diff options
Diffstat (limited to 'synapse/rest')
| -rw-r--r-- | synapse/rest/client/v2_alpha/keys.py | 50 | ||||
| -rw-r--r-- | synapse/rest/media/v1/preview_url_resource.py | 86 |
2 files changed, 80 insertions, 56 deletions
diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index dc1d4d8fc6..c5ff16adf3 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -130,9 +130,7 @@ class KeyUploadServlet(RestServlet): # old access_token without an associated device_id. Either way, we # need to double-check the device is registered to avoid ending up with # keys without a corresponding device. - self.device_handler.check_device_registered( - user_id, device_id, "unknown device" - ) + self.device_handler.check_device_registered(user_id, device_id) result = yield self.store.count_e2e_one_time_keys(user_id, device_id) defer.returnValue((200, {"one_time_key_counts": result})) @@ -186,17 +184,19 @@ class KeyQueryServlet(RestServlet): ) def __init__(self, hs): + """ + Args: + hs (synapse.server.HomeServer): + """ super(KeyQueryServlet, self).__init__() - self.store = hs.get_datastore() self.auth = hs.get_auth() - self.federation = hs.get_replication_layer() - self.is_mine = hs.is_mine + self.e2e_keys_handler = hs.get_e2e_keys_handler() @defer.inlineCallbacks def on_POST(self, request, user_id, device_id): yield self.auth.get_user_by_req(request) body = parse_json_object_from_request(request) - result = yield self.handle_request(body) + result = yield self.e2e_keys_handler.query_devices(body) defer.returnValue(result) @defer.inlineCallbacks @@ -205,45 +205,11 @@ class KeyQueryServlet(RestServlet): auth_user_id = requester.user.to_string() user_id = user_id if user_id else auth_user_id device_ids = [device_id] if device_id else [] - result = yield self.handle_request( + result = yield self.e2e_keys_handler.query_devices( {"device_keys": {user_id: device_ids}} ) defer.returnValue(result) - @defer.inlineCallbacks - def handle_request(self, body): - local_query = [] - remote_queries = {} - for user_id, device_ids in body.get("device_keys", {}).items(): - user = UserID.from_string(user_id) - if self.is_mine(user): - if not device_ids: - local_query.append((user_id, None)) - else: - for device_id in device_ids: - local_query.append((user_id, device_id)) - else: - remote_queries.setdefault(user.domain, {})[user_id] = list( - device_ids - ) - results = yield self.store.get_e2e_device_keys(local_query) - - json_result = {} - for user_id, device_keys in results.items(): - for device_id, json_bytes in device_keys.items(): - json_result.setdefault(user_id, {})[device_id] = json.loads( - json_bytes - ) - - for destination, device_keys in remote_queries.items(): - remote_result = yield self.federation.query_client_keys( - destination, {"device_keys": device_keys} - ) - for user_id, keys in remote_result["device_keys"].items(): - if user_id in device_keys: - json_result[user_id] = keys - defer.returnValue((200, {"device_keys": json_result})) - class OneTimeKeyServlet(RestServlet): """ diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 74c64f1371..4060593f7f 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -29,6 +29,8 @@ from synapse.http.server import ( from synapse.util.async import ObservableDeferred from synapse.util.stringutils import is_ascii +from copy import deepcopy + import os import re import fnmatch @@ -329,20 +331,23 @@ class PreviewUrlResource(Resource): # ...or if they are within a <script/> or <style/> tag. # This is a very very very coarse approximation to a plain text # render of the page. - text_nodes = tree.xpath("//text()[not(ancestor::header | ancestor::nav | " - "ancestor::aside | ancestor::footer | " - "ancestor::script | ancestor::style)]" + - "[ancestor::body]") - text = '' - for text_node in text_nodes: - if len(text) < 500: - text += text_node + ' ' - else: - break - text = re.sub(r'[\t ]+', ' ', text) - text = re.sub(r'[\t \r\n]*[\r\n]+', '\n', text) - text = text.strip()[:500] - og['og:description'] = text if text else None + + # We don't just use XPATH here as that is slow on some machines. + + # We clone `tree` as we modify it. + cloned_tree = deepcopy(tree.find("body")) + + TAGS_TO_REMOVE = ("header", "nav", "aside", "footer", "script", "style",) + for el in cloned_tree.iter(TAGS_TO_REMOVE): + el.getparent().remove(el) + + # Split all the text nodes into paragraphs (by splitting on new + # lines) + text_nodes = ( + re.sub(r'\s+', '\n', el.text).strip() + for el in cloned_tree.iter() if el.text + ) + og['og:description'] = summarize_paragraphs(text_nodes) # TODO: delete the url downloads to stop diskfilling, # as we only ever cared about its OG @@ -450,3 +455,56 @@ class PreviewUrlResource(Resource): content_type.startswith("application/xhtml") ): return True + + +def summarize_paragraphs(text_nodes, min_size=200, max_size=500): + # Try to get a summary of between 200 and 500 words, respecting + # first paragraph and then word boundaries. + # TODO: Respect sentences? + + description = '' + + # Keep adding paragraphs until we get to the MIN_SIZE. + for text_node in text_nodes: + if len(description) < min_size: + text_node = re.sub(r'[\t \r\n]+', ' ', text_node) + description += text_node + '\n\n' + else: + break + + description = description.strip() + description = re.sub(r'[\t ]+', ' ', description) + description = re.sub(r'[\t \r\n]*[\r\n]+', '\n\n', description) + + # If the concatenation of paragraphs to get above MIN_SIZE + # took us over MAX_SIZE, then we need to truncate mid paragraph + if len(description) > max_size: + new_desc = "" + + # This splits the paragraph into words, but keeping the + # (preceeding) whitespace intact so we can easily concat + # words back together. + for match in re.finditer("\s*\S+", description): + word = match.group() + + # Keep adding words while the total length is less than + # MAX_SIZE. + if len(word) + len(new_desc) < max_size: + new_desc += word + else: + # At this point the next word *will* take us over + # MAX_SIZE, but we also want to ensure that its not + # a huge word. If it is add it anyway and we'll + # truncate later. + if len(new_desc) < min_size: + new_desc += word + break + + # Double check that we're not over the limit + if len(new_desc) > max_size: + new_desc = new_desc[:max_size] + + # We always add an ellipsis because at the very least + # we chopped mid paragraph. + description = new_desc.strip() + "…" + return description if description else None |