Add Open Graph metadata support

app.py

@@ -238,6 +238,11 @@ def get_attachment_url(url, size):
     return _get_file_url(url, size, Kind.ATTACHMENT)
 
 
+@app.template_filter()
+def get_og_image_url(url, size=100):
+    return _get_file_url(url, size, Kind.OG_IMAGE)
+
+
 @app.template_filter()
 def permalink_id(val):
     return str(hash(val))

sass/base_theme.scss

@@ -189,9 +189,11 @@ a:hover {
     h3 { margin: 0; }
   }
 }
+.note-box {
+  margin-bottom: 70px;
+}
 .note {
   display: flex;
-  margin-bottom: 70px;
   .l {
     color: $color-note-link;
   }
@@ -229,7 +231,11 @@ a:hover {
     padding:10px 0;
   }
 }
-
+.color-menu-background {
+  background: $color-menu-background;
+}
+.og-link { text-decoration: none; }
+.og-link:hover { text-decoration: none; }
 .bar-item-no-hover {
   background: $color-menu-background;
   padding: 5px;

tasks.py

@@ -19,6 +19,7 @@ from config import ID
 from config import KEY
 from config import MEDIA_CACHE
 from config import USER_AGENT
+from utils import opengraph
 from utils.media import Kind
 
 log = logging.getLogger(__name__)
@@ -103,12 +104,49 @@ def process_new_activity(self, iri: str) -> None:
         self.retry(exc=err, countdown=int(random.uniform(2, 4) ** self.request.retries))
 
 
+@app.task(bind=True, max_retries=12)  # noqa: C901
+def fetch_og_metadata(self, iri: str) -> None:
+    try:
+        activity = ap.fetch_remote_activity(iri)
+        log.info(f"activity={activity!r}")
+        if activity.has_type(ap.ActivityType.CREATE):
+            note = activity.get_object()
+            links = opengraph.links_from_note(note.to_dict())
+            og_metadata = opengraph.fetch_og_metadata(USER_AGENT, links)
+            for og in og_metadata:
+                if not og.get("image"):
+                    continue
+                MEDIA_CACHE.cache_og_image(og["image"])
+
+            log.debug(f"OG metadata {og_metadata!r}")
+            DB.activities.update_one(
+                {"remote_id": iri}, {"$set": {"meta.og_metadata": og_metadata}}
+            )
+
+        log.info(f"OG metadata fetched for {iri}")
+    except (ActivityGoneError, ActivityNotFoundError):
+        log.exception(f"dropping activity {iri}, skip OG metedata")
+    except requests.exceptions.HTTPError as http_err:
+        if 400 <= http_err.response.status_code < 500:
+            log.exception("bad request, no retry")
+        log.exception("failed to fetch OG metadata")
+        self.retry(
+            exc=http_err, countdown=int(random.uniform(2, 4) ** self.request.retries)
+        )
+    except Exception as err:
+        log.exception(f"failed to fetch OG metadata for {iri}")
+        self.retry(exc=err, countdown=int(random.uniform(2, 4) ** self.request.retries))
+
+
 @app.task(bind=True, max_retries=12)
 def cache_actor(self, iri: str, also_cache_attachments: bool = True) -> None:
     try:
         activity = ap.fetch_remote_activity(iri)
         log.info(f"activity={activity!r}")
 
+        if activity.has_type(ap.ActivityType.CREATE):
+            fetch_og_metadata.delay(iri)
+
         actor = activity.get_actor()
 
         cache_actor_with_inbox = False

templates/utils.html

@@ -21,6 +21,7 @@
 {% else %}
 {% set actor = obj.attributedTo | get_actor  %}
 {% endif %}
+<div class="note-box">
 <div class="note h-entry" id="activity-{{ obj.id | permalink_id }}">
 
 <div class="h-card p-author">
@@ -63,6 +64,26 @@
 </div>
 	{% endif %}
 
+
+
+{% if meta and meta.og_metadata %}
+{% for og in meta.og_metadata %}
+<a href="{{ og.url }}" class="og-link" style="margin:30px 0;clear:both;display: flex;">
+<div>
+<img  style="width:100px;border-radius:3px;" src="{{ og.image | get_og_image_url }}">
+</div>
+<div style="padding:0 20px;">
+<strong>{{ og.title }}</strong>
+<p>{{ og.description | truncate(80) }}</p>
+<small>{{ og.site_name }}</small>
+</div>
+</a>
+{% endfor %}
+{% endif %}
+
+
+
+
 <div class="bottom-bar">
 {% if perma %}<span class="perma-item">{{ obj.published | format_time }}</span>
 {% if not (obj.id | is_from_outbox) %}
@@ -163,10 +184,10 @@
 {% endif %}
 
 
-
+</div>
+</div>                       
 </div>
 
-</div>                       
 
 {%- endmacro %}
 

utils/media.py

@@ -31,6 +31,7 @@ class Kind(Enum):
     ATTACHMENT = "attachment"
     ACTOR_ICON = "actor_icon"
     UPLOAD = "upload"
+    OG_IMAGE = "og"
 
 
 class MediaCache(object):
@@ -38,6 +39,24 @@ class MediaCache(object):
         self.fs = gridfs.GridFS(gridfs_db)
         self.user_agent = user_agent
 
+    def cache_og_image(self, url: str) -> None:
+        if self.fs.find_one({"url": url, "kind": Kind.OG_IMAGE.value}):
+            return
+        i = load(url, self.user_agent)
+        # Save the original attachment (gzipped)
+        i.thumbnail((100, 100))
+        with BytesIO() as buf:
+            with GzipFile(mode="wb", fileobj=buf) as f1:
+                i.save(f1, format=i.format)
+            buf.seek(0)
+            self.fs.put(
+                buf,
+                url=url,
+                size=100,
+                content_type=i.get_format_mimetype(),
+                kind=Kind.OG_IMAGE.value,
+            )
+
     def cache_attachment(self, url: str) -> None:
         if self.fs.find_one({"url": url, "kind": Kind.ATTACHMENT.value}):
             return
@@ -141,6 +160,8 @@ class MediaCache(object):
     def cache(self, url: str, kind: Kind) -> None:
         if kind == Kind.ACTOR_ICON:
             self.cache_actor_icon(url)
+        elif kind == Kind.OG_IMAGE:
+            self.cache_og_image(url)
         else:
             self.cache_attachment(url)
 

utils/opengraph.py

@@ -23,24 +23,11 @@ def links_from_note(note):
     return links
 
 
-def fetch_og_metadata(user_agent, col, remote_id):
+def fetch_og_metadata(user_agent, links):
-    doc = col.find_one({"remote_id": remote_id})
-    if not doc:
-        raise ValueError
-    note = doc["activity"]["object"]
-    print(note)
-    links = links_from_note(note)
-    if not links:
-        return 0
-    # FIXME(tsileo): set the user agent by giving HTML directly to OpenGraph
     htmls = []
     for l in links:
         check_url(l)
-        r = requests.get(l, headers={"User-Agent": user_agent})
+        r = requests.get(l, headers={"User-Agent": user_agent}, timeout=15)
         r.raise_for_status()
         htmls.append(r.text)
-    links_og_metadata = [dict(opengraph.OpenGraph(html=html)) for html in htmls]
+    return [dict(opengraph.OpenGraph(html=html)) for html in htmls]
-    col.update_one(
-        {"remote_id": remote_id}, {"$set": {"meta.og_metadata": links_og_metadata}}
-    )
-    return len(links)