Added id-type attribute to attachments to specify a type of hash (currently only md5). Added management command to delete global id caches

boards/management/commands/invalidate_sync_cache.py

@@ -0,0 +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.all():
+            if global_id.is_local() and global_id.content is not None:
+                global_id.content = None
+                global_id.save()
+                count += 1
+        print('Invalidated {} caches.'.format(count))

boards/models/__init__.py

@@ -1,14 +1,14 b''
 STATUS_ACTIVE = 'active'
 STATUS_BUMPLIMIT = 'bumplimit'
 STATUS_ARCHIVE = 'archived'
 
 
+from boards.models.sync_key import KeyPair
 from boards.models.signature import GlobalId, Signature
-from boards.models.sync_key import KeyPair
 from boards.models.image import PostImage
 from boards.models.attachment import Attachment
 from boards.models.thread import Thread
 from boards.models.post import Post
 from boards.models.tag import Tag
 from boards.models.user import Ban
 from boards.models.banner import Banner

boards/models/post/sync.py

@@ -1,276 +1,280 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 {}'
 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'
 
 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)
 
                 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)
                 signatures = SyncManager._verify_model(content_str, 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.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_content.find(TAG_TRIPCODE).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 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)
         else:
             raise SyncException(EXCEPTION_NODE.format(tag_status.text))
 
     @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(content_str, 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)
 
             if not KeyPair.objects.verify(signature, content_str):
                 raise SyncException(EXCEPTION_SIGNATURE.format(content_str))
 
             signatures.append(signature)
 
         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

@@ -1,144 +1,150 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_PULL = 'pull'
 
 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_pull(self):
         """
         Form a pull request from a list of ModelId objects.
         """
 
         request = et.Element(TAG_REQUEST)
         request.set(ATTR_TYPE, TYPE_PULL)
         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()
+
 
 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')