Improve video caching

blueprints/tasks.py

@@ -39,6 +39,7 @@ from core.shared import p
 from core.tasks import Tasks
 from utils import now
 from utils import opengraph
+from utils.media import is_video
 
 blueprint = flask.Blueprint("tasks", __name__)
 
@@ -216,6 +217,25 @@ def task_finish_post_to_inbox() -> _Response:
     return ""
 
 
+def select_video_to_cache(links):
+    """Try to find the 360p version from a video urls, or return the smallest one."""
+    videos = []
+    for link in links:
+        if link.get("mimeType", "").startswith("video/") or is_video(link["href"]):
+            videos.append({"href": link["href"], "height": link["height"]})
+
+    if not videos:
+        app.logger.warning(f"failed to select a video from {links!r}")
+        return None
+
+    videos = sorted(videos, key=lambda l: l["height"])
+    for video in videos:
+        if video["height"] == 360:
+            return video
+
+    return videos[0]
+
+
 @blueprint.route("/task/cache_attachments", methods=["POST"])
 def task_cache_attachments() -> _Response:
     task = p.parse(flask.request)
@@ -230,10 +250,10 @@ def task_cache_attachments() -> _Response:
 
         if obj.has_type(ap.ActivityType.VIDEO):
             if isinstance(obj.url, list):
-                for link in obj.url:
+                # TODO: filter only videogt
-                    if link.get("mimeType", "").startswith("video/"):
+                link = select_video_to_cache(obj.url)
-                        config.MEDIA_CACHE.cache_attachment({"url": link["href"]}, iri)
+                if link:
-                        break
+                    config.MEDIA_CACHE.cache_attachment({"url": link["href"]}, iri)
             elif isinstance(obj.url, str):
                 config.MEDIA_CACHE.cache_attachment({"url": obj.url}, iri)
             else:
@@ -290,9 +310,13 @@ def task_cache_actor() -> _Response:
         if not activity.has_type([ap.ActivityType.CREATE, ap.ActivityType.ANNOUNCE]):
             return ""
 
-        if activity.get_object()._data.get(
+        if (
-            "attachment", []
+            activity.has_type(ap.ActivityType.CREATE)
-        ) or activity.get_object().has_type(ap.ActivityType.VIDEO):
+            and activity.get_object()._data.get("attachment", [])
+        ) or (
+            activity.has_type(ap.ActivityType.ANNOUNCE)
+            and activity.get_object().has_type(ap.ActivityType.VIDEO)
+        ):
             Tasks.cache_attachments(iri)
 
     except (ActivityGoneError, ActivityNotFoundError):

utils/media.py

@@ -2,6 +2,7 @@ import base64
 import mimetypes
 from enum import Enum
 from enum import unique
+from functools import lru_cache
 from gzip import GzipFile
 from io import BytesIO
 from typing import Any
@@ -14,6 +15,22 @@ from little_boxes import activitypub as ap
 from PIL import Image
 
 
+@lru_cache(2048)
+def _is_img(filename):
+    mimetype, _ = mimetypes.guess_type(filename.lower())
+    if mimetype and mimetype.split("/")[0] in ["image"]:
+        return True
+    return False
+
+
+@lru_cache(2048)
+def is_video(filename):
+    mimetype, _ = mimetypes.guess_type(filename.lower())
+    if mimetype and mimetype.split("/")[0] in ["video"]:
+        return True
+    return False
+
+
 def load(url: str, user_agent: str) -> Image:
     """Initializes a `PIL.Image` from the URL."""
     with requests.get(url, stream=True, headers={"User-Agent": user_agent}) as resp:
@@ -74,10 +91,7 @@ class MediaCache(object):
 
         # If it's an image, make some thumbnails
         if (
-            url.endswith(".png")
+            _is_img(url)
-            or url.endswith(".jpg")
-            or url.endswith(".jpeg")
-            or url.endswith(".gif")
             or attachment.get("mediaType", "").startswith("image/")
             or ap._has_type(attachment.get("type"), ap.ActivityType.IMAGE)
         ):
@@ -123,8 +137,9 @@ class MediaCache(object):
             resp.raise_for_status()
             with BytesIO() as buf:
                 with GzipFile(mode="wb", fileobj=buf) as f1:
-                    for chunk in resp.iter_content():
+                    for chunk in resp.iter_content(chunk_size=2 << 20):
                         if chunk:
+                            print(len(chunk))
                             f1.write(chunk)
                 buf.seek(0)
                 self.fs.put(

utils/template_filters.py

@@ -1,5 +1,4 @@
 import logging
-import mimetypes
 import urllib
 from datetime import datetime
 from datetime import timezone
@@ -22,6 +21,7 @@ from config import MEDIA_CACHE
 from core.activitypub import _answer_key
 from utils import parse_datetime
 from utils.media import Kind
+from utils.media import _is_img
 
 _logger = logging.getLogger(__name__)
 
@@ -312,14 +312,6 @@ def has_actor_type(doc):
     return False
 
 
-@lru_cache(512)
-def _is_img(filename):
-    mimetype, _ = mimetypes.guess_type(filename.lower())
-    if mimetype and mimetype.split("/")[0] in ["image"]:
-        return True
-    return False
-
-
 @filters.app_template_filter()
 def not_only_imgs(attachment):
     for a in attachment: