github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[kemonoparty:discord] improve 'inline' extraction (#1940)

Browse Source 
- extract media.discordapp.*NET* URLs
- rewrite media.discordapp.net to cdn.discordapp.com
- use a more restricted set of characters for the URL path
This commit is contained in:
Mike Fährmann
2021-10-24 21:15:21 +02:00
parent 02a247f4e5
commit f1487a3cfa
1 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
gallery_dl/extractor/kemonoparty.py
View File

@@ -237,7 +237,8 @@ class KemonopartyDiscordExtractor(KemonopartyExtractor):
self._prepare_ddosguard_cookies() self._prepare_ddosguard_cookies()
find_inline = re.compile( find_inline = re.compile(
r"https?://(?:cdn|media)\.discordapp.com/\S+").findall r"https?://(?:cdn\.discordapp.com|media\.discordapp\.net)"
r"(/[A-Za-z0-9-._~:/?#\[\]@!$&'()*+,;%=]+)").findall
posts = self.posts() posts = self.posts()
max_posts = self.config("max-posts") max_posts = self.config("max-posts")
@@ -251,7 +252,8 @@ class KemonopartyDiscordExtractor(KemonopartyExtractor):
attachment["type"] = "attachment" attachment["type"] = "attachment"
append(attachment) append(attachment)
for path in find_inline(post["content"] or ""): for path in find_inline(post["content"] or ""):
append({"path": path, "name": path, "type": "inline"}) append({"path": "https://cdn.discordapp.com" + path,
"name": path, "type": "inline"})
post["channel_name"] = self.channel_name post["channel_name"] = self.channel_name
post["date"] = text.parse_datetime( post["date"] = text.parse_datetime(