Convert \r\n and \r to \n in the post text used in sync

boards/models/post/__init__.py

@@ -1,454 +1,455 b''
 import logging
 import re
 import uuid
 
 from django.core.exceptions import ObjectDoesNotExist
 from django.core.urlresolvers import reverse
 from django.db import models
 from django.db.models import TextField, QuerySet
 from django.template.defaultfilters import truncatewords, striptags
 from django.template.loader import render_to_string
 from django.utils import timezone
 from django.dispatch import receiver
 from django.db.models.signals import pre_save, post_save
 
 from boards import settings
 from boards.abstracts.tripcode import Tripcode
 from boards.mdx_neboard import get_parser
 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.models.user import Notification
 
 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)
 
     tripcode = models.CharField(max_length=50, blank=True, default='')
     opening = models.BooleanField(db_index=True)
     hidden = models.BooleanField(default=False)
 
     def __str__(self):
         return 'P#{}/{}'.format(self.id, self.get_title())
 
     def get_referenced_posts(self):
         threads = self.get_threads().all()
         return self.referenced_posts.filter(threads__in=threads)\
             .order_by('pub_time').distinct().all()
 
     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_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 delete(self, using=None):
         """
         Deletes all post images and the post itself.
         """
 
         for image in self.images.all():
             image_refs_count = image.post_images.count()
             if image_refs_count == 1:
                 image.delete()
 
         for attachment in self.attachments.all():
             attachment_refs_count = attachment.attachment_posts.count()
             if attachment_refs_count == 1:
                 attachment.delete()
 
         if self.global_id:
             self.global_id.delete()
 
         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_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()):
             absolute_post_id = str(Post.objects.get(id=post_id).global_id)
             replacements[post_id] = absolute_post_id
 
         text = self.get_raw_text()
         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 get_absolute_id(self) -> str:
         """
         If the post has many threads, shows its main thread OP id in the post
         ID.
         """
 
         if self.get_threads().count() > 1:
             return '{}/{}'.format(self.get_thread().get_opening_post_id(), self.id)
         else:
             return str(self.id)
 
 
     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
 
 
 # SIGNALS (Maybe move to other module?)
 @receiver(post_save, sender=Post)
 def connect_replies(instance, **kwargs):
     for reply_number in re.finditer(REGEX_REPLY, instance.get_raw_text()):
         post_id = reply_number.group(1)
 
         try:
             referenced_post = Post.objects.get(id=post_id)
 
             referenced_post.referenced_posts.add(instance)
             referenced_post.last_edit_time = instance.pub_time
             referenced_post.build_refmap()
             referenced_post.save(update_fields=['refmap', 'last_edit_time'])
         except ObjectDoesNotExist:
             pass
 
 
 @receiver(post_save, sender=Post)
 def connect_notifications(instance, **kwargs):
     for reply_number in re.finditer(REGEX_NOTIFICATION, instance.get_raw_text()):
         user_name = reply_number.group(1).lower()
         Notification.objects.get_or_create(name=user_name, post=instance)
 
 
 @receiver(pre_save, sender=Post)
 def preparse_text(instance, **kwargs):
     instance._text_rendered = get_parser().parse(instance.get_raw_text())

boards/models/post/sync.py

@@ -1,201 +1,201 b''
 import xml.etree.ElementTree as et
 from django.db import transaction
 from boards.models import KeyPair, GlobalId, Signature, Post, Tag
 
 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'
 
 TYPE_GET = 'get'
 
 ATTR_VERSION = 'version'
 ATTR_TYPE = 'type'
 ATTR_NAME = 'name'
 ATTR_VALUE = 'value'
 ATTR_MIMETYPE = 'mimetype'
 ATTR_KEY = 'key'
 
 STATUS_SUCCESS = 'success'
 
 
 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')
 
             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)
             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())
 
             signatures_tag = et.SubElement(model, TAG_SIGNATURES)
             post_signatures = post.global_id.signature_set.all()
             if post_signatures:
                 signatures = post_signatures
                 # TODO Adding signature to a post is not yet added. For now this
                 # block is useless
             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)
                 signature = Signature(
                     key_type=key.key_type,
                     key=key.public_key,
-                    signature=key.sign(et.tostring(content_tag, ENCODING_UNICODE)),
+                    signature=key.sign(et.tostring(content_tag, encoding=ENCODING_UNICODE)),
                     global_id=post.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):
         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)
 
                 signatures = SyncManager._verify_model(tag_content, tag_model)
 
                 tag_id = tag_content.find(TAG_ID)
                 global_id, exists = GlobalId.from_xml_element(tag_id)
 
                 if exists:
                     print('Post with same ID already exists')
                 else:
                     global_id.save()
                     for signature in signatures:
                         signature.global_id = global_id
                         signature.save()
 
                     title = tag_content.find(TAG_TITLE).text
                     text = tag_content.find(TAG_TEXT).text
                     pub_time = tag_content.find(TAG_PUB_TIME).text
 
                     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 Exception('Load the OP first')
                     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
 
                     # TODO Get images
 
                     post = Post.objects.import_post(
                         title=title, text=text, pub_time=pub_time,
                         opening_post=opening_post, tags=tags,
                         global_id=global_id)
         else:
             # TODO Throw an exception?
             pass
 
     @staticmethod
     def generate_response_pull():
         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.all():
             tag_id = et.SubElement(models, TAG_ID)
             post.global_id.to_xml_element(tag_id)
 
         return et.tostring(response, ENCODING_UNICODE)
 
     @staticmethod
     def _verify_model(tag_content, tag_model):
         """
         Verifies all signatures for a single model.
         """
 
         signatures = []
 
         tag_signatures = tag_model.find(TAG_SIGNATURES)
         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)
 
             signature = Signature(key_type=signature_type,
                                   key=signature_key,
                                   signature=signature_value)
 
             content = et.tostring(tag_content, ENCODING_UNICODE)
 
             if not KeyPair.objects.verify(
                     signature, content):
                 raise Exception('Invalid model signature for {}'.format(content))
 
             signatures.append(signature)
 
         return signatures

boards/tests/test_sync.py

@@ -1,101 +1,102 b''
 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',
+        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>'
             '</content>' % (
                 post.global_id.key,
                 post.global_id.local_id,
                 post.global_id.key_type,
                 post.title,
-                post.get_raw_text(),
+                post.get_sync_text(),
                 post.get_thread().get_tags().first().name,
                 post.get_pub_time_str(),
             ) in response,
             'Wrong response generated for the GET request.')
 
         post.delete()
         key.delete()
 
         KeyPair.objects.generate_key(primary=True)
 
         SyncManager.parse_response_get(response)
         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)
         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>'
             '</content>' % (
                 post.global_id.key,
                 post.global_id.local_id,
                 post.global_id.key_type,
                 post.title,
-                post.get_raw_text(),
+                post.get_sync_text(),
                 post.get_thread().get_tags().first().name,
                 post.get_pub_time_str(),
             ) in response,
             'Wrong response generated for the GET request.')