Clean tracking info from youtube links

app/activitypub/util.py

@@ -24,7 +24,7 @@ import pytesseract
 
 from app.utils import get_request, allowlist_html, html_to_markdown, get_setting, ap_datetime, markdown_to_html, \
     is_image_url, domain_from_url, gibberish, ensure_directory_exists, markdown_to_text, head_request, post_ranking, \
-    shorten_string, reply_already_exists, reply_is_just_link_to_gif_reaction, confidence
+    shorten_string, reply_already_exists, reply_is_just_link_to_gif_reaction, confidence, clean_link
 
 
 def public_key():
@@ -1227,6 +1227,7 @@ def create_post(activity_log: ActivityPubLog, community: Community, request_json
                 post.image = image
             else:
                 post.type = POST_TYPE_LINK
+                post.url = clean_link(post.url)
             domain = domain_from_url(post.url)
             # notify about links to banned websites.
             already_notified = set()  # often admins and mods are the same people - avoid notifying them twice

app/community/util.py

@@ -15,7 +15,7 @@ from app.constants import POST_TYPE_ARTICLE, POST_TYPE_LINK, POST_TYPE_IMAGE
 from app.models import Community, File, BannedInstances, PostReply, PostVote, Post, utcnow, CommunityMember, Site, \
     Instance, Notification, User
 from app.utils import get_request, gibberish, markdown_to_html, domain_from_url, allowlist_html, \
-    html_to_markdown, is_image_url, ensure_directory_exists, inbox_domain, post_ranking, shorten_string, parse_page
+    html_to_markdown, is_image_url, ensure_directory_exists, inbox_domain, post_ranking, shorten_string, parse_page, clean_link
 from sqlalchemy import func
 import os
 
@@ -177,7 +177,7 @@ def save_post(form, post: Post):
         post.body = form.link_body.data
         post.body_html = markdown_to_html(post.body)
         url_changed = post.id is None or form.link_url.data != post.url
-        post.url = form.link_url.data
+        post.url = clean_link(form.link_url.data)
         post.type = POST_TYPE_LINK
         domain = domain_from_url(form.link_url.data)
         domain.post_count += 1

app/utils.py

@@ -745,3 +745,14 @@ def sha256_digest(input_string):
     sha256_hash = hashlib.sha256()
     sha256_hash.update(input_string.encode('utf-8'))
     return sha256_hash.hexdigest()
+
+
+def clean_link(url):
+    # strip ?si=abcDEFgh from youtu.be links
+    clean = re.search(r"(https://youtu.be/\w+)", url)
+
+    if clean is not None:
+        return clean.group(1)
+    else:
+        return url
+