+1);
                var hour = this.shared_date.getHours();
                var day = this.shared_date.getDay();
                emit(this.story_hash, {'month': date, 'hour': hour, 'day': day});
            }
        """
        reduce_f = """
            function(key, values) {
                return values;
            }
        """
        dates = defaultdict(int)
        hours = defaultdict(int)
        days = defaultdict(int)
        results = MSharedStory.objects(user_id=self.user_id).map_reduce(map_f, reduce_f, output='inline')
        for result in results:
            dates[result.value['month']] += 1
            hours[str(int(result.value['hour']))] += 1
            days[str(int(result.value['day']))] += 1
            year = int(re.findall(r"(\d{4})-\d{1,2}", result.value['month'])[0])
            if year < min_year:
                min_year = year
        
        # Assemble a list with 0's filled in for missing months, 
        # trimming left and right 0's.
        months = []
        start = False
        for year in range(min_year, now.year+1):
            for month in range(1, 12+1):
                if datetime.datetime(year, month, 1) < now:
                    key = '%s-%s' % (year, month)
                    if dates.get(key) or start:
                        start = True
                        months.append((key, dates.get(key, 0)))
                        total += dates.get(key, 0)
                        month_count += 1

        self.story_count_history = months
        self.story_days_history = days
        self.story_hours_history = hours
        self.average_stories_per_month = total / max(1, month_count)
        self.save()
    
    def save_classifier_counts(self):
        
        def calculate_scores(cls, facet):
            map_f = """
                function() {
                    emit(this["%s"], {
                        pos: this.score>0 ? this.score : 0, 
                        neg: this.score<0 ? Math.abs(this.score) : 0
                    });
                }
            """ % (facet)
            reduce_f = """
                function(key, values) {
                    var result = {pos: 0, neg: 0};
                    values.forEach(function(value) {
                        result.pos += value.pos;
                        result.neg += value.neg;
                    });
                    return result;
                }
            """
            scores = []
            res = cls.objects(social_user_id=self.user_id).map_reduce(map_f, reduce_f, output='inline')
            for r in res:
                facet_values = dict([(k, int(v)) for k,v in list(r.value.items())])
                facet_values[facet] = r.key
                scores.append(facet_values)
            scores = sorted(scores, key=lambda v: v['neg'] - v['pos'])

            return scores
        
        scores = {}
        for cls, facet in [(MClassifierTitle, 'title'), 
                           (MClassifierAuthor, 'author'), 
                           (MClassifierTag, 'tag'), 
                           (MClassifierFeed, 'feed_id')]:
            scores[facet] = calculate_scores(cls, facet)
            if facet == 'feed_id' and scores[facet]:
                scores['feed'] = scores[facet]
                del scores['feed_id']
            elif not scores[facet]:
                del scores[facet]
                
        if scores:
            self.feed_classifier_counts = scores
            self.save()
    
    def save_sent_email(self, max_quota=20):
        if not self.story_email_history:
            self.story_email_history = []
        
        self.story_email_history.insert(0, datetime.datetime.now())
        self.story_email_history = self.story_email_history[:max_quota]
        
        self.save()
        
    def over_story_email_quota(self, quota=1, hours=24):
        counted = 0
        day_ago = datetime.datetime.now() - datetime.timedelta(hours=hours)
        sent_emails = self.story_email_history

        if not sent_emails:
            sent_emails = []

        for sent_date in sent_emails:
            if sent_date > day_ago:
                counted += 1
        
        if counted >= quota:
            return True
        
        return False
        
class MSocialSubscription(mongo.Document):
    UNREAD_CUTOFF = datetime.datetime.utcnow() - datetime.timedelta(days=settings.DAYS_OF_UNREAD)

    user_id = mongo.IntField()
    subscription_user_id = mongo.IntField(unique_with='user_id')
    follow_date = mongo.DateTimeField(default=datetime.datetime.utcnow())
    last_read_date = mongo.DateTimeField(default=UNREAD_CUTOFF)
    mark_read_date = mongo.DateTimeField(default=UNREAD_CUTOFF)
    unread_count_neutral = mongo.IntField(default=0)
    unread_count_positive = mongo.IntField(default=0)
    unread_count_negative = mongo.IntField(default=0)
    unread_count_updated = mongo.DateTimeField()
    oldest_unread_story_date = mongo.DateTimeField()
    needs_unread_recalc = mongo.BooleanField(default=False)
    feed_opens = mongo.IntField(default=0)
    is_trained = mongo.BooleanField(default=False)
    
    meta = {
        'collection': 'social_subscription',
        'indexes': [('user_id', 'subscription_user_id')],
        'allow_inheritance': False,
    }

    def __str__(self):
        user = User.objects.get(pk=self.user_id)
        subscription_user = User.objects.get(pk=self.subscription_user_id)
        return "Socialsub %s:%s" % (user, subscription_user)
    
    @classmethod
    def feeds(cls, user_id=None, subscription_user_id=None, calculate_all_scores=False,
              update_counts=False, *args, **kwargs):
        params = {
            'user_id': user_id,
        }
        if subscription_user_id:
            params["subscription_user_id"] = subscription_user_id
        social_subs = cls.objects.filter(**params)

        social_feeds = []
        if social_subs:
            if calculate_all_scores:
                for s in social_subs: s.calculate_feed_scores()

            # Fetch user profiles of subscriptions
            social_user_ids = [sub.subscription_user_id for sub in social_subs]
            social_profiles = MSocialProfile.profile_feeds(social_user_ids)
            for social_sub in social_subs:
                user_id = social_sub.subscription_user_id
                if social_profiles[user_id]['shared_stories_count'] <= 0:
                    continue
                if update_counts and social_sub.needs_unread_recalc:
                    social_sub.calculate_feed_scores()
                    
                # Combine subscription read counts with feed/user info
                feed = dict(list(social_sub.canonical().items()) + list(social_profiles[user_id].items()))
                social_feeds.append(feed)

        return social_feeds
    
    @classmethod
    def feeds_with_updated_counts(cls, user, social_feed_ids=None):
        feeds = {}
        
        # Get social subscriptions for user
        user_subs = cls.objects.filter(user_id=user.pk)
        if social_feed_ids:
            social_user_ids = [int(f.replace('social:', '')) for f in social_feed_ids]
            user_subs = user_subs.filter(subscription_user_id__in=social_user_ids)
            profiles = MSocialProfile.objects.filter(user_id__in=social_user_ids)
            profiles = dict((p.user_id, p) for p in profiles)
        
        for i, sub in enumerate(user_subs):
            # Count unreads if subscription is stale.
            if (sub.needs_unread_recalc or 
                (sub.unread_count_updated and
                 sub.unread_count_updated < user.profile.unread_cutoff) or 
                (sub.oldest_unread_story_date and
                 sub.oldest_unread_story_date < user.profile.unread_cutoff)):
                sub = sub.calculate_feed_scores(force=True, silent=True)

            feed_id = "social:%s" % sub.subscription_user_id
            feeds[feed_id] = {
                'ps': sub.unread_count_positive,
                'nt': sub.unread_count_neutral,
                'ng': sub.unread_count_negative,
                'id': feed_id,
            }
            if social_feed_ids and sub.subscription_user_id in profiles:
                feeds[feed_id]['shared_stories_count'] = profiles[sub.subscription_user_id].shared_stories_count

        return feeds
        
    def canonical(self):
        return {
            'user_id': self.user_id,
            'subscription_user_id': self.subscription_user_id,
            'nt': self.unread_count_neutral,
            'ps': self.unread_count_positive,
            'ng': self.unread_count_negative,
            'is_trained': self.is_trained,
            'feed_opens': self.feed_opens,
        }

    @classmethod
    def subs_for_users(cls, user_id, subscription_user_ids=None, read_filter="unread"):
        socialsubs = cls.objects
        if read_filter == "unread":
            socialsubs = socialsubs.filter(Q(unread_count_neutral__gt=0) |
                                           Q(unread_count_positive__gt=0))
        if not subscription_user_ids:
            socialsubs = socialsubs.filter(user_id=user_id)\
                                   .only('subscription_user_id', 'mark_read_date', 'is_trained')
        else:
            socialsubs = socialsubs.filter(user_id=user_id, 
                                           subscription_user_id__in=subscription_user_ids)\
                                   .only('subscription_user_id', 'mark_read_date', 'is_trained')
        
        return socialsubs

    @classmethod
    def story_hashes(cls, user_id, relative_user_id, subscription_user_ids=None, socialsubs=None,
                     read_filter="unread", order="newest", 
                     include_timestamps=False, group_by_user=True, cutoff_date=None):
        r = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL)
        pipeline = r.pipeline()
        story_hashes = {} if group_by_user else []
        
        if not socialsubs:
            socialsubs = cls.subs_for_users(relative_user_id, 
                                            subscription_user_ids=subscription_user_ids,
                                            read_filter=read_filter)
            subscription_user_ids = [sub.subscription_user_id for sub in socialsubs]
            if not subscription_user_ids:
                return story_hashes
        
        current_time = int(time.time() + 60*60*24)
        if not cutoff_date:
            cutoff_date = datetime.datetime.now() - datetime.timedelta(days=settings.DAYS_OF_STORY_HASHES)
        unread_timestamp = int(time.mktime(cutoff_date.timetuple()))-1000
        feed_counter = 0

        read_dates = dict()
        for us in socialsubs:
            read_dates[us.subscription_user_id] = int(max(us.mark_read_date, cutoff_date).strftime('%s'))

        for sub_user_id_group in chunks(subscription_user_ids, 20):
            pipeline = r.pipeline()
            for sub_user_id in sub_user_id_group:
                stories_key                 = 'B:%s' % (sub_user_id)
                sorted_stories_key          = 'zB:%s' % (sub_user_id)
                read_stories_key            = 'RS:%s' % (user_id)
                read_social_stories_key     = 'RS:%s:B:%s' % (user_id, sub_user_id)
                unread_stories_key          = 'UB:%s:%s' % (user_id, sub_user_id)
                sorted_stories_key          = 'zB:%s' % (sub_user_id)
                unread_ranked_stories_key   = 'zUB:%s:%s' % (user_id, sub_user_id)
                expire_unread_stories_key = False
            
                max_score = current_time
                if read_filter == 'unread':
                    # +1 for the intersection b/w zF and F, which carries an implicit score of 1.
                    min_score = read_dates[sub_user_id] + 1
                    pipeline.sdiffstore(unread_stories_key, stories_key, read_stories_key)
                    pipeline.sdiffstore(unread_stories_key, unread_stories_key, read_social_stories_key)
                    expire_unread_stories_key = True
                else:
                    min_score = unread_timestamp
                    unread_stories_key = stories_key

                if order == 'oldest':
                    byscorefunc = pipeline.zrangebyscore
                else:
                    byscorefunc = pipeline.zrevrangebyscore
                    min_score, max_score = max_score, min_score
            
                pipeline.zinterstore(unread_ranked_stories_key, [sorted_stories_key, unread_stories_key])
                byscorefunc(unread_ranked_stories_key, min_score, max_score, withscores=include_timestamps)
                pipeline.delete(unread_ranked_stories_key)
                if expire_unread_stories_key:
                    pipeline.delete(unread_stories_key)

        
            results = pipeline.execute()
        
            for hashes in results:
                if not isinstance(hashes, list): continue
                if group_by_user:
                    story_hashes[subscription_user_ids[feed_counter]] = hashes
                    feed_counter += 1
                else:
                    story_hashes.extend(hashes)
        
        return story_hashes
        
    def get_stories(self, offset=0, limit=6, order='newest', read_filter='all', 
                    withscores=False, hashes_only=False, cutoff_date=None, 
                    mark_read_complement=False):
        r = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL)
        ignore_user_stories = False
        
        stories_key             = 'B:%s' % (self.subscription_user_id)
        read_stories_key        = 'RS:%s' % (self.user_id)
        read_social_stories_key = 'RS:%s:B:%s' % (self.user_id, self.subscription_user_id)
        unread_stories_key      = 'UB:%s:%s' % (self.user_id, self.subscription_user_id)

        if not r.exists(stories_key):
            return []
        elif read_filter != 'unread' or not r.exists(read_stories_key):
            ignore_user_stories = True
            unread_stories_key = stories_key
        else:
            r.sdiffstore(unread_stories_key, stories_key, read_stories_key)
            r.sdiffstore(unread_stories_key, unread_stories_key, read_social_stories_key)

        sorted_stories_key          = 'zB:%s' % (self.subscription_user_id)
        unread_ranked_stories_key   = 'z%sUB:%s:%s' % ('h' if hashes_only else '', 
                                                       self.user_id, self.subscription_user_id)
        r.zinterstore(unread_ranked_stories_key, [sorted_stories_key, unread_stories_key])
        
        now             = datetime.datetime.now()
        current_time    = int(time.time() + 60*60*24)
        mark_read_time  = int(time.mktime(self.mark_read_date.timetuple())) + 1
        if cutoff_date:
            mark_read_time  = int(time.mktime(cutoff_date.timetuple())) + 1
        
        if order == 'oldest':
            byscorefunc = r.zrangebyscore
            min_score = mark_read_time
            max_score = current_time
        else: # newest
            byscorefunc = r.zrevrangebyscore
            min_score = current_time
            if mark_read_complement:
                min_score = mark_read_time
            now = datetime.datetime.now()
            unread_cutoff = cutoff_date
            if not unread_cutoff:
                unread_cutoff = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
            max_score = int(time.mktime(unread_cutoff.timetuple()))-1

        story_ids = byscorefunc(unread_ranked_stories_key, min_score, 
                                  max_score, start=offset, num=limit,
                                  withscores=withscores)
        
        if withscores:
            story_ids = [(s[0], int(s[1])) for s in story_ids]
        
        r.expire(unread_ranked_stories_key, 1*60*60)

        if not ignore_user_stories:
            r.delete(unread_stories_key)

        return story_ids
        
    @classmethod
    def feed_stories(cls, user_id, social_user_ids, offset=0, limit=6, 
                     order='newest', read_filter='all', relative_user_id=None, cache=True,
                     socialsubs=None, cutoff_date=None):
        rt = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_TEMP_POOL)
        
        if not relative_user_id:
            relative_user_id = user_id
            
        if order == 'oldest':
            range_func = rt.zrange
        else:
            range_func = rt.zrevrange
            
        if not isinstance(social_user_ids, list):
            social_user_ids = [social_user_ids]

        ranked_stories_keys  = 'zU:%s:social' % (user_id)
        unread_ranked_stories_keys  = 'zhU:%s:social' % (user_id)
        if (offset and cache and 
            rt.exists(ranked_stories_keys) and 
            rt.exists(unread_ranked_stories_keys)):
            story_hashes_and_dates = range_func(ranked_stories_keys, offset, limit, withscores=True)
            if not story_hashes_and_dates:
                return [], [], []
            story_hashes, story_dates = list(zip(*story_hashes_and_dates))
            if read_filter == "unread":
                unread_story_hashes = story_hashes
            else:
                unread_story_hashes = range_func(unread_ranked_stories_keys, 0, offset+limit)
            return story_hashes, story_dates, unread_story_hashes
        else:
            rt.delete(ranked_stories_keys)
            rt.delete(unread_ranked_stories_keys)
        
        story_hashes = cls.story_hashes(user_id, relative_user_id, 
                                        subscription_user_ids=social_user_ids, 
                                        read_filter=read_filter, order=order, 
                                        include_timestamps=True,
                                        group_by_user=False,
                                        socialsubs=socialsubs,
                                        cutoff_date=cutoff_date)
        if not story_hashes:
            return [], [], []
        
        pipeline = rt.pipeline()
        for story_hash_group in chunks(story_hashes, 100):
            pipeline.zadd(ranked_stories_keys, dict(story_hash_group))
        pipeline.execute()
        story_hashes_and_dates = range_func(ranked_stories_keys, offset, limit, withscores=True)
        if not story_hashes_and_dates:
            return [], [], []
        story_hashes, story_dates = list(zip(*story_hashes_and_dates))

        if read_filter == "unread":
            unread_feed_story_hashes = story_hashes
            rt.zunionstore(unread_ranked_stories_keys, [ranked_stories_keys])
        else:
            unread_story_hashes = cls.story_hashes(user_id, relative_user_id, 
                                                   subscription_user_ids=social_user_ids, 
                                                   read_filter="unread", order=order, 
                                                   include_timestamps=True,
                                                   group_by_user=False,
                                                   socialsubs=socialsubs,
                                                   cutoff_date=cutoff_date)
            if unread_story_hashes:
                pipeline = rt.pipeline()
                for unread_story_hash_group in chunks(unread_story_hashes, 100):
                    pipeline.zadd(unread_ranked_stories_keys, dict(unread_story_hash_group))
                pipeline.execute()
            unread_feed_story_hashes = range_func(unread_ranked_stories_keys, offset, limit)
        
        rt.expire(ranked_stories_keys, 60*60)
        rt.expire(unread_ranked_stories_keys, 60*60)
        
        return story_hashes, story_dates, unread_feed_story_hashes

    def mark_newer_stories_read(self, cutoff_date):
        if (self.unread_count_negative == 0
            and self.unread_count_neutral == 0
            and self.unread_count_positive == 0
            and not self.needs_unread_recalc):
            return
        
        cutoff_date = cutoff_date - datetime.timedelta(seconds=1)
        story_hashes = self.get_stories(limit=500, order="newest", cutoff_date=cutoff_date,
                                        read_filter="unread", hashes_only=True)
        data = self.mark_story_ids_as_read(story_hashes, aggregated=True)
        return data
    
    def mark_story_ids_as_read(self, story_hashes, feed_id=None, mark_all_read=False, request=None, aggregated=False):
        data = dict(code=0, payload=story_hashes)
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        if not request:
            request = User.objects.get(pk=self.user_id)
    
        if not self.needs_unread_recalc and not mark_all_read:
            self.needs_unread_recalc = True
            self.save()
    
        sub_username = MSocialProfile.get_user(self.subscription_user_id).username
        
        if len(story_hashes) > 1:
            logging.user(request, "~FYRead %s stories in social subscription: %s" % (len(story_hashes), sub_username))
        else:
            logging.user(request, "~FYRead story in social subscription: %s" % (sub_username))
            
        
        for story_hash in set(story_hashes):
            if feed_id is not None:
                story_hash = MStory.ensure_story_hash(story_hash, story_feed_id=feed_id)
            if feed_id is None:
                feed_id, _ = MStory.split_story_hash(story_hash)
            
            if len(story_hashes) == 1:
                RUserStory.aggregate_mark_read(feed_id)
                
            # Find other social feeds with this story to update their counts
            friend_key = "F:%s:F" % (self.user_id)
            share_key = "S:%s" % (story_hash)
            friends_with_shares = [int(f) for f in r.sinter(share_key, friend_key)]
            
            RUserStory.mark_read(self.user_id, feed_id, story_hash, social_user_ids=friends_with_shares,
                                 aggregated=(mark_all_read or aggregated))
            
            if self.user_id in friends_with_shares:
                friends_with_shares.remove(self.user_id)
            if friends_with_shares:
                socialsubs = MSocialSubscription.objects.filter(
                                user_id=self.user_id,
                                subscription_user_id__in=friends_with_shares)
                for socialsub in socialsubs:
                    if not socialsub.needs_unread_recalc and not mark_all_read:
                        socialsub.needs_unread_recalc = True
                        socialsub.save()
            
            # Also count on original subscription
            usersubs = UserSubscription.objects.filter(user=self.user_id, feed=feed_id)
            if usersubs:
                usersub = usersubs[0]
                if not usersub.needs_unread_recalc:
                    usersub.needs_unread_recalc = True
                    usersub.save()
        
        return data
        
    @classmethod
    def mark_unsub_story_ids_as_read(cls, user_id, social_user_id, story_ids, feed_id=None,
                                     request=None):
        data = dict(code=0, payload=story_ids)
        r = redis.Redis(connection_pool=settings.REDIS_POOL)

        if not request:
            request = User.objects.get(pk=user_id)
    
        if len(story_ids) > 1:
            logging.user(request, "~FYRead %s social stories from global" % (len(story_ids)))
        else:
            logging.user(request, "~FYRead social story from global")
        
        for story_id in set(story_ids):
            try:
                story = MSharedStory.objects.get(user_id=social_user_id,
                                                 story_guid=story_id)
            except MSharedStory.DoesNotExist:
                continue
                
            # Find other social feeds with this story to update their counts
            friend_key = "F:%s:F" % (user_id)
            share_key = "S:%s" % (story.story_hash)
            friends_with_shares = [int(f) for f in r.sinter(share_key, friend_key)]
            
            RUserStory.mark_read(user_id, story.story_feed_id, story.story_hash,
                                 social_user_ids=friends_with_shares)
            
            # Also count on original subscription
            usersubs = UserSubscription.objects.filter(user=user_id, feed=story.story_feed_id)
            if usersubs:
                usersub = usersubs[0]
                if not usersub.needs_unread_recalc:
                    usersub.needs_unread_recalc = True
                    usersub.save()
                # XXX TODO: Real-time notification, just for this user
        return data
    
    def mark_feed_read(self, cutoff_date=None):
        user_profile = Profile.objects.get(user_id=self.user_id)
        recount = True
        
        if cutoff_date:
            cutoff_date = cutoff_date + datetime.timedelta(seconds=1)
        else:
            # Use the latest story to get last read time.
            latest_shared_story = MSharedStory.objects(user_id=self.subscription_user_id,
                                                       shared_date__gte=user_profile.unread_cutoff
                                  ).order_by('-shared_date').only('shared_date').first()
            if latest_shared_story:
                cutoff_date = latest_shared_story['shared_date'] + datetime.timedelta(seconds=1)
            else:
                cutoff_date = datetime.datetime.utcnow()
                recount = False
        
        self.last_read_date = cutoff_date
        self.mark_read_date = cutoff_date
        self.oldest_unread_story_date = cutoff_date
        if not recount:
            self.unread_count_negative = 0
            self.unread_count_positive = 0
            self.unread_count_neutral = 0
            self.unread_count_updated = datetime.datetime.utcnow()
            self.needs_unread_recalc = False
        else:
            self.needs_unread_recalc = True
        
        # Manually mark all shared stories as read.
        unread_story_hashes = self.get_stories(read_filter='unread', limit=500, hashes_only=True,
                                               mark_read_complement=True)
        self.mark_story_ids_as_read(unread_story_hashes, mark_all_read=True)
        
        self.save()
    
    def calculate_feed_scores(self, force=False, silent=False):
        if not self.needs_unread_recalc and not force:
            return self
            
        now = datetime.datetime.now()
        user_profile = Profile.objects.get(user_id=self.user_id)

        if user_profile.last_seen_on < user_profile.unread_cutoff:
            # if not silent:
            #     logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
            return self
            
        feed_scores = dict(negative=0, neutral=0, positive=0)
        
        # Two weeks in age. If mark_read_date is older, mark old stories as read.
        date_delta = user_profile.unread_cutoff
        if date_delta < self.mark_read_date:
            date_delta = self.mark_read_date
        else:
            self.mark_read_date = date_delta

        unread_story_hashes = self.get_stories(read_filter='unread', limit=500, hashes_only=True,
                                               cutoff_date=date_delta)
        stories_db = MSharedStory.objects(user_id=self.subscription_user_id,
                                          story_hash__in=unread_story_hashes)
        story_feed_ids = set()
        for s in stories_db:
            story_feed_ids.add(s['story_feed_id'])
        story_feed_ids = list(story_feed_ids)

        usersubs = UserSubscription.objects.filter(user__pk=self.user_id, feed__pk__in=story_feed_ids)
        usersubs_map = dict((sub.feed_id, sub) for sub in usersubs)
        
        oldest_unread_story_date = now
        unread_stories_db = []

        for story in stories_db:
            if story['story_hash'] not in unread_story_hashes:
                continue
            feed_id = story.story_feed_id
            if usersubs_map.get(feed_id) and story.shared_date < usersubs_map[feed_id].mark_read_date:
                continue
                
            unread_stories_db.append(story)
            if story.shared_date < oldest_unread_story_date:
                oldest_unread_story_date = story.shared_date
        stories = Feed.format_stories(unread_stories_db)

        classifier_feeds   = list(MClassifierFeed.objects(user_id=self.user_id, social_user_id=self.subscription_user_id))
        classifier_authors = list(MClassifierAuthor.objects(user_id=self.user_id, social_user_id=self.subscription_user_id))
        classifier_titles  = list(MClassifierTitle.objects(user_id=self.user_id, social_user_id=self.subscription_user_id))
        classifier_tags    = list(MClassifierTag.objects(user_id=self.user_id, social_user_id=self.subscription_user_id))
        # Merge with feed specific classifiers
        if story_feed_ids:
            classifier_feeds   = classifier_feeds + list(MClassifierFeed.objects(user_id=self.user_id,
                                                                                 feed_id__in=story_feed_ids))
            classifier_authors = classifier_authors + list(MClassifierAuthor.objects(user_id=self.user_id,
                                                                                     feed_id__in=story_feed_ids))
            classifier_titles  = classifier_titles + list(MClassifierTitle.objects(user_id=self.user_id,
                                                                                   feed_id__in=story_feed_ids))
            classifier_tags    = classifier_tags + list(MClassifierTag.objects(user_id=self.user_id,
                                                                               feed_id__in=story_feed_ids))

        for story in stories:
            scores = {
                'feed'   : apply_classifier_feeds(classifier_feeds, story['story_feed_id'],
                                                  social_user_ids=self.subscription_user_id),
                'author' : apply_classifier_authors(classifier_authors, story),
                'tags'   : apply_classifier_tags(classifier_tags, story),
                'title'  : apply_classifier_titles(classifier_titles, story),
            }
            
            max_score = max(scores['author'], scores['tags'], scores['title'])
            min_score = min(scores['author'], scores['tags'], scores['title'])
            
            if max_score > 0:
                feed_scores['positive'] += 1
            elif min_score < 0:
                feed_scores['negative'] += 1
            else:
                if scores['feed'] > 0:
                    feed_scores['positive'] += 1
                elif scores['feed'] < 0:
                    feed_scores['negative'] += 1
                else:
                    feed_scores['neutral'] += 1
                
        
        self.unread_count_positive = feed_scores['positive']
        self.unread_count_neutral = feed_scores['neutral']
        self.unread_count_negative = feed_scores['negative']
        self.unread_count_updated = datetime.datetime.now()
        self.oldest_unread_story_date = oldest_unread_story_date
        self.needs_unread_recalc = False
        
        self.save()

        if (self.unread_count_positive == 0 and 
            self.unread_count_neutral == 0):
            self.mark_feed_read()
        
        if not silent:
            logging.info(' ---> [%s] Computing social scores: %s (%s/%s/%s)' % (user_profile, self.subscription_user_id, feed_scores['negative'], feed_scores['neutral'], feed_scores['positive']))
            
        return self
    
    @classmethod
    def mark_dirty_sharing_story(cls, user_id, story_feed_id, story_guid_hash):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        friends_key = "F:%s:F" % (user_id)
        share_key = "S:%s:%s" % (story_feed_id, story_guid_hash)
        following_user_ids = r.sinter(friends_key, share_key)
        following_user_ids = [int(f) for f in following_user_ids]
        if not following_user_ids:
            return None

        social_subs = cls.objects.filter(user_id=user_id, subscription_user_id__in=following_user_ids)
        for social_sub in social_subs:
            social_sub.needs_unread_recalc = True
            social_sub.save()
        return social_subs

class MCommentReply(mongo.EmbeddedDocument):
    reply_id      = mongo.ObjectIdField()
    user_id       = mongo.IntField()
    publish_date  = mongo.DateTimeField()
    comments      = mongo.StringField()
    email_sent    = mongo.BooleanField(default=False)
    liking_users  = mongo.ListField(mongo.IntField())
    
    def canonical(self):
        reply = {
            'reply_id': self.reply_id,
            'user_id': self.user_id,
            'publish_date': relative_timesince(self.publish_date),
            'date': self.publish_date,
            'comments': self.comments,
        }
        return reply
        
    meta = {
        'ordering': ['publish_date'],
        'id_field': 'reply_id',
        'allow_inheritance': False,
        'strict': False,
    }


class MSharedStory(mongo.DynamicDocument):
    user_id                  = mongo.IntField()
    shared_date              = mongo.DateTimeField()
    comments                 = mongo.StringField()
    has_comments             = mongo.BooleanField(default=False)
    has_replies              = mongo.BooleanField(default=False)
    replies                  = mongo.ListField(mongo.EmbeddedDocumentField(MCommentReply))
    source_user_id           = mongo.IntField()
    story_hash               = mongo.StringField()
    story_feed_id            = mongo.IntField()
    story_date               = mongo.DateTimeField()
    story_title              = mongo.StringField(max_length=1024)
    story_content            = mongo.StringField()
    story_content_z          = mongo.BinaryField()
    story_original_content   = mongo.StringField()
    story_original_content_z = mongo.BinaryField()
    original_text_z          = mongo.BinaryField()
    original_page_z          = mongo.BinaryField()
    story_content_type       = mongo.StringField(max_length=255)
    story_author_name        = mongo.StringField()
    story_permalink          = mongo.StringField()
    story_guid               = mongo.StringField(unique_with=('user_id',))
    story_guid_hash          = mongo.StringField(max_length=6)
    image_urls               = mongo.ListField(mongo.StringField(max_length=1024))
    story_tags               = mongo.ListField(mongo.StringField(max_length=250))
    posted_to_services       = mongo.ListField(mongo.StringField(max_length=20))
    mute_email_users         = mongo.ListField(mongo.IntField())
    liking_users             = mongo.ListField(mongo.IntField())
    emailed_reshare          = mongo.BooleanField(default=False)
    emailed_replies          = mongo.ListField(mongo.ObjectIdField())
    image_count              = mongo.IntField()
    image_sizes              = mongo.ListField(mongo.DictField())
    
    meta = {
        'collection': 'shared_stories',
        'indexes': [('user_id', '-shared_date'), ('user_id', 'story_feed_id'), 
                    'shared_date', 'story_guid', 'story_feed_id', 'story_hash'],
        'ordering': ['-shared_date'],
        'allow_inheritance': False,
        'strict': False,
    }

    def __str__(self):
        user = User.objects.get(pk=self.user_id)
        return "%s: %s (%s)%s%s" % (user.username, 
                                    self.decoded_story_title[:20], 
                                    self.story_feed_id, 
                                    ': ' if self.has_comments else '', 
                                    self.comments[:20])

    @property
    def guid_hash(self):
        return hashlib.sha1(self.story_guid.encode('utf-8')).hexdigest()[:6]
    
    @property
    def feed_guid_hash(self):
        return "%s:%s" % (self.story_feed_id or "0", self.guid_hash)
    
    @property
    def decoded_story_title(self):
        h = html.parser.HTMLParser()
        return h.unescape(self.story_title)
        
    def canonical(self):
        return {
            "user_id": self.user_id,
            "shared_date": self.shared_date,
            "story_title": self.story_title,
            "story_content": self.story_content_z and zlib.decompress(self.story_content_z),
            "comments": self.comments,
        }
        
    def save(self, *args, **kwargs):
        scrubber = SelectiveScriptScrubber()

        if self.story_content:
            self.story_content = scrubber.scrub(self.story_content)
            self.story_content_z = zlib.compress(self.story_content)
            self.story_content = None
        if self.story_original_content:
            self.story_original_content_z = zlib.compress(self.story_original_content)
            self.story_original_content = None

        self.story_guid_hash = self.guid_hash
        self.story_title = strip_tags(self.story_title)
        self.story_hash = self.feed_guid_hash

        self.comments = linkify(strip_tags(self.comments))
        for reply in self.replies:
            reply.comments = linkify(strip_tags(reply.comments))

        self.shared_date = self.shared_date or datetime.datetime.utcnow()
        self.has_replies = bool(len(self.replies))
        
        super(MSharedStory, self).save(*args, **kwargs)
        
        author = MSocialProfile.get_user(self.user_id)
        author.count_follows()
        
        self.sync_redis()
        
        MActivity.new_shared_story(user_id=self.user_id, source_user_id=self.source_user_id, 
                                   story_title=self.story_title, 
                                   comments=self.comments, story_feed_id=self.story_feed_id,
                                   story_id=self.story_guid, share_date=self.shared_date)
        return self
        
    def delete(self, *args, **kwargs):
        MActivity.remove_shared_story(user_id=self.user_id, story_feed_id=self.story_feed_id,
                                      story_id=self.story_guid)

        self.remove_from_redis()

        super(MSharedStory, self).delete(*args, **kwargs)

    @classmethod
    def trim_old_stories(cls, stories=10, days=90, dryrun=False):
        print(" ---> Fetching shared story counts...")
        stats = settings.MONGODB.newsblur.shared_stories.aggregate([{
            "$group": {
                "_id":      "$user_id",
                "stories":  {"$sum": 1},
            },
        }, {
            "$match": {
                "stories": {"$gte": stories}
            },
        }])
        month_ago = datetime.datetime.now() - datetime.timedelta(days=days)
        user_ids = list(stats)
        user_ids = sorted(user_ids, key=lambda x:x['stories'], reverse=True)
        print(" ---> Found %s users with more than %s starred stories" % (len(user_ids), stories))

        total = 0
        for stat in user_ids:
            try:
                user = User.objects.select_related('profile').get(pk=stat['_id'])
            except User.DoesNotExist:
                user = None
            
            if user and (user.profile.is_premium or user.profile.last_seen_on > month_ago):
                continue
            
            total += stat['stories']
            username = "%s (%s)" % (user and user.username or " - ", stat['_id'])
            print(" ---> %19.19s: %-20.20s %s stories" % (user and user.profile.last_seen_on or "Deleted",
                                                          username, 
                                                          stat['stories']))
            if not dryrun and stat['_id']:
                cls.objects.filter(user_id=stat['_id']).delete()
            elif not dryrun and stat['_id'] == 0:
                print(" ---> Deleting unshared stories (user_id = 0)")
                cls.objects.filter(user_id=stat['_id']).delete()
                    
        
        print(" ---> Deleted %s stories in total." % total)
        
    def unshare_story(self):
        socialsubs = MSocialSubscription.objects.filter(subscription_user_id=self.user_id,
                                                        needs_unread_recalc=False)
        for socialsub in socialsubs:
            socialsub.needs_unread_recalc = True
            socialsub.save()

        self.delete()
    
    @classmethod
    def feed_quota(cls, user_id, story_hash, feed_id=None, days=1, quota=1):
        day_ago = datetime.datetime.now()-datetime.timedelta(days=days)
        params = dict(user_id=user_id,
                      shared_date__gte=day_ago, 
                      story_hash__nin=[story_hash])
        if feed_id:
            params['story_feed_id'] = feed_id
        shared_count = cls.objects.filter(**params).count()

        return shared_count >= quota
    
    @classmethod
    def count_potential_spammers(cls, days=1, destroy=False):
        try:
            guaranteed_spammers = detect_spammers(days=days, destroy=destroy)
        except NameError:
            logging.debug(" ---> ~FR~SNMissing ~SBspam.py~SN")
            guaranteed_spammers = []
        
        return guaranteed_spammers
        
    @classmethod
    def get_shared_stories_from_site(cls, feed_id, user_id, story_url, limit=3):
        your_story = cls.objects.filter(story_feed_id=feed_id,
                                        story_permalink=story_url,
                                        user_id=user_id).limit(1).first()
        same_stories = cls.objects.filter(story_feed_id=feed_id,
                                          story_permalink=story_url,
                                          user_id__ne=user_id
                                          ).order_by('-shared_date')

        same_stories = [{
            "user_id": story.user_id,
            "comments": story.comments,
            "relative_date": relative_timesince(story.shared_date),
            "blurblog_permalink": story.blurblog_permalink(),
        } for story in same_stories]
        
        other_stories = []
        if feed_id:
            other_stories = cls.objects.filter(story_feed_id=feed_id,
                                               story_permalink__ne=story_url
                                               ).order_by('-shared_date').limit(limit)
            other_stories = [{
                "user_id": story.user_id,
                "story_title": story.story_title,
                "story_permalink": story.story_permalink,
                "comments": story.comments,
                "relative_date": relative_timesince(story.shared_date),
                "blurblog_permalink": story.blurblog_permalink(),
            } for story in other_stories]
        
        return your_story, same_stories, other_stories
    
    def set_source_user_id(self, source_user_id):
        if source_user_id == self.user_id:
            return
            
        def find_source(source_user_id, seen_user_ids):
            parent_shared_story = MSharedStory.objects.filter(user_id=source_user_id, 
                                                              story_guid=self.story_guid, 
                                                              story_feed_id=self.story_feed_id).limit(1)
            if parent_shared_story and parent_shared_story[0].source_user_id:
                user_id = parent_shared_story[0].source_user_id
                if user_id in seen_user_ids:
                    return source_user_id
                else:
                    seen_user_ids.append(user_id)
                    return find_source(user_id, seen_user_ids)
            else:
                return source_user_id
        
        if source_user_id:
            source_user_id = find_source(source_user_id, [])
            if source_user_id == self.user_id:
                return
            elif not self.source_user_id or source_user_id != self.source_user_id:
                self.source_user_id = source_user_id
                logging.debug("   ---> Re-share from %s." % source_user_id)
                self.save()
                
                MInteraction.new_reshared_story(user_id=self.source_user_id,
                                                reshare_user_id=self.user_id,
                                                comments=self.comments,
                                                story_title=self.story_title,
                                                story_feed_id=self.story_feed_id,
                                                story_id=self.story_guid)
    
    def mute_for_user(self, user_id):
        if user_id not in self.mute_email_users:
            self.mute_email_users.append(user_id)
            self.save()
        
    @classmethod
    def switch_feed(cls, original_feed_id, duplicate_feed_id):
        shared_stories = cls.objects.filter(story_feed_id=duplicate_feed_id)
        logging.info(" ---> %s shared stories" % shared_stories.count())
        for story in shared_stories:
            story.story_feed_id = original_feed_id
            story.save()
        
    @classmethod
    def collect_popular_stories(cls, cutoff=None, days=None, shared_feed_ids=None):
        if not days:
            days = 3
        if not cutoff:
            cutoff = 6
        if not shared_feed_ids:
            shared_feed_ids = []
        # shared_stories_count = sum(json.decode(MStatistics.get('stories_shared')))
        # cutoff = cutoff or max(math.floor(.025 * shared_stories_count), 3)
        today = datetime.datetime.now() - datetime.timedelta(days=days)
        
        map_f = """
            function() {
                emit(this.story_hash, {
                    'story_hash': this.story_hash, 
                    'feed_id': this.story_feed_id, 
                    'title': this.story_title,
                    'count': 1
                });
            }
        """
        reduce_f = """
            function(key, values) {
                var r = {'story_hash': key, 'count': 0};
                for (var i=0; i < values.length; i++) {
                    r.feed_id = values[i].feed_id;
                    r.title = values[i].title;
                    r.count += values[i].count;
                }
                return r;
            }
        """
        finalize_f = """
            function(key, value) {
                if (value.count >= %(cutoff)s && [%(shared_feed_ids)s].indexOf(value.feed_id) == -1) {
                    var english_title = value.title.replace(/[^\\062-\\177]/g, "");
                    if (english_title.length < 5) return;
            
                    return value;
                }
            }
        """ % {'cutoff': cutoff, 'shared_feed_ids': ', '.join(shared_feed_ids)}
        res = cls.objects(shared_date__gte=today).map_reduce(map_f, reduce_f, 
                                                             finalize_f=finalize_f, 
                                                             output='inline')
        stories = dict([(r.key, r.value) for r in res if r.value])
        return stories, cutoff
        
    @classmethod
    def share_popular_stories(cls, cutoff=None, days=None, interactive=True):
        publish_new_stories = False
        popular_profile = MSocialProfile.objects.get(username='popular')
        popular_user = User.objects.get(pk=popular_profile.user_id)
        week_ago = datetime.datetime.now() - datetime.timedelta(days=7)
        shared_feed_ids = [str(s.story_feed_id) 
                           for s in MSharedStory.objects(user_id=popular_profile.user_id,
                                                         shared_date__gte=week_ago).only('story_feed_id')]
        shared_stories_today, cutoff = cls.collect_popular_stories(cutoff=cutoff, days=days, 
                                                                   shared_feed_ids=shared_feed_ids)
        shared = 0
        
        for story_hash, story_info in list(shared_stories_today.items()):
            story, _ = MStory.find_story(story_info['feed_id'], story_info['story_hash'])
            if not story:
                logging.user(popular_user, "~FRPopular stories, story not found: %s" % story_info)
                continue
            if story.story_feed_id in shared_feed_ids:
                logging.user(popular_user, "~FRPopular stories, story feed just shared: %s" % story_info)
                continue
            
            if interactive:
                feed = Feed.get_by_id(story.story_feed_id)
                accept_story = eval(input("%s / %s [Y/n]: " % (story.decoded_story_title, feed.title)))
                if accept_story in ['n', 'N']: continue
                
            story_db = dict([(k, v) for k, v in list(story._data.items()) 
                                if k is not None and v is not None])
            story_db.pop('user_id', None)
            story_db.pop('id', None)
            story_db.pop('comments', None)
            story_db.pop('replies', None)
            story_db['has_comments'] = False
            story_db['has_replies'] = False
            story_db['shared_date'] = datetime.datetime.now()
            story_values = {
                'user_id': popular_profile.user_id,
                'story_guid': story_db['story_guid'],
            }
            try:
                shared_story = MSharedStory.objects.get(**story_values)
            except MSharedStory.DoesNotExist:
                story_values.update(story_db)
                shared_story = MSharedStory.objects.create(**story_values)
                shared_story.post_to_service('twitter')
                shared += 1
                shared_feed_ids.append(story.story_feed_id)
                publish_new_stories = True
                logging.user(popular_user, "~FCSharing: ~SB~FM%s (%s shares, %s min)" % (
                    story.decoded_story_title[:50],
                    story_info['count'],
                    cutoff))
            
        if publish_new_stories:
            socialsubs = MSocialSubscription.objects.filter(subscription_user_id=popular_user.pk)
            for socialsub in socialsubs:
                socialsub.needs_unread_recalc = True
                socialsub.save()
            shared_story.publish_update_to_subscribers()
        
        return shared

    @staticmethod
    def check_shared_story_hashes(user_id, story_hashes, r=None):
        if not r:
            r = redis.Redis(connection_pool=settings.REDIS_POOL)
        pipeline = r.pipeline()
        
        for story_hash in story_hashes:
            feed_id, guid_hash = MStory.split_story_hash(story_hash)
            share_key = "S:%s:%s" % (feed_id, guid_hash)
            pipeline.sismember(share_key, user_id)
        shared_hashes = pipeline.execute()
        
        return [story_hash for s, story_hash in enumerate(story_hashes) if shared_hashes[s]]

    @classmethod
    def sync_all_redis(cls, drop=False):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        h = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL)
        # h2 = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL2)
        if drop:
            for key_name in ["C", "S"]:
                keys = r.keys("%s:*" % key_name)
                print(" ---> Removing %s keys named %s:*" % (len(keys), key_name))
                for key in keys:
                    r.delete(key)
        for story in cls.objects.all():
            story.sync_redis_shares(r=r)
            story.sync_redis_story(r=h)
    
    def sync_redis(self):
        self.sync_redis_shares()
        self.sync_redis_story()

    def sync_redis_shares(self, r=None):
        if not r:
            r = redis.Redis(connection_pool=settings.REDIS_POOL)
        
        share_key   = "S:%s:%s" % (self.story_feed_id, self.guid_hash)
        comment_key = "C:%s:%s" % (self.story_feed_id, self.guid_hash)
        r.sadd(share_key, self.user_id)
        if self.has_comments:
            r.sadd(comment_key, self.user_id)
        else:
            r.srem(comment_key, self.user_id)

    def sync_redis_story(self, r=None):
        if not r:
            r = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL)
        # if not r2:
        #     r2 = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL2)
        
        r.sadd('B:%s' % self.user_id, self.feed_guid_hash)
        # r2.sadd('B:%s' % self.user_id, self.feed_guid_hash)
        redis_data = {
            self.feed_guid_hash : time.mktime(self.shared_date.timetuple())
        }
        r.zadd('zB:%s' % self.user_id, redis_data)
        # r2.zadd('zB:%s' % self.user_id, {self.feed_guid_hash:
        #        time.mktime(self.shared_date.timetuple())})
        r.expire('B:%s' % self.user_id, settings.DAYS_OF_STORY_HASHES*24*60*60)
        # r2.expire('B:%s' % self.user_id, settings.DAYS_OF_STORY_HASHES*24*60*60)
        r.expire('zB:%s' % self.user_id, settings.DAYS_OF_STORY_HASHES*24*60*60)
        # r2.expire('zB:%s' % self.user_id, settings.DAYS_OF_STORY_HASHES*24*60*60)
    
    def remove_from_redis(self):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        share_key = "S:%s:%s" % (self.story_feed_id, self.guid_hash)
        r.srem(share_key, self.user_id)

        comment_key = "C:%s:%s" % (self.story_feed_id, self.guid_hash)
        r.srem(comment_key, self.user_id)

        h = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL)
        # h2 = redis.Redis(connection_pool=settings.REDIS_STORY_HASH_POOL2)
        h.srem('B:%s' % self.user_id, self.feed_guid_hash)
        # h2.srem('B:%s' % self.user_id, self.feed_guid_hash)
        h.zrem('zB:%s' % self.user_id, self.feed_guid_hash)
        # h2.zrem('zB:%s' % self.user_id, self.feed_guid_hash)

    def publish_update_to_subscribers(self):
        try:
            r = redis.Redis(connection_pool=settings.REDIS_PUBSUB_POOL)
            feed_id = "social:%s" % self.user_id
            listeners_count = r.publish("%s:story" % feed_id, 'story:new:%s' % self.story_hash)
            if listeners_count:
                logging.debug("   ---> ~FMPublished to %s subscribers" % (listeners_count))
        except redis.ConnectionError:
            logging.debug("   ***> ~BMRedis is unavailable for real-time.")

    def comments_with_author(self):
        comments = {
            'id': self.id,
            'user_id': self.user_id,
            'comments': self.comments,
            'shared_date': relative_timesince(self.shared_date),
            'date': self.shared_date,
            'replies': [reply.canonical() for reply in self.replies],
            'liking_users': self.liking_users and list(self.liking_users),
            'source_user_id': self.source_user_id,
        }
        return comments
    
    def comment_with_author_and_profiles(self):
        comment = self.comments_with_author()
        profile_user_ids = set([comment['user_id']])
        reply_user_ids = [reply['user_id'] for reply in comment['replies']]
        profile_user_ids = profile_user_ids.union(reply_user_ids)
        profile_user_ids = profile_user_ids.union(comment['liking_users'])
        if comment['source_user_id']:
            profile_user_ids.add(comment['source_user_id'])
        profiles = MSocialProfile.objects.filter(user_id__in=list(profile_user_ids))
        profiles = [profile.canonical(compact=True) for profile in profiles]

        return comment, profiles
        
    @classmethod
    def stories_with_comments_and_profiles(cls, stories, user_id, check_all=False):
        r = redis.Redis(connection_pool=settings.REDIS_POOL)
        friend_key = "F:%s:F" % (user_id)
        profile_user_ids = set()
        for story in stories: 
            story['friend_comments'] = []
            story['friend_shares'] = []
            story['public_comments'] = []
            story['reply_count'] = 0
            if check_all or story['comment_count']:
                comment_key = "C:%s:%s" % (story['story_feed_id'], story['guid_hash'])
                story['comment_count'] = r.scard(comment_key)
                friends_with_comments = [int(f) for f in r.sinter(comment_key, friend_key)]
                sharer_user_ids = [int(f) for f in r.smembers(comment_key)]
                shared_stories = []
                if sharer_user_ids:
                    params = {
                        'story_hash': story['story_hash'],
                        'user_id__in': sharer_user_ids,
                    }
                    if 'story_db_id' in params:
                        params.pop('story_db_id')
                    shared_stories = cls.objects.filter(**params)\
                                                .hint([('story_hash', 1)])
                for shared_story in shared_stories:
                    comments = shared_story.comments_with_author()
                    story['reply_count'] += len(comments['replies'])
                    if shared_story.user_id in friends_with_comments:
                        story['friend_comments'].append(comments)
                    else:
                        story['public_comments'].append(comments)
                    if comments.get('source_user_id'):
                        profile_user_ids.add(comments['source_user_id'])
                    if comments.get('liking_users'):
                        profile_user_ids = profile_user_ids.union(comments['liking_users'])
                all_comments = story['friend_comments'] + story['public_comments']
                profile_user_ids = profile_user_ids.union([reply['user_id'] 
                                                           for c in all_comments
                                                           for reply in c['replies']])
                if story.get('source_user_id'):
                    profile_user_ids.add(story['source_user_id'])
                story['comment_count_friends'] = len(friends_with_comments)
                story['comment_count_public'] = story['comment_count'] - len(friends_with_comments)
                
            if check_all or story['share_count']:
                share_key = "S:%s:%s" % (story['story_feed_id'], story['guid_hash'])
                story['share_count'] = r.scard(share_key)
                friends_with_shares = [int(f) for f in r.sinter(share_key, friend_key)]
                nonfriend_user_ids = [int(f) for f in r.sdiff(share_key, friend_key)]
                profile_user_ids.update(nonfriend_user_ids)
                profile_user_ids.update(friends_with_shares)
                story['commented_by_public']  = [c['user_id'] for c in story['public_comments']]
                story['commented_by_friends'] = [c['user_id'] for c in story['friend_comments']]
                story['shared_by_public']     = list(set(nonfriend_user_ids) - 
                                                    set(story['commented_by_public']))
                story['shared_by_friends']    = list(set(friends_with_shares) - 
                                                     set(story['commented_by_friends']))
                story['share_count_public']  = story['share_count'] - len(friends_with_shares)
                story['share_count_friends'] = len(friends_with_shares)
                story['friend_user_ids'] = list(set(story['commented_by_friends'] + story['shared_by_friends']))
                story['public_user_ids'] = list(set(story['commented_by_public'] + story['shared_by_public']))
                if not story['share_user_ids']:
                    story['share_user_ids'] = story['friend_user_ids'] + story['public_user_ids']
                if story.get('source_user_id'):
                    profile_user_ids.add(story['source_user_id'])
                shared_stories = []
                if story['shared_by_friends']:
                    params = {
                        'story_hash': story['story_hash'],
                        'user_id__in': story['shared_by_friends'],
                    }
                    shared_stories = cls.objects.filter(**params)\
                                                .hint([('story_hash', 1)])
                for shared_story in shared_stories:
                    comments = shared_story.comments_with_author()
                    story['reply_count'] += len(comments['replies'])
                    story['friend_shares'].append(comments)
                    profile_user_ids = profile_user_ids.union([reply['user_id'] 
                                                               for reply in comments['replies']])
                    if comments.get('source_user_id'):
                        profile_user_ids.add(comments['source_user_id'])
                    if comments.get('liking_users'):
                        profile_user_ids = profile_user_ids.union(comments['liking_users'])
            
        profiles = MSocialProfile.objects.filter(user_id__in=list(profile_user_ids))
        profiles = [profile.canonical(compact=True) for profile in profiles]
        
        # Toss public comments by private profiles
        profiles_dict = dict((profile['user_id'], profile) for profile in profiles)
        for story in stories:
            commented_by_public = story.get('commented_by_public') or [c['user_id'] for c in story['public_comments']]
            for user_id in commented_by_public:
                if profiles_dict[user_id]['private']:
                    story['public_comments'] = [c for c in story['public_comments'] if c['user_id'] != user_id]
                    story['comment_count_public'] -= 1
        
        return stories, profiles
    
    @staticmethod
    def attach_users_to_stories(stories, profiles):
        profiles = dict([(p['user_id'], p) for p in profiles])
        for s, story in enumerate(stories):
            for u, user_id in enumerate(story['shared_by_friends']):
                if user_id not in profiles: continue
                stories[s]['shared_by_friends'][u] = profiles[user_id]
            for u, user_id in enumerate(story['shared_by_public']):
                if user_id not in profiles: continue
                stories[s]['shared_by_public'][u] = profiles[user_id]
            for comment_set in ['friend_comments', 'public_comments', 'friend_shares']:
                for c, comment in enumerate(story[comment_set]):
                    if comment['user_id'] not in profiles: continue
                    stories[s][comment_set][c]['user'] = profiles[comment['user_id']]
                    if comment['source_user_id'] and comment['source_user_id'] in profiles:
                        stories[s][comment_set][c]['source_user'] = profiles[comment['source_user_id']]
                    for r, reply in enumerate(comment['replies']):
                        if reply['user_id'] not in profiles: continue
                        stories[s][comment_set][c]['replies'][r]['user'] = profiles[reply['user_id']]
                    stories[s][comment_set][c]['liking_user_ids'] = list(comment['liking_users'])
                    for u, user_id in enumerate(comment['liking_users']):
                        if user_id not in profiles: continue
                        stories[s][comment_set][c]['liking_users'][u] = profiles[user_id]

        return stories
    
    @staticmethod
    def attach_users_to_comment(comment, profiles):
        profiles = dict([(p['user_id'], p) for p in profiles])

        if comment['user_id'] not in profiles: return comment
        comment['user'] = profiles[comment['user_id']]

        if comment['source_user_id']:
            comment['source_user'] = profiles[comment['source_user_id']]

        for r, reply in enumerate(comment['replies']):
            if reply['user_id'] not in profiles: continue
            comment['replies'][r]['user'] = profiles[reply['user_id']]
        comment['liking_user_ids'] = list(comment['liking_users'])
        for u, user_id in enumerate(comment['liking_users']):
            if user_id not in profiles: continue
            comment['liking_users'][u] = profiles[user_id]

        return comment
        
    def add_liking_user(self, user_id):
        if user_id not in self.liking_users:
            self.liking_users.append(user_id)
            self.save()

    def remove_liking_user(self, user_id):
        if user_id in self.liking_users:
            self.liking_users.remove(user_id)
            self.save()
        
    def blurblog_permalink(self):
        profile = MSocialProfile.get_user(self.user_id)
        return "%sstory/%s/%s" % (
            profile.blurblog_url,
            slugify(self.story_title)[:20],
            self.story_hash
        )
    
    def generate_post_to_service_message(self, truncate=None, include_url=True):
        message = strip_tags(self.comments)
        if not message or len(message) < 1:
            message = self.decoded_story_title
            if include_url and truncate:
                message = truncate_chars(message, truncate - 18 - 30)
            feed = Feed.get_by_id(self.story_feed_id)
            if feed:
                if truncate:
                    message += " (%s)" % truncate_chars(feed.feed_title, 18)
                else:
                    message += " (%s)" % truncate_chars(feed.feed_title, 30)
            if include_url:
                message += " " + self.blurblog_permalink()
        elif include_url:
            if truncate:
                message = truncate_chars(message, truncate - 24)
            message += " " + self.blurblog_permalink()
        
        return message
        
    def post_to_service(self, service):
        user = User.objects.get(pk=self.user_id)
        
        if service in self.posted_to_services:
            logging.user(user, "~BM~FRAlready posted to %s." % (service))
            return
        
        posts_last_hour = MSharedStory.objects.filter(user_id=self.user_id,
                                                      posted_to_services__contains=service,
                                                      shared_date__gte=datetime.datetime.now() -
                                                                       datetime.timedelta(hours=1)).count()
        if posts_last_hour >= 3:
            logging.user(user, "~BM~FRPosted to %s > 3 times in past hour" % service)
            return
            
        posted = False
        social_service = MSocialServices.objects.get(user_id=self.user_id)
        
        message = self.generate_post_to_service_message()
        logging.user(user, "~BM~FGPosting to %s: ~SB%s" % (service, message))
        
        if service == 'twitter':
            posted = social_service.post_to_twitter(self)
        elif service == 'facebook':
            posted = social_service.post_to_facebook(self)
        
        if posted:
            self.posted_to_services.append(service)
            self.save()
            
    def notify_user_ids(self, include_parent=True):
        user_ids = set()
        for reply in self.replies:
            if reply.user_id not in self.mute_email_users:
                user_ids.add(reply.user_id)
            
        if include_parent and self.user_id not in self.mute_email_users:
            user_ids.add(self.user_id)
        
        return list(user_ids)
    
    def reply_for_id(self, reply_id):
        for reply in self.replies:
            if reply.reply_id == reply_id:
                return reply
                
    def send_emails_for_new_reply(self, reply_id):
        if reply_id in self.emailed_replies:
            logging.debug(" ***> Already sent reply email: %s on %s" % (reply_id, self))
            return

        reply = self.reply_for_id(reply_id)
        if not reply:
            logging.debug(" ***> Reply doesn't exist: %s on %s" % (reply_id, self))
            return
            
        notify_user_ids = self.notify_user_ids()
        if reply.user_id in notify_user_ids:
            notify_user_ids.remove(reply.user_id)
        reply_user = User.objects.get(pk=reply.user_id)
        reply_user_profile = MSocialProfile.get_user(reply.user_id)
        sent_emails = 0

        story_feed = Feed.get_by_id(self.story_feed_id)
        comment = self.comments_with_author()
        profile_user_ids = set([comment['user_id']])
        reply_user_ids = list(r['user_id'] for r in comment['replies'])
        profile_user_ids = profile_user_ids.union(reply_user_ids)
        if self.source_user_id:
            profile_user_ids.add(self.source_user_id)
        profiles = MSocialProfile.objects.filter(user_id__in=list(profile_user_ids))
        profiles = [profile.canonical(compact=True) for profile in profiles]
        comment = MSharedStory.attach_users_to_comment(comment, profiles)
        
        for user_id in notify_user_ids:
            user = User.objects.get(pk=user_id)

            if not user.email or not user.profile.send_emails:
                if not user.email:
                    logging.user(user, "~FMNo email to send to, skipping.")
                elif not user.profile.send_emails:
                    logging.user(user, "~FMDisabled emails, skipping.")
                continue
            
            mute_url = "http://%s%s" % (
                Site.objects.get_current().domain,
                reverse('social-mute-story', kwargs={
                    'secret_token': user.profile.secret_token,
                    'shared_story_id': self.id,
                })
            )
            data = {
                'reply_user_profile': reply_user_profile,
                'comment': comment,
                'shared_story': self,
                'story_feed': story_feed,
                'mute_url': mute_url,
            }
            story_title = self.decoded_story_title.replace('\n', ' ')
            
            text    = render_to_string('mail/email_reply.txt', data)
            html    = pynliner.fromString(render_to_string('mail/email_reply.xhtml', data))
            subject = "%s replied to you on \"%s\" on NewsBlur" % (reply_user.username, story_title)
            msg     = EmailMultiAlternatives(subject, text, 
                                             from_email='NewsBlur <%s>' % settings.HELLO_EMAIL,
                                             to=['%s <%s>' % (user.username, user.email)])
            msg.attach_alternative(html, "text/html")
            msg.send()
            sent_emails += 1
                
        logging.user(reply_user, "~BB~FM~SBSending %s/%s email%s for new reply: %s" % (
            sent_emails, len(notify_user_ids), 
            '' if len(notify_user_ids) == 1 else 's', 
            self.decoded_story_title[:30]))
        
        self.emailed_replies.append(reply.reply_id)
        self.save()
    
    def send_email_for_reshare(self):
        if self.emailed_reshare:
            logging.debug(" ***> Already sent reply email: %s" % self)
            return
            
        reshare_user = User.objects.get(pk=self.user_id)
        reshare_user_profile = MSocialProfile.get_user(self.user_id)
        original_user = User.objects.get(pk=self.source_user_id)
        original_shared_story = MSharedStory.objects.get(user_id=self.source_user_id,
                                                         story_guid=self.story_guid)
                                                         
        if not original_user.email or not original_user.profile.send_emails:
            if not original_user.email:
                logging.user(original_user, "~FMNo email to send to, skipping.")
            elif not original_user.profile.send_emails:
                logging.user(original_user, "~FMDisabled emails, skipping.")
            return
            
        story_feed = Feed.get_by_id(self.story_feed_id)
        comment = self.comments_with_author()
        profile_user_ids = set([comment['user_id']])
        reply_user_ids = [reply['user_id'] for reply in comment['replies']]
        profile_user_ids = profile_user_ids.union(reply_user_ids)
        if self.source_user_id:
            profile_user_ids.add(self.source_user_id)
        profiles = MSocialProfile.objects.filter(user_id__in=list(profile_user_ids))
        profiles = [profile.canonical(compact=True) for profile in profiles]
        comment = MSharedStory.attach_users_to_comment(comment, profiles)
        
        mute_url = "http://%s%s" % (
            Site.objects.get_current().domain,
            reverse('social-mute-story', kwargs={
                'secret_token': original_user.profile.secret_token,
                'shared_story_id': original_shared_story.id,
            })
        )
        data = {
            'comment': comment,
            'shared_story': self,
            'reshare_user_profile': reshare_user_profile,
            'original_shared_story': original_shared_story,
            'story_feed': story_feed,
            'mute_url': mute_url,
        }
        story_title = self.decoded_story_title.replace('\n', ' ')
        
        text    = render_to_string('mail/email_reshare.txt', data)
        html    = pynliner.fromString(render_to_string('mail/email_reshare.xhtml', data))
        subject = "%s re-shared \"%s\" from you on NewsBlur" % (reshare_user.username, story_title)
        msg     = EmailMultiAlternatives(subject, text, 
                                         from_email='NewsBlur <%s>' % settings.HELLO_EMAIL,
                                         to=['%s <%s>' % (original_user.username, original_user.email)])
        msg.attach_alternative(html, "text/html")
        msg.send()
        
        self.emailed_reshare = True
        self.save()
            
        logging.user(reshare_user, "~BB~FM~SBSending %s email for story re-share: %s" % (
            original_user.username,
            self.decoded_story_title[:30]))
    
    def extract_image_urls(self, force=False):
        if not self.story_content_z:
            return

        if self.image_urls and not force:
            return
            
        soup = BeautifulSoup(zlib.decompress(self.story_content_z), features="lxml")
        image_sources = [img.get('src') for img in soup.findAll('img') if img and img.get('src')]
        if len(image_sources) > 0:
            self.image_urls = image_sources
            self.save()
            
    def calculate_image_sizes(self, force=False):
        if not self.story_content_z:
            return
        
        if not force and self.image_count:
            return self.image_sizes
            
        headers = {
            'User-Agent': 'NewsBlur Image Fetcher - %s '
                          '(Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_1) '
                          'AppleWebKit/534.48.3 (KHTML, like Gecko) Version/5.1 '
                          'Safari/534.48.3)' % (
                settings.NEWSBLUR_URL
            ),
        }
        
        self.extract_image_urls()
        image_sizes = []
        
        for image_source in self.image_urls[:10]:
            if any(ignore in image_source for ignore in IGNORE_IMAGE_SOURCES):
                continue
            req = requests.get(image_source, headers=headers, stream=True)
            try:
                datastream = BytesIO(req.content)
                width, height = ImageOps.image_size(datastream)
            except IOError as e:
                logging.debug(" ***> Couldn't read image: %s / %s" % (e, image_source))
                datastream = BytesIO(req.content[:100])
                _, width, height = image_size(datastream)
            # if width <= 16 or height <= 16:
            #     continue
            image_sizes.append({'src': image_source, 'size': (width, height)})
        
        if image_sizes:
            image_sizes = sorted(image_sizes, key=lambda i: i['size'][0] * i['size'][1],
                                              reverse=True)
            self.image_sizes = image_sizes
        self.image_count = len(image_sizes)
        self.save()
        
        logging.debug(" ---> ~SN~FGFetched image sizes on shared story: ~SB%s/%s images" % 
                      (self.image_count, len(self.image_urls)))
        
        return image_sizes
    
    def fetch_original_text(self, force=False, request=None, debug=False):
        original_text_z = self.original_text_z
        feed = Feed.get_by_id(self.story_feed_id)
        
        if not original_text_z or force:
            ti = TextImporter(self, feed=feed, request=request, debug=False)
            original_text = ti.fetch()
        else:
            logging.user(request, "~FYFetching ~FGoriginal~FY story text, ~SBfound.")
            original_text = zlib.decompress(original_text_z)
        
        return original_text

    def fetch_original_page(self, force=False, request=None, debug=False):
        if not self.original_page_z or force:
            feed = Feed.get_by_id(self.story_feed_id)
            importer = PageImporter(request=request, feed=feed, story=self)
            original_page = importer.fetch_story()
        else:
            logging.user(request, "~FYFetching ~FGoriginal~FY story page, ~SBfound.")
            original_page = zlib.decompress(self.original_page_z)
        
        return original_page

class MSocialServices(mongo.Document):
    user_id               = mongo.IntField()
    autofollow            = mongo.BooleanField(default=True)
    twitter_uid           = mongo.StringField()
    twitter_access_key    = mongo.StringField()
    twitter_access_secret = mongo.StringField()
    twitter_friend_ids    = mongo.ListField(mongo.StringField())
    twitter_picture_url   = mongo.StringField()
    twitter_username      = mongo.StringField()
    twitter_refresh_date  = mongo.DateTimeField()
    facebook_uid          = mongo.StringField()
    facebook_access_token = mongo.StringField()
    facebook_friend_ids   = mongo.ListField(mongo.StringField())
    facebook_picture_url  = mongo.StringField()
    facebook_refresh_date = mongo.DateTimeField()
    upload_picture_url    = mongo.StringField()
    syncing_twitter       = mongo.BooleanField(default=False)
    syncing_facebook      = mongo.BooleanField(default=False)
    
    meta = {
        'collection': 'social_services',
        'indexes': ['user_id', 'twitter_friend_ids', 'facebook_friend_ids', 'twitter_uid', 'facebook_uid'],
        'allow_inheritance': False,
        'strict': False,
    }
    
    def __str__(self):
        user = User.objects.get(pk=self.user_id)
        return "%s (Twitter: %s, FB: %s)" % (user.username, self.twitter_uid, self.facebook_uid)
        
    def canonical(self):
        user = User.objects.get(pk=self.user_id)
        return {
            'twitter': {
                'twitter_username': self.twitter_username,
                'twitter_picture_url': self.twitter_picture_url,
                'twitter_uid': self.twitter_uid,
                'syncing': self.syncing_twitter,
            },
            'facebook': {
                'facebook_uid': self.facebook_uid,
                'facebook_picture_url': self.facebook_picture_url,
                'syncing': self.syncing_facebook,
            },
            'gravatar': {
                'gravatar_picture_url': "https://www.gravatar.com/avatar/" + \
                                        hashlib.md5(user.email.lower().encode('utf-8')).hexdigest()
            },
            'upload': {
                'upload_picture_url': self.upload_picture_url
            }
        }
    
    @classmethod
    def get_user(cls, user_id):
        try:
            profile = cls.objects.get(user_id=user_id)
            created = False
        except cls.DoesNotExist:
            profile = cls.objects.create(user_id=user_id)
            created = True
        except cls.MultipleObjectsReturned:
            dupes = cls.objects.filter(user_id=user_id)
            logging.debug(" ---> ~FRDeleting dupe social services. %s found." % dupes.count())
            for dupe in dupes[1:]:
                dupe.delete()
            profile = dupes[0]
            created = False

        if created:
            profile.save()
        return profile
        
    @classmethod
    def profile(cls, user_id):
        profile = cls.get_user(user_id=user_id)
        return profile.canonical()
    
    def save_uploaded_photo(self, photo):
        photo_body = photo.read()
        filename = photo.name

        s3 = s3_utils.S3Store()
        image_name = s3.save_profile_picture(self.user_id, filename, photo_body)
        if image_name:        
            self.upload_picture_url = "https://s3.amazonaws.com/%s/avatars/%s/thumbnail_%s" % (
                settings.S3_AVATARS_BUCKET_NAME,
                self.user_id,
                image_name,
            )
            self.save()
        
        return image_name and self.upload_picture_url

    def twitter_api(self):
        twitter_consumer_key = settings.TWITTER_CONSUMER_KEY
        twitter_consumer_secret = settings.TWITTER_CONSUMER_SECRET
        auth = tweepy.OAuthHandler(twitter_consumer_key, twitter_consumer_secret)
        auth.set_access_token(self.twitter_access_key, self.twitter_access_secret)
        api = tweepy.API(auth)
        return api
    
    def facebook_api(self):
        graph = facebook.GraphAPI(access_token=self.facebook_access_token, version="3.1")
        return graph
    
    def sync_twitter_friends(self):
        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~BG~FMTwitter import starting...")
        
        api = self.twitter_api()
        try:
            twitter_user = api.me()
        except tweepy.TweepError as e:
            api = None
        
        if not api:
            logging.user(user, "~BG~FMTwitter import ~SBfailed~SN: no api access.")
            self.syncing_twitter = False
            self.save()
            return
            
        self.twitter_picture_url = twitter_user.profile_image_url_https
        self.twitter_username = twitter_user.screen_name
        self.twitter_refreshed_date = datetime.datetime.utcnow()
        self.syncing_twitter = False
        self.save()
        
        profile = MSocialProfile.get_user(self.user_id)
        profile.location = profile.location or twitter_user.location
        profile.bio = profile.bio or twitter_user.description
        profile.website = profile.website or twitter_user.url
        profile.save()
        profile.count_follows()
        
        if not profile.photo_url or not profile.photo_service:
            self.set_photo('twitter')

        try:
            friend_ids = list(str(friend.id) for friend in list(tweepy.Cursor(api.friends).items()))
        except tweepy.TweepError as e:
            logging.user(user, "~BG~FMTwitter import ~SBfailed~SN: %s" % e)
            return
        if not friend_ids:
            logging.user(user, "~BG~FMTwitter import ~SBfailed~SN: no friend_ids.")
        self.twitter_friend_ids = friend_ids
        self.save()
        
        following = self.follow_twitter_friends()
        
        if not following:
            logging.user(user, "~BG~FMTwitter import finished.")
        
    def follow_twitter_friends(self):
        social_profile = MSocialProfile.get_user(self.user_id)
        following = []
        followers = 0
        
        if not self.autofollow:
            return following

        # Follow any friends already on NewsBlur
        user_social_services = MSocialServices.objects.filter(twitter_uid__in=self.twitter_friend_ids)
        for user_social_service in user_social_services:
            followee_user_id = user_social_service.user_id
            socialsub = social_profile.follow_user(followee_user_id)
            if socialsub:
                following.append(followee_user_id)
    
        # Friends already on NewsBlur should follow back
        # following_users = MSocialServices.objects.filter(twitter_friend_ids__contains=self.twitter_uid)
        # for following_user in following_users:
        #     if following_user.autofollow:
        #         following_user_profile = MSocialProfile.get_user(following_user.user_id)
        #         following_user_profile.follow_user(self.user_id, check_unfollowed=True)
        #         followers += 1
        
        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~BG~FMTwitter import: %s users, now following ~SB%s~SN with ~SB%s~SN follower-backs" % (len(self.twitter_friend_ids), len(following), followers))
        
        return following
        
    def sync_facebook_friends(self):
        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~BG~FMFacebook import starting...")
        
        graph = self.facebook_api()
        if not graph:
            logging.user(user, "~BG~FMFacebook import ~SBfailed~SN: no api access.")
            self.syncing_facebook = False
            self.save()
            return

        friends = graph.get_connections("me", "friends")
        if not friends:
            logging.user(user, "~BG~FMFacebook import ~SBfailed~SN: no friend_ids.")
            self.syncing_facebook = False
            self.save()
            return

        facebook_friend_ids = [str(friend["id"]) for friend in friends["data"]]
        self.facebook_friend_ids = facebook_friend_ids
        self.facebook_refresh_date = datetime.datetime.utcnow()
        self.facebook_picture_url = "https://graph.facebook.com/%s/picture" % self.facebook_uid
        self.syncing_facebook = False
        self.save()
        
        facebook_user = graph.request('me', args={'fields':'website,about,location'})
        profile = MSocialProfile.get_user(self.user_id)
        profile.location = profile.location or (facebook_user.get('location') and facebook_user['location']['name'])
        profile.bio = profile.bio or facebook_user.get('about')
        if not profile.website and facebook_user.get('website'):
            profile.website = facebook_user.get('website').split()[0]
        profile.save()
        profile.count_follows()
        if not profile.photo_url or not profile.photo_service:
            self.set_photo('facebook')
        
        self.follow_facebook_friends()
        
    def follow_facebook_friends(self):
        social_profile = MSocialProfile.get_user(self.user_id)
        following = []
        followers = 0
        
        if not self.autofollow:
            return following

        # Follow any friends already on NewsBlur
        user_social_services = MSocialServices.objects.filter(facebook_uid__in=self.facebook_friend_ids)
        for user_social_service in user_social_services:
            followee_user_id = user_social_service.user_id
            socialsub = social_profile.follow_user(followee_user_id)
            if socialsub:
                following.append(followee_user_id)
    
        # Friends already on NewsBlur should follow back
        # following_users = MSocialServices.objects.filter(facebook_friend_ids__contains=self.facebook_uid)
        # for following_user in following_users:
        #     if following_user.autofollow:
        #         following_user_profile = MSocialProfile.get_user(following_user.user_id)
        #         following_user_profile.follow_user(self.user_id, check_unfollowed=True)
        #         followers += 1
        
        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~BG~FMFacebook import: %s users, now following ~SB%s~SN with ~SB%s~SN follower-backs" % (len(self.facebook_friend_ids), len(following), followers))
        
        return following
    
    def disconnect_twitter(self):
        self.syncing_twitter = False
        self.twitter_uid = None
        self.save()
    
    def disconnect_facebook(self):
        self.syncing_facebook = False
        self.facebook_uid = None
        self.save()
    
    def set_photo(self, service):
        profile = MSocialProfile.get_user(self.user_id)
        if service == 'nothing':
            service = None

        profile.photo_service = service
        if not service:
            profile.photo_url = None
        elif service == 'twitter':
            profile.photo_url = self.twitter_picture_url
        elif service == 'facebook':
            profile.photo_url = self.facebook_picture_url
        elif service == 'upload':
            profile.photo_url = self.upload_picture_url
        elif service == 'gravatar':
            user = User.objects.get(pk=self.user_id)
            profile.photo_url = "https://www.gravatar.com/avatar/" + \
                                hashlib.md5(user.email.encode('utf-8')).hexdigest()
        profile.save()
        return profile
    
    @classmethod
    def sync_all_twitter_photos(cls, days=14, everybody=False):
        if everybody:
            sharers = [ss.user_id for ss in MSocialServices.objects.all().only('user_id')]
        elif days:
            week_ago = datetime.datetime.now() - datetime.timedelta(days=days)
            shares = MSharedStory.objects.filter(shared_date__gte=week_ago)
            sharers = sorted(set([s.user_id for s in shares]))
        print(" ---> %s sharing user_ids" % len(sorted(sharers)))

        for user_id in sharers:
            try:
                profile = MSocialProfile.objects.get(user_id=user_id)
            except MSocialProfile.DoesNotExist:
                continue
            if not profile.photo_service == 'twitter': continue
            ss = MSocialServices.objects.get(user_id=user_id)
            try:
                ss.sync_twitter_photo()
                print(" ---> Syncing %s" % user_id)
            except Exception as e:
                print(" ***> Exception on %s: %s" % (user_id, e))

    def sync_twitter_photo(self):
        profile = MSocialProfile.get_user(self.user_id)

        if profile.photo_service != "twitter":
            return
        
        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~FCSyncing Twitter profile photo...")
        
        try:
            api = self.twitter_api()
            me = api.me()
        except (tweepy.TweepError, TypeError) as e:
            logging.user(user, "~FRException (%s): ~FCsetting to blank profile photo" % e)
            self.twitter_picture_url = None
            self.set_photo("nothing")
            return

        self.twitter_picture_url = me.profile_image_url_https
        self.save()
        self.set_photo('twitter')
        
    def post_to_twitter(self, shared_story):
        message = shared_story.generate_post_to_service_message(truncate=280)
        shared_story.calculate_image_sizes()
        
        try:
            api = self.twitter_api()
            filename = self.fetch_image_file_for_twitter(shared_story)
            if filename:
                api.update_with_media(filename, status=message)
                os.remove(filename)
            else:
                api.update_status(status=message)
        except tweepy.TweepError as e:
            user = User.objects.get(pk=self.user_id)
            logging.user(user, "~FRTwitter error: ~SB%s" % e)
            return
            
        return True
    
    def fetch_image_file_for_twitter(self, shared_story):
        if not shared_story.image_urls: return

        user = User.objects.get(pk=self.user_id)
        logging.user(user, "~FCFetching image for twitter: ~SB%s" % shared_story.image_urls[0])
        
        url = shared_story.image_urls[0]
        image_filename = os.path.basename(urllib.parse.urlparse(url).path)
        req = requests.get(url, stream=True)
        filename = "/tmp/%s-%s" % (shared_story.story_hash, image_filename)
        
        if req.status_code == 200:
            f = open(filename, "wb")
            for chunk in req:
                f.write(chunk)
            f.close()
        
            return filename
                
    def post_to_facebook(self, shared_story):
        message = shared_story.generate_post_to_service_message(include_url=False)
        shared_story.calculate_image_sizes()
        content = zlib.decompress(shared_story.story_content_z)[:1024]
        
        try:
            api = self.facebook_api()
            # api.put_wall_post(message=message)
            api.put_object('me', '%s:share' % settings.FACEBOOK_NAMESPACE, 
                           link=shared_story.blurblog_permalink(), 
                           type="link", 
                           name=shared_story.decoded_story_title, 
                           description=content, 
                           website=shared_story.blurblog_permalink(),
                           message=message,
                           )
        except facebook.GraphAPIError as e:
            logging.debug("---> ~SN~FMFacebook posting error, disconnecting: ~SB~FR%s" % e)
            self.disconnect_facebook()
            return
            
        return True
        

class MInteraction(mongo.Document):
    user_id      = mongo.IntField()
    date         = mongo.DateTimeField(default=datetime.datetime.now)
    category     = mongo.StringField()
    title        = mongo.StringField()
    content      = mongo.StringField()
    with_user_id = mongo.IntField()
    feed_id      = mongo.DynamicField()
    story_feed_id= mongo.IntField()
    content_id   = mongo.StringField()
    
    meta = {
        'collection': 'interactions',
        'indexes': [('user_id', '-date'), 'category', 'with_user_id'],
        'allow_inheritance': False,
        'ordering': ['-date'],
    }
    
    def __str__(self):
        user = User.objects.get(pk=self.user_id)
        with_user = self.with_user_id and User.objects.get(pk=self.with_user_id)
        return "<%s> %s on %s: %s - %s" % (user.username, with_user and with_user.username, self.date, 
                                           self.category, self.content and self.content[:20])
    
    def canonical(self):
        story_hash = None
        if self.story_feed_id:
            story_hash = MStory.ensure_story_hash(self.content_id, story_feed_id=self.story_feed_id)

        return {
            'date': self.date,
            'category': self.category,
            'title': self.title,
            'content': self.content,
            'with_user_id': self.with_user_id,
            'feed_id': self.feed_id,
            'story_feed_id': self.story_feed_id,
            'content_id': self.content_id,
            'story_hash': story_hash,
        }
    
    @classmethod
    def trim(cls, user_id, limit=100):
        user = User.objects.get(pk=user_id)
        interactions = cls.objects.filter(user_id=user_id).skip(limit)
        interaction_count = interactions.count(True)

        if interaction_count == 0:
            interaction_count = cls.objects.filter(user_id=user_id).count()
            logging.user(user, "~FBNot trimming interactions, only ~SB%s~SN interactions found" % interaction_count)
            return
        
        logging.user(user, "~FBTrimming ~SB%s~SN interactions..." % interaction_count)

        for interaction in interactions:
            interaction.delete()

        logging.user(user, "~FBDone trimming ~SB%s~SN interactions" % interaction_count)
    
    @classmethod
    def publish_update_to_subscribers(self, user_id):
        user = User.objects.get(pk=user_id)
        try:
            r = redis.Redis(connection_pool=settings.REDIS_POOL)
            listeners_count = r.publish(user.username, 'interaction:new')
            if listeners_count:
                logging.debug("   ---> ~FMPublished to %s subscribers" % (listeners_count))
        except redis.ConnectionError:
            logging.debug("   ***> ~BMRedis is unavailable for real-time.")

    @classmethod
    def user(cls, user_id, page=1, limit=None, categories=None):
        user_profile = Profile.objects.get(user=user_id)
        dashboard_date = user_profile.dashboard_date or user_profile.last_seen_on
        page = max(1, page)
        limit = int(limit) if limit else 4
        offset = (page-1) * limit
        
        interactions_db = cls.objects.filter(user_id=user_id)
        if categories:
            interactions_db = interactions_db.filter(category__in=categories)
        interactions_db = interactions_db[offset:offset+limit+1]
        
        has_next_page = len(interactions_db) > limit
        interactions_db = interactions_db[offset:offset+limit]
        with_user_ids = [i.with_user_id for i in interactions_db if i.with_user_id]
        social_profiles = dict((p.user_id, p) for p in MSocialProfile.objects.filter(user_id__in=with_user_ids))
        
        interactions = []
        for interaction_db in interactions_db:
            interaction = interaction_db.canonical()
            social_profile = social_profiles.get(interaction_db.with_user_id)
            if social_profile:
                interaction['photo_url'] = social_profile.profile_photo_url
            interaction['with_user'] = social_profiles.get(interaction_db.with_user_id)
            interaction['time_since'] = relative_timesince(interaction_db.date)
            interaction['date'] = interaction_db.date
            interaction['is_new'] = interaction_db.date > dashboard_date
            interactions.append(interaction)

        return interactions, has_next_page
    
    @classmethod
    def user_unread_count(cls, user_id):
        user_profile = Profile.objects.get(user=user_id)
        dashboard_date = user_profile.dashboard_date or user_profile.last_seen_on
        
        interactions_count = cls.objects.filter(user_id=user_id, date__gte=dashboard_date).count()
        
        return interactions_count
        
    @classmethod
    def new_follow(cls, follower_user_id, followee_user_id):
        params = {
            'user_id': followee_user_id, 
            'with_user_id': follower_user_id,
            'category': 'follow',
        }
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            cls.objects.create(**params)
        except cls.MultipleObjectsReturned:
            dupes = cls.objects.filter(**params).order_by('-date')
            logging.debug(" ---> ~FRDeleting dupe follow interactions. %s found." % dupes.count())
            for dupe in dupes[1:]:
                dupe.delete()
        
        cls.publish_update_to_subscribers(followee_user_id)
    
    @classmethod
    def new_comment_reply(cls, user_id, reply_user_id, reply_content, story_id, story_feed_id, story_title=None, original_message=None):
        params = {
            'user_id': user_id,
            'with_user_id': reply_user_id,
            'category': 'comment_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % user_id,
            'story_feed_id': story_feed_id,
            'title': story_title,
            'content_id': story_id,
        }
        if original_message:
            params['content'] = original_message
            original = cls.objects.filter(**params).limit(1)
            if original:
                original = original[0]
                original.content = linkify(strip_tags(reply_content))
                original.save()
            else:
                original_message = None

        if not original_message:
            cls.objects.create(**params)
        
        cls.publish_update_to_subscribers(user_id)
            
    @classmethod
    def remove_comment_reply(cls, user_id, reply_user_id, reply_content, story_id, story_feed_id):
        params = {
            'user_id': user_id,
            'with_user_id': reply_user_id,
            'category': 'comment_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % user_id,
            'story_feed_id': story_feed_id,
            'content_id': story_id,
        }
        original = cls.objects.filter(**params)
        original.delete()
        
        cls.publish_update_to_subscribers(user_id)
    
    @classmethod
    def new_comment_like(cls, liking_user_id, comment_user_id, story_id, story_feed_id, story_title, comments):
        params = dict(user_id=comment_user_id,
                      with_user_id=liking_user_id,
                      category="comment_like",
                      feed_id="social:%s" % comment_user_id,
                      story_feed_id=story_feed_id,
                      content_id=story_id)
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            params.update(dict(title=story_title, content=comments))
            cls.objects.create(**params)
        
        cls.publish_update_to_subscribers(comment_user_id)

    @classmethod
    def new_reply_reply(cls, user_id, comment_user_id, reply_user_id, reply_content, story_id, story_feed_id, story_title=None, original_message=None):
        params = {
            'user_id': user_id,
            'with_user_id': reply_user_id,
            'category': 'reply_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % comment_user_id,
            'story_feed_id': story_feed_id,
            'title': story_title,
            'content_id': story_id,
        }
        if original_message:
            params['content'] = original_message
            original = cls.objects.filter(**params).limit(1)
            if original:
                original = original[0]
                original.content = reply_content
                original.save()
            else:
                original_message = None

        if not original_message:
            cls.objects.create(**params)
        
        cls.publish_update_to_subscribers(user_id)
            
    @classmethod
    def remove_reply_reply(cls, user_id, comment_user_id, reply_user_id, reply_content, story_id, story_feed_id):
        params = {
            'user_id': user_id,
            'with_user_id': reply_user_id,
            'category': 'reply_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % comment_user_id,
            'story_feed_id': story_feed_id,
            'content_id': story_id,
        }
        original = cls.objects.filter(**params)
        original.delete()
        
        cls.publish_update_to_subscribers(user_id)
        
    @classmethod
    def new_reshared_story(cls, user_id, reshare_user_id, comments, story_title, story_feed_id, story_id, original_comments=None):
        params = {
            'user_id': user_id,
            'with_user_id': reshare_user_id,
            'category': 'story_reshare',
            'content': comments,
            'title': story_title,
            'feed_id': "social:%s" % reshare_user_id,
            'story_feed_id': story_feed_id,
            'content_id': story_id,
        }
        if original_comments:
            params['content'] = original_comments
            original = cls.objects.filter(**params).limit(1)
            if original:
                interaction = original[0]
                interaction.content = comments
                interaction.save()
            else:
                original_comments = None

        if not original_comments:
            cls.objects.create(**params)
        
        cls.publish_update_to_subscribers(user_id)

class MActivity(mongo.Document):
    user_id      = mongo.IntField()
    date         = mongo.DateTimeField(default=datetime.datetime.now)
    category     = mongo.StringField()
    title        = mongo.StringField()
    content      = mongo.StringField()
    with_user_id = mongo.IntField()
    feed_id      = mongo.DynamicField()
    story_feed_id= mongo.IntField()
    content_id   = mongo.StringField()
    
    meta = {
        'collection': 'activities',
        'indexes': [('user_id', '-date'), 'category', 'with_user_id'],
        'allow_inheritance': False,
        'ordering': ['-date'],
    }
    
    def __str__(self):
        user = User.objects.get(pk=self.user_id)
        return "<%s> %s - %s" % (user.username, self.category, self.content and self.content[:20])
    
    def canonical(self):
        story_hash = None
        if self.story_feed_id:
            story_hash = MStory.ensure_story_hash(self.content_id, story_feed_id=self.story_feed_id)

        return {
            'date': self.date,
            'category': self.category,
            'title': self.title,
            'content': self.content,
            'user_id': self.user_id,
            'with_user_id': self.with_user_id or self.user_id,
            'feed_id': self.feed_id or self.story_feed_id,
            'story_feed_id': self.story_feed_id or self.feed_id,
            'content_id': self.content_id,
            'story_hash': story_hash,
        }
    
    @classmethod
    def trim(cls, user_id, limit=100):
        user = User.objects.get(pk=user_id)
        activities = cls.objects.filter(user_id=user_id).skip(limit)
        activity_count = activities.count(True)

        if activity_count == 0:
            activity_count = cls.objects.filter(user_id=user_id).count()
            logging.user(user, "~FBNot trimming activities, only ~SB%s~SN activities found" % activity_count)
            return
        
        logging.user(user, "~FBTrimming ~SB%s~SN activities..." % activity_count)

        for activity in activities:
            activity.delete()

        logging.user(user, "~FBDone trimming ~SB%s~SN activities" % activity_count)
        
    @classmethod
    def user(cls, user_id, page=1, limit=4, public=False, categories=None):
        user_profile = Profile.objects.get(user=user_id)
        dashboard_date = user_profile.dashboard_date or user_profile.last_seen_on
        page = max(1, page)
        limit = int(limit)
        offset = (page-1) * limit
        
        activities_db = cls.objects.filter(user_id=user_id)
        if categories:
            activities_db = activities_db.filter(category__in=categories)
        if public:
            activities_db = activities_db.filter(category__nin=['star', 'feedsub'])
        activities_db = activities_db[offset:offset+limit+1]
        
        has_next_page = len(activities_db) > limit
        activities_db = activities_db[offset:offset+limit]
        with_user_ids = [a.with_user_id for a in activities_db if a.with_user_id]
        social_profiles = dict((p.user_id, p) for p in MSocialProfile.objects.filter(user_id__in=with_user_ids))
        activities = []
        for activity_db in activities_db:
            activity = activity_db.canonical()
            activity['date'] = activity_db.date
            activity['time_since'] = relative_timesince(activity_db.date)
            social_profile = social_profiles.get(activity_db.with_user_id)
            if social_profile:
                activity['photo_url'] = social_profile.profile_photo_url
            activity['is_new'] = activity_db.date > dashboard_date
            activity['with_user'] = social_profiles.get(activity_db.with_user_id or activity_db.user_id)
            activities.append(activity)
        
        return activities, has_next_page
            
    @classmethod
    def new_starred_story(cls, user_id, story_title, story_feed_id, story_id):
        params = dict(user_id=user_id,
                      category='star',
                      story_feed_id=story_feed_id,
                      content_id=story_id)
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            params.update(dict(content=story_title))
            cls.objects.create(**params)

    @classmethod
    def remove_starred_story(cls, user_id, story_feed_id, story_id):
        params = {
            'user_id': user_id,
            'category': 'star',
            'story_feed_id': story_feed_id,
            'content_id': story_id,
        }
        original = cls.objects.filter(**params)
        original.delete()
    
    @classmethod
    def new_feed_subscription(cls, user_id, feed_id, feed_title):
        params = {
            "user_id": user_id,
            "category": 'feedsub',
            "feed_id": feed_id,
        }
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            params.update(dict(content=feed_title))
            cls.objects.create(**params)
        except cls.MultipleObjectsReturned:
            dupes = cls.objects.filter(**params).order_by('-date')
            logging.debug(" ---> ~FRDeleting dupe feed subscription activities. %s found." % dupes.count())
            for dupe in dupes[1:]:
                dupe.delete()

                           
    @classmethod
    def new_follow(cls, follower_user_id, followee_user_id):
        params = {
            'user_id': follower_user_id, 
            'with_user_id': followee_user_id,
            'category': 'follow',
        }
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            cls.objects.create(**params)
        except cls.MultipleObjectsReturned:
            dupes = cls.objects.filter(**params).order_by('-date')
            logging.debug(" ---> ~FRDeleting dupe follow activities. %s found." % dupes.count())
            for dupe in dupes[1:]:
                dupe.delete()
    
    @classmethod
    def new_comment_reply(cls, user_id, comment_user_id, reply_content, story_id, story_feed_id, story_title=None, original_message=None):
        params = {
            'user_id': user_id,
            'with_user_id': comment_user_id,
            'category': 'comment_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % comment_user_id,
            'story_feed_id': story_feed_id,
            'title': story_title,
            'content_id': story_id,
        }
        if original_message:
            params['content'] = original_message
            original = cls.objects.filter(**params).limit(1)
            if original:
                original = original[0]
                original.content = linkify(strip_tags(reply_content))
                original.save()
            else:
                original_message = None

        if not original_message:
            cls.objects.create(**params)
            
    @classmethod
    def remove_comment_reply(cls, user_id, comment_user_id, reply_content, story_id, story_feed_id):
        params = {
            'user_id': user_id,
            'with_user_id': comment_user_id,
            'category': 'comment_reply',
            'content': linkify(strip_tags(reply_content)),
            'feed_id': "social:%s" % comment_user_id,
            'story_feed_id': story_feed_id,
            'content_id': story_id,
        }
        original = cls.objects.filter(**params)
        original.delete()
            
    @classmethod
    def new_comment_like(cls, liking_user_id, comment_user_id, story_id, story_feed_id, story_title, comments):
        params = dict(user_id=liking_user_id,
                      with_user_id=comment_user_id,
                      category="comment_like",
                      feed_id="social:%s" % comment_user_id,
                      story_feed_id=story_feed_id,
                      content_id=story_id)
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            params.update(dict(title=story_title, content=comments))
            cls.objects.create(**params)
    
    @classmethod
    def new_shared_story(cls, user_id, source_user_id, story_title, comments, story_feed_id, story_id, share_date=None):
        data = {
            "user_id": user_id,
            "category": 'sharedstory',
            "feed_id": "social:%s" % user_id,
            "story_feed_id": story_feed_id,
            "content_id": story_id,
        }
        extradata = {'with_user_id': source_user_id,
                     'title': story_title,
                     'content': comments}

        try:
            a = cls.objects.get(**data)
        except cls.DoesNotExist:
            merged = {}
            merged.update(data)
            merged.update(extradata)
            a = cls.objects.create(**merged)
        except cls.MultipleObjectsReturned:
            dupes = cls.objects.filter(**data)
            logging.debug(" ---> ~FRDeleting dupe shared story activities. %s found." % dupes.count())
            a = dupes[0]
            for dupe in dupes[1:]:
                dupe.delete()

        if a.content != comments:
            a.content = comments
            a.save()
        if source_user_id and a.with_user_id != source_user_id:
            a.source_user_id = source_user_id
            a.save()
        if share_date:
            a.date = share_date
            a.save()

    @classmethod
    def remove_shared_story(cls, user_id, story_feed_id, story_id):
        params = dict(user_id=user_id,
                      category='sharedstory',
                      feed_id="social:%s" % user_id,
                      story_feed_id=story_feed_id,
                      content_id=story_id)
        try:
            a = cls.objects.get(**params)
        except cls.DoesNotExist:
            return
        except cls.MultipleObjectsReturned:
            a = cls.objects.filter(**params)
        
        a.delete()
        
    @classmethod
    def new_signup(cls, user_id):
        params = dict(user_id=user_id, with_user_id=user_id, category="signup")
        try:
            return cls.objects.get(**params)
        except cls.DoesNotExist:
            return cls.objects.create(**params)


class MFollowRequest(mongo.Document):
    follower_user_id    = mongo.IntField(unique_with='followee_user_id')
    followee_user_id    = mongo.IntField()
    date                = mongo.DateTimeField(default=datetime.datetime.now)
    
    meta = {
        'collection': 'follow_request',
        'indexes': ['follower_user_id', 'followee_user_id'],
        'ordering': ['-date'],
        'allow_inheritance': False,
    }
    
    @classmethod
    def add(cls, follower_user_id, followee_user_id):
        params = dict(follower_user_id=follower_user_id, followee_user_id=followee_user_id)
        try:
            cls.objects.get(**params)
        except cls.DoesNotExist:
            cls.objects.create(**params)
    
    @classmethod
    def remove(cls, follower_user_id, followee_user_id):
        cls.objects.filter(follower_user_id=follower_user_id, 
                           followee_user_id=followee_user_id).delete()