Process updated posts from sync server

boards/management/commands/invalidate_sync_cache.py

@@ -1,22 +1,20 b''
 from django.core.management import BaseCommand
 from django.db import transaction
 
 from boards.models import GlobalId
 
 __author__ = 'neko259'
 
 
 class Command(BaseCommand):
     help = 'Removes local global ID cache'
 
     @transaction.atomic
     def handle(self, *args, **options):
         count = 0
         for global_id in GlobalId.objects.exclude(content__isnull=True).exclude(
                 content=''):
             if global_id.is_local():
-                global_id.content = None
-                global_id.save()
-                global_id.signature_set.all().delete()
+                global_id.clear_cache()
                 count += 1
         print('Invalidated {} caches.'.format(count))

boards/management/commands/sync_with_server.py

@@ -1,88 +1,90 b''
 import re
 import xml.etree.ElementTree as ET
 
 import httplib2
 from django.core.management import BaseCommand
 
 from boards.models import GlobalId
 from boards.models.post.sync import SyncManager, TAG_ID, TAG_VERSION
 
 __author__ = 'neko259'
 
 
 REGEX_GLOBAL_ID = re.compile(r'(\w+)::([\w\+/]+)::(\d+)')
 
 
 class Command(BaseCommand):
     help = 'Send a sync or get request to the server.'
 
     def add_arguments(self, parser):
         parser.add_argument('url', type=str, help='Server root url')
         parser.add_argument('--global-id', type=str, default='',
                             help='Post global ID')
         parser.add_argument('--split-query', type=int,
                             help='Split GET query into separate by the given'
                                  ' number of posts in one')
 
     def handle(self, *args, **options):
         url = options.get('url')
 
         list_url = url + 'api/sync/list/'
         get_url = url + 'api/sync/get/'
         file_url = url[:-1]
 
         global_id_str = options.get('global_id')
         if global_id_str:
             match = REGEX_GLOBAL_ID.match(global_id_str)
             if match:
                 key_type = match.group(1)
                 key = match.group(2)
                 local_id = match.group(3)
 
                 global_id = GlobalId(key_type=key_type, key=key,
                                      local_id=local_id)
 
                 xml = GlobalId.objects.generate_request_get([global_id])
                 h = httplib2.Http()
                 response, content = h.request(get_url, method="POST", body=xml)
 
                 SyncManager.parse_response_get(content, file_url)
             else:
                 raise Exception('Invalid global ID')
         else:
-            print('Running LIST request')
+            print('Running LIST request...')
             h = httplib2.Http()
             xml = GlobalId.objects.generate_request_list()
             response, content = h.request(list_url, method="POST", body=xml)
+            print('Processing response...')
 
             root = ET.fromstring(content)
             status = root.findall('status')[0].text
             if status == 'success':
                 ids_to_sync = list()
 
                 models = root.findall('models')[0]
                 for model in models:
                     tag_id = model.find(TAG_ID)
                     global_id, exists = GlobalId.from_xml_element(tag_id)
                     tag_version = model.find(TAG_VERSION)
                     if tag_version is not None:
                         version = int(tag_version.text) or 1
                     else:
                         version = 1
                     if not exists or global_id.post.version < version:
                         ids_to_sync.append(global_id)
                 print('Starting sync...')
 
                 if len(ids_to_sync) > 0:
                     limit = options.get('split_query', len(ids_to_sync))
                     for offset in range(0, len(ids_to_sync), limit):
                         xml = GlobalId.objects.generate_request_get(ids_to_sync[offset:offset+limit])
                         h = httplib2.Http()
-                        print('Running GET request')
+                        print('Running GET request...')
                         response, content = h.request(get_url, method="POST", body=xml)
+                        print('Processing response...')
 
                         SyncManager.parse_response_get(content, file_url)
                 else:
                     print('Nothing to get, everything synced')
             else:
                 raise Exception('Invalid response status')

boards/models/post/__init__.py

@@ -1,396 +1,397 b''
 import uuid
 
 import re
 from boards import settings
 from boards.abstracts.tripcode import Tripcode
 from boards.models import PostImage, Attachment, KeyPair, GlobalId
 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'
 
 IMAGE_THUMB_SIZE = (200, 150)
 
 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()
     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='post_images', db_index=True)
     attachments = models.ManyToManyField(Attachment, null=True, blank=True,
                                          related_name='attachment_posts')
 
     poster_ip = models.GenericIPAddressField()
 
     # TODO This field can be removed cause UID is used for update now
     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) -> 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.
         """
 
         post_urls = [refpost.get_link_view()
                      for refpost in self.referenced_posts.all()]
 
         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
 
         if url is None:
             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(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.
         """
 
         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 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) -> PostImage:
         return self.images.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.content = None
-            global_id.save()
-            global_id.signature_set.all().delete()
+            global_id.clear_cache()

boards/models/post/manager.py

@@ -1,161 +1,184 b''
 import logging
 
 from datetime import datetime, timedelta, date
 from datetime import time as dtime
 
 from django.db import models, transaction
 from django.utils import timezone
 
 import boards
 
 from boards.models.user import Ban
 from boards.mdx_neboard import Parser
 from boards.models import PostImage, Attachment
 from boards import utils
 
 __author__ = 'neko259'
 
 IMAGE_TYPES = (
     'jpeg',
     'jpg',
     'png',
     'bmp',
     'gif',
 )
 
 POSTS_PER_DAY_RANGE = 7
 NO_IP = '0.0.0.0'
 
 
 class PostManager(models.Manager):
     @transaction.atomic
     def create_post(self, title: str, text: str, file=None, thread=None,
                     ip=NO_IP, tags: list=None, opening_posts: list=None,
                     tripcode='', monochrome=False, images=[]):
         """
         Creates new post
         """
 
         if thread is not None and thread.is_archived():
             raise Exception('Cannot post into an archived thread')
 
         if not utils.is_anonymous_mode():
             is_banned = Ban.objects.filter(ip=ip).exists()
         else:
             is_banned = False
 
         # TODO Raise specific exception and catch it in the views
         if is_banned:
             raise Exception("This user is banned")
 
         if not tags:
             tags = []
         if not opening_posts:
             opening_posts = []
 
         posting_time = timezone.now()
         new_thread = False
         if not thread:
             thread = boards.models.thread.Thread.objects.create(
                 bump_time=posting_time, last_edit_time=posting_time,
                     monochrome=monochrome)
             list(map(thread.tags.add, tags))
             boards.models.thread.Thread.objects.process_oldest_threads()
             new_thread = True
 
         pre_text = Parser().preparse(text)
 
         post = self.create(title=title,
                            text=pre_text,
                            pub_time=posting_time,
                            poster_ip=ip,
                            thread=thread,
                            last_edit_time=posting_time,
                            tripcode=tripcode,
                            opening=new_thread)
         post.threads.add(thread)
 
         logger = logging.getLogger('boards.post.create')
 
         logger.info('Created post [{}] with text [{}] by {}'.format(post,
             post.get_text(),post.poster_ip))
 
         if file:
             self._add_file_to_post(file, post)
         for image in images:
             post.images.add(image)
 
         post.connect_threads(opening_posts)
         post.set_global_id()
 
         # Thread needs to be bumped only when the post is already created
         if not new_thread:
             thread.last_edit_time = posting_time
             thread.bump()
             thread.save()
 
         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()
 
     @utils.cached_result()
     def get_posts_per_day(self) -> float:
         """
         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)
 
         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
 
         return ppd
 
     @transaction.atomic
     def import_post(self, title: str, text: str, pub_time: str, global_id,
                     opening_post=None, tags=list(), files=list(),
-                    tripcode=None):
+                    tripcode=None, version=1):
         is_opening = opening_post is None
         if is_opening:
             thread = boards.models.thread.Thread.objects.create(
                 bump_time=pub_time, last_edit_time=pub_time)
             list(map(thread.tags.add, tags))
         else:
             thread = opening_post.get_thread()
 
-        post = self.create(title=title, text=text,
+        post = self.create(title=title,
+                           text=text,
                            pub_time=pub_time,
                            poster_ip=NO_IP,
                            last_edit_time=pub_time,
                            global_id=global_id,
                            opening=is_opening,
-                           thread=thread, tripcode=tripcode)
+                           thread=thread,
+                           tripcode=tripcode,
+                           version=version)
 
-        # TODO Add files
         for file in files:
             self._add_file_to_post(file, post)
 
         post.threads.add(thread)
 
+    @transaction.atomic
+    def update_post(self, post, title: str, text: str, pub_time: str,
+                    tags=list(), files=list(), tripcode=None, version=1):
+        post.title = title
+        post.text = text
+        post.pub_time = pub_time
+        post.tripcode = tripcode
+        post.version = version
+        post.save()
+
+        post.clear_cache()
+
+        post.images.clear()
+        post.attachments.clear()
+        for file in files:
+            self._add_file_to_post(file, post)
+
+        thread = post.get_thread()
+        thread.tags.clear()
+        list(map(thread.tags.add, tags))
+
     def _add_file_to_post(self, file, post):
         file_type = file.name.split('.')[-1].lower()
         if file_type in IMAGE_TYPES:
             post.images.add(PostImage.objects.create_with_hash(file))
         else:
             post.attachments.add(Attachment.objects.create_with_hash(file))

boards/models/post/sync.py

@@ -1,299 +1,310 b''
 import xml.etree.ElementTree as et
 
 from boards.models.attachment.downloaders import download
 from boards.utils import get_file_mimetype, get_file_hash
 from django.db import transaction
 from boards.models import KeyPair, GlobalId, Signature, Post, Tag
 
 EXCEPTION_NODE = 'Sync node returned an error: {}.'
 EXCEPTION_OP = 'Load the OP first.'
 EXCEPTION_DOWNLOAD = 'File was not downloaded.'
 EXCEPTION_HASH = 'File hash does not match attachment hash.'
 EXCEPTION_SIGNATURE = 'Invalid model signature for {}.'
 EXCEPTION_AUTHOR_SIGNATURE = 'Model {} has no author signature.'
 ENCODING_UNICODE = 'unicode'
 
 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'
 TAG_TAGS = 'tags'
 TAG_TAG = 'tag'
 TAG_ATTACHMENT_REFS = 'attachment-refs'
 TAG_ATTACHMENT_REF = 'attachment-ref'
 TAG_TRIPCODE = 'tripcode'
 TAG_VERSION = 'version'
 
 TYPE_GET = 'get'
 
 ATTR_VERSION = 'version'
 ATTR_TYPE = 'type'
 ATTR_NAME = 'name'
 ATTR_VALUE = 'value'
 ATTR_MIMETYPE = 'mimetype'
 ATTR_KEY = 'key'
 ATTR_REF = 'ref'
 ATTR_URL = 'url'
 ATTR_ID_TYPE = 'id-type'
 
 ID_TYPE_MD5 = 'md5'
 
 STATUS_SUCCESS = 'success'
 
 
 class SyncException(Exception):
     pass
 
 
 class SyncManager:
     @staticmethod
     def generate_response_get(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')
 
             global_id = post.global_id
 
             images = post.images.all()
             attachments = post.attachments.all()
             if global_id.content:
                 model.append(et.fromstring(global_id.content))
                 if len(images) > 0 or len(attachments) > 0:
                     attachment_refs = et.SubElement(model, TAG_ATTACHMENT_REFS)
                     for image in images:
                         SyncManager._attachment_to_xml(
                             None, attachment_refs, image.image.file,
                             image.hash, image.image.url)
                     for file in attachments:
                         SyncManager._attachment_to_xml(
                             None, attachment_refs, file.file.file,
                             file.hash, file.file.url)
             else:
                 content_tag = et.SubElement(model, TAG_CONTENT)
 
                 tag_id = et.SubElement(content_tag, TAG_ID)
                 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)
                 text.text = post.get_sync_text()
 
                 thread = post.get_thread()
                 if post.is_opening():
                     tag_tags = et.SubElement(content_tag, TAG_TAGS)
                     for tag in thread.get_tags():
                         tag_tag = et.SubElement(tag_tags, TAG_TAG)
                         tag_tag.text = tag.name
                 else:
                     tag_thread = et.SubElement(content_tag, TAG_THREAD)
                     thread_id = et.SubElement(tag_thread, TAG_ID)
                     thread.get_opening_post().global_id.to_xml_element(thread_id)
 
                 pub_time = et.SubElement(content_tag, TAG_PUB_TIME)
                 pub_time.text = str(post.get_pub_time_str())
 
                 if post.tripcode:
                     tripcode = et.SubElement(content_tag, TAG_TRIPCODE)
                     tripcode.text = post.tripcode
 
                 if len(images) > 0 or len(attachments) > 0:
                     attachments_tag = et.SubElement(content_tag, TAG_ATTACHMENTS)
                     attachment_refs = et.SubElement(model, TAG_ATTACHMENT_REFS)
 
                     for image in images:
                         SyncManager._attachment_to_xml(
                             attachments_tag, attachment_refs, image.image.file,
                             image.hash, image.image.url)
                     for file in attachments:
                         SyncManager._attachment_to_xml(
                             attachments_tag, attachment_refs, file.file.file,
                             file.hash, file.file.url)
                 version_tag = et.SubElement(content_tag, TAG_VERSION)
                 version_tag.text = str(post.version)
 
                 global_id.content = et.tostring(content_tag, ENCODING_UNICODE)
                 global_id.save()
 
             signatures_tag = et.SubElement(model, TAG_SIGNATURES)
             post_signatures = global_id.signature_set.all()
             if post_signatures:
                 signatures = post_signatures
             else:
                 key = KeyPair.objects.get(public_key=global_id.key)
                 signature = Signature(
                     key_type=key.key_type,
                     key=key.public_key,
                     signature=key.sign(global_id.content),
                     global_id=global_id,
                 )
                 signature.save()
                 signatures = [signature]
             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)
                 signature_tag.set(ATTR_KEY, signature.key)
 
         return et.tostring(response, ENCODING_UNICODE)
 
     @staticmethod
     @transaction.atomic
     def parse_response_get(response_xml, hostname):
         tag_root = et.fromstring(response_xml)
         tag_status = tag_root.find(TAG_STATUS)
         if STATUS_SUCCESS == tag_status.text:
             tag_models = tag_root.find(TAG_MODELS)
             for tag_model in tag_models:
                 tag_content = tag_model.find(TAG_CONTENT)
 
                 content_str = et.tostring(tag_content, ENCODING_UNICODE)
 
                 tag_id = tag_content.find(TAG_ID)
                 global_id, exists = GlobalId.from_xml_element(tag_id)
                 signatures = SyncManager._verify_model(global_id, content_str, tag_model)
 
-                if exists:
-                    print('Post with same ID already exists')
+                version = int(tag_content.find(TAG_VERSION).text)
+                is_old = exists and global_id.post.version < version
+                if exists and not is_old:
+                    print('Post with same ID exists and is up to date.')
                 else:
                     global_id.content = content_str
                     global_id.save()
                     for signature in signatures:
                         signature.global_id = global_id
                         signature.save()
 
                     title = tag_content.find(TAG_TITLE).text or ''
                     text = tag_content.find(TAG_TEXT).text or ''
                     pub_time = tag_content.find(TAG_PUB_TIME).text
                     tripcode_tag = tag_content.find(TAG_TRIPCODE)
                     if tripcode_tag is not None:
                         tripcode = tripcode_tag.text or ''
                     else:
                         tripcode = ''
 
                     thread = tag_content.find(TAG_THREAD)
                     tags = []
                     if thread:
                         thread_id = thread.find(TAG_ID)
                         op_global_id, exists = GlobalId.from_xml_element(thread_id)
                         if exists:
                             opening_post = Post.objects.get(global_id=op_global_id)
                         else:
                             raise SyncException(EXCEPTION_OP)
                     else:
                         opening_post = None
                         tag_tags = tag_content.find(TAG_TAGS)
                         for tag_tag in tag_tags:
                             tag, created = Tag.objects.get_or_create(
                                 name=tag_tag.text)
                             tags.append(tag)
 
                     # TODO Check that the replied posts are already present
                     # before adding new ones
 
                     files = []
                     tag_attachments = tag_content.find(TAG_ATTACHMENTS) or list()
                     tag_refs = tag_model.find(TAG_ATTACHMENT_REFS)
                     for attachment in tag_attachments:
                         tag_ref = tag_refs.find("{}[@ref='{}']".format(
                             TAG_ATTACHMENT_REF, attachment.text))
                         url = tag_ref.get(ATTR_URL)
                         attached_file = download(hostname + url)
                         if attached_file is None:
                             raise SyncException(EXCEPTION_DOWNLOAD)
 
                         hash = get_file_hash(attached_file)
                         if hash != attachment.text:
                             raise SyncException(EXCEPTION_HASH)
 
                         files.append(attached_file)
 
-                    Post.objects.import_post(
-                        title=title, text=text, pub_time=pub_time,
-                        opening_post=opening_post, tags=tags,
-                        global_id=global_id, files=files, tripcode=tripcode)
-                print('Parsed post {}'.format(global_id))
+                    if is_old:
+                        post = global_id.post
+                        Post.objects.update_post(
+                            post, title=title, text=text, pub_time=pub_time,
+                            tags=tags, files=files, tripcode=tripcode,
+                            version=version)
+                        print('Parsed updated post {}'.format(global_id))
+                    else:
+                        Post.objects.import_post(
+                            title=title, text=text, pub_time=pub_time,
+                            opening_post=opening_post, tags=tags,
+                            global_id=global_id, files=files, tripcode=tripcode,
+                            version=version)
+                        print('Parsed new post {}'.format(global_id))
         else:
             raise SyncException(EXCEPTION_NODE.format(tag_status.text))
 
     @staticmethod
     def generate_response_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 Post.objects.prefetch_related('global_id').all():
             tag_model = et.SubElement(models, TAG_MODEL)
             tag_id = et.SubElement(tag_model, TAG_ID)
             post.global_id.to_xml_element(tag_id)
             tag_version = et.SubElement(tag_model, TAG_VERSION)
             tag_version.text = str(post.version)
 
         return et.tostring(response, ENCODING_UNICODE)
 
     @staticmethod
     def _verify_model(global_id, content_str, tag_model):
         """
         Verifies all signatures for a single model.
         """
 
         signatures = []
 
         tag_signatures = tag_model.find(TAG_SIGNATURES)
         has_author_signature = False
         for tag_signature in tag_signatures:
             signature_type = tag_signature.get(ATTR_TYPE)
             signature_value = tag_signature.get(ATTR_VALUE)
             signature_key = tag_signature.get(ATTR_KEY)
 
             if global_id.key_type == signature_type and\
                     global_id.key == signature_key:
                 has_author_signature = True
 
             signature = Signature(key_type=signature_type,
                                   key=signature_key,
                                   signature=signature_value)
 
             if not KeyPair.objects.verify(signature, content_str):
                 raise SyncException(EXCEPTION_SIGNATURE.format(content_str))
 
             signatures.append(signature)
         if not has_author_signature:
             raise SyncException(EXCEPTION_AUTHOR_SIGNATURE.format(content_str))
 
         return signatures
 
     @staticmethod
     def _attachment_to_xml(tag_attachments, tag_refs, file, hash, url):
         if tag_attachments is not None:
             mimetype = get_file_mimetype(file)
             attachment = et.SubElement(tag_attachments, TAG_ATTACHMENT)
             attachment.set(ATTR_MIMETYPE, mimetype)
             attachment.set(ATTR_ID_TYPE, ID_TYPE_MD5)
             attachment.text = hash
 
         attachment_ref = et.SubElement(tag_refs, TAG_ATTACHMENT_REF)
         attachment_ref.set(ATTR_REF, hash)
         attachment_ref.set(ATTR_URL, url)

boards/models/signature.py

@@ -1,150 +1,158 b''
 import xml.etree.ElementTree as et
 from django.db import models
 from boards.models import KeyPair
 
 
 TAG_MODEL = 'model'
 TAG_REQUEST = 'request'
 TAG_ID = 'id'
 
 TYPE_GET = 'get'
 TYPE_LIST = 'list'
 
 ATTR_VERSION = 'version'
 ATTR_TYPE = 'type'
 ATTR_NAME = 'name'
 
 ATTR_KEY = 'key'
 ATTR_KEY_TYPE = 'type'
 ATTR_LOCAL_ID = 'local-id'
 
 
 class GlobalIdManager(models.Manager):
     def generate_request_get(self, global_id_list: list):
         """
         Form a get request from a list of ModelId objects.
         """
 
         request = et.Element(TAG_REQUEST)
         request.set(ATTR_TYPE, TYPE_GET)
         request.set(ATTR_VERSION, '1.0')
 
         model = et.SubElement(request, TAG_MODEL)
         model.set(ATTR_VERSION, '1.0')
         model.set(ATTR_NAME, 'post')
 
         for global_id in global_id_list:
             tag_id = et.SubElement(model, TAG_ID)
             global_id.to_xml_element(tag_id)
 
         return et.tostring(request, 'unicode')
 
     def generate_request_list(self):
         """
         Form a pull request from a list of ModelId objects.
         """
 
         request = et.Element(TAG_REQUEST)
         request.set(ATTR_TYPE, TYPE_LIST)
         request.set(ATTR_VERSION, '1.0')
 
         model = et.SubElement(request, TAG_MODEL)
         model.set(ATTR_VERSION, '1.0')
         model.set(ATTR_NAME, 'post')
 
         return et.tostring(request, 'unicode')
 
     def global_id_exists(self, global_id):
         """
         Checks if the same global id already exists in the system.
         """
 
         return self.filter(key=global_id.key,
                            key_type=global_id.key_type,
                            local_id=global_id.local_id).exists()
 
 
 class GlobalId(models.Model):
     """
     Global model ID and cache.
     Key, key type and local ID make a single global identificator of the model.
     Content is an XML cache of the model that can be passed along between nodes
     without manual serialization each time.
     """
     class Meta:
         app_label = 'boards'
 
     objects = GlobalIdManager()
 
     def __init__(self, *args, **kwargs):
         models.Model.__init__(self, *args, **kwargs)
 
         if 'key' in kwargs and 'key_type' in kwargs and 'local_id' in kwargs:
             self.key = kwargs['key']
             self.key_type = kwargs['key_type']
             self.local_id = kwargs['local_id']
 
     key = models.TextField()
     key_type = models.TextField()
     local_id = models.IntegerField()
     content = models.TextField(blank=True, null=True)
 
     def __str__(self):
         return '%s::%s::%d' % (self.key_type, self.key, self.local_id)
 
     def to_xml_element(self, element: et.Element):
         """
         Exports global id to an XML element.
         """
 
         element.set(ATTR_KEY, self.key)
         element.set(ATTR_KEY_TYPE, self.key_type)
         element.set(ATTR_LOCAL_ID, str(self.local_id))
 
     @staticmethod
     def from_xml_element(element: et.Element):
         """
         Parses XML id tag and gets global id from it.
 
         Arguments:
         element -- the XML 'id' element
 
         Returns:
         global_id -- id itself
         exists -- True if the global id was taken from database, False if it
         did not exist and was created.
         """
 
         try:
             return GlobalId.objects.get(key=element.get(ATTR_KEY),
                                         key_type=element.get(ATTR_KEY_TYPE),
                                         local_id=int(element.get(
                                             ATTR_LOCAL_ID))), True
         except GlobalId.DoesNotExist:
             return GlobalId(key=element.get(ATTR_KEY),
                             key_type=element.get(ATTR_KEY_TYPE),
                             local_id=int(element.get(ATTR_LOCAL_ID))), False
 
     def is_local(self):
         """Checks fo the ID is local model's"""
         return KeyPair.objects.filter(
                 key_type=self.key_type, public_key=self.key).exists()
 
+    def clear_cache(self):
+        """
+        Removes content cache and signatures.
+        """
+        self.content = None
+        self.save()
+        self.signature_set.all().delete()
+
 
 class Signature(models.Model):
     class Meta:
         app_label = 'boards'
 
     def __init__(self, *args, **kwargs):
         models.Model.__init__(self, *args, **kwargs)
 
         if 'key' in kwargs and 'key_type' in kwargs and 'signature' in kwargs:
             self.key_type = kwargs['key_type']
             self.key = kwargs['key']
             self.signature = kwargs['signature']
 
     key_type = models.TextField()
     key = models.TextField()
     signature = models.TextField()
 
     global_id = models.ForeignKey('GlobalId')

boards/tests/test_sync.py

@@ -1,106 +1,105 b''
+from django.test import TestCase
+
 from boards.models import KeyPair, Post, Tag
 from boards.models.post.sync import SyncManager
 from boards.tests.mocks import MockRequest
 from boards.views.sync import response_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.
         """
 
         key = KeyPair.objects.generate_key(primary=True)
         tag = Tag.objects.create(name='tag1')
         post = Post.objects.create_post(title='test_title',
                                         text='test_text\rline two',
                                         tags=[tag])
 
         request = MockRequest()
         request.body = (
             '<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)
         )
 
         response = response_get(request).content.decode()
         self.assertTrue(
             '<status>success</status>'
             '<models>'
             '<model name="post">'
             '<content>'
             '<id key="%s" local-id="%d" type="%s" />'
             '<title>%s</title>'
             '<text>%s</text>'
             '<tags><tag>%s</tag></tags>'
             '<pub-time>%s</pub-time>'
             '<version>%s</version>'
             '</content>' % (
                 post.global_id.key,
                 post.global_id.local_id,
                 post.global_id.key_type,
                 post.title,
                 post.get_sync_text(),
                 post.get_thread().get_tags().first().name,
                 post.get_pub_time_str(),
                 post.version,
             ) in response,
             'Wrong response generated for the GET request.')
 
         post.delete()
         key.delete()
 
         KeyPair.objects.generate_key(primary=True)
 
         SyncManager.parse_response_get(response, None)
         self.assertEqual(1, Post.objects.count(),
                          'Post was not created from XML response.')
 
         parsed_post = Post.objects.first()
         self.assertEqual('tag1',
                          parsed_post.get_thread().get_tags().first().name,
                          'Invalid tag was parsed.')
 
         SyncManager.parse_response_get(response, None)
         self.assertEqual(1, Post.objects.count(),
                          'The same post was imported twice.')
 
         self.assertEqual(1, parsed_post.global_id.signature_set.count(),
                          'Signature was not saved.')
 
         post = parsed_post
 
         # Trying to sync the same once more
         response = response_get(request).content.decode()
 
         self.assertTrue(
             '<status>success</status>'
             '<models>'
             '<model name="post">'
             '<content>'
             '<id key="%s" local-id="%d" type="%s" />'
             '<title>%s</title>'
             '<text>%s</text>'
             '<tags><tag>%s</tag></tags>'
             '<pub-time>%s</pub-time>'
             '<version>%s</version>'
             '</content>' % (
                 post.global_id.key,
                 post.global_id.local_id,
                 post.global_id.key_type,
                 post.title,
                 post.get_sync_text(),
                 post.get_thread().get_tags().first().name,
                 post.get_pub_time_str(),
                 post.version,
             ) in response,
             'Wrong response generated for the GET request.')