Add Open Graph metadata support

This commit is contained in:
Thomas Sileo
2018-07-21 23:16:40 +02:00
parent 23faef985b
commit 648e385c49
6 changed files with 98 additions and 20 deletions

View File

@@ -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)

View File

@@ -23,24 +23,11 @@ def links_from_note(note):
return links
def fetch_og_metadata(user_agent, col, remote_id):
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
def fetch_og_metadata(user_agent, links):
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]
col.update_one(
{"remote_id": remote_id}, {"$set": {"meta.og_metadata": links_og_metadata}}
)
return len(links)
return [dict(opengraph.OpenGraph(html=html)) for html in htmls]