purge unviewable videos after 30 days

files/classes/comment.py

@@ -237,6 +237,7 @@ class Comment(Base):
 	casino_game = relationship("CasinoGame")
 	wall_user = relationship("User", primaryjoin="User.id==Comment.wall_user_id")
 	edits = relationship("CommentEdit", order_by="CommentEdit.id.desc()")
+	media_usages = relationship("MediaUsage")
 
 	def __init__(self, *args, **kwargs):
 		if "created_utc" not in kwargs:

files/classes/media.py

@@ -1,5 +1,6 @@
 import time
 from sqlalchemy import Column, ForeignKey
+from sqlalchemy.orm import relationship
 from sqlalchemy.sql.sqltypes import *
 from flask import request, has_request_context
 from files.classes import Base
@@ -23,3 +24,23 @@ class Media(Base):
 
 	def __repr__(self):
 		return f"<{self.__class__.__name__}(kind={self.kind}, filename={self.filename})>"
+
+class MediaUsage(Base):
+	__tablename__ = "media_usages"
+	id = Column(Integer, primary_key=True)
+	filename = Column(String, ForeignKey("media.filename"))
+	post_id = Column(Integer, ForeignKey("posts.id"))
+	comment_id = Column(Integer, ForeignKey("comments.id"))
+	created_utc = Column(Integer)
+	deleted_utc = Column(Integer)
+	removed_utc = Column(Integer)
+
+	media = relationship("Media")
+
+	def __init__(self, *args, **kwargs):
+		if "created_utc" not in kwargs:
+			kwargs["created_utc"] = int(time.time())
+		super().__init__(*args, **kwargs)
+
+	def __repr__(self):
+		return f"<{self.__class__.__name__}(id={self.id})>"

files/classes/post.py

@@ -87,6 +87,7 @@ class Post(Base):
 	hole_obj = relationship("Hole", primaryjoin="foreign(Post.hole)==remote(Hole.name)")
 	options = relationship("PostOption", order_by="PostOption.id")
 	edits = relationship("PostEdit", order_by="PostEdit.id.desc()")
+	media_usages = relationship("MediaUsage")
 
 	def __init__(self, *args, **kwargs):
 		if "created_utc" not in kwargs:

files/helpers/actions.py

@@ -430,6 +430,11 @@ def execute_antispam_post_check(title, v, url):
 
 		for post in similar_posts + similar_urls:
 			post.is_banned = True
+			
+			for media_usage in post.media_usages:
+				media_usage.removed_utc = time.time()
+				g.db.add(media_usage)
+
 			post.profile_pinned = False
 			post.ban_reason = "AutoJanny for spamming"
 			g.db.add(post)
@@ -491,6 +496,11 @@ def execute_antispam_comment_check(body, v):
 
 	for comment in similar_comments:
 		comment.is_banned = True
+
+		for media_usage in comment.media_usages:
+			media_usage.removed_utc = time.time()
+			g.db.add(media_usage)
+
 		comment.ban_reason = "AutoJanny for spamming"
 		g.db.add(comment)
 		ma = ModAction(

files/helpers/cron.py

@@ -12,6 +12,7 @@ from sqlalchemy import or_
 
 import click
 import requests
+import humanize
 
 from files.helpers.stats import *
 from files.routes.front import frontlist
@@ -421,4 +422,26 @@ def _set_top_poster_of_the_day_id():
 	cache.set("top_poster_of_the_day_id", user.id, timeout=86400)
 
 def _cleanup_videos():
-	subprocess.call("scripts/cleanup_videos.sh", timeout=3000)
+	subprocess.call("scripts/cleanup_videos.sh", timeout=3000)
+
+	db = db_session()
+
+	clean = [x[0] for x in db.query(MediaUsage.filename).filter_by(deleted_utc=None, removed_utc=None)]
+
+	one_month_ago = time.time() - 2592000
+
+	to_delete = db.query(MediaUsage.filename, Media.size).join(MediaUsage.media).filter(
+		MediaUsage.filename.notin_(clean),
+		or_(
+			MediaUsage.deleted_utc < one_month_ago,
+			MediaUsage.removed_utc < one_month_ago,
+		),
+	).order_by(Media.size.desc())
+
+	total_saved = 0
+	for filename, size in to_delete:
+		total_saved += size
+		print(filename, humanize.naturalsize(size, binary=True), flush=True)
+
+	total_saved = humanize.naturalsize(total_saved, binary=True)
+	print(f"Total saved: {total_saved}")

files/helpers/media.py

@@ -163,7 +163,7 @@ def reencode_video(old, new, check_sizes=False):
 
 
 
-def process_video(file, v):
+def process_video(file, v, post=None):
 	if isinstance(file, str):
 		old = file
 	else:
@@ -224,6 +224,12 @@ def process_video(file, v):
 	)
 	g.db.add(media)
 
+	if post:
+		media_usage = MediaUsage(filename=new)
+		media_usage.post_id = post.id
+		g.db.add(media_usage)
+
+
 	url = SITE_FULL_VIDEOS + new.split('/videos')[1]
 
 	name = f'/images/{time.time()}'.replace('.','') + '.webp'

files/helpers/regex.py

@@ -2,7 +2,7 @@ import random
 import re
 from flask import g
 
-from files.classes.media import Media
+from files.classes.media import *
 from .config.const import *
 
 NOT_IN_CODE_OR_LINKS = '(?!([^<]*<\/(code|pre|a)>|[^`\n]*`))'
@@ -109,14 +109,30 @@ image_check_regex = re.compile(f'!\[\]\(((?!(https:\/\/({hosts})\/|\/)).*?)\)',
 video_regex_extensions = '|'.join(VIDEO_FORMATS)
 video_sub_regex = re.compile(f'(?<!")(https:\/\/({hosts})\/[\w:~,()\-.#&\/=?@%;+]*?\.({video_regex_extensions}))' + NOT_IN_CODE_OR_LINKS_OR_SPOILER, flags=re.A|re.I)
 
-def video_sub_regex_matcher(match):
+def video_sub_regex_matcher(match, obj):
 	url = match.group(1)
 	if url.startswith(SITE_FULL_VIDEOS):
 		filename = '/videos/' + url.split(f'{SITE_FULL_VIDEOS}/')[1]
 		g.db.flush()
-		posterurl = g.db.query(Media.posterurl).filter_by(filename=filename).one_or_none()
-		if posterurl:
-			return 	f'<p class="resizable"><video poster="{posterurl[0]}" controls preload="none" src="{url}"></video></p>'
+		media = g.db.get(Media, filename)
+		if media:
+			if obj:
+				if not obj.id: raise Exception("The thing that never happens happened again")
+				if str(obj.__class__) == "<class 'files.classes.post.Post'>":
+					existing = g.db.query(MediaUsage.id).filter_by(filename=filename, post_id=obj.id).one_or_none()
+					if not existing:
+						media_usage = MediaUsage(filename=filename)
+						media_usage.post_id = obj.id
+						g.db.add(media_usage)
+				else:			
+					existing = g.db.query(MediaUsage.id).filter_by(filename=filename, comment_id=obj.id).one_or_none()
+					if not existing:
+						media_usage = MediaUsage(filename=filename)
+						media_usage.comment_id = obj.id
+						g.db.add(media_usage)
+
+			if media.posterurl:
+				return 	f'<p class="resizable"><video poster="{media.posterurl[0]}" controls preload="none" src="{url}"></video></p>'
 	return f'<p class="resizable"><video controls preload="none" src="{url}"></video></p>'
 
 audio_regex_extensions = '|'.join(AUDIO_FORMATS)

files/helpers/sanitize.py

@@ -487,7 +487,7 @@ def sanitize(sanitized, golden=True, limit_pings=0, showmore=False, count_emojis
 
 	sanitized = sanitized.replace('&','&')
 
-	sanitized = video_sub_regex.sub(video_sub_regex_matcher, sanitized)
+	sanitized = video_sub_regex.sub(lambda match: video_sub_regex_matcher(match, obj), sanitized)
 	sanitized = audio_sub_regex.sub(r'<audio controls preload="none" src="\1"></audio>', sanitized)
 
 	if count_emojis:

files/routes/admin.py

@@ -245,6 +245,11 @@ def revert_actions(v, username):
 
 	for item in posts + comments:
 		item.is_banned = False
+
+		for media_usage in item.media_usages:
+			media_usage.removed_utc = None
+			g.db.add(media_usage)
+
 		item.ban_reason = None
 		item.is_approved = v.id
 		g.db.add(item)
@@ -1377,6 +1382,11 @@ def unprogstack_comment(comment_id, v):
 def remove_post(post_id, v):
 	post = get_post(post_id)
 	post.is_banned = True
+
+	for media_usage in post.media_usages:
+		media_usage.removed_utc = time.time()
+		g.db.add(media_usage)
+
 	post.is_approved = None
 
 	if not FEATURES['AWARDS'] or not post.pinned or not post.pinned.endswith(PIN_AWARD_TEXT):
@@ -1423,6 +1433,11 @@ def approve_post(post_id, v):
 		g.db.add(ma)
 
 	post.is_banned = False
+
+	for media_usage in post.media_usages:
+		media_usage.removed_utc = None
+		g.db.add(media_usage)
+
 	post.ban_reason = None
 	post.is_approved = v.id
 
@@ -1595,6 +1610,11 @@ def remove_comment(c_id, v):
 	comment = get_comment(c_id)
 
 	comment.is_banned = True
+
+	for media_usage in comment.media_usages:
+		media_usage.removed_utc = time.time()
+		g.db.add(media_usage)
+
 	comment.is_approved = None
 	comment.ban_reason = v.username
 	g.db.add(comment)
@@ -1633,6 +1653,11 @@ def approve_comment(c_id, v):
 		g.db.add(ma)
 
 	comment.is_banned = False
+
+	for media_usage in comment.media_usages:
+		media_usage.removed_utc = None
+		g.db.add(media_usage)
+
 	comment.ban_reason = None
 	comment.is_approved = v.id
 
@@ -1721,6 +1746,11 @@ def admin_nuke_user(v):
 			continue
 
 		post.is_banned = True
+
+		for media_usage in post.media_usages:
+			media_usage.removed_utc = time.time()
+			g.db.add(media_usage)
+
 		post.ban_reason = v.username
 		g.db.add(post)
 
@@ -1729,6 +1759,11 @@ def admin_nuke_user(v):
 			continue
 
 		comment.is_banned = True
+
+		for media_usage in comment.media_usages:
+			media_usage.removed_utc = time.time()
+			g.db.add(media_usage)
+
 		comment.ban_reason = v.username
 		g.db.add(comment)
 
@@ -1757,6 +1792,11 @@ def admin_nunuke_user(v):
 			continue
 
 		post.is_banned = False
+
+		for media_usage in post.media_usages:
+			media_usage.removed_utc = None
+			g.db.add(media_usage)
+
 		post.ban_reason = None
 		post.is_approved = v.id
 		g.db.add(post)
@@ -1766,6 +1806,11 @@ def admin_nunuke_user(v):
 			continue
 
 		comment.is_banned = False
+
+		for media_usage in comment.media_usages:
+			media_usage.removed_utc = None
+			g.db.add(media_usage)
+
 		comment.ban_reason = None
 		comment.is_approved = v.id
 		g.db.add(comment)

files/routes/comments.py

@@ -232,6 +232,7 @@ def comment(v):
 
 	c.upvotes = 1
 
+	g.db.add(c)
 	body_html = sanitize(body, limit_pings=5, showmore=(not v.hieroglyphs), count_emojis=not v.marsify, commenters_ping_post_id=commenters_ping_post_id, obj=c, author=v)
 
 	if post_target.id not in ADMIGGER_THREADS and not (v.chud and v.chud_phrase.lower() in body.lower()):
@@ -256,7 +257,6 @@ def comment(v):
 
 	c.body_html = body_html
 
-	g.db.add(c)
 	g.db.flush()
 
 	if not posting_to_post and v.admin_level >= PERMS['ADMIN_NOTES'] and request.values.get('admin_note') == 'true' :
@@ -277,6 +277,11 @@ def comment(v):
 
 	if not complies_with_chud(c):
 		c.is_banned = True
+
+		for media_usage in c.media_usages:
+			media_usage.removed_utc = time.time()
+			g.db.add(media_usage)
+
 		c.ban_reason = "AutoJanny for lack of chud phrase"
 		g.db.add(c)
 
@@ -425,6 +430,10 @@ def delete_comment(cid, v):
 			v.comment_count -= 1
 			g.db.add(v)
 
+		for media_usage in c.media_usages:
+			media_usage.deleted_utc = c.deleted_utc
+			g.db.add(media_usage)
+
 		cache.delete_memoized(comment_idlist)
 
 		if c.parent_post:
@@ -459,6 +468,10 @@ def undelete_comment(cid, v):
 			v.comment_count += 1
 			g.db.add(v)
 
+		for media_usage in c.media_usages:
+			media_usage.deleted_utc = None
+			g.db.add(media_usage)
+
 		cache.delete_memoized(comment_idlist)
 
 		if c.parent_post:

files/routes/posts.py

@@ -641,6 +641,7 @@ def submit_post(v, hole=None):
 
 	p.title_html = title_html
 
+	g.db.add(p)
 	body_html = sanitize(body, count_emojis=True, limit_pings=100, obj=p, author=v)
 
 	if v.hieroglyphs and not p.distinguished and marseyaward_body_regex.search(body_html):
@@ -684,7 +685,7 @@ def submit_post(v, hole=None):
 			copyfile(name, name2)
 			p.thumburl = process_image(name2, v, resize=199)
 		elif file.content_type.startswith('video/'):
-			p.url, p.posterurl, name = process_video(file, v)
+			p.url, p.posterurl, name = process_video(file, v, post=p)
 			if p.posterurl:
 				name2 = name.replace('.webp', 'r.webp')
 				copyfile(name, name2)
@@ -693,9 +694,27 @@ def submit_post(v, hole=None):
 			p.url = process_audio(file, v)
 		else:
 			stop(415)
+	elif p.url and p.url.startswith(SITE_FULL_VIDEOS):
+		filename = p.url.split(SITE_FULL_VIDEOS)[0]
+		print(filename, flush=True)
+		media = g.db.get(Media, filename)
+		print(media, flush=True)
+		if media:
+			media_usage = MediaUsage(
+				filename=filename,
+				post_id=p.id,
+			)
+			g.db.add(media_usage)
+			if media.posterurl:
+				p.posterurl = media.posterurl
 
 	if not p.draft and not complies_with_chud(p):
 		p.is_banned = True
+
+		for media_usage in p.media_usages:
+			media_usage.removed_utc = time.time()
+			g.db.add(media_usage)
+
 		p.ban_reason = "AutoJanny for lack of chud phrase"
 
 		body = random.choice(CHUD_MSGS).format(username=v.username, type='post', CHUD_PHRASE=v.chud_phrase)
@@ -802,6 +821,10 @@ def delete_post_pid(pid, v):
 		v.post_count -= 1
 		g.db.add(v)
 
+		for media_usage in p.media_usages:
+			media_usage.deleted_utc = p.deleted_utc
+			g.db.add(media_usage)
+
 		for sort in COMMENT_SORTS.keys():
 			cache.delete(f'post_{p.id}_{sort}')
 
@@ -830,6 +853,10 @@ def undelete_post_pid(pid, v):
 		v.post_count += 1
 		g.db.add(v)
 
+		for media_usage in p.media_usages:
+			media_usage.deleted_utc = None
+			g.db.add(media_usage)
+
 		for sort in COMMENT_SORTS.keys():
 			cache.delete(f'post_{p.id}_{sort}')
 

migrations/20241022-cleanup-videos.sql

@@ -0,0 +1,28 @@
+create table media_usages (
+    id integer primary key,
+    filename character varying(55) NOT NULL,
+	post_id integer,
+	comment_id integer,
+	created_utc integer not null,
+	deleted_utc integer,
+	removed_utc integer
+);
+
+CREATE SEQUENCE public.media_usages_id_seq
+    AS integer
+    START WITH 1
+    INCREMENT BY 1
+    NO MINVALUE
+    NO MAXVALUE
+    CACHE 1;
+
+ALTER SEQUENCE public.media_usages_id_seq OWNED BY public.media_usages.id;
+
+ALTER TABLE ONLY public.media_usages ALTER COLUMN id SET DEFAULT nextval('public.media_usages_id_seq'::regclass);
+
+
+alter table media_usages
+    add constraint media_usages_post_fkey foreign key (post_id) references posts(id);
+
+alter table media_usages
+    add constraint media_usages_comment_fkey foreign key (comment_id) references comments(id);

requirements.txt

@@ -13,6 +13,7 @@ gevent-websocket
 gevent_inotifyx
 greenlet
 gunicorn
+humanize
 imagehash
 inotify
 isodate