Post URL loading optimizations

boards/models/post/__init__.py

@@ -1,414 +1,410 b''
 import uuid
 import hashlib
 import re
 
 from boards import settings
 from boards.abstracts.tripcode import Tripcode
 from boards.models import Attachment, KeyPair, GlobalId
 from boards.models.attachment import FILE_TYPES_IMAGE
 from boards.models.base import Viewable
 from boards.models.post.export import get_exporter, DIFF_TYPE_JSON
 from boards.models.post.manager import PostManager
 from boards.utils import datetime_to_epoch
 from django.core.exceptions import ObjectDoesNotExist
 from django.core.urlresolvers import reverse
 from django.db import models
 from django.db.models import TextField, QuerySet, F
 from django.template.defaultfilters import truncatewords, striptags
 from django.template.loader import render_to_string
 
 CSS_CLS_HIDDEN_POST = 'hidden_post'
 CSS_CLS_DEAD_POST = 'dead_post'
 CSS_CLS_ARCHIVE_POST = 'archive_post'
 CSS_CLS_POST = 'post'
 CSS_CLS_MONOCHROME = 'monochrome'
 
 TITLE_MAX_WORDS = 10
 
 APP_LABEL_BOARDS = 'boards'
 
 BAN_REASON_AUTO = 'Auto'
 
 TITLE_MAX_LENGTH = 200
 
 REGEX_REPLY = re.compile(r'\[post\](\d+)\[/post\]')
 REGEX_GLOBAL_REPLY = re.compile(r'\[post\](\w+)::([^:]+)::(\d+)\[/post\]')
 REGEX_URL = re.compile(r'https?\://[a-zA-Z0-9\-\.]+\.[a-zA-Z]{2,3}(/\S*)?')
 REGEX_NOTIFICATION = re.compile(r'\[user\](\w+)\[/user\]')
 
 PARAMETER_TRUNCATED = 'truncated'
 PARAMETER_TAG = 'tag'
 PARAMETER_OFFSET = 'offset'
 PARAMETER_DIFF_TYPE = 'type'
 PARAMETER_CSS_CLASS = 'css_class'
 PARAMETER_THREAD = 'thread'
 PARAMETER_IS_OPENING = 'is_opening'
 PARAMETER_POST = 'post'
 PARAMETER_OP_ID = 'opening_post_id'
 PARAMETER_NEED_OPEN_LINK = 'need_open_link'
 PARAMETER_REPLY_LINK = 'reply_link'
 PARAMETER_NEED_OP_DATA = 'need_op_data'
 
 POST_VIEW_PARAMS = (
     'need_op_data',
     'reply_link',
     'need_open_link',
     'truncated',
     'mode_tree',
     'perms',
     'tree_depth',
 )
 
 
 class Post(models.Model, Viewable):
     """A post is a message."""
 
     objects = PostManager()
 
     class Meta:
         app_label = APP_LABEL_BOARDS
         ordering = ('id',)
 
     title = models.CharField(max_length=TITLE_MAX_LENGTH, null=True, blank=True)
     pub_time = models.DateTimeField(db_index=True)
     text = TextField(blank=True, null=True)
     _text_rendered = TextField(blank=True, null=True, editable=False)
 
     attachments = models.ManyToManyField(Attachment, null=True, blank=True,
                                          related_name='attachment_posts')
 
     poster_ip = models.GenericIPAddressField()
 
     # Used for cache and threads updating
     last_edit_time = models.DateTimeField()
 
     referenced_posts = models.ManyToManyField('Post', symmetrical=False,
                                               null=True,
                                               blank=True, related_name='refposts',
                                               db_index=True)
     refmap = models.TextField(null=True, blank=True)
     threads = models.ManyToManyField('Thread', db_index=True,
                                      related_name='multi_replies')
     thread = models.ForeignKey('Thread', db_index=True, related_name='pt+')
 
     url = models.TextField()
     uid = models.TextField(db_index=True)
 
     # Global ID with author key. If the message was downloaded from another
     # server, this indicates the server.
     global_id = models.OneToOneField(GlobalId, null=True, blank=True,
                                      on_delete=models.CASCADE)
 
     tripcode = models.CharField(max_length=50, blank=True, default='')
     opening = models.BooleanField(db_index=True)
     hidden = models.BooleanField(default=False)
     version = models.IntegerField(default=1)
 
     def __str__(self):
         return 'P#{}/{}'.format(self.id, self.get_title())
 
     def get_title(self) -> str:
         return self.title
 
     def get_title_or_text(self):
         title = self.get_title()
         if not title:
             title = truncatewords(striptags(self.get_text()), TITLE_MAX_WORDS)
 
         return title
 
     def build_refmap(self, excluded_ids=None) -> None:
         """
         Builds a replies map string from replies list. This is a cache to stop
         the server from recalculating the map on every post show.
         """
 
         replies = self.referenced_posts
         if excluded_ids is not None:
             replies = replies.exclude(id__in=excluded_ids)
         else:
             replies = replies.all()
 
         post_urls = [refpost.get_link_view() for refpost in replies]
 
         self.refmap = ', '.join(post_urls)
 
     def is_referenced(self) -> bool:
         return self.refmap and len(self.refmap) > 0
 
     def is_opening(self) -> bool:
         """
         Checks if this is an opening post or just a reply.
         """
 
         return self.opening
 
     def get_absolute_url(self, thread=None):
-        url = None
-
-        if thread is None:
-            thread = self.get_thread()
-
-            # Url is cached only for the "main" thread. When getting url
-            # for other threads, do it manually.
-            if self.url:
-                url = self.url
+        # Url is cached only for the "main" thread. When getting url
+        # for other threads, do it manually.
+        url = self.url
 
         if url is None:
+            if thread is None:
+                thread = self.get_thread()
             opening = self.is_opening()
             opening_id = self.id if opening else thread.get_opening_post_id()
             url = reverse('thread', kwargs={'post_id': opening_id})
             if not opening:
                 url += '#' + str(self.id)
 
         return url
 
     def get_thread(self):
         return self.thread
 
     def get_thread_id(self):
         return self.thread_id
 
     def get_threads(self) -> QuerySet:
         """
         Gets post's thread.
         """
 
         return self.threads
 
     def _get_cache_key(self):
         return [datetime_to_epoch(self.last_edit_time)]
 
     def get_view_params(self, *args, **kwargs):
         """
         Gets the parameters required for viewing the post based on the arguments
         given and the post itself.
         """
         thread = self.get_thread()
 
         css_classes = [CSS_CLS_POST]
         if thread.is_archived():
             css_classes.append(CSS_CLS_ARCHIVE_POST)
         elif not thread.can_bump():
             css_classes.append(CSS_CLS_DEAD_POST)
         if self.is_hidden():
             css_classes.append(CSS_CLS_HIDDEN_POST)
         if thread.is_monochrome():
             css_classes.append(CSS_CLS_MONOCHROME)
 
         params = dict()
         for param in POST_VIEW_PARAMS:
             if param in kwargs:
                 params[param] = kwargs[param]
 
         params.update({
             PARAMETER_POST: self,
             PARAMETER_IS_OPENING: self.is_opening(),
             PARAMETER_THREAD: thread,
             PARAMETER_CSS_CLASS: ' '.join(css_classes),
         })
 
         return params
 
     def get_view(self, *args, **kwargs) -> str:
         """
         Renders post's HTML view. Some of the post params can be passed over
         kwargs for the means of caching (if we view the thread, some params
         are same for every post and don't need to be computed over and over.
         """
         params = self.get_view_params(*args, **kwargs)
 
         return render_to_string('boards/post.html', params)
 
     def get_search_view(self, *args, **kwargs):
         return self.get_view(need_op_data=True, *args, **kwargs)
 
     def get_first_image(self) -> Attachment:
         return self.attachments.filter(mimetype__in=FILE_TYPES_IMAGE).earliest('id')
 
     def set_global_id(self, key_pair=None):
         """
         Sets global id based on the given key pair. If no key pair is given,
         default one is used.
         """
 
         if key_pair:
             key = key_pair
         else:
             try:
                 key = KeyPair.objects.get(primary=True)
             except KeyPair.DoesNotExist:
                 # Do not update the global id because there is no key defined
                 return
         global_id = GlobalId(key_type=key.key_type,
                              key=key.public_key,
                              local_id=self.id)
         global_id.save()
 
         self.global_id = global_id
 
         self.save(update_fields=['global_id'])
 
     def get_pub_time_str(self):
         return str(self.pub_time)
 
     def get_replied_ids(self):
         """
         Gets ID list of the posts that this post replies.
         """
 
         raw_text = self.get_raw_text()
 
         local_replied = REGEX_REPLY.findall(raw_text)
         global_replied = []
         for match in REGEX_GLOBAL_REPLY.findall(raw_text):
             key_type = match[0]
             key = match[1]
             local_id = match[2]
 
             try:
                 global_id = GlobalId.objects.get(key_type=key_type,
                                                  key=key, local_id=local_id)
                 for post in Post.objects.filter(global_id=global_id).only('id'):
                     global_replied.append(post.id)
             except GlobalId.DoesNotExist:
                 pass
         return local_replied + global_replied
 
     def get_post_data(self, format_type=DIFF_TYPE_JSON, request=None,
                       include_last_update=False) -> str:
         """
         Gets post HTML or JSON data that can be rendered on a page or used by
         API.
         """
 
         return get_exporter(format_type).export(self, request,
                                                 include_last_update)
 
     def notify_clients(self, recursive=True):
         """
         Sends post HTML data to the thread web socket.
         """
 
         if not settings.get_bool('External', 'WebsocketsEnabled'):
             return
 
         thread_ids = list()
         for thread in self.get_threads().all():
             thread_ids.append(thread.id)
 
             thread.notify_clients()
 
         if recursive:
             for reply_number in re.finditer(REGEX_REPLY, self.get_raw_text()):
                 post_id = reply_number.group(1)
 
                 try:
                     ref_post = Post.objects.get(id=post_id)
 
                     if ref_post.get_threads().exclude(id__in=thread_ids).exists():
                         # If post is in this thread, its thread was already notified.
                         # Otherwise, notify its thread separately.
                         ref_post.notify_clients(recursive=False)
                 except ObjectDoesNotExist:
                     pass
 
     def build_url(self):
         self.url = self.get_absolute_url()
         self.save(update_fields=['url'])
 
     def save(self, force_insert=False, force_update=False, using=None,
              update_fields=None):
         new_post = self.id is None
 
         self.uid = str(uuid.uuid4())
         if update_fields is not None and 'uid' not in update_fields:
             update_fields += ['uid']
 
         if not new_post:
             for thread in self.get_threads().all():
                 thread.last_edit_time = self.last_edit_time
 
                 thread.save(update_fields=['last_edit_time', 'status'])
 
         super().save(force_insert, force_update, using, update_fields)
 
         if self.url is None:
             self.build_url()
 
     def get_text(self) -> str:
         return self._text_rendered
 
     def get_raw_text(self) -> str:
         return self.text
 
     def get_sync_text(self) -> str:
         """
         Returns text applicable for sync. It has absolute post reflinks.
         """
 
         replacements = dict()
         for post_id in REGEX_REPLY.findall(self.get_raw_text()):
             try:
                 absolute_post_id = str(Post.objects.get(id=post_id).global_id)
                 replacements[post_id] = absolute_post_id
             except Post.DoesNotExist:
                 pass
 
         text = self.get_raw_text() or ''
         for key in replacements:
             text = text.replace('[post]{}[/post]'.format(key),
                                 '[post]{}[/post]'.format(replacements[key]))
         text = text.replace('\r\n', '\n').replace('\r', '\n')
 
         return text
 
     def connect_threads(self, opening_posts):
         for opening_post in opening_posts:
             threads = opening_post.get_threads().all()
             for thread in threads:
                 if thread.can_bump():
                     thread.update_bump_status()
 
                 thread.last_edit_time = self.last_edit_time
                 thread.save(update_fields=['last_edit_time', 'status'])
             self.threads.add(opening_post.get_thread())
 
     def get_tripcode(self):
         if self.tripcode:
             return Tripcode(self.tripcode)
 
     def get_link_view(self):
         """
         Gets view of a reflink to the post.
         """
         result = '<a href="{}">&gt;&gt;{}</a>'.format(self.get_absolute_url(),
                                                       self.id)
         if self.is_opening():
             result = '<b>{}</b>'.format(result)
 
         return result
 
     def is_hidden(self) -> bool:
         return self.hidden
 
     def set_hidden(self, hidden):
         self.hidden = hidden
 
     def increment_version(self):
         self.version = F('version') + 1
 
     def clear_cache(self):
         """
         Clears sync data (content cache, signatures etc).
         """
         global_id = self.global_id
         if global_id is not None and global_id.is_local()\
                 and global_id.content is not None:
             global_id.clear_cache()
 
     def get_tags(self):
         return self.get_thread().get_tags()
 
     def get_ip_color(self):
         return hashlib.md5(self.poster_ip.encode()).hexdigest()[:6]

boards/models/thread.py

@@ -1,328 +1,328 b''
 import logging
 from adjacent import Client
 from datetime import timedelta
 
 
 from django.db.models import Count, Sum, QuerySet, Q
 from django.utils import timezone
 from django.db import models, transaction
 
 from boards.models.attachment import FILE_TYPES_IMAGE
 from boards.models import STATUS_BUMPLIMIT, STATUS_ACTIVE, STATUS_ARCHIVE
 
 from boards import settings
 import boards
 from boards.utils import cached_result, datetime_to_epoch
 from boards.models.post import Post
 from boards.models.tag import Tag
 
 FAV_THREAD_NO_UPDATES = -1
 
 
 __author__ = 'neko259'
 
 
 logger = logging.getLogger(__name__)
 
 
 WS_NOTIFICATION_TYPE_NEW_POST = 'new_post'
 WS_NOTIFICATION_TYPE = 'notification_type'
 
 WS_CHANNEL_THREAD = "thread:"
 
 STATUS_CHOICES = (
         (STATUS_ACTIVE, STATUS_ACTIVE),
         (STATUS_BUMPLIMIT, STATUS_BUMPLIMIT),
         (STATUS_ARCHIVE, STATUS_ARCHIVE),
 )
 
 
 class ThreadManager(models.Manager):
     def process_old_threads(self):
         """
          Preserves maximum thread count. If there are too many threads,
          archive or delete the old ones.
          """
         old_time_delta = settings.get_int('Messages', 'ThreadArchiveDays')
         old_time = timezone.now() - timedelta(days=old_time_delta)
         old_ops = Post.objects.filter(opening=True, pub_time__lte=old_time).exclude(thread__status=STATUS_ARCHIVE)
 
         for op in old_ops:
             thread = op.get_thread()
             if settings.get_bool('Storage', 'ArchiveThreads'):
                 self._archive_thread(thread)
             else:
                 thread.delete()
             logger.info('Processed old thread {}'.format(thread))
 
 
     def _archive_thread(self, thread):
         thread.status = STATUS_ARCHIVE
         thread.last_edit_time = timezone.now()
         thread.update_posts_time()
         thread.save(update_fields=['last_edit_time', 'status'])
 
     def get_new_posts(self, datas):
         query = None
         # TODO Use classes instead of dicts
         for data in datas:
             if data['last_id'] != FAV_THREAD_NO_UPDATES:
                 q = (Q(id=data['op'].get_thread_id())
                      & Q(multi_replies__id__gt=data['last_id']))
                 if query is None:
                     query = q
                 else:
                     query = query | q
         if query is not None:
             return self.filter(query).annotate(
                 new_post_count=Count('multi_replies'))
 
     def get_new_post_count(self, datas):
         new_posts = self.get_new_posts(datas)
         return new_posts.aggregate(total_count=Count('multi_replies'))\
                 ['total_count'] if new_posts else 0
 
 
 def get_thread_max_posts():
     return settings.get_int('Messages', 'MaxPostsPerThread')
 
 
 class Thread(models.Model):
     objects = ThreadManager()
 
     class Meta:
         app_label = 'boards'
 
     tags = models.ManyToManyField('Tag', related_name='thread_tags')
     bump_time = models.DateTimeField(db_index=True)
     last_edit_time = models.DateTimeField()
     max_posts = models.IntegerField(default=get_thread_max_posts)
     status = models.CharField(max_length=50, default=STATUS_ACTIVE,
             choices=STATUS_CHOICES, db_index=True)
     monochrome = models.BooleanField(default=False)
 
     def get_tags(self) -> QuerySet:
         """
         Gets a sorted tag list.
         """
 
         return self.tags.order_by('name')
 
     def bump(self):
         """
         Bumps (moves to up) thread if possible.
         """
 
         if self.can_bump():
             self.bump_time = self.last_edit_time
 
             self.update_bump_status()
 
             logger.info('Bumped thread %d' % self.id)
 
     def has_post_limit(self) -> bool:
         return self.max_posts > 0
 
     def update_bump_status(self, exclude_posts=None):
         if self.has_post_limit() and self.get_reply_count() >= self.max_posts:
             self.status = STATUS_BUMPLIMIT
             self.update_posts_time(exclude_posts=exclude_posts)
 
     def _get_cache_key(self):
         return [datetime_to_epoch(self.last_edit_time)]
 
     @cached_result(key_method=_get_cache_key)
     def get_reply_count(self) -> int:
         return self.get_replies().count()
 
     @cached_result(key_method=_get_cache_key)
     def get_images_count(self) -> int:
         return self.get_replies().filter(
             attachments__mimetype__in=FILE_TYPES_IMAGE)\
             .annotate(images_count=Count(
             'attachments')).aggregate(Sum('images_count'))['images_count__sum'] or 0
 
     def can_bump(self) -> bool:
         """
         Checks if the thread can be bumped by replying to it.
         """
 
         return self.get_status() == STATUS_ACTIVE
 
     def get_last_replies(self) -> QuerySet:
         """
         Gets several last replies, not including opening post
         """
 
         last_replies_count = settings.get_int('View', 'LastRepliesCount')
 
         if last_replies_count > 0:
             reply_count = self.get_reply_count()
 
             if reply_count > 0:
                 reply_count_to_show = min(last_replies_count,
                                           reply_count - 1)
                 replies = self.get_replies()
                 last_replies = replies[reply_count - reply_count_to_show:]
 
                 return last_replies
 
     def get_skipped_replies_count(self) -> int:
         """
         Gets number of posts between opening post and last replies.
         """
         reply_count = self.get_reply_count()
         last_replies_count = min(settings.get_int('View', 'LastRepliesCount'),
                                  reply_count - 1)
         return reply_count - last_replies_count - 1
 
     # TODO Remove argument, it is not used
     def get_replies(self, view_fields_only=True) -> QuerySet:
         """
         Gets sorted thread posts
         """
         query = self.multi_replies.order_by('pub_time').prefetch_related(
             'thread', 'attachments')
         return query
 
     def get_viewable_replies(self) -> QuerySet:
         """
         Gets replies with only fields that are used for viewing.
         """
         return self.get_replies().defer('poster_ip', 'text', 'last_edit_time',
                                         'version')
 
     def get_top_level_replies(self) -> QuerySet:
         return self.get_replies().exclude(refposts__threads__in=[self])
 
     def get_replies_with_images(self, view_fields_only=False) -> QuerySet:
         """
         Gets replies that have at least one image attached
         """
         return self.get_replies(view_fields_only).filter(
             attachments__mimetype__in=FILE_TYPES_IMAGE).annotate(images_count=Count(
             'attachments')).filter(images_count__gt=0)
 
     def get_opening_post(self, only_id=False) -> Post:
         """
         Gets the first post of the thread
         """
 
         query = self.get_replies().filter(opening=True)
         if only_id:
             query = query.only('id')
         opening_post = query.first()
 
         return opening_post
 
     @cached_result()
     def get_opening_post_id(self) -> int:
         """
         Gets ID of the first thread post.
         """
 
         return self.get_opening_post(only_id=True).id
 
     def get_pub_time(self):
         """
         Gets opening post's pub time because thread does not have its own one.
         """
 
         return self.get_opening_post().pub_time
 
     def __str__(self):
-        return 'T#{}/{}'.format(self.id, self.get_opening_post_id())
+        return 'T#{}'.format(self.id)
 
     def get_tag_url_list(self) -> list:
         return boards.models.Tag.objects.get_tag_url_list(self.get_tags())
 
     def update_posts_time(self, exclude_posts=None):
         last_edit_time = self.last_edit_time
 
         for post in self.multi_replies.all():
             if exclude_posts is None or post not in exclude_posts:
                 # Manual update is required because uids are generated on save
                 post.last_edit_time = last_edit_time
                 post.save(update_fields=['last_edit_time'])
 
                 post.get_threads().update(last_edit_time=last_edit_time)
 
     def notify_clients(self):
         if not settings.get_bool('External', 'WebsocketsEnabled'):
             return
 
         client = Client()
 
         channel_name = WS_CHANNEL_THREAD + str(self.get_opening_post_id())
         client.publish(channel_name, {
             WS_NOTIFICATION_TYPE: WS_NOTIFICATION_TYPE_NEW_POST,
             })
         client.send()
 
     def get_absolute_url(self):
         return self.get_opening_post().get_absolute_url()
 
     def get_required_tags(self):
         return self.get_tags().filter(required=True)
 
     def get_replies_newer(self, post_id):
         return self.get_replies().filter(id__gt=post_id)
 
     def is_archived(self):
         return self.get_status() == STATUS_ARCHIVE
 
     def get_status(self):
         return self.status
 
     def is_monochrome(self):
         return self.monochrome
 
     # If tags have parent, add them to the tag list
     @transaction.atomic
     def refresh_tags(self):
         for tag in self.get_tags().all():
             parents = tag.get_all_parents()
             if len(parents) > 0:
                 self.tags.add(*parents)
 
     def get_reply_tree(self):
         replies = self.get_replies().prefetch_related('refposts')
         tree = []
         for reply in replies:
             parents = reply.refposts.all()
 
             found_parent = False
             searching_for_index = False
 
             if len(parents) > 0:
                 index = 0
                 parent_depth = 0
 
                 indexes_to_insert = []
 
                 for depth, element in tree:
                     index += 1
 
                     # If this element is next after parent on the same level,
                     # insert child before it
                     if searching_for_index and depth <= parent_depth:
                         indexes_to_insert.append((index - 1, parent_depth))
                         searching_for_index = False
 
                     if element in parents:
                         found_parent = True
                         searching_for_index = True
                         parent_depth = depth
 
             if not found_parent:
                 tree.append((0, reply))
             else:
                 if searching_for_index:
                     tree.append((parent_depth + 1, reply))
 
                 offset = 0
                 for last_index, parent_depth in indexes_to_insert:
                     tree.insert(last_index + offset, (parent_depth + 1, reply))
                     offset += 1
 
         return tree