Image deduplication (BB-53). When an image with the same hash is uploaded, it

won't be saved. Instead, post will be linked to the same PostImage object

boards/forms.py

@@ -1,246 +1,240 b''
 import re
 import time
 
 from django import forms
 from django.forms.util import ErrorList
 from django.utils.translation import ugettext_lazy as _
 
 from boards.mdx_neboard import formatters
 from boards.models.post import TITLE_MAX_LENGTH
 from boards.models import PostImage, Tag
 from neboard import settings
 import boards.settings as board_settings
 
 REGEX_TAGS = re.compile(r'^[\w\s\d]+$', re.UNICODE)
 
 VETERAN_POSTING_DELAY = 5
 
 ATTRIBUTE_PLACEHOLDER = 'placeholder'
 
 LAST_POST_TIME = 'last_post_time'
 LAST_LOGIN_TIME = 'last_login_time'
 TEXT_PLACEHOLDER = _('Type message here. Use formatting panel for more advanced usage.')
 TAGS_PLACEHOLDER = _('tag1 several_words_tag')
 
-ERROR_IMAGE_DUPLICATE = _('Such image was already posted')
-
 LABEL_TITLE = _('Title')
 LABEL_TEXT = _('Text')
 LABEL_TAG = _('Tag')
 LABEL_SEARCH = _('Search')
 
 TAG_MAX_LENGTH = 20
 
 
 class FormatPanel(forms.Textarea):
     """
     Panel for text formatting. Consists of buttons to add different tags to the
     form text area.
     """
 
     def render(self, name, value, attrs=None):
         output = '<div id="mark-panel">'
         for formatter in formatters:
             output += '<span class="mark_btn"' + \
                       ' onClick="addMarkToMsg(\'' + formatter.format_left + \
                       '\', \'' + formatter.format_right + '\')">' + \
                       formatter.preview_left + formatter.name + \
                       formatter.preview_right + '</span>'
 
         output += '</div>'
         output += super(FormatPanel, self).render(name, value, attrs=None)
 
         return output
 
 
 class PlainErrorList(ErrorList):
     def __unicode__(self):
         return self.as_text()
 
     def as_text(self):
         return ''.join(['(!) %s ' % e for e in self])
 
 
 class NeboardForm(forms.Form):
     """
     Form with neboard-specific formatting.
     """
 
     def as_div(self):
         """
         Returns this form rendered as HTML <as_div>s.
         """
 
         return self._html_output(
             # TODO Do not show hidden rows in the list here
             normal_row='<div class="form-row"><div class="form-label">'
                        '%(label)s'
                        '</div></div>'
                        '<div class="form-row"><div class="form-input">'
                        '%(field)s'
                        '</div></div>'
                        '<div class="form-row">'
                        '%(help_text)s'
                        '</div>',
             error_row='<div class="form-row">'
                       '<div class="form-label"></div>'
                       '<div class="form-errors">%s</div>'
                       '</div>',
             row_ender='</div>',
             help_text_html='%s',
             errors_on_separate_row=True)
 
     def as_json_errors(self):
         errors = []
 
         for name, field in list(self.fields.items()):
             if self[name].errors:
                 errors.append({
                     'field': name,
                     'errors': self[name].errors.as_text(),
                 })
 
         return errors
 
 
 class PostForm(NeboardForm):
 
     title = forms.CharField(max_length=TITLE_MAX_LENGTH, required=False,
                             label=LABEL_TITLE)
     text = forms.CharField(
         widget=FormatPanel(attrs={ATTRIBUTE_PLACEHOLDER: TEXT_PLACEHOLDER}),
         required=False, label=LABEL_TEXT)
     image = forms.ImageField(required=False, label=_('Image'),
                              widget=forms.ClearableFileInput(
                                  attrs={'accept': 'image/*'}))
 
     # This field is for spam prevention only
     email = forms.CharField(max_length=100, required=False, label=_('e-mail'),
                             widget=forms.TextInput(attrs={
                                 'class': 'form-email'}))
 
     session = None
     need_to_ban = False
 
     def clean_title(self):
         title = self.cleaned_data['title']
         if title:
             if len(title) > TITLE_MAX_LENGTH:
                 raise forms.ValidationError(_('Title must have less than %s '
                                               'characters') %
                                             str(TITLE_MAX_LENGTH))
         return title
 
     def clean_text(self):
         text = self.cleaned_data['text'].strip()
         if text:
             if len(text) > board_settings.MAX_TEXT_LENGTH:
                 raise forms.ValidationError(_('Text must have less than %s '
                                               'characters') %
                                             str(board_settings
                                             .MAX_TEXT_LENGTH))
         return text
 
     def clean_image(self):
         image = self.cleaned_data['image']
         if image:
             if image.size > board_settings.MAX_IMAGE_SIZE:
                 raise forms.ValidationError(
                     _('Image must be less than %s bytes')
                     % str(board_settings.MAX_IMAGE_SIZE))
 
-            image_hash = PostImage.get_hash(image)
-            if PostImage.objects.filter(hash=image_hash).exists():
-                raise forms.ValidationError(ERROR_IMAGE_DUPLICATE)
-
         return image
 
     def clean(self):
         cleaned_data = super(PostForm, self).clean()
 
         if not self.session:
             raise forms.ValidationError('Humans have sessions')
 
         if cleaned_data['email']:
             self.need_to_ban = True
             raise forms.ValidationError('A human cannot enter a hidden field')
 
         if not self.errors:
             self._clean_text_image()
 
         if not self.errors and self.session:
             self._validate_posting_speed()
 
         return cleaned_data
 
     def _clean_text_image(self):
         text = self.cleaned_data.get('text')
         image = self.cleaned_data.get('image')
 
         if (not text) and (not image):
             error_message = _('Either text or image must be entered.')
             self._errors['text'] = self.error_class([error_message])
 
     def _validate_posting_speed(self):
         can_post = True
 
         posting_delay = settings.POSTING_DELAY
 
         if board_settings.LIMIT_POSTING_SPEED and LAST_POST_TIME in \
                 self.session:
             now = time.time()
             last_post_time = self.session[LAST_POST_TIME]
 
             current_delay = int(now - last_post_time)
 
             if current_delay < posting_delay:
                 error_message = _('Wait %s seconds after last posting') % str(
                     posting_delay - current_delay)
                 self._errors['text'] = self.error_class([error_message])
 
                 can_post = False
 
         if can_post:
             self.session[LAST_POST_TIME] = time.time()
 
 
 class ThreadForm(PostForm):
 
     tags = forms.CharField(
         widget=forms.TextInput(attrs={ATTRIBUTE_PLACEHOLDER: TAGS_PLACEHOLDER}),
         max_length=100, label=_('Tags'), required=True)
 
     def clean_tags(self):
         tags = self.cleaned_data['tags'].strip()
 
         if not tags or not REGEX_TAGS.match(tags):
             raise forms.ValidationError(
                 _('Inappropriate characters in tags.'))
 
         required_tag_exists = False
         for tag in tags.split():
             tag_model = Tag.objects.filter(name=tag.strip().lower(),
                                            required=True)
             if tag_model.exists():
                 required_tag_exists = True
                 break
 
         if not required_tag_exists:
             raise forms.ValidationError(_('Need at least 1 required tag.'))
 
         return tags
 
     def clean(self):
         cleaned_data = super(ThreadForm, self).clean()
 
         return cleaned_data
 
 
 class SettingsForm(NeboardForm):
 
     theme = forms.ChoiceField(choices=settings.THEMES,
                               label=_('Theme'))
 
 
 class SearchForm(NeboardForm):
     query = forms.CharField(max_length=500, label=LABEL_SEARCH, required=False)

boards/models/post.py

@@ -1,441 +1,451 b''
 from datetime import datetime, timedelta, date
 from datetime import time as dtime
 import logging
 import re
 
 from adjacent import Client
 from django.core.cache import cache
 from django.core.urlresolvers import reverse
 from django.db import models, transaction
 from django.db.models import TextField
 from django.template.loader import render_to_string
 from django.utils import timezone
 
 from boards import settings
 from boards.mdx_neboard import bbcode_extended
 from boards.models import PostImage
 from boards.models.base import Viewable
 from boards.models.thread import Thread
 from boards.utils import datetime_to_epoch
 
 
 WS_NOTIFICATION_TYPE_NEW_POST = 'new_post'
 WS_NOTIFICATION_TYPE = 'notification_type'
 
 WS_CHANNEL_THREAD = "thread:"
 
 APP_LABEL_BOARDS = 'boards'
 
 CACHE_KEY_PPD = 'ppd'
 CACHE_KEY_POST_URL = 'post_url'
 
 POSTS_PER_DAY_RANGE = 7
 
 BAN_REASON_AUTO = 'Auto'
 
 IMAGE_THUMB_SIZE = (200, 150)
 
 TITLE_MAX_LENGTH = 200
 
 # TODO This should be removed
 NO_IP = '0.0.0.0'
 
 # TODO Real user agent should be saved instead of this
 UNKNOWN_UA = ''
 
 REGEX_REPLY = re.compile(r'\[post\](\d+)\[/post\]')
 
 PARAMETER_TRUNCATED = 'truncated'
 PARAMETER_TAG = 'tag'
 PARAMETER_OFFSET = 'offset'
 PARAMETER_DIFF_TYPE = 'type'
 PARAMETER_BUMPABLE = 'bumpable'
 PARAMETER_THREAD = 'thread'
 PARAMETER_IS_OPENING = 'is_opening'
 PARAMETER_MODERATOR = 'moderator'
 PARAMETER_POST = 'post'
 PARAMETER_OP_ID = 'opening_post_id'
 PARAMETER_NEED_OPEN_LINK = 'need_open_link'
 
 DIFF_TYPE_HTML = 'html'
 DIFF_TYPE_JSON = 'json'
 
 PREPARSE_PATTERNS = {
     r'>>(\d+)': r'[post]\1[/post]',  # Reflink ">>123"
     r'^>([^>].+)': r'[quote]\1[/quote]',  # Quote ">text"
     r'^//(.+)': r'[comment]\1[/comment]',  # Comment "//text"
 }
 
 
 class PostManager(models.Manager):
     @transaction.atomic
     def create_post(self, title: str, text: str, image=None, thread=None,
                     ip=NO_IP, tags: list=None):
         """
         Creates new post
         """
 
         if not tags:
             tags = []
 
         posting_time = timezone.now()
         if not thread:
             thread = Thread.objects.create(bump_time=posting_time,
                                            last_edit_time=posting_time)
             new_thread = True
         else:
             new_thread = False
 
         pre_text = self._preparse_text(text)
 
         post = self.create(title=title,
                            text=pre_text,
                            pub_time=posting_time,
                            thread_new=thread,
                            poster_ip=ip,
                            poster_user_agent=UNKNOWN_UA,  # TODO Get UA at
                            # last!
                            last_edit_time=posting_time)
 
         logger = logging.getLogger('boards.post.create')
 
         logger.info('Created post {} by {}'.format(
             post, post.poster_ip))
 
         if image:
-            post_image = PostImage.objects.create(image=image)
+            # Try to find existing image. If it exists, assign it to the post
+            # instead of createing the new one
+            image_hash = PostImage.get_hash(image)
+            existing = PostImage.objects.filter(hash=image_hash)
+            if len(existing) > 0:
+                post_image = existing[0]
+            else:
+                post_image = PostImage.objects.create(image=image)
+                logger.info('Created new image #{} for post #{}'.format(
+                            post_image.id, post.id))
             post.images.add(post_image)
-            logger.info('Created image #{} for post #{}'.format(
-                post_image.id, post.id))
 
         thread.replies.add(post)
         list(map(thread.add_tag, tags))
 
         if new_thread:
             Thread.objects.process_oldest_threads()
         else:
             thread.bump()
             thread.last_edit_time = posting_time
             thread.save()
 
         self.connect_replies(post)
 
         return post
 
     def delete_posts_by_ip(self, ip):
         """
         Deletes all posts of the author with same IP
         """
 
         posts = self.filter(poster_ip=ip)
         for post in posts:
             post.delete()
 
     def  connect_replies(self, post):
         """
         Connects replies to a post to show them as a reflink map
         """
 
         for reply_number in re.finditer(REGEX_REPLY, post.get_raw_text()):
             post_id = reply_number.group(1)
             ref_post = self.filter(id=post_id)
             if ref_post.count() > 0:
                 referenced_post = ref_post[0]
                 referenced_post.referenced_posts.add(post)
                 referenced_post.last_edit_time = post.pub_time
                 referenced_post.build_refmap()
                 referenced_post.save(update_fields=['refmap', 'last_edit_time'])
 
                 referenced_thread = referenced_post.get_thread()
                 referenced_thread.last_edit_time = post.pub_time
                 referenced_thread.save(update_fields=['last_edit_time'])
 
     def get_posts_per_day(self):
         """
         Gets average count of posts per day for the last 7 days
         """
 
         day_end = date.today()
         day_start = day_end - timedelta(POSTS_PER_DAY_RANGE)
 
         cache_key = CACHE_KEY_PPD + str(day_end)
         ppd = cache.get(cache_key)
         if ppd:
             return ppd
 
         day_time_start = timezone.make_aware(datetime.combine(
             day_start, dtime()), timezone.get_current_timezone())
         day_time_end = timezone.make_aware(datetime.combine(
             day_end, dtime()), timezone.get_current_timezone())
 
         posts_per_period = float(self.filter(
             pub_time__lte=day_time_end,
             pub_time__gte=day_time_start).count())
 
         ppd = posts_per_period / POSTS_PER_DAY_RANGE
 
         cache.set(cache_key, ppd)
         return ppd
 
     def _preparse_text(self, text):
         """
         Preparses text to change patterns like '>>' to a proper bbcode
         tags.
         """
 
         for key, value in PREPARSE_PATTERNS.items():
             text = re.sub(key, value, text, flags=re.MULTILINE)
 
         return text
 
 
 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)
     pub_time = models.DateTimeField()
     text = TextField(blank=True, null=True)
     _text_rendered = TextField(blank=True, null=True, editable=False)
 
     images = models.ManyToManyField(PostImage, null=True, blank=True,
                                     related_name='ip+', db_index=True)
 
     poster_ip = models.GenericIPAddressField()
     poster_user_agent = models.TextField()
 
     thread_new = models.ForeignKey('Thread', null=True, default=None,
                                    db_index=True)
     last_edit_time = models.DateTimeField()
 
     referenced_posts = models.ManyToManyField('Post', symmetrical=False,
                                               null=True,
                                               blank=True, related_name='rfp+',
                                               db_index=True)
     refmap = models.TextField(null=True, blank=True)
 
     def __str__(self):
         return 'P#{}/{}'.format(self.id, self.title)
 
     def get_title(self) -> str:
         """
         Gets original post title or part of its text.
         """
 
         title = self.title
         if not title:
             title = self.get_text()
 
         return title
 
     def build_refmap(self) -> 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.
         """
         map_string = ''
 
         first = True
         for refpost in self.referenced_posts.all():
             if not first:
                 map_string += ', '
             map_string += '<a href="%s">&gt;&gt;%s</a>' % (refpost.get_url(),
                                                            refpost.id)
             first = False
 
         self.refmap = map_string
 
     def get_sorted_referenced_posts(self):
         return self.refmap
 
     def is_referenced(self) -> bool:
         if not self.refmap:
             return False
         else:
             return len(self.refmap) > 0
 
     def is_opening(self) -> bool:
         """
         Checks if this is an opening post or just a reply.
         """
 
         return self.get_thread().get_opening_post_id() == self.id
 
     @transaction.atomic
     def add_tag(self, tag):
         edit_time = timezone.now()
 
         thread = self.get_thread()
         thread.add_tag(tag)
         self.last_edit_time = edit_time
         self.save(update_fields=['last_edit_time'])
 
         thread.last_edit_time = edit_time
         thread.save(update_fields=['last_edit_time'])
 
     def get_url(self, thread=None):
         """
         Gets full url to the post.
         """
 
         cache_key = CACHE_KEY_POST_URL + str(self.id)
         link = cache.get(cache_key)
 
         if not link:
             if not thread:
                 thread = self.get_thread()
 
             opening_id = thread.get_opening_post_id()
 
             if self.id != opening_id:
                 link = reverse('thread', kwargs={
                     'post_id': opening_id}) + '#' + str(self.id)
             else:
                 link = reverse('thread', kwargs={'post_id': self.id})
 
             cache.set(cache_key, link)
 
         return link
 
     def get_thread(self) -> Thread:
         """
         Gets post's thread.
         """
 
         return self.thread_new
 
     def get_referenced_posts(self):
         return self.referenced_posts.only('id', 'thread_new')
 
     def get_view(self, moderator=False, need_open_link=False,
                  truncated=False, *args, **kwargs):
         """
         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.
         """
 
         is_opening = kwargs.get(PARAMETER_IS_OPENING, self.is_opening())
         thread = kwargs.get(PARAMETER_THREAD, self.get_thread())
         can_bump = kwargs.get(PARAMETER_BUMPABLE, thread.can_bump())
 
         if is_opening:
             opening_post_id = self.id
         else:
             opening_post_id = thread.get_opening_post_id()
 
         return render_to_string('boards/post.html', {
             PARAMETER_POST: self,
             PARAMETER_MODERATOR: moderator,
             PARAMETER_IS_OPENING: is_opening,
             PARAMETER_THREAD: thread,
             PARAMETER_BUMPABLE: can_bump,
             PARAMETER_NEED_OPEN_LINK: need_open_link,
             PARAMETER_TRUNCATED: truncated,
             PARAMETER_OP_ID: opening_post_id,
         })
 
     def get_search_view(self, *args, **kwargs):
         return self.get_view(args, kwargs)
 
     def get_first_image(self) -> PostImage:
         return self.images.earliest('id')
 
     def delete(self, using=None):
         """
         Deletes all post images and the post itself. If the post is opening,
         thread with all posts is deleted.
         """
 
-        self.images.all().delete()
+        for image in self.images.all():
+            image_refs_count = Post.objects.filter(images__in=[image]).count()
+            if image_refs_count == 1:
+                image.delete()
 
         if self.is_opening():
             self.get_thread().delete()
         else:
             thread = self.get_thread()
             thread.last_edit_time = timezone.now()
             thread.save()
 
         super(Post, self).delete(using)
 
         logging.getLogger('boards.post.delete').info(
             'Deleted post {}'.format(self))
 
     def get_post_data(self, format_type=DIFF_TYPE_JSON, request=None,
                       include_last_update=False):
         """
         Gets post HTML or JSON data that can be rendered on a page or used by
         API.
         """
 
         if format_type == DIFF_TYPE_HTML:
             params = dict()
             params['post'] = self
             if PARAMETER_TRUNCATED in request.GET:
                 params[PARAMETER_TRUNCATED] = True
 
             return render_to_string('boards/api_post.html', params)
         elif format_type == DIFF_TYPE_JSON:
             post_json = {
                 'id': self.id,
                 'title': self.title,
                 'text': self._text_rendered,
                 }
             if self.images.exists():
                 post_image = self.get_first_image()
                 post_json['image'] = post_image.image.url
                 post_json['image_preview'] = post_image.image.url_200x150
             if include_last_update:
                 post_json['bump_time'] = datetime_to_epoch(
                     self.thread_new.bump_time)
             return post_json
 
     def send_to_websocket(self, request, recursive=True):
         """
         Sends post HTML data to the thread web socket.
         """
 
         if not settings.WEBSOCKETS_ENABLED:
             return
 
         client = Client()
 
         thread = self.get_thread()
         thread_id = thread.id
         channel_name = WS_CHANNEL_THREAD + str(thread.get_opening_post_id())
         client.publish(channel_name, {
             WS_NOTIFICATION_TYPE: WS_NOTIFICATION_TYPE_NEW_POST,
         })
         client.send()
 
         logger = logging.getLogger('boards.post.websocket')
 
         logger.info('Sent notification from post #{} to channel {}'.format(
             self.id, channel_name))
 
         if recursive:
             for reply_number in re.finditer(REGEX_REPLY, self.get_raw_text()):
                 post_id = reply_number.group(1)
                 ref_post = Post.objects.filter(id=post_id)[0]
 
                 # If post is in this thread, its thread was already notified.
                 # Otherwise, notify its thread separately.
                 if ref_post.thread_new_id != thread_id:
                     ref_post.send_to_websocket(request, recursive=False)
 
     def save(self, force_insert=False, force_update=False, using=None,
              update_fields=None):
         self._text_rendered = bbcode_extended(self.get_raw_text())
 
         super().save(force_insert, force_update, using, update_fields)
 
     def get_text(self) -> str:
         return self._text_rendered
 
     def get_raw_text(self) -> str:
         return self.text