Do not run validation on files while performing sync

boards/models/attachment/downloaders.py

@@ -1,101 +1,101 b''
 import re
 
 import requests
 from django.core.files.uploadedfile import TemporaryUploadedFile
 from pytube import YouTube
 
 from boards.utils import validate_file_size
 
 YOUTUBE_VIDEO_FORMAT = 'webm'
 
 HTTP_RESULT_OK = 200
 
 HEADER_CONTENT_LENGTH = 'content-length'
 HEADER_CONTENT_TYPE = 'content-type'
 
 FILE_DOWNLOAD_CHUNK_BYTES = 200000
 
 REGEX_YOUTUBE_URL = re.compile(r'https?://((www\.)?youtube\.com/watch\?v=|youtu.be/)[-\w]+')
 REGEX_MAGNET = re.compile(r'magnet:\?xt=urn:(btih:)?[a-z0-9]{20,50}.*')
 
 TYPE_URL_ONLY = (
     'application/xhtml+xml',
     'text/html',
 )
 
 
 class Downloader:
     @staticmethod
     def handles(url: str) -> bool:
         return True
 
     @staticmethod
-    def download(url: str):
+    def download(url: str, validate):
         # Verify content headers
         response_head = requests.head(url, verify=False)
         content_type = response_head.headers[HEADER_CONTENT_TYPE].split(';')[0]
-        if content_type in TYPE_URL_ONLY:
+        if validate and content_type in TYPE_URL_ONLY:
             return None
 
         length_header = response_head.headers.get(HEADER_CONTENT_LENGTH)
-        if length_header:
+        if validate and length_header:
             length = int(length_header)
             validate_file_size(length)
         # Get the actual content into memory
         response = requests.get(url, verify=False, stream=True)
 
         if response.status_code == HTTP_RESULT_OK:
             # Download file, stop if the size exceeds limit
             size = 0
 
             # Set a dummy file name that will be replaced
             # anyway, just keep the valid extension
             filename = 'file.' + content_type.split('/')[1]
 
             file = TemporaryUploadedFile(filename, content_type, 0, None, None)
             for chunk in response.iter_content(FILE_DOWNLOAD_CHUNK_BYTES):
                 size += len(chunk)
                 validate_file_size(size)
                 file.write(chunk)
 
             return file
 
 
 class YouTubeDownloader(Downloader):
     @staticmethod
-    def download(url: str):
+    def download(url: str, validate):
         yt = YouTube()
         yt.from_url(url)
         videos = yt.filter(YOUTUBE_VIDEO_FORMAT)
         if len(videos) > 0:
             video = videos[0]
             return Downloader.download(video.url)
 
     @staticmethod
     def handles(url: str) -> bool:
         return REGEX_YOUTUBE_URL.match(url) is not None
 
 
 class NothingDownloader(Downloader):
     @staticmethod
     def handles(url: str) -> bool:
         return REGEX_MAGNET.match(url)
 
     @staticmethod
-    def download(url: str):
+    def download(url: str, validate):
         return None
 
 
 DOWNLOADERS = (
     YouTubeDownloader,
     NothingDownloader,
     Downloader,
 )
 
 
-def download(url):
+def download(url, validate=True):
     for downloader in DOWNLOADERS:
         if downloader.handles(url):
-            return downloader.download(url)
+            return downloader.download(url, validate=validate)
     raise Exception('No downloader supports this URL.')
 

boards/models/post/sync.py

@@ -1,331 +1,331 b''
 import xml.etree.ElementTree as et
 import logging
 
 from boards.abstracts.exceptions import SyncException
 from boards.models import KeyPair, GlobalId, Signature, Post, Tag
 from boards.models.attachment.downloaders import download
 from boards.utils import get_file_mimetype, get_file_hash
 from django.db import transaction
 
 EXCEPTION_NODE = 'Sync node returned an error: {}.'
 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'
 ID_TYPE_URL = 'url'
 
 STATUS_SUCCESS = 'success'
 
 
 logger = logging.getLogger('boards.sync')
 
 
 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
 
             attachments = post.attachments.all()
             if global_id.content:
                 model.append(et.fromstring(global_id.content))
                 if len(attachments) > 0:
                     internal_attachments = False
                     for attachment in attachments:
                         if attachment.is_internal():
                             internal_attachments = True
                             break
 
                     if internal_attachments:
                         attachment_refs = et.SubElement(model, TAG_ATTACHMENT_REFS)
                         for file in attachments:
                             SyncManager._attachment_to_xml(
                                 None, attachment_refs, file)
             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(attachments) > 0:
                     attachments_tag = et.SubElement(content_tag, TAG_ATTACHMENTS)
                     
                     internal_attachments = False
                     for attachment in attachments:
                         if attachment.is_internal():
                             internal_attachments = True
                             break
 
                     if internal_attachments:
                         attachment_refs = et.SubElement(model, TAG_ATTACHMENT_REFS)
                     else:
                         attachment_refs = None
 
                     for file in attachments:
                         SyncManager._attachment_to_xml(
                             attachments_tag, attachment_refs, file)
                 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
     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:
                 SyncManager.parse_post(tag_model, hostname)
         else:
             raise SyncException(EXCEPTION_NODE.format(tag_status.text))
 
     @staticmethod
     @transaction.atomic
     def parse_post(tag_model, hostname):
         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)
 
         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:
                     logger.debug('No thread exists for post {}'.format(global_id))
             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 = []
             urls = []
             tag_attachments = tag_content.find(TAG_ATTACHMENTS) or list()
             tag_refs = tag_model.find(TAG_ATTACHMENT_REFS)
             for attachment in tag_attachments:
                 if attachment.get(ATTR_ID_TYPE) == ID_TYPE_URL:
                     urls.append(attachment.text)
                 else:
                     tag_ref = tag_refs.find("{}[@ref='{}']".format(
                         TAG_ATTACHMENT_REF, attachment.text))
                     url = tag_ref.get(ATTR_URL)
-                    attached_file = download(hostname + url)
+                    attached_file = download(hostname + url, validate=False)
                     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)
 
             if is_old:
                 post = global_id.post
                 Post.objects.update_post(
                     post, title=title, text=text, pub_time=pub_time,
                     tags=tags, files=files, file_urls=urls,
                     tripcode=tripcode, version=version)
                 logger.debug('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,
                     file_urls=urls, tripcode=tripcode,
                     version=version)
                 logger.debug('Parsed new post {}'.format(global_id))
 
     @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, attachment):
         if tag_attachments is not None:
             attachment_tag = et.SubElement(tag_attachments, TAG_ATTACHMENT)
             if attachment.is_internal():
                 mimetype = get_file_mimetype(attachment.file.file)
                 attachment_tag.set(ATTR_MIMETYPE, mimetype)
                 attachment_tag.set(ATTR_ID_TYPE, ID_TYPE_MD5)
                 attachment_tag.text = attachment.hash
             else:
                 attachment_tag.set(ATTR_ID_TYPE, ID_TYPE_URL)
                 attachment_tag.text = attachment.url
 
         if tag_refs is not None:
             attachment_ref = et.SubElement(tag_refs, TAG_ATTACHMENT_REF)
             attachment_ref.set(ATTR_REF, attachment.hash)
             attachment_ref.set(ATTR_URL, attachment.file.url)