def classifiers(self, usersub):
classifiers = {}
if usersub.is_trained:
classifiers['feeds'] = list(MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id,
social_user_id=0))
classifiers['authors'] = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifiers['titles'] = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifiers['tags'] = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
return classifiers
def classifiers(self, usersub):
classifiers = {}
if usersub.is_trained:
classifiers['feeds'] = list(MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id,
social_user_id=0))
classifiers['authors'] = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifiers['titles'] = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifiers['tags'] = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
return classifiers
def sort_by_feed(classifiers):
feed_classifiers = defaultdict(list)
for classifier in classifiers:
feed_classifiers[classifier.feed_id].append(classifier)
return feed_classifiers
classifiers = {}
try:
classifier_feeds = sort_by_feed(
MClassifierFeed.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_authors = sort_by_feed(
MClassifierAuthor.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_titles = sort_by_feed(
MClassifierTitle.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_tags = sort_by_feed(
MClassifierTag.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
except OperationFailure:
logging.info(" ***> Classifiers failure")
else:
for feed_id in found_feed_ids:
classifiers[feed_id] = get_classifiers_for_user(
user, feed_id, classifier_feeds[feed_id],
classifier_authors[feed_id], classifier_titles[feed_id],
classifier_tags[feed_id])
# Just need to format stories
for story in stories:
story_date = localtime_for_timezone(story['story_date'],
def load_river_stories(request):
limit = 18
offset = 0
start = datetime.datetime.utcnow()
user = get_user(request)
feed_ids = [int(feed_id) for feed_id in request.REQUEST.getlist('feeds') if feed_id]
original_feed_ids = list(feed_ids)
page = int(request.REQUEST.get('page', 1))
read_stories_count = int(request.REQUEST.get('read_stories_count', 0))
new_flag = request.REQUEST.get('new_flag', False)
bottom_delta = datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if not feed_ids:
logging.user(request, "~FCLoading empty river stories: page %s" % (page))
return dict(stories=[])
# Fetch all stories at and before the page number.
# Not a single page, because reading stories can move them up in the unread order.
# `read_stories_count` is an optimization, works best when all 25 stories before have been read.
limit = limit * page - read_stories_count
# Read stories to exclude
read_stories = MUserStory.objects(user_id=user.pk, feed_id__in=feed_ids).only('story_id')
read_stories = [rs.story_id for rs in read_stories]
# Determine mark_as_read dates for all feeds to ignore all stories before this date.
# max_feed_count = 0
feed_counts = {}
feed_last_reads = {}
for feed_id in feed_ids:
try:
usersub = UserSubscription.objects.get(feed__pk=feed_id, user=user)
except UserSubscription.DoesNotExist:
continue
if not usersub: continue
feed_counts[feed_id] = (usersub.unread_count_negative * 1 +
usersub.unread_count_neutral * 10 +
usersub.unread_count_positive * 20)
# if feed_counts[feed_id] > max_feed_count:
# max_feed_count = feed_counts[feed_id]
feed_last_reads[feed_id] = int(time.mktime(usersub.mark_read_date.timetuple()))
feed_counts = sorted(feed_counts.items(), key=itemgetter(1))[:50]
feed_ids = [f[0] for f in feed_counts]
feed_last_reads = dict([(str(feed_id), feed_last_reads[feed_id]) for feed_id in feed_ids
if feed_id in feed_last_reads])
feed_counts = dict(feed_counts)
# After excluding read stories, all that's left are stories
# past the mark_read_date. Everything returned is guaranteed to be unread.
mstories = MStory.objects(
story_guid__nin=read_stories,
story_feed_id__in=feed_ids,
# story_date__gte=start - bottom_delta
).map_reduce("""function() {
var d = feed_last_reads[this[~story_feed_id]];
if (this[~story_date].getTime()/1000 > d) {
emit(this[~id], this);
}
}""",
"""function(key, values) {
return values[0];
}""",
output='inline',
scope={
'feed_last_reads': feed_last_reads
}
)
mstories = [story.value for story in mstories if story and story.value]
mstories = sorted(mstories, cmp=lambda x, y: cmp(story_score(y, bottom_delta), story_score(x, bottom_delta)))
# story_feed_counts = defaultdict(int)
# mstories_pruned = []
# for story in mstories:
# print story['story_title'], story_feed_counts[story['story_feed_id']]
# if story_feed_counts[story['story_feed_id']] >= 3: continue
# mstories_pruned.append(story)
# story_feed_counts[story['story_feed_id']] += 1
stories = []
for i, story in enumerate(mstories):
if i < offset: continue
if i >= offset + limit: break
stories.append(bunch(story))
stories = Feed.format_stories(stories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
# Find starred stories
starred_stories = MStarredStory.objects(
user_id=user.pk,
story_feed_id__in=found_feed_ids
).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date)
for story in starred_stories])
# Intelligence classifiers for all feeds involved
def sort_by_feed(classifiers):
feed_classifiers = defaultdict(list)
for classifier in classifiers:
feed_classifiers[classifier.feed_id].append(classifier)
return feed_classifiers
classifier_feeds = sort_by_feed(MClassifierFeed.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_authors = sort_by_feed(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_titles = sort_by_feed(MClassifierTitle.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_tags = sort_by_feed(MClassifierTag.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifiers = {}
for feed_id in found_feed_ids:
classifiers[feed_id] = get_classifiers_for_user(user, feed_id, classifier_feeds[feed_id],
classifier_authors[feed_id],
classifier_titles[feed_id],
classifier_tags[feed_id])
# Just need to format stories
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds[story['story_feed_id']], story['story_feed_id']),
'author': apply_classifier_authors(classifier_authors[story['story_feed_id']], story),
'tags': apply_classifier_tags(classifier_tags[story['story_feed_id']], story),
'title': apply_classifier_titles(classifier_titles[story['story_feed_id']], story),
}
diff = datetime.datetime.utcnow() - start
timediff = float("%s.%.2s" % (diff.seconds, (diff.microseconds / 1000)))
logging.user(request, "~FCLoading river stories: page %s - ~SB%s/%s "
"stories ~SN(%s/%s/%s feeds) ~FB(%s seconds)" %
(page, len(stories), len(mstories), len(found_feed_ids),
len(feed_ids), len(original_feed_ids), timediff))
if new_flag:
return dict(stories=stories, classifiers=classifiers)
else:
logging.user(request, "~BR~FCNo new flag on river")
return dict(stories=stories)
def api_shared_story(request):
user = request.user
body = request.body_json
after = body.get('after', None)
before = body.get('before', None)
limit = body.get('limit', 50)
fields = body.get('triggerFields')
blurblog_user = fields['blurblog_user']
entries = []
if isinstance(blurblog_user, int) or blurblog_user.isdigit():
social_user_ids = [int(blurblog_user)]
elif blurblog_user == "all":
socialsubs = MSocialSubscription.objects.filter(user_id=user.pk)
social_user_ids = [ss.subscription_user_id for ss in socialsubs]
mstories = MSharedStory.objects(
user_id__in=social_user_ids
).order_by('-shared_date')[:limit]
stories = Feed.format_stories(mstories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
share_user_ids = list(set([story['user_id'] for story in stories]))
users = dict([(u.pk, u.username)
for u in User.objects.filter(pk__in=share_user_ids).only('pk', 'username')])
feeds = dict([(f.pk, {
"title": f.feed_title,
"website": f.feed_link,
"address": f.feed_address,
}) for f in Feed.objects.filter(pk__in=found_feed_ids)])
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
# Merge with feed specific classifiers
classifier_feeds = classifier_feeds + list(MClassifierFeed.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_authors = classifier_authors + list(MClassifierAuthor.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_titles = classifier_titles + list(MClassifierTitle.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_tags = classifier_tags + list(MClassifierTag.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
for story in stories:
if before and int(story['shared_date'].strftime("%s")) > before: continue
if after and int(story['shared_date'].strftime("%s")) < after: continue
score = compute_story_score(story, classifier_titles=classifier_titles,
classifier_authors=classifier_authors,
classifier_tags=classifier_tags,
classifier_feeds=classifier_feeds)
if score < 0: continue
feed = feeds.get(story['story_feed_id'], None)
entries.append({
"StoryTitle": story['story_title'],
"StoryContent": story['story_content'],
"StoryURL": story['story_permalink'],
"StoryAuthor": story['story_authors'],
"PublishedAt": story['story_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"StoryScore": score,
"Comments": story['comments'],
"Username": users.get(story['user_id']),
"SharedAt": story['shared_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"Site": feed and feed['title'],
"SiteURL": feed and feed['website'],
"SiteRSS": feed and feed['address'],
"ifttt": {
"id": story['story_hash'],
"timestamp": int(story['shared_date'].strftime("%s"))
},
})
if after:
entries = sorted(entries, key=lambda s: s['ifttt']['timestamp'])
logging.user(request, "~FMChecking shared stories from ~SB~FCIFTTT~SN~FM: ~SB~FM%s~FM~SN - ~SB%s~SN stories" % (blurblog_user, len(entries)))
return {"data": entries}
starred_stories = MStarredStory.objects(
user_id=user.pk,
story_feed_id__in=found_feed_ids
).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date)
for story in starred_stories])
# Intelligence classifiers for all feeds involved
def sort_by_feed(classifiers):
feed_classifiers = defaultdict(list)
for classifier in classifiers:
feed_classifiers[classifier.feed_id].append(classifier)
return feed_classifiers
classifier_feeds = sort_by_feed(MClassifierFeed.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_authors = sort_by_feed(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_titles = sort_by_feed(MClassifierTitle.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_tags = sort_by_feed(MClassifierTag.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifiers = {}
for feed_id in found_feed_ids:
classifiers[feed_id] = get_classifiers_for_user(user, feed_id, classifier_feeds[feed_id],
classifier_authors[feed_id],
classifier_titles[feed_id],
classifier_tags[feed_id])
# Just need to format stories
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
def calculate_feed_scores(self, silent=False, stories=None, force=False):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF and not force:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
# if not self.feed.fetched_once:
# if not silent:
# logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
# self.needs_unread_recalc = False
# self.save()
# return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
if not stories:
stories = cache.get('S:%s' % self.feed_id)
unread_story_hashes = self.get_stories(read_filter='unread', limit=500, hashes_only=True)
if not stories:
stories_db = MStory.objects(story_hash__in=unread_story_hashes)
stories = Feed.format_stories(stories_db, self.feed_id)
oldest_unread_story_date = now
unread_stories = []
for story in stories:
if story['story_date'] < date_delta:
continue
if story['story_hash'] in unread_story_hashes:
unread_stories.append(story)
if story['story_date'] < oldest_unread_story_date:
oldest_unread_story_date = story['story_date']
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id, social_user_id=0))
classifier_authors = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in unread_stories:
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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.user(self.user, '~FC~SNComputing scores: %s (~SB%s~SN/~SB%s~SN/~SB%s~SN)' % (self.feed, feed_scores['negative'], feed_scores['neutral'], feed_scores['positive']))
return self
def load_single_feed(request, feed_id):
start = time.time()
user = get_user(request)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 12))
page = int(request.REQUEST.get('page', 1))
dupe_feed_id = None
userstories_db = None
if page: offset = limit * (page-1)
if not feed_id: raise Http404
try:
feed = Feed.objects.get(id=feed_id)
except Feed.DoesNotExist:
feed_address = request.REQUEST.get('feed_address')
dupe_feed = DuplicateFeed.objects.filter(duplicate_address=feed_address)
if dupe_feed:
feed = dupe_feed[0].feed
dupe_feed_id = feed_id
else:
raise Http404
stories = feed.get_stories(offset, limit)
# Get intelligence classifier for user
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk, feed_id=feed_id))
checkpoint1 = time.time()
usersub = UserSubscription.objects.get(user=user, feed=feed)
userstories = []
if usersub:
userstories_db = MUserStory.objects(user_id=user.pk,
feed_id=feed.pk,
read_date__gte=usersub.mark_read_date)
starred_stories = MStarredStory.objects(user_id=user.pk, story_feed_id=feed_id).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
for us in userstories_db:
if hasattr(us.story, 'story_guid') and isinstance(us.story.story_guid, unicode):
userstories.append(us.story.story_guid)
elif hasattr(us.story, 'id') and isinstance(us.story.id, unicode):
userstories.append(us.story.id) # TODO: Remove me after migration from story.id->guid
checkpoint2 = time.time()
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
if usersub:
if story['id'] in userstories:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] < usersub.mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] > usersub.last_read_date:
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
else:
story['read_status'] = 1
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, feed),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
checkpoint3 = time.time()
# Intelligence
feed_tags = json.decode(feed.data.popular_tags) if feed.data.popular_tags else []
feed_authors = json.decode(feed.data.popular_authors) if feed.data.popular_authors else []
classifiers = get_classifiers_for_user(user, feed_id, classifier_feeds,
classifier_authors, classifier_titles, classifier_tags)
if usersub:
usersub.feed_opens += 1
usersub.save()
timediff = time.time()-start
last_update = relative_timesince(feed.last_update)
logging.user(request.user, "~FYLoading feed: ~SB%s%s ~SN(%.4s seconds)" % (
feed, ('~SN/p%s' % page) if page > 1 else '', timediff))
FeedLoadtime.objects.create(feed=feed, loadtime=timediff)
if timediff >= 1:
diff1 = checkpoint1-start
diff2 = checkpoint2-start
diff3 = checkpoint3-start
logging.user(request.user, "~FYSlow feed load: ~SB%.4s/%.4s(%s)/%.4s" % (
diff1, diff2, userstories_db and userstories_db.count(), diff3))
data = dict(stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers,
last_update=last_update,
feed_id=feed.pk)
if dupe_feed_id: data['dupe_feed_id'] = dupe_feed_id
if not usersub:
data.update(feed.canonical())
return data
def api_unread_story(request, trigger_slug=None):
user = request.user
body = request.body_json
after = body.get('after', None)
before = body.get('before', None)
limit = body.get('limit', 50)
fields = body.get('triggerFields')
feed_or_folder = fields['feed_or_folder']
entries = []
if isinstance(feed_or_folder, int) or feed_or_folder.isdigit():
feed_id = int(feed_or_folder)
try:
usersub = UserSubscription.objects.get(user=user, feed_id=feed_id)
except UserSubscription.DoesNotExist:
return dict(data=[])
found_feed_ids = [feed_id]
found_trained_feed_ids = [feed_id] if usersub.is_trained else []
stories = usersub.get_stories(order="newest", read_filter="unread",
offset=0, limit=limit,
default_cutoff_date=user.profile.unread_cutoff)
else:
folder_title = feed_or_folder
if folder_title == "Top Level":
folder_title = " "
usf = UserSubscriptionFolders.objects.get(user=user)
flat_folders = usf.flatten_folders()
feed_ids = None
if folder_title != "all":
feed_ids = flat_folders.get(folder_title)
usersubs = UserSubscription.subs_for_feeds(user.pk, feed_ids=feed_ids,
read_filter="unread")
feed_ids = [sub.feed_id for sub in usersubs]
params = {
"user_id": user.pk,
"feed_ids": feed_ids,
"offset": 0,
"limit": limit,
"order": "newest",
"read_filter": "unread",
"usersubs": usersubs,
"cutoff_date": user.profile.unread_cutoff,
}
story_hashes, unread_feed_story_hashes = UserSubscription.feed_stories(**params)
mstories = MStory.objects(story_hash__in=story_hashes).order_by('-story_date')
stories = Feed.format_stories(mstories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
trained_feed_ids = [sub.feed_id for sub in usersubs if sub.is_trained]
found_trained_feed_ids = list(set(trained_feed_ids) & set(found_feed_ids))
if found_trained_feed_ids:
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
feeds = dict([(f.pk, {
"title": f.feed_title,
"website": f.feed_link,
"address": f.feed_address,
}) for f in Feed.objects.filter(pk__in=found_feed_ids)])
for story in stories:
if before and int(story['story_date'].strftime("%s")) > before: continue
if after and int(story['story_date'].strftime("%s")) < after: continue
score = 0
if found_trained_feed_ids and story['story_feed_id'] in found_trained_feed_ids:
score = compute_story_score(story, classifier_titles=classifier_titles,
classifier_authors=classifier_authors,
classifier_tags=classifier_tags,
classifier_feeds=classifier_feeds)
if score < 0: continue
if trigger_slug == "new-unread-focus-story" and score < 1: continue
feed = feeds.get(story['story_feed_id'], None)
entries.append({
"StoryTitle": story['story_title'],
"StoryContent": story['story_content'],
"StoryURL": story['story_permalink'],
"StoryAuthor": story['story_authors'],
"PublishedAt": story['story_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"StoryScore": score,
"Site": feed and feed['title'],
"SiteURL": feed and feed['website'],
"SiteRSS": feed and feed['address'],
"meta": {
"id": story['story_hash'],
"timestamp": int(story['story_date'].strftime("%s"))
},
})
if after:
entries = sorted(entries, key=lambda s: s['meta']['timestamp'])
logging.user(request, "~FYChecking unread%s stories with ~SB~FCIFTTT~SN~FY: ~SB%s~SN - ~SB%s~SN stories" % (" ~SBfocus~SN" if trigger_slug == "new-unread-focus-story" else "", feed_or_folder, len(entries)))
return {"data": entries[:limit]}
for story in starred_stories])
except OperationFailure:
logging.info(" ***> Starred stories failure")
starred_stories = {}
# Intelligence classifiers for all feeds involved
def sort_by_feed(classifiers):
feed_classifiers = defaultdict(list)
for classifier in classifiers:
feed_classifiers[classifier.feed_id].append(classifier)
return feed_classifiers
classifiers = {}
try:
classifier_feeds = sort_by_feed(MClassifierFeed.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_authors = sort_by_feed(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_titles = sort_by_feed(MClassifierTitle.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_tags = sort_by_feed(MClassifierTag.objects(user_id=user.pk, feed_id__in=found_feed_ids))
except OperationFailure:
logging.info(" ***> Classifiers failure")
else:
for feed_id in found_feed_ids:
classifiers[feed_id] = get_classifiers_for_user(user, feed_id, classifier_feeds[feed_id],
classifier_authors[feed_id],
classifier_titles[feed_id],
classifier_tags[feed_id])
# Just need to format stories
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
def calculate_feed_scores(self, silent=False, stories_db=None):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(' ---> [%s] NOT Computing scores: %s' %
(self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(user_id=self.user_id,
feed_id=self.feed_id,
read_date__gte=self.mark_read_date)
# if not silent:
# logging.info(' ---> [%s] Read stories: %s' % (self.user, datetime.datetime.now() - now))
read_stories_ids = [us.story_id for us in read_stories]
stories_db = stories_db or MStory.objects(story_feed_id=self.feed_id,
story_date__gte=date_delta)
# if not silent:
# logging.info(' ---> [%s] MStory: %s' % (self.user, datetime.datetime.now() - now))
oldest_unread_story_date = now
unread_stories_db = []
for story in stories_db:
if story.story_date < date_delta:
continue
if hasattr(
story,
'story_guid') and story.story_guid not in read_stories_ids:
unread_stories_db.append(story)
if story.story_date < oldest_unread_story_date:
oldest_unread_story_date = story.story_date
stories = Feed.format_stories(unread_stories_db, self.feed_id)
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(
MClassifierFeed.objects(user_id=self.user_id,
feed_id=self.feed_id,
social_user_id=0))
classifier_authors = list(
MClassifierAuthor.objects(user_id=self.user_id,
feed_id=self.feed_id))
classifier_titles = list(
MClassifierTitle.objects(user_id=self.user_id,
feed_id=self.feed_id))
classifier_tags = list(
MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in stories:
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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
and self.unread_count_negative == 0):
self.mark_feed_read()
if not silent:
logging.info(' ---> [%s] Computing scores: %s (%s/%s/%s)' %
(self.user, self.feed, feed_scores['negative'],
feed_scores['neutral'], feed_scores['positive']))
return self
def load_river_blurblog(request):
limit = 10
start = time.time()
user = get_user(request)
social_user_ids = [int(uid) for uid in request.REQUEST.getlist("social_user_ids") if uid]
original_user_ids = list(social_user_ids)
page = int(request.REQUEST.get("page", 1))
order = request.REQUEST.get("order", "newest")
read_filter = request.REQUEST.get("read_filter", "unread")
relative_user_id = request.REQUEST.get("relative_user_id", None)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
UNREAD_CUTOFF = datetime.datetime.utcnow() - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if not relative_user_id:
relative_user_id = get_user(request).pk
if not social_user_ids:
socialsubs = MSocialSubscription.objects.filter(user_id=user.pk)
social_user_ids = [s.subscription_user_id for s in socialsubs]
offset = (page - 1) * limit
limit = page * limit - 1
story_ids, story_dates = MSocialSubscription.feed_stories(
user.pk, social_user_ids, offset=offset, limit=limit, order=order, read_filter=read_filter
)
mstories = MStory.objects(id__in=story_ids)
story_id_to_dates = dict(zip(story_ids, story_dates))
def sort_stories_by_id(a, b):
return int(story_id_to_dates[str(b.id)]) - int(story_id_to_dates[str(a.id)])
sorted_mstories = sorted(mstories, cmp=sort_stories_by_id)
stories = Feed.format_stories(sorted_mstories)
for s, story in enumerate(stories):
story["story_date"] = datetime.datetime.fromtimestamp(story_dates[s])
stories, user_profiles = MSharedStory.stories_with_comments_and_profiles(stories, relative_user_id, check_all=True)
story_feed_ids = list(set(s["story_feed_id"] for s in stories))
usersubs = UserSubscription.objects.filter(user__pk=user.pk, feed__pk__in=story_feed_ids)
usersubs_map = dict((sub.feed_id, sub) for sub in usersubs)
unsub_feed_ids = list(set(story_feed_ids).difference(set(usersubs_map.keys())))
unsub_feeds = Feed.objects.filter(pk__in=unsub_feed_ids)
unsub_feeds = [feed.canonical(include_favicon=False) for feed in unsub_feeds]
# Find starred stories
if story_feed_ids:
story_ids = [story["id"] for story in stories]
starred_stories = MStarredStory.objects(user_id=user.pk, story_guid__in=story_ids).only(
"story_guid", "starred_date"
)
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
shared_stories = MSharedStory.objects(user_id=user.pk, story_guid__in=story_ids).only(
"story_guid", "shared_date", "comments"
)
shared_stories = dict(
[
(story.story_guid, dict(shared_date=story.shared_date, comments=story.comments))
for story in shared_stories
]
)
userstories_db = MUserStory.objects(user_id=user.pk, feed_id__in=story_feed_ids, story_id__in=story_ids).only(
"story_id"
)
userstories = set(us.story_id for us in userstories_db)
else:
starred_stories = {}
shared_stories = {}
userstories = []
# Intelligence classifiers for all feeds involved
if story_feed_ids:
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk, feed_id__in=story_feed_ids))
else:
classifier_feeds = []
classifier_authors = []
classifier_titles = []
classifier_tags = []
classifiers = sort_classifiers_by_feed(
user=user,
feed_ids=story_feed_ids,
classifier_feeds=classifier_feeds,
classifier_authors=classifier_authors,
classifier_titles=classifier_titles,
classifier_tags=classifier_tags,
)
# Just need to format stories
for story in stories:
if story["id"] in userstories:
story["read_status"] = 1
elif story["story_date"] < UNREAD_CUTOFF:
story["read_status"] = 1
else:
story["read_status"] = 0
story_date = localtime_for_timezone(story["story_date"], user.profile.timezone)
story["short_parsed_date"] = format_story_link_date__short(story_date, now)
story["long_parsed_date"] = format_story_link_date__long(story_date, now)
if story["id"] in starred_stories:
story["starred"] = True
starred_date = localtime_for_timezone(starred_stories[story["id"]], user.profile.timezone)
story["starred_date"] = format_story_link_date__long(starred_date, now)
story["intelligence"] = {
"feed": apply_classifier_feeds(classifier_feeds, story["story_feed_id"]),
"author": apply_classifier_authors(classifier_authors, story),
"tags": apply_classifier_tags(classifier_tags, story),
"title": apply_classifier_titles(classifier_titles, story),
}
if story["id"] in shared_stories:
story["shared"] = True
shared_date = localtime_for_timezone(shared_stories[story["id"]]["shared_date"], user.profile.timezone)
story["shared_date"] = format_story_link_date__long(shared_date, now)
story["shared_comments"] = strip_tags(shared_stories[story["id"]]["comments"])
diff = time.time() - start
timediff = round(float(diff), 2)
logging.user(
request,
"~FYLoading ~FCriver blurblogs stories~FY: ~SBp%s~SN (%s/%s "
"stories, ~SN%s/%s/%s feeds)"
% (page, len(stories), len(mstories), len(story_feed_ids), len(social_user_ids), len(original_user_ids)),
)
return {
"stories": stories,
"user_profiles": user_profiles,
"feeds": unsub_feeds,
"classifiers": classifiers,
"elapsed_time": timediff,
}
def load_single_feed(request, feed_id):
start = datetime.datetime.utcnow()
user = get_user(request)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 12))
page = int(request.REQUEST.get('page', 1))
if page:
offset = limit * (page-1)
dupe_feed_id = None
if not feed_id:
raise Http404
try:
feed = Feed.objects.get(id=feed_id)
except Feed.DoesNotExist:
feed_address = request.REQUEST.get('feed_address')
dupe_feed = DuplicateFeed.objects.filter(duplicate_address=feed_address)
if dupe_feed:
feed = dupe_feed[0].feed
dupe_feed_id = feed_id
else:
raise Http404
stories = feed.get_stories(offset, limit)
# Get intelligence classifier for user
classifier_feeds = MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id)
classifier_authors = MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id)
classifier_titles = MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id)
classifier_tags = MClassifierTag.objects(user_id=user.pk, feed_id=feed_id)
usersub = UserSubscription.objects.get(user=user, feed=feed)
userstories = []
if usersub:
userstories_db = MUserStory.objects(user_id=user.pk,
feed_id=feed.pk,
read_date__gte=usersub.mark_read_date)
starred_stories = MStarredStory.objects(user_id=user.pk, story_feed_id=feed_id).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
for us in userstories_db:
if hasattr(us.story, 'story_guid') and isinstance(us.story.story_guid, unicode):
userstories.append(us.story.story_guid)
elif hasattr(us.story, 'id') and isinstance(us.story.id, unicode):
userstories.append(us.story.id) # TODO: Remove me after migration from story.id->guid
for story in stories:
[x.rewind() for x in [classifier_feeds, classifier_authors, classifier_tags, classifier_titles]]
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
if usersub:
if story['id'] in userstories:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] < usersub.mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] > usersub.last_read_date:
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
else:
story['read_status'] = 1
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, feed),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
# Intelligence
feed_tags = json.decode(feed.data.popular_tags) if feed.data.popular_tags else []
feed_authors = json.decode(feed.data.popular_authors) if feed.data.popular_authors else []
classifiers = get_classifiers_for_user(user, feed_id, classifier_feeds,
classifier_authors, classifier_titles, classifier_tags)
if usersub:
usersub.feed_opens += 1
usersub.save()
diff = datetime.datetime.utcnow()-start
timediff = float("%s.%.2s" % (diff.seconds, (diff.microseconds / 1000)))
last_update = relative_timesince(feed.last_update)
logging.user(request.user, "~FYLoading feed: ~SB%s%s ~SN(%s seconds)" % (
feed, ('~SN/p%s' % page) if page > 1 else '', timediff))
FeedLoadtime.objects.create(feed=feed, loadtime=timediff)
data = dict(stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers,
last_update=last_update,
feed_id=feed.pk)
if dupe_feed_id: data['dupe_feed_id'] = dupe_feed_id
if not usersub:
data.update(feed.canonical())
return data
def calculate_feed_scores(self, silent=False, stories=None):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(" ---> [%s] NOT Computing scores: %s" % (self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(
user_id=self.user_id, feed_id=self.feed_id, read_date__gte=self.mark_read_date
)
read_stories_ids = [us.story_id for us in read_stories]
if not stories:
stories_db = MStory.objects(story_feed_id=self.feed_id, story_date__gte=date_delta)
stories = Feed.format_stories(stories_db, self.feed_id)
oldest_unread_story_date = now
unread_stories = []
for story in stories:
if story["story_date"] < date_delta:
continue
if story["id"] not in read_stories_ids:
unread_stories.append(story)
if story["story_date"] < oldest_unread_story_date:
oldest_unread_story_date = story["story_date"]
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id, social_user_id=0))
classifier_authors = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {"feed": apply_classifier_feeds(classifier_feeds, self.feed)}
for story in unread_stories:
scores.update(
{
"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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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 and self.unread_count_negative == 0:
self.mark_feed_read()
if not silent:
logging.info(
" ---> [%s] Computing scores: %s (%s/%s/%s)"
% (self.user, self.feed, feed_scores["negative"], feed_scores["neutral"], feed_scores["positive"])
)
return self
def load_single_feed(request):
user = get_user(request)
offset = int(request.REQUEST.get("offset", 0))
limit = int(request.REQUEST.get("limit", 30))
page = int(request.REQUEST.get("page", 0))
if page:
offset = limit * page
feed_id = int(request.REQUEST.get("feed_id", 0))
if feed_id == 0:
raise Http404
try:
feed = Feed.objects.get(id=feed_id)
except Feed.DoesNotExist:
feed_address = request.REQUEST.get("feed_address")
dupe_feed = DuplicateFeed.objects.filter(duplicate_address=feed_address)
if dupe_feed:
feed = dupe_feed[0].feed
else:
raise Http404
force_update = request.GET.get("force_update", False)
now = datetime.datetime.utcnow()
stories = feed.get_stories(offset, limit)
if force_update:
feed.update(force_update)
# Get intelligence classifier for user
classifier_feeds = MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id)
classifier_authors = MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id)
classifier_titles = MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id)
classifier_tags = MClassifierTag.objects(user_id=user.pk, feed_id=feed_id)
usersub = UserSubscription.objects.get(user=user, feed=feed)
userstories = []
userstories_db = MUserStory.objects(user_id=user.pk, feed_id=feed.pk, read_date__gte=usersub.mark_read_date)
starred_stories = MStarredStory.objects(user_id=user.pk, story_feed_id=feed_id).only("story_guid", "starred_date")
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
for us in userstories_db:
if hasattr(us.story, "story_guid") and isinstance(us.story.story_guid, unicode):
userstories.append(us.story.story_guid)
elif hasattr(us.story, "id") and isinstance(us.story.id, unicode):
userstories.append(us.story.id) # TODO: Remove me after migration from story.id->guid
for story in stories:
classifier_feeds.rewind()
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
story_date = localtime_for_timezone(story["story_date"], user.profile.timezone)
story["short_parsed_date"] = format_story_link_date__short(story_date)
story["long_parsed_date"] = format_story_link_date__long(story_date)
if story["id"] in userstories:
story["read_status"] = 1
elif not story.get("read_status") and story["story_date"] < usersub.mark_read_date:
story["read_status"] = 1
elif not story.get("read_status") and story["story_date"] > usersub.last_read_date:
story["read_status"] = 0
if story["id"] in starred_stories:
story["starred"] = True
starred_date = localtime_for_timezone(starred_stories[story["id"]], user.profile.timezone)
story["starred_date"] = format_story_link_date__long(starred_date)
story["intelligence"] = {
"feed": apply_classifier_feeds(classifier_feeds, feed),
"author": apply_classifier_authors(classifier_authors, story),
"tags": apply_classifier_tags(classifier_tags, story),
"title": apply_classifier_titles(classifier_titles, story),
}
# Intelligence
feed_tags = json.decode(feed.popular_tags) if feed.popular_tags else []
feed_authors = json.decode(feed.popular_authors) if feed.popular_authors else []
classifiers = get_classifiers_for_user(
user, feed_id, classifier_feeds, classifier_authors, classifier_titles, classifier_tags
)
usersub.feed_opens += 1
usersub.save()
diff = datetime.datetime.utcnow() - now
timediff = float("%s.%s" % (diff.seconds, (diff.microseconds / 1000)))
last_update = relative_timesince(feed.last_update)
logging.info(" ---> [%s] ~FYLoading feed: ~SB%s ~SN(%s seconds)" % (request.user, feed, timediff))
FeedLoadtime.objects.create(feed=feed, loadtime=timediff)
data = dict(
stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers,
last_update=last_update,
feed_id=feed.pk,
)
return data
def calculate_feed_scores(self, silent=False, stories=None, force=False):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF and not force:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
# if not self.feed.fetched_once:
# if not silent:
# logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
# self.needs_unread_recalc = False
# self.save()
# return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
if not stories:
stories = cache.get('S:%s' % self.feed_id)
unread_story_hashes = self.get_stories(read_filter='unread',
limit=500,
hashes_only=True)
if not stories:
stories_db = MStory.objects(story_hash__in=unread_story_hashes)
stories = Feed.format_stories(stories_db, self.feed_id)
oldest_unread_story_date = now
unread_stories = []
for story in stories:
if story['story_date'] < date_delta:
continue
if story['story_hash'] in unread_story_hashes:
unread_stories.append(story)
if story['story_date'] < oldest_unread_story_date:
oldest_unread_story_date = story['story_date']
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(
MClassifierFeed.objects(user_id=self.user_id,
feed_id=self.feed_id,
social_user_id=0))
classifier_authors = list(
MClassifierAuthor.objects(user_id=self.user_id,
feed_id=self.feed_id))
classifier_titles = list(
MClassifierTitle.objects(user_id=self.user_id,
feed_id=self.feed_id))
classifier_tags = list(
MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in unread_stories:
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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.user(
self.user,
'~FC~SNComputing scores: %s (~SB%s~SN/~SB%s~SN/~SB%s~SN)' %
(self.feed, feed_scores['negative'], feed_scores['neutral'],
feed_scores['positive']))
return self
def calculate_feed_scores(self, silent=False, stories=None, force=False):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
oldest_unread_story_date = now
if self.user.profile.last_seen_on < UNREAD_CUTOFF and not force:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return self
ong = self.unread_count_negative
ont = self.unread_count_neutral
ops = self.unread_count_positive
# if not self.feed.fetched_once:
# if not silent:
# logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
# self.needs_unread_recalc = False
# self.save()
# return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
if self.is_trained:
if not stories:
stories = cache.get("S:%s" % self.feed_id)
unread_story_hashes = self.get_stories(read_filter="unread", limit=500, hashes_only=True)
if not stories:
stories_db = MStory.objects(story_hash__in=unread_story_hashes)
stories = Feed.format_stories(stories_db, self.feed_id)
unread_stories = []
for story in stories:
if story["story_date"] < date_delta:
continue
if story["story_hash"] in unread_story_hashes:
unread_stories.append(story)
if story["story_date"] < oldest_unread_story_date:
oldest_unread_story_date = story["story_date"]
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(
MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id, social_user_id=0)
)
classifier_authors = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
if (
not len(classifier_feeds)
and not len(classifier_authors)
and not len(classifier_titles)
and not len(classifier_tags)
):
self.is_trained = False
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {"feed": apply_classifier_feeds(classifier_feeds, self.feed)}
for story in unread_stories:
scores.update(
{
"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
else:
unread_story_hashes = self.get_stories(read_filter="unread", limit=500, hashes_only=True, withscores=True)
feed_scores["neutral"] = len(unread_story_hashes)
if feed_scores["neutral"]:
oldest_unread_story_date = datetime.datetime.fromtimestamp(unread_story_hashes[-1][1])
if not silent:
logging.user(
self.user,
"~FBUnread count (~SB%s~SN%s): ~SN(~FC%s~FB/~FC%s~FB/~FC%s~FB) ~SBto~SN (~FC%s~FB/~FC%s~FB/~FC%s~FB)"
% (
self.feed_id,
"/~FMtrained~FB" if self.is_trained else "",
ong,
ont,
ops,
feed_scores["negative"],
feed_scores["neutral"],
feed_scores["positive"],
),
)
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.user(
self.user,
"~FC~SNComputing scores: %s (~SB%s~SN/~SB%s~SN/~SB%s~SN)"
% (self.feed, feed_scores["negative"], feed_scores["neutral"], feed_scores["positive"]),
)
return self
def api_shared_story(request):
user = request.user
body = request.body_json
after = body.get('after', None)
before = body.get('before', None)
limit = body.get('limit', 50)
fields = body.get('triggerFields')
blurblog_user = fields['blurblog_user']
entries = []
if isinstance(blurblog_user, int) or blurblog_user.isdigit():
social_user_ids = [int(blurblog_user)]
elif blurblog_user == "all":
socialsubs = MSocialSubscription.objects.filter(user_id=user.pk)
social_user_ids = [ss.subscription_user_id for ss in socialsubs]
mstories = MSharedStory.objects(
user_id__in=social_user_ids
).order_by('-shared_date')[:limit]
stories = Feed.format_stories(mstories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
share_user_ids = list(set([story['user_id'] for story in stories]))
users = dict([(u.pk, u.username)
for u in User.objects.filter(pk__in=share_user_ids).only('pk', 'username')])
feeds = dict([(f.pk, {
"title": f.feed_title,
"website": f.feed_link,
"address": f.feed_address,
}) for f in Feed.objects.filter(pk__in=found_feed_ids)])
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk,
social_user_id__in=social_user_ids))
# Merge with feed specific classifiers
classifier_feeds = classifier_feeds + list(MClassifierFeed.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_authors = classifier_authors + list(MClassifierAuthor.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_titles = classifier_titles + list(MClassifierTitle.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
classifier_tags = classifier_tags + list(MClassifierTag.objects(user_id=user.pk,
feed_id__in=found_feed_ids))
for story in stories:
if before and int(story['shared_date'].strftime("%s")) > before: continue
if after and int(story['shared_date'].strftime("%s")) < after: continue
score = compute_story_score(story, classifier_titles=classifier_titles,
classifier_authors=classifier_authors,
classifier_tags=classifier_tags,
classifier_feeds=classifier_feeds)
if score < 0: continue
feed = feeds.get(story['story_feed_id'], None)
entries.append({
"StoryTitle": story['story_title'],
"StoryContent": story['story_content'],
"StoryURL": story['story_permalink'],
"StoryAuthor": story['story_authors'],
"PublishedAt": story['story_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"StoryScore": score,
"Comments": story['comments'],
"Username": users.get(story['user_id']),
"SharedAt": story['shared_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"Site": feed and feed['title'],
"SiteURL": feed and feed['website'],
"SiteRSS": feed and feed['address'],
"meta": {
"id": story['story_hash'],
"timestamp": int(story['shared_date'].strftime("%s"))
},
})
if after:
entries = sorted(entries, key=lambda s: s['meta']['timestamp'])
logging.user(request, "~FMChecking shared stories from ~SB~FCIFTTT~SN~FM: ~SB~FM%s~FM~SN - ~SB%s~SN stories" % (blurblog_user, len(entries)))
return {"data": entries}
def load_single_feed(request):
user = get_user(request)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 30))
page = int(request.REQUEST.get('page', 0))
if page:
offset = limit * page
feed_id = int(request.REQUEST['feed_id'])
feed = Feed.objects.get(id=feed_id)
force_update = request.GET.get('force_update', False)
now = datetime.datetime.now()
stories = feed.get_stories(offset, limit)
if force_update:
feed.update(force_update)
# Get intelligence classifier for user
classifier_feeds = MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id)
classifier_authors = MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id)
classifier_titles = MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id)
classifier_tags = MClassifierTag.objects(user_id=user.pk, feed_id=feed_id)
try:
usersub = UserSubscription.objects.get(user=user, feed=feed)
except UserSubscription.DoesNotExist:
# FIXME: Why is this happening for `conesus` when logged into another account?!
logging.info(" ***> [%s] UserSub DNE, creating: %s" % (user, feed))
usersub = UserSubscription.objects.create(user=user, feed=feed)
userstories = MUserStory.objects(user_id=user.pk,
feed_id=feed.pk,
read_date__gte=usersub.mark_read_date)
userstories = [us.story.id for us in userstories]
for story in stories:
classifier_feeds.rewind()
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
if story.get('id') in userstories:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] < usersub.mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] > usersub.last_read_date:
story['read_status'] = 0
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, feed),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
# Intelligence
feed_tags = json.decode(feed.popular_tags) if feed.popular_tags else []
feed_authors = json.decode(feed.popular_authors) if feed.popular_authors else []
classifiers = get_classifiers_for_user(user, feed_id, classifier_feeds,
classifier_authors, classifier_titles, classifier_tags)
usersub.feed_opens += 1
usersub.save()
diff = datetime.datetime.now()-now
logging.info(" ---> [%s] Loading feed: %s (%s.%s seconds)" % (request.user, feed,
diff.seconds,
diff.microseconds / 1000))
data = dict(stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers)
return data
def load_river_stories(request):
limit = 18
offset = 0
start = datetime.datetime.utcnow()
user = get_user(request)
feed_ids = [int(feed_id) for feed_id in request.REQUEST.getlist('feeds') if feed_id]
original_feed_ids = list(feed_ids)
page = int(request.REQUEST.get('page', 0))+1
read_stories_count = int(request.REQUEST.get('read_stories_count', 0))
bottom_delta = datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if not feed_ids:
logging.user(request.user, "~FCLoading empty river stories: page %s" % (page))
return dict(stories=[])
# Fetch all stories at and before the page number.
# Not a single page, because reading stories can move them up in the unread order.
# `read_stories_count` is an optimization, works best when all 25 stories before have been read.
limit = limit * page - read_stories_count
# Read stories to exclude
read_stories = MUserStory.objects(user_id=user.pk, feed_id__in=feed_ids).only('story')
read_stories = [rs.story.id for rs in read_stories]
# Determine mark_as_read dates for all feeds to ignore all stories before this date.
# max_feed_count = 0
feed_counts = {}
feed_last_reads = {}
for feed_id in feed_ids:
try:
usersub = UserSubscription.objects.get(feed__pk=feed_id, user=user)
except UserSubscription.DoesNotExist:
continue
if not usersub: continue
feed_counts[feed_id] = (usersub.unread_count_negative * 1 +
usersub.unread_count_neutral * 10 +
usersub.unread_count_positive * 20)
# if feed_counts[feed_id] > max_feed_count:
# max_feed_count = feed_counts[feed_id]
feed_last_reads[feed_id] = int(time.mktime(usersub.mark_read_date.timetuple()))
feed_counts = sorted(feed_counts.items(), key=itemgetter(1))[:50]
feed_ids = [f[0] for f in feed_counts]
feed_last_reads = dict([(str(feed_id), feed_last_reads[feed_id]) for feed_id in feed_ids])
feed_counts = dict(feed_counts)
# After excluding read stories, all that's left are stories
# past the mark_read_date. Everything returned is guaranteed to be unread.
mstories = MStory.objects(
id__nin=read_stories,
story_feed_id__in=feed_ids,
story_date__gte=start - bottom_delta
).map_reduce("""function() {
var d = feed_last_reads[this[~story_feed_id]];
if (this[~story_date].getTime()/1000 > d) {
emit(this[~id], this);
}
}""",
"""function(key, values) {
return values[0];
}""",
output='inline',
scope={
'feed_last_reads': feed_last_reads
}
)
mstories = [story.value for story in mstories]
mstories = sorted(mstories, cmp=lambda x, y: cmp(story_score(y, bottom_delta), story_score(x, bottom_delta)))
# story_feed_counts = defaultdict(int)
# mstories_pruned = []
# for story in mstories:
# print story['story_title'], story_feed_counts[story['story_feed_id']]
# if story_feed_counts[story['story_feed_id']] >= 3: continue
# mstories_pruned.append(story)
# story_feed_counts[story['story_feed_id']] += 1
stories = []
for i, story in enumerate(mstories):
if i < offset: continue
if i >= offset + limit: break
stories.append(bunch(story))
stories = Feed.format_stories(stories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
# Find starred stories
starred_stories = MStarredStory.objects(
user_id=user.pk,
story_feed_id__in=found_feed_ids
).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date)
for story in starred_stories])
# Intelligence classifiers for all feeds involved
def sort_by_feed(classifiers):
feed_classifiers = defaultdict(list)
for classifier in classifiers:
feed_classifiers[classifier.feed_id].append(classifier)
return feed_classifiers
classifier_feeds = sort_by_feed(MClassifierFeed.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_authors = sort_by_feed(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_titles = sort_by_feed(MClassifierTitle.objects(user_id=user.pk, feed_id__in=found_feed_ids))
classifier_tags = sort_by_feed(MClassifierTag.objects(user_id=user.pk, feed_id__in=found_feed_ids))
# Just need to format stories
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds[story['story_feed_id']], story['story_feed_id']),
'author': apply_classifier_authors(classifier_authors[story['story_feed_id']], story),
'tags': apply_classifier_tags(classifier_tags[story['story_feed_id']], story),
'title': apply_classifier_titles(classifier_titles[story['story_feed_id']], story),
}
diff = datetime.datetime.utcnow() - start
timediff = float("%s.%.2s" % (diff.seconds, (diff.microseconds / 1000)))
logging.user(request.user, "~FCLoading river stories: page %s - ~SB%s/%s "
"stories ~SN(%s/%s/%s feeds) ~FB(%s seconds)" %
(page, len(stories), len(mstories), len(found_feed_ids),
len(feed_ids), len(original_feed_ids), timediff))
return dict(stories=stories)
def calculate_feed_scores(self, silent=False, stories_db=None):
UNREAD_CUTOFF = datetime.datetime.utcnow() - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
if not silent:
logging.info(' ---> [%s] Computing scores: %s' % (self.user, self.feed))
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(user_id=self.user.pk,
feed_id=self.feed.pk,
read_date__gte=self.mark_read_date)
# if not silent:
# logging.info(' ---> [%s] Read stories: %s' % (self.user, datetime.datetime.now() - now))
read_stories_ids = []
for us in read_stories:
if hasattr(us.story, 'story_guid') and isinstance(us.story.story_guid, unicode):
read_stories_ids.append(us.story.story_guid)
elif hasattr(us.story, 'id') and isinstance(us.story.id, unicode):
read_stories_ids.append(us.story.id) # TODO: Remove me after migration from story.id->guid
stories_db = stories_db or MStory.objects(story_feed_id=self.feed.pk,
story_date__gte=date_delta)
# if not silent:
# logging.info(' ---> [%s] MStory: %s' % (self.user, datetime.datetime.now() - now))
unread_stories_db = []
for story in stories_db:
if story.story_date < date_delta:
continue
if hasattr(story, 'story_guid') and story.story_guid not in read_stories_ids:
unread_stories_db.append(story)
elif isinstance(story.id, unicode) and story.id not in read_stories_ids:
unread_stories_db.append(story)
stories = Feed.format_stories(unread_stories_db, self.feed.pk)
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = MClassifierFeed.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_authors = MClassifierAuthor.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_titles = MClassifierTitle.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_tags = MClassifierTag.objects(user_id=self.user.pk, feed_id=self.feed.pk)
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in stories:
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
self.unread_count_positive = feed_scores['positive']
self.unread_count_neutral = feed_scores['neutral']
self.unread_count_negative = feed_scores['negative']
self.needs_unread_recalc = False
self.save()
if (self.unread_count_positive == 0 and
self.unread_count_neutral == 0):
self.mark_feed_read()
cache.delete('usersub:%s' % self.user.id)
return
def calculate_feed_scores(self, silent=False, stories_db=None):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(user_id=self.user.pk,
feed_id=self.feed.pk,
read_date__gte=self.mark_read_date)
# if not silent:
# logging.info(' ---> [%s] Read stories: %s' % (self.user, datetime.datetime.now() - now))
read_stories_ids = []
for us in read_stories:
read_stories_ids.append(us.story_id)
stories_db = stories_db or MStory.objects(story_feed_id=self.feed.pk,
story_date__gte=date_delta)
# if not silent:
# logging.info(' ---> [%s] MStory: %s' % (self.user, datetime.datetime.now() - now))
oldest_unread_story_date = now
unread_stories_db = []
for story in stories_db:
if story.story_date < date_delta:
continue
if hasattr(story, 'story_guid') and story.story_guid not in read_stories_ids:
unread_stories_db.append(story)
if story.story_date < oldest_unread_story_date:
oldest_unread_story_date = story.story_date
stories = Feed.format_stories(unread_stories_db, self.feed.pk)
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(MClassifierFeed.objects(user_id=self.user.pk, feed_id=self.feed.pk))
classifier_authors = list(MClassifierAuthor.objects(user_id=self.user.pk, feed_id=self.feed.pk))
classifier_titles = list(MClassifierTitle.objects(user_id=self.user.pk, feed_id=self.feed.pk))
classifier_tags = list(MClassifierTag.objects(user_id=self.user.pk, feed_id=self.feed.pk))
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in stories:
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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()
cache.delete('usersub:%s' % self.user.id)
if not silent:
logging.info(' ---> [%s] Computing scores: %s (%s/%s/%s)' % (self.user, self.feed, feed_scores['negative'], feed_scores['neutral'], feed_scores['positive']))
return
def calculate_feed_scores(self, silent=False):
if self.user.profile.last_seen_on < MONTH_AGO:
if not silent:
logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 month+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
if not silent:
logging.info(' ---> [%s] Computing scores: %s' % (self.user, self.feed))
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 = datetime.datetime.now()-datetime.timedelta(days=DAYS_OF_UNREAD)
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(user_id=self.user.pk,
feed_id=self.feed.pk)
read_stories_ids = [rs.story.id for rs in read_stories]
from django.db import connection
connection.queries = []
stories_db = MStory.objects(story_feed_id=self.feed.pk,
story_date__gte=date_delta)
stories_db = [story for story in stories_db if story.id not in read_stories_ids]
stories = self.feed.format_stories(stories_db)
classifier_feeds = MClassifierFeed.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_authors = MClassifierAuthor.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_titles = MClassifierTitle.objects(user_id=self.user.pk, feed_id=self.feed.pk)
classifier_tags = MClassifierTag.objects(user_id=self.user.pk, feed_id=self.feed.pk)
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in stories:
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
scores.update({
'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['feed'], scores['author'], scores['tags'], scores['title'])
min_score = min(scores['feed'], scores['author'], scores['tags'], scores['title'])
if max_score > 0:
feed_scores['positive'] += 1
if min_score < 0:
feed_scores['negative'] += 1
if max_score == 0 and min_score == 0:
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.needs_unread_recalc = False
self.save()
if (self.unread_count_positive == 0 and
self.unread_count_neutral == 0):
self.mark_feed_read()
cache.delete('usersub:%s' % self.user.id)
return
def api_unread_story(request, trigger_slug=None):
user = request.user
body = request.body_json
after = body.get('after', None)
before = body.get('before', None)
limit = body.get('limit', 50)
fields = body.get('triggerFields')
feed_or_folder = fields['feed_or_folder']
entries = []
if isinstance(feed_or_folder, int) or feed_or_folder.isdigit():
feed_id = int(feed_or_folder)
usersub = UserSubscription.objects.get(user=user, feed_id=feed_id)
found_feed_ids = [feed_id]
found_trained_feed_ids = [feed_id] if usersub.is_trained else []
stories = usersub.get_stories(order="newest", read_filter="unread",
offset=0, limit=limit,
default_cutoff_date=user.profile.unread_cutoff)
else:
folder_title = feed_or_folder
if folder_title == "Top Level":
folder_title = " "
usf = UserSubscriptionFolders.objects.get(user=user)
flat_folders = usf.flatten_folders()
feed_ids = None
if folder_title != "all":
feed_ids = flat_folders.get(folder_title)
usersubs = UserSubscription.subs_for_feeds(user.pk, feed_ids=feed_ids,
read_filter="unread")
feed_ids = [sub.feed_id for sub in usersubs]
params = {
"user_id": user.pk,
"feed_ids": feed_ids,
"offset": 0,
"limit": limit,
"order": "newest",
"read_filter": "unread",
"usersubs": usersubs,
"cutoff_date": user.profile.unread_cutoff,
}
story_hashes, unread_feed_story_hashes = UserSubscription.feed_stories(**params)
mstories = MStory.objects(story_hash__in=story_hashes).order_by('-story_date')
stories = Feed.format_stories(mstories)
found_feed_ids = list(set([story['story_feed_id'] for story in stories]))
trained_feed_ids = [sub.feed_id for sub in usersubs if sub.is_trained]
found_trained_feed_ids = list(set(trained_feed_ids) & set(found_feed_ids))
if found_trained_feed_ids:
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk,
feed_id__in=found_trained_feed_ids))
feeds = dict([(f.pk, {
"title": f.feed_title,
"website": f.feed_link,
"address": f.feed_address,
}) for f in Feed.objects.filter(pk__in=found_feed_ids)])
for story in stories:
if before and int(story['story_date'].strftime("%s")) > before: continue
if after and int(story['story_date'].strftime("%s")) < after: continue
score = 0
if found_trained_feed_ids and story['story_feed_id'] in found_trained_feed_ids:
score = compute_story_score(story, classifier_titles=classifier_titles,
classifier_authors=classifier_authors,
classifier_tags=classifier_tags,
classifier_feeds=classifier_feeds)
if score < 0: continue
if trigger_slug == "new-unread-focus-story" and score < 1: continue
feed = feeds.get(story['story_feed_id'], None)
entries.append({
"StoryTitle": story['story_title'],
"StoryContent": story['story_content'],
"StoryURL": story['story_permalink'],
"StoryAuthor": story['story_authors'],
"PublishedAt": story['story_date'].strftime("%Y-%m-%dT%H:%M:%SZ"),
"StoryScore": score,
"Site": feed and feed['title'],
"SiteURL": feed and feed['website'],
"SiteRSS": feed and feed['address'],
"ifttt": {
"id": story['story_hash'],
"timestamp": int(story['story_date'].strftime("%s"))
},
})
if after:
entries = sorted(entries, key=lambda s: s['ifttt']['timestamp'])
logging.user(request, "~FYChecking unread%s stories with ~SB~FCIFTTT~SN~FY: ~SB%s~SN - ~SB%s~SN stories" % (" ~SBfocus~SN" if trigger_slug == "new-unread-focus-story" else "", feed_or_folder, len(entries)))
return {"data": entries[:limit]}
def load_social_stories(request, user_id, username=None):
start = time.time()
user = get_user(request)
social_user_id = int(user_id)
social_user = get_object_or_404(User, pk=social_user_id)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 6))
page = request.REQUEST.get('page')
order = request.REQUEST.get('order', 'newest')
read_filter = request.REQUEST.get('read_filter', 'all')
stories = []
if page: offset = limit * (int(page) - 1)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
UNREAD_CUTOFF = datetime.datetime.utcnow() - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
social_profile = MSocialProfile.get_user(social_user.pk)
try:
socialsub = MSocialSubscription.objects.get(user_id=user.pk, subscription_user_id=social_user_id)
except MSocialSubscription.DoesNotExist:
socialsub = None
mstories = MSharedStory.objects(user_id=social_user.pk).order_by('-shared_date')[offset:offset+limit]
stories = Feed.format_stories(mstories)
if socialsub and (read_filter == 'unread' or order == 'oldest'):
story_ids = socialsub.get_stories(order=order, read_filter=read_filter, offset=offset, limit=limit)
story_date_order = "%sshared_date" % ('' if order == 'oldest' else '-')
if story_ids:
mstories = MSharedStory.objects(user_id=social_user.pk,
story_db_id__in=story_ids).order_by(story_date_order)
stories = Feed.format_stories(mstories)
else:
mstories = MSharedStory.objects(user_id=social_user.pk).order_by('-shared_date')[offset:offset+limit]
stories = Feed.format_stories(mstories)
if not stories:
return dict(stories=[])
checkpoint1 = time.time()
stories, user_profiles = MSharedStory.stories_with_comments_and_profiles(stories, user.pk, check_all=True)
story_feed_ids = list(set(s['story_feed_id'] for s in stories))
usersubs = UserSubscription.objects.filter(user__pk=user.pk, feed__pk__in=story_feed_ids)
usersubs_map = dict((sub.feed_id, sub) for sub in usersubs)
unsub_feed_ids = list(set(story_feed_ids).difference(set(usersubs_map.keys())))
unsub_feeds = Feed.objects.filter(pk__in=unsub_feed_ids)
unsub_feeds = [feed.canonical(include_favicon=False) for feed in unsub_feeds]
date_delta = UNREAD_CUTOFF
if socialsub and date_delta < socialsub.mark_read_date:
date_delta = socialsub.mark_read_date
# Get intelligence classifier for user
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk, social_user_id=social_user_id))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk, social_user_id=social_user_id))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk, social_user_id=social_user_id))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk, social_user_id=social_user_id))
# Merge with feed specific classifiers
classifier_feeds = classifier_feeds + list(MClassifierFeed.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_authors = classifier_authors + list(MClassifierAuthor.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_titles = classifier_titles + list(MClassifierTitle.objects(user_id=user.pk, feed_id__in=story_feed_ids))
classifier_tags = classifier_tags + list(MClassifierTag.objects(user_id=user.pk, feed_id__in=story_feed_ids))
checkpoint2 = time.time()
story_ids = [story['id'] for story in stories]
userstories_db = MUserStory.objects(user_id=user.pk,
feed_id__in=story_feed_ids,
story_id__in=story_ids).only('story_id')
userstories = set(us.story_id for us in userstories_db)
starred_stories = MStarredStory.objects(user_id=user.pk,
story_feed_id__in=story_feed_ids,
story_guid__in=story_ids).only('story_guid', 'starred_date')
shared_stories = MSharedStory.objects(user_id=user.pk,
story_feed_id__in=story_feed_ids,
story_guid__in=story_ids)\
.only('story_guid', 'shared_date', 'comments')
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
shared_stories = dict([(story.story_guid, dict(shared_date=story.shared_date, comments=story.comments))
for story in shared_stories])
for story in stories:
story['social_user_id'] = social_user_id
story_feed_id = story['story_feed_id']
# story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
shared_date = localtime_for_timezone(story['shared_date'], user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(shared_date, now)
story['long_parsed_date'] = format_story_link_date__long(shared_date, now)
if not socialsub:
story['read_status'] = 1
elif story['id'] in userstories:
story['read_status'] = 1
elif story['shared_date'] < date_delta:
story['read_status'] = 1
elif not usersubs_map.get(story_feed_id):
story['read_status'] = 0
elif not story.get('read_status') and story['story_date'] < usersubs_map[story_feed_id].mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['shared_date'] < date_delta:
story['read_status'] = 1
# elif not story.get('read_status') and socialsub and story['shared_date'] > socialsub.last_read_date:
# story['read_status'] = 0
else:
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
if story['id'] in shared_stories:
story['shared'] = True
shared_date = localtime_for_timezone(shared_stories[story['id']]['shared_date'],
user.profile.timezone)
story['shared_date'] = format_story_link_date__long(shared_date, now)
story['shared_comments'] = strip_tags(shared_stories[story['id']]['comments'])
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, story['story_feed_id'],
social_user_id=social_user_id),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
classifiers = sort_classifiers_by_feed(user=user, feed_ids=story_feed_ids,
classifier_feeds=classifier_feeds,
classifier_authors=classifier_authors,
classifier_titles=classifier_titles,
classifier_tags=classifier_tags)
if socialsub:
socialsub.feed_opens += 1
socialsub.save()
diff1 = checkpoint1-start
diff2 = checkpoint2-start
logging.user(request, "~FYLoading ~FMshared stories~FY: ~SB%s%s ~SN(~SB%.4ss/%.4ss~SN)" % (
social_profile.title[:22], ('~SN/p%s' % page) if page > 1 else '', diff1, diff2))
return {
"stories": stories,
"user_profiles": user_profiles,
"feeds": unsub_feeds,
"classifiers": classifiers,
}
def load_single_feed(request, feed_id):
start = time.time()
user = get_user(request)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 12))
page = int(request.REQUEST.get('page', 1))
dupe_feed_id = None
userstories_db = None
if page: offset = limit * (page-1)
if not feed_id: raise Http404
try:
feed = Feed.objects.get(id=feed_id)
except Feed.DoesNotExist:
feed_address = request.REQUEST.get('feed_address')
dupe_feed = DuplicateFeed.objects.filter(duplicate_address=feed_address)
if dupe_feed:
feed = dupe_feed[0].feed
dupe_feed_id = feed_id
else:
raise Http404
stories = feed.get_stories(offset, limit)
# Get intelligence classifier for user
classifier_feeds = list(MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id))
classifier_authors = list(MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=user.pk, feed_id=feed_id))
checkpoint1 = time.time()
usersub = UserSubscription.objects.get(user=user, feed=feed)
userstories = []
if usersub and stories:
story_ids = [story['id'] for story in stories]
userstories_db = MUserStory.objects(user_id=user.pk,
feed_id=feed.pk,
story_id__in=story_ids).only('story_id')
starred_stories = MStarredStory.objects(user_id=user.pk,
story_feed_id=feed_id,
story_guid__in=story_ids).only('story_guid', 'starred_date')
starred_stories = dict([(story.story_guid, story.starred_date) for story in starred_stories])
userstories = set(us.story_id for us in userstories_db)
checkpoint2 = time.time()
for story in stories:
story_date = localtime_for_timezone(story['story_date'], user.profile.timezone)
now = localtime_for_timezone(datetime.datetime.now(), user.profile.timezone)
story['short_parsed_date'] = format_story_link_date__short(story_date, now)
story['long_parsed_date'] = format_story_link_date__long(story_date, now)
if usersub:
if story['id'] in userstories:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] < usersub.mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] > usersub.last_read_date:
story['read_status'] = 0
if story['id'] in starred_stories:
story['starred'] = True
starred_date = localtime_for_timezone(starred_stories[story['id']], user.profile.timezone)
story['starred_date'] = format_story_link_date__long(starred_date, now)
else:
story['read_status'] = 1
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, feed),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
checkpoint3 = time.time()
# Intelligence
feed_tags = json.decode(feed.data.popular_tags) if feed.data.popular_tags else []
feed_authors = json.decode(feed.data.popular_authors) if feed.data.popular_authors else []
classifiers = get_classifiers_for_user(user, feed_id, classifier_feeds,
classifier_authors, classifier_titles, classifier_tags)
if usersub:
usersub.feed_opens += 1
usersub.save()
diff1 = checkpoint1-start
diff2 = checkpoint2-start
diff3 = checkpoint3-start
timediff = time.time()-start
last_update = relative_timesince(feed.last_update)
logging.user(request, "~FYLoading feed: ~SB%s%s ~SN(%.4s seconds, ~SB%.4s/%.4s(%s)/%.4s~SN)" % (
feed.feed_title[:32], ('~SN/p%s' % page) if page > 1 else '', timediff,
diff1, diff2, userstories_db and userstories_db.count() or '~SN0~SB', diff3))
FeedLoadtime.objects.create(feed=feed, loadtime=timediff)
data = dict(stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers,
last_update=last_update,
feed_id=feed.pk)
if dupe_feed_id: data['dupe_feed_id'] = dupe_feed_id
if not usersub:
data.update(feed.canonical())
return data
def load_single_feed(request):
user = get_user(request)
offset = int(request.REQUEST.get('offset', 0))
limit = int(request.REQUEST.get('limit', 30))
page = int(request.REQUEST.get('page', 0))
if page:
offset = limit * page
feed_id = int(request.REQUEST.get('feed_id', 0))
if feed_id == 0:
raise Http404
try:
feed = Feed.objects.get(id=feed_id)
except Feed.DoesNotExist:
feed_address = request.REQUEST.get('feed_address')
dupe_feed = DuplicateFeed.objects.filter(duplicate_address=feed_address)
if dupe_feed:
feed = dupe_feed[0].feed
else:
raise Http404
force_update = request.GET.get('force_update', False)
now = datetime.datetime.utcnow()
stories = feed.get_stories(offset, limit)
if force_update:
feed.update(force_update)
# Get intelligence classifier for user
classifier_feeds = MClassifierFeed.objects(user_id=user.pk, feed_id=feed_id)
classifier_authors = MClassifierAuthor.objects(user_id=user.pk, feed_id=feed_id)
classifier_titles = MClassifierTitle.objects(user_id=user.pk, feed_id=feed_id)
classifier_tags = MClassifierTag.objects(user_id=user.pk, feed_id=feed_id)
try:
usersub = UserSubscription.objects.get(user=user, feed=feed)
except UserSubscription.DoesNotExist:
# FIXME: Why is this happening for `conesus` when logged into another account?!
logging.info(" ***> [%s] UserSub DNE, creating: %s" % (user, feed))
usersub = UserSubscription.objects.create(user=user, feed=feed)
userstories = []
userstories_db = MUserStory.objects(user_id=user.pk,
feed_id=feed.pk,
read_date__gte=usersub.mark_read_date)
for us in userstories_db:
if hasattr(us.story, 'story_guid') and isinstance(us.story.story_guid, unicode):
userstories.append(us.story.story_guid)
elif hasattr(us.story, 'id') and isinstance(us.story.id, unicode):
userstories.append(us.story.id) # TODO: Remove me after migration from story.id->guid
for story in stories:
classifier_feeds.rewind()
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
if story['id'] in userstories:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] < usersub.mark_read_date:
story['read_status'] = 1
elif not story.get('read_status') and story['story_date'] > usersub.last_read_date:
story['read_status'] = 0
story['intelligence'] = {
'feed': apply_classifier_feeds(classifier_feeds, feed),
'author': apply_classifier_authors(classifier_authors, story),
'tags': apply_classifier_tags(classifier_tags, story),
'title': apply_classifier_titles(classifier_titles, story),
}
# Intelligence
feed_tags = json.decode(feed.popular_tags) if feed.popular_tags else []
feed_authors = json.decode(feed.popular_authors) if feed.popular_authors else []
classifiers = get_classifiers_for_user(user, feed_id, classifier_feeds,
classifier_authors, classifier_titles, classifier_tags)
usersub.feed_opens += 1
usersub.save()
diff = datetime.datetime.utcnow()-now
timediff = float("%s.%s" % (diff.seconds, (diff.microseconds / 1000)))
logging.info(" ---> [%s] Loading feed: %s (%s seconds)" % (request.user, feed, timediff))
FeedLoadtime.objects.create(feed=feed, loadtime=timediff)
last_update = relative_timesince(feed.last_update)
data = dict(stories=stories,
feed_tags=feed_tags,
feed_authors=feed_authors,
classifiers=classifiers,
last_update=last_update,
feed_id=feed.pk)
return data
def calculate_feed_scores(self, silent=False, stories=None, force=False):
# now = datetime.datetime.strptime("2009-07-06 22:30:03", "%Y-%m-%d %H:%M:%S")
now = datetime.datetime.now()
oldest_unread_story_date = now
if self.user.profile.last_seen_on < self.user.profile.unread_cutoff and not force:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return self
ong = self.unread_count_negative
ont = self.unread_count_neutral
ops = self.unread_count_positive
# if not self.feed.fetched_once:
# if not silent:
# logging.info(' ---> [%s] NOT Computing scores: %s' % (self.user, self.feed))
# self.needs_unread_recalc = False
# self.save()
# return
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 = self.user.profile.unread_cutoff
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
if self.is_trained:
if not stories:
stories = cache.get('S:%s' % self.feed_id)
unread_story_hashes = self.story_hashes(user_id=self.user_id, feed_ids=[self.feed_id],
usersubs=[self],
read_filter='unread', group_by_feed=False,
cutoff_date=self.user.profile.unread_cutoff)
if not stories:
stories_db = MStory.objects(story_hash__in=unread_story_hashes)
stories = Feed.format_stories(stories_db, self.feed_id)
unread_stories = []
for story in stories:
if story['story_date'] < date_delta:
continue
if story['story_hash'] in unread_story_hashes:
unread_stories.append(story)
if story['story_date'] < oldest_unread_story_date:
oldest_unread_story_date = story['story_date']
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = list(MClassifierFeed.objects(user_id=self.user_id, feed_id=self.feed_id, social_user_id=0))
classifier_authors = list(MClassifierAuthor.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_titles = list(MClassifierTitle.objects(user_id=self.user_id, feed_id=self.feed_id))
classifier_tags = list(MClassifierTag.objects(user_id=self.user_id, feed_id=self.feed_id))
if (not len(classifier_feeds) and
not len(classifier_authors) and
not len(classifier_titles) and
not len(classifier_tags)):
self.is_trained = False
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in unread_stories:
scores.update({
'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
else:
unread_story_hashes = self.story_hashes(user_id=self.user_id, feed_ids=[self.feed_id],
usersubs=[self],
read_filter='unread', group_by_feed=False,
include_timestamps=True,
cutoff_date=self.user.profile.unread_cutoff)
feed_scores['neutral'] = len(unread_story_hashes)
if feed_scores['neutral']:
oldest_unread_story_date = datetime.datetime.fromtimestamp(unread_story_hashes[-1][1])
if not silent:
logging.user(self.user, '~FBUnread count (~SB%s~SN%s): ~SN(~FC%s~FB/~FC%s~FB/~FC%s~FB) ~SBto~SN (~FC%s~FB/~FC%s~FB/~FC%s~FB)' % (self.feed_id, '/~FMtrained~FB' if self.is_trained else '', ong, ont, ops, feed_scores['negative'], feed_scores['neutral'], feed_scores['positive']))
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.user(self.user, '~FC~SNComputing scores: %s (~SB%s~SN/~SB%s~SN/~SB%s~SN)' % (self.feed, feed_scores['negative'], feed_scores['neutral'], feed_scores['positive']))
self.trim_read_stories()
return self
def calculate_feed_scores(self, silent=False, stories_db=None):
now = datetime.datetime.utcnow()
UNREAD_CUTOFF = now - datetime.timedelta(days=settings.DAYS_OF_UNREAD)
if self.user.profile.last_seen_on < UNREAD_CUTOFF:
# if not silent:
# logging.info(' ---> [%s] SKIPPING Computing scores: %s (1 week+)' % (self.user, self.feed))
return
if not self.feed.fetched_once:
if not silent:
logging.info(' ---> [%s] NOT Computing scores: %s' %
(self.user, self.feed))
self.needs_unread_recalc = False
self.save()
return
if not silent:
logging.info(' ---> [%s] Computing scores: %s' %
(self.user, self.feed))
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 = UNREAD_CUTOFF
if date_delta < self.mark_read_date:
date_delta = self.mark_read_date
else:
self.mark_read_date = date_delta
read_stories = MUserStory.objects(user_id=self.user.pk,
feed_id=self.feed.pk,
read_date__gte=self.mark_read_date)
# if not silent:
# logging.info(' ---> [%s] Read stories: %s' % (self.user, datetime.datetime.now() - now))
read_stories_ids = []
for us in read_stories:
if hasattr(us.story, 'story_guid') and isinstance(
us.story.story_guid, unicode):
read_stories_ids.append(us.story.story_guid)
elif hasattr(us.story, 'id') and isinstance(us.story.id, unicode):
read_stories_ids.append(
us.story.id
) # TODO: Remove me after migration from story.id->guid
stories_db = stories_db or MStory.objects(story_feed_id=self.feed.pk,
story_date__gte=date_delta)
# if not silent:
# logging.info(' ---> [%s] MStory: %s' % (self.user, datetime.datetime.now() - now))
oldest_unread_story_date = now
unread_stories_db = []
for story in stories_db:
if story.story_date < date_delta:
continue
if hasattr(
story,
'story_guid') and story.story_guid not in read_stories_ids:
unread_stories_db.append(story)
if story.story_date < oldest_unread_story_date:
oldest_unread_story_date = story.story_date
stories = Feed.format_stories(unread_stories_db, self.feed.pk)
# if not silent:
# logging.info(' ---> [%s] Format stories: %s' % (self.user, datetime.datetime.now() - now))
classifier_feeds = MClassifierFeed.objects(user_id=self.user.pk,
feed_id=self.feed.pk)
classifier_authors = MClassifierAuthor.objects(user_id=self.user.pk,
feed_id=self.feed.pk)
classifier_titles = MClassifierTitle.objects(user_id=self.user.pk,
feed_id=self.feed.pk)
classifier_tags = MClassifierTag.objects(user_id=self.user.pk,
feed_id=self.feed.pk)
# if not silent:
# logging.info(' ---> [%s] Classifiers: %s (%s)' % (self.user, datetime.datetime.now() - now, classifier_feeds.count() + classifier_authors.count() + classifier_tags.count() + classifier_titles.count()))
scores = {
'feed': apply_classifier_feeds(classifier_feeds, self.feed),
}
for story in stories:
classifier_authors.rewind()
classifier_tags.rewind()
classifier_titles.rewind()
scores.update({
'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
# if not silent:
# logging.info(' ---> [%s] End classifiers: %s' % (self.user, datetime.datetime.now() - now))
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()
cache.delete('usersub:%s' % self.user.id)
return