From 148fe58a247d61ffb76c566ba397285480d93f74 Mon Sep 17 00:00:00 2001 From: Patrick Cloke Date: Mon, 6 Jun 2022 07:46:04 -0400 Subject: Do not break URL previews if an image is unreachable. (#12950) Avoid breaking a URL preview completely if the chosen image 404s or is unreachable for some other reason (e.g. DNS). --- synapse/rest/media/v1/preview_url_resource.py | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) (limited to 'synapse/rest/media/v1') diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 2b2db63bf7..54a849eac9 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -586,12 +586,16 @@ class PreviewUrlResource(DirectServeJsonResource): og: The Open Graph dictionary. This is modified with image information. """ # If there's no image or it is blank, there's nothing to do. - if "og:image" not in og or not og["og:image"]: + if "og:image" not in og: + return + + # Remove the raw image URL, this will be replaced with an MXC URL, if successful. + image_url = og.pop("og:image") + if not image_url: return # The image URL from the HTML might be relative to the previewed page, # convert it to an URL which can be requested directly. - image_url = og["og:image"] url_parts = urlparse(image_url) if url_parts.scheme != "data": image_url = urljoin(media_info.uri, image_url) @@ -599,7 +603,16 @@ class PreviewUrlResource(DirectServeJsonResource): # FIXME: it might be cleaner to use the same flow as the main /preview_url # request itself and benefit from the same caching etc. But for now we # just rely on the caching on the master request to speed things up. - image_info = await self._handle_url(image_url, user, allow_data_urls=True) + try: + image_info = await self._handle_url(image_url, user, allow_data_urls=True) + except Exception as e: + # Pre-caching the image failed, don't block the entire URL preview. + logger.warning( + "Pre-caching image failed during URL preview: %s errored with %s", + image_url, + e, + ) + return if _is_media(image_info.media_type): # TODO: make sure we don't choke on white-on-transparent images @@ -611,13 +624,11 @@ class PreviewUrlResource(DirectServeJsonResource): og["og:image:width"] = dims["width"] og["og:image:height"] = dims["height"] else: - logger.warning("Couldn't get dims for %s", og["og:image"]) + logger.warning("Couldn't get dims for %s", image_url) og["og:image"] = f"mxc://{self.server_name}/{image_info.filesystem_id}" og["og:image:type"] = image_info.media_type og["matrix:image:size"] = image_info.media_length - else: - del og["og:image"] async def _handle_oembed_response( self, url: str, media_info: MediaInfo, expiration_ms: int -- cgit 1.4.1