Fixed connecting replies to posts

boards/models/post.py

@@ -1,623 +1,621 b''
 from datetime import datetime, timedelta, date
 from datetime import time as dtime
 import logging
 import re
 import xml.etree.ElementTree as et
 
 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.models import PostImage, KeyPair, GlobalId, Signature
 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 import utils
 from boards.utils import datetime_to_epoch
 
 ENCODING_UNICODE = 'unicode'
 
 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+)\[/post\]')
 REGEX_GLOBAL_REPLY = re.compile(r'\[post\](\w+)::([^:]+)::(\d+)\[/post\]')
 
 TAG_MODEL = 'model'
 TAG_REQUEST = 'request'
 TAG_RESPONSE = 'response'
 TAG_ID = 'id'
 TAG_STATUS = 'status'
 TAG_MODELS = 'models'
 TAG_TITLE = 'title'
 TAG_TEXT = 'text'
 TAG_THREAD = 'thread'
 TAG_PUB_TIME = 'pub-time'
 TAG_SIGNATURES = 'signatures'
 TAG_SIGNATURE = 'signature'
 TAG_CONTENT = 'content'
 TAG_ATTACHMENTS = 'attachments'
 TAG_ATTACHMENT = 'attachment'
 
 TYPE_GET = 'get'
 
 ATTR_VERSION = 'version'
 ATTR_TYPE = 'type'
 ATTR_NAME = 'name'
 ATTR_VALUE = 'value'
 ATTR_MIMETYPE = 'mimetype'
 
 STATUS_SUCCESS = 'success'
 
 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)
 
         post.set_global_id()
 
         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)
             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()
 
     # TODO This can be moved into a post
     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)
+        for reply_number in post.get_replied_ids():
+            ref_post = self.filter(id=reply_number)
             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
 
     # TODO Make a separate sync facade?
     def generate_response_get(self, model_list: list):
         response = et.Element(TAG_RESPONSE)
 
         status = et.SubElement(response, TAG_STATUS)
         status.text = STATUS_SUCCESS
 
         models = et.SubElement(response, TAG_MODELS)
 
         for post in model_list:
             model = et.SubElement(models, TAG_MODEL)
             model.set(ATTR_NAME, 'post')
 
             content_tag = et.SubElement(model, TAG_CONTENT)
 
             tag_id = et.SubElement(content_tag, TAG_ID)
             post.global_id.to_xml_element(tag_id)
 
             title = et.SubElement(content_tag, TAG_TITLE)
             title.text = post.title
 
             text = et.SubElement(content_tag, TAG_TEXT)
             # TODO Replace local links by global ones in the text
-            text.text = post.text.raw
+            text.text = post.get_raw_text()
 
             if not post.is_opening():
                 thread = et.SubElement(content_tag, TAG_THREAD)
                 thread.text = str(post.get_thread().get_opening_post_id())
             else:
                 # TODO Output tags here
                 pass
 
             pub_time = et.SubElement(content_tag, TAG_PUB_TIME)
             pub_time.text = str(post.get_pub_time_epoch())
 
             signatures_tag = et.SubElement(model, TAG_SIGNATURES)
             post_signatures = post.signature.all()
             if post_signatures:
                 signatures = post.signatures
             else:
                 # TODO Maybe the signature can be computed only once after
                 # the post is added? Need to add some on_save signal queue
                 # and add this there.
                 key = KeyPair.objects.get(public_key=post.global_id.key)
                 signatures = [Signature(
                     key_type=key.key_type,
                     key=key.public_key,
                     signature=key.sign(et.tostring(model, ENCODING_UNICODE)),
                     )]
             for signature in signatures:
                 signature_tag = et.SubElement(signatures_tag, TAG_SIGNATURE)
                 signature_tag.set(ATTR_TYPE, signature.key_type)
                 signature_tag.set(ATTR_VALUE, signature.signature)
 
         return et.tostring(response, ENCODING_UNICODE)
 
     def parse_response_get(self, response_xml):
         tag_root = et.fromstring(response_xml)
         tag_status = tag_root[0]
         if 'success' == tag_status.text:
             tag_models = tag_root[1]
             for tag_model in tag_models:
                 tag_content = tag_model[0]
                 tag_id = tag_content[1]
                 try:
                     GlobalId.from_xml_element(tag_id, existing=True)
                     # If this post already exists, just continue
                     # TODO Compare post content and update the post if necessary
                     pass
                 except GlobalId.DoesNotExist:
                     global_id = GlobalId.from_xml_element(tag_id)
 
                     title = tag_content.find(TAG_TITLE).text
                     text = tag_content.find(TAG_TEXT).text
                     # TODO Check that the replied posts are already present
                     # before adding new ones
 
                     # TODO Pub time, thread, tags
 
                     post = Post.objects.create(title=title, text=text)
         else:
             # TODO Throw an exception?
             pass
 
     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()
 
     # Replies to the post
     referenced_posts = models.ManyToManyField('Post', symmetrical=False,
                                               null=True,
                                               blank=True, related_name='rfp+',
                                               db_index=True)
 
     # Replies map. This is built from the referenced posts list to speed up
     # page loading (no need to get all the referenced posts from the database).
     refmap = models.TextField(null=True, blank=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)
 
     # One post can be signed by many nodes that give their trust to it
     signature = models.ManyToManyField('Signature', 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()
         self.signature.all().delete()
         if self.global_id:
             self.global_id.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 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_epoch(self):
         return utils.datetime_to_epoch(self.pub_time)
 
-    # TODO Use this to connect replies
     def get_replied_ids(self):
         """
         Gets ID list of the posts that this post replies.
         """
 
-        local_replied = REGEX_REPLY.findall(self.text.raw)
+        raw_text = self.get_raw_text()
+
+        local_replied = REGEX_REPLY.findall(raw_text)
         global_replied = []
-        # TODO Similar code is used in mdx_neboard, maybe it can be extracted
-        # into a method?
-        for match in REGEX_GLOBAL_REPLY.findall(self.text.raw):
+        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):
         """
         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

boards/tests/test_sync.py

@@ -1,48 +1,48 b''
 from boards.models import KeyPair, Post
 from boards.tests.mocks import MockRequest
 from boards.views.sync import respond_get
 
 __author__ = 'neko259'
 
 
 from django.test import TestCase
 
 
 class SyncTest(TestCase):
     def test_get(self):
         """
         Forms a GET request of a post and checks the response.
         """
 
         KeyPair.objects.generate_key(primary=True)
         post = Post.objects.create_post(title='test_title', text='test_text')
 
         request = MockRequest()
         request.POST['xml'] = (
             '<request type="get" version="1.0">'
             '<model name="post" version="1.0">'
             '<id key="%s" local-id="%d" type="%s" />'
             '</model>'
             '</request>' % (post.global_id.key,
                             post.id,
                             post.global_id.key_type)
         )
 
         self.assertTrue(
             '<status>success</status>'
             '<models>'
             '<model name="post">'
             '<content>'
             '<id key="%s" local-id="%d" type="%s" />'
             '<title>%s</title>'
             '<text>%s</text>'
             '<pub-time>%d</pub-time>'
             '</content>' % (
                 post.global_id.key,
                 post.id,
                 post.global_id.key_type,
                 post.title,
-                post.text.raw,
+                post.get_raw_text(),
                 post.get_pub_time_epoch(),
             ) in respond_get(request).content.decode(),
-            'Wrong response generated for the GET request.')
  No newline at end of file
+            'Wrong response generated for the GET request.')