snapshot

clean_vids.py

@@ -0,0 +1,782 @@
+import contextlib
+import hashlib
+import json
+import logging
+import re
+import sqlite3
+import subprocess
+import typing
+from collections import Counter, defaultdict
+from pathlib import Path
+
+import paramiko
+import psycopg
+import psycopg.rows
+
+import ffmpeg
+
+VIDS_DIR = Path(r"/Volumes/FIVER/_ingress/")
+schema = """
+create table if not exists videos(
+    file_path text not null primary key,
+    size_bytes int not null,
+    hash_partial text not null,
+    ffprobe json not null
+);
+
+create table if not exists scan_state(
+    file_path text not null primary key,
+    stderr text
+);
+"""
+
+
+def find_vids(root: Path):
+    extensions = [".mkv", ".mp4"]
+    for ext in extensions:
+        for f in root.rglob(f"*{ext}"):
+            yield f
+
+
+def hash_partial(f: Path) -> str:
+    sha1 = hashlib.sha1()
+    chunk_size = 1024 * 1024 * 10  # 10MB chunk size
+
+    total_read = 0
+    with f.open("rb") as file:
+        while chunk := file.read(chunk_size):
+            total_read += len(chunk)
+            sha1.update(chunk)
+            break  # Only reads the first 10MB
+
+    return f"sha1:{total_read}:{sha1.hexdigest()}"
+
+
+@contextlib.contextmanager
+def connect_db_sqlite() -> typing.Generator[psycopg.Connection, typing.Any, typing.Any]:
+    with sqlite3.connect("vids.sqlite3") as conn:
+        conn.row_factory = sqlite3.Row
+        conn.execute("PRAGMA foreign_keys = 1")
+        yield conn
+
+
+@contextlib.contextmanager
+def connect_db() -> typing.Generator[psycopg.Connection, typing.Any, typing.Any]:
+    with psycopg.connect("postgres://abdus:abdus@db.abdus.dev:5444/smut?sslmode=disable") as conn:
+        conn.row_factory = psycopg.rows.dict_row
+        yield conn
+
+
+def video_exists(con: psycopg.Connection, file_path: str) -> bool:
+    stmt = con.execute(
+        "select exists(select 1 from videos where file_path = %s) as exists",
+        [file_path],
+    )
+    row = stmt.fetchone()
+    return row["exists"]
+
+
+def delete_video(con: psycopg.Connection, file_path: str) -> bool:
+    con.execute("delete from videos where file_path = %s", [file_path])
+
+
+def save_video(
+    con: psycopg.Connection,
+    category: str,
+    file_path: str,
+    ffprobe: dict,
+    hash_partial: str,
+):
+    sql = """insert into videos(category, file_path, ffprobe, hash_partial) values (%s, %s, %s, %s);"""
+    con.execute(
+        sql,
+        [
+            category,
+            file_path,
+            json.dumps(ffprobe),
+            hash_partial,
+        ],
+    )
+
+
+def save_error(con: psycopg.Connection, file_path: str, stderr: str):
+    sql = """insert into scan_state(file_path, stderr) values (%s, %s) on conflict do nothing"""
+    con.execute(
+        sql,
+        [
+            file_path,
+            stderr,
+        ],
+    )
+
+
+def scan_videos():
+    logging.basicConfig(format=logging.BASIC_FORMAT, level=logging.DEBUG)
+    with connect_db() as con:
+        # with con:
+        #     con.executescript(schema)
+
+        all_vids = list(find_vids(VIDS_DIR))
+        for i, f in enumerate(all_vids):
+            if "/_picks/" in str(f):
+                continue
+            progress = f"[{i+1}/{len(all_vids)}]"
+            if f.is_dir():
+                continue
+
+            file_path = str(f.absolute())
+
+            if video_exists(con, file_path=file_path):
+                logging.debug(f"{progress} video already exists {f=}")
+                continue
+
+            logging.debug(f"probing {f=}")
+            try:
+                probed = ffmpeg.ffprobe(f)
+            except subprocess.CalledProcessError as e:
+                save_error(con, file_path=file_path, stderr=e.stderr)
+                continue
+
+            file_hash = hash_partial(f)
+
+            save_video(
+                category="downloaded.fiver",
+                con=con,
+                file_path=file_path,
+                ffprobe=probed.__dict__,
+                hash_partial=file_hash,
+            )
+            logging.info(f"{progress} saved {f=}")
+            con.commit()
+
+
+def clean_records():
+    all_vids = set(map(str, find_vids(VIDS_DIR)))
+
+    with connect_db() as con:
+        for row in con.execute("select file_path from videos"):
+            file_path = row["file_path"]
+            if file_path not in all_vids:
+                delete_video(con, file_path=row["file_path"])
+
+
+def unwanted(dry_run: bool):
+    take = set(
+        r"""
+aidra fox
+rachel starr
+abella danger
+sata jones
+octavia red
+paige owens
+lily lou
+cj miles
+arabelle raphael
+chloe amour
+erin everheart
+kenzie reeves
+alex grey
+alicia trece
+angel youngs
+angela white
+april olsen
+blake blossom
+brandy renee
+chantal
+demi sutra
+eden ivy
+ema karter
+emma hix
+gianna dior
+hazel moore
+jazmin luv
+jessy jey
+josephine jackson
+katsaros
+kira noir
+kitana lure
+kitana montana
+kristy black
+leal
+lia lin
+lily labeau
+liya silver
+maddy may
+nata ocean
+rebel rhyder
+rein
+savannah bond
+scarlett alexis
+scarlit scandal
+shalina
+simon kitty
+slimthick vic
+stratton
+tatum
+tommy king
+vanna bardot
+vic marie
+violet starr
+vittoria
+emily willis
+THUNDERCOCK
+lucky bee
+charlotte sins
+mandy muse
+skye blue
+alexis tae
+willow
+TRUEANAL 
+ALLANAL
+avery cristy
+ANALONLY
+SWALLOWED
+kelly collins
+nicole black
+xxlayna marie
+katrina colt
+allie addison
+DIRTYAUDITIONS
+scarlet chase
+#sloppiest kiss
+liz ocean
+medusa
+nicole aria
+stacy bloom
+lara frost
+jamie jett
+brooklyn gray
+kimmy granger
+kelsey kane
+kay lovely
+ivy wolfe
+stefany kyler
+leo ahsoka
+lydia black
+renee rose
+emily jade
+anya olsen
+alexis crystal
+sybil
+megan rain
+charlotte sartre
+madison ivy
+maddy black
+gina valentina
+lacey jayne
+hailey rose
+graycee baybee
+katie kush
+lilu moon
+juis wild
+kylie rocket
+dolly dyson
+nichole saphir
+Octokuro
+reislin
+isabella de laa
+qween goddess
+INTHECRACK
+athenea rose
+daniela garcia
+kama oxi
+mina
+baby nicols
+jade amor
+daniela ortiz
+NYMPHO
+""".strip().splitlines(keepends=False)
+    )
+    drop = set(
+        r"""
+ADDICTED2GIRLS
+ASIANSEEPLOITED
+BANANAFEVER
+BBCPOVD
+BFFS
+# BRAZZERSEXXTRA
+BRIDE4K
+CHLOELAMOUR
+DRKINLA
+ETERNALDESIRE
+EVERYTHINGBUTT
+FAKEHOSTEL
+FAKEHUBORIGINALS
+FAKETAXI
+FAMILYSINNERS
+FEETISHPOV
+GIRLGIRLXXX
+GOTFILLED
+HOOKUPHOTSHOT
+HOWWOMENORGASM
+IMMORALLIVE
+IWANTCLIPS
+JACQUIEETMICHELTV
+JERKAOKE
+NICONICE
+NURUMASSAGE
+OHMYHOLES
+OYELOCA
+PASCALSSUBSLUTS
+SCHOOLOFCOCK
+SHOTHERFIRST
+SLIPPERYMASSAGE
+TABOOHEAT
+TEAMSKEETX
+WAXXXED
+WOODMANCASTINGX
+PISSVIDS
+LEGALPORNO
+ANALVIDS
+PORNBOX
+\btap\b
+alice xo
+alicia williams
+mona azar
+anna de ville
+baby kxtten
+bella rolland
+betzz
+charlie forde
+charlie red
+cheyla collins
+claudia garcia
+cory chase
+dani blue
+dellai
+eimy zoren
+emily pink
+gia dibella
+greta foss 
+danielle renae
+harley king
+ivy maddox
+sasha rose
+olive glass
+katherin moore
+kristina grace
+kylie page
+larissa leite
+lily starfire
+lina arian
+luna legend
+luna rishi
+luna wolfs
+lya cutie
+margan lee
+marie berger
+marilyn johnson
+mia cheers
+monika fox
+moona snake
+nichole saphir
+nicole murkovski
+olivia sparkle
+polly petrova
+pornworld
+publicagent
+publicbang
+stacy cruz
+sussy sweet
+triple anal
+vanessa vega
+venera maxima
+victoria nyx
+whitney wright
+zirael rem
+               """.lower()
+        .strip()
+        .splitlines(keepends=False)
+    )
+
+    take_patterns = [re.compile(p.strip().replace(" ", "."), flags=re.I) for p in take if not p.strip().startswith("#")]
+    drop_patterns = [re.compile(p.strip().replace(" ", "."), flags=re.I) for p in drop if not p.strip().startswith("#")]
+
+    to_delete = set()
+    with connect_db() as con:
+        for row in con.execute("select file_path from videos where category = 'box'"):
+            file_path = Path(row["file_path"])
+            text = re.sub(r"[.]+", " ", file_path.stem.lower())
+
+            if any(p.search(text) for p in drop_patterns):
+                to_delete.add(file_path)
+            if any(p.search(text) for p in take_patterns):
+                to_delete.discard(file_path)
+
+    for it in sorted(to_delete):
+        print(it)
+
+    return
+
+    with connect_db() as con:
+        for path in sorted(to_delete, key=str):
+            try:
+                print(path.stem)
+                con.execute(
+                    "update videos set marked_for_deletion_at = now() where file_path = %s",
+                    [str(path)],
+                )
+                if not dry_run:
+                    path.unlink()
+                    delete_video(con, file_path=str(path))
+                con.commit()
+            except:
+                pass
+
+    print(len(to_delete))
+
+
+def large_unwanted():
+    with connect_db() as con:
+        for row in con.execute("select file_path from videos order by size_bytes desc limit 500"):
+            file_path = Path(row["file_path"])
+            print(file_path)
+
+
+def list_studios():
+    re_studio = re.compile(r"^(.+?)\.\d+", flags=re.IGNORECASE)
+    seen = defaultdict(int)
+    with connect_db() as con:
+        for row in con.execute("select file_path from videos"):
+            file_path = Path(row["file_path"])
+            m = re_studio.search(file_path.stem.lower())
+            if m:
+                seen[m.group(1)] += 1
+
+    for it, n in Counter(seen).most_common(20):
+        print(it, n)
+
+
+def lower_resolutions(dry_run: bool):
+    with connect_db() as con:
+        for row in con.execute("""select file_path from videos where file_path like '%720p%';"""):
+            file_path = Path(row["file_path"])
+            video_1080p_path = file_path.with_stem(file_path.stem.replace("720p", "1080p"))
+            if video_exists(con, file_path=str(video_1080p_path)) and video_1080p_path.is_file():
+                print(file_path)
+                if not dry_run:
+                    file_path.unlink()
+
+
+def dupes():
+    sql = """
+with durationed as (
+    select v.*, json_extract(v.ffprobe, '$.duration_sec') as duration_sec from videos v
+)
+select v.file_path, dupe.file_path from durationed v
+join durationed dupe on v.file_path != dupe.file_path and abs(dupe.duration_sec - v.duration_sec) < 3
+where abs(dupe.size_bytes - v.size_bytes) < 3000000
+"""
+    with connect_db() as con:
+        for row in con.execute(sql):
+            print(row[0], row[1])
+            print()
+
+
+def different_format(dry_run: bool):
+    sql = """
+    select file_path from videos where file_path like '%.mkv'
+    """
+    with connect_db() as con:
+        for row in con.execute(sql):
+            mkv_path = Path(row["file_path"])
+            mp4_path = mkv_path.with_suffix(".mp4")
+            if mkv_path.is_file() and mp4_path.is_file():
+                if abs(mp4_path.stat().st_size - mkv_path.stat().st_size) < 5_000_000:
+                    print(mkv_path)
+                    if not dry_run:
+                        mkv_path.unlink()
+
+
+def count_shared_words(a: str, b: str) -> int:
+    a_words = set(re.split(r"[\s\-.]+", a.lower()))
+    b_words = set(re.split(r"[\s\-.]+", b.lower()))
+
+    return len(a_words.intersection(b_words))
+
+
+def reencoded_self(dry_run: bool):
+    sql = """
+    with 
+    durationed_v as (
+        select v.*, json_extract(v.ffprobe, '$.duration_sec') as duration_sec from videos v
+    ),
+    durationed_dl as (
+        select v.*, json_extract(v.ffprobe, '$.duration_sec') as duration_sec from downloaded_videos v
+    )
+    select v.file_path, dlv.file_path from durationed_v v
+    join durationed_dl dlv on abs(v.duration_sec - dlv.duration_sec) < 0.1
+    where dlv.file_path like '%/_reenc/%'
+"""
+
+    with connect_db() as con:
+        for row in con.execute(sql):
+            v_path = Path(row[0])
+            dlv_path = Path(row[1])
+            if count_shared_words(v_path.stem, dlv_path.stem) < 2:
+                continue
+
+            print(v_path.name, dlv_path.name, sep="\n")
+            if not dry_run:
+                try:
+                    v_path.unlink()
+                except:
+                    pass
+
+
+def local_duped(dry_run: bool):
+    sql = """
+with durationed_dl as (
+    select v.*, json_extract(v.ffprobe, '$.duration_sec') as duration_sec from downloaded_videos v
+)
+select v.file_path, dupe.file_path 
+from durationed_dl v
+    join durationed_dl dupe on 
+        v.id != dupe.id
+        and v.id > dupe.id
+        and abs(dupe.duration_sec - v.duration_sec) < 0.1
+        and abs(dupe.size_bytes - v.size_bytes) < 3000000
+    """
+
+    root = Path(r"/Volumes/FIVER/_ingress/")
+    with connect_db() as con:
+        for row in con.execute(sql):
+            a_path = Path(row[0])
+            b_path = Path(row[1])
+            if count_shared_words(a_path.stem, b_path.stem) < 6:
+                continue
+
+            # print(a_path.name, b_path.name, sep='\n')
+            keep, remove = a_path, b_path
+            if len(b_path.stem) > len(a_path.stem):
+                keep, remove = b_path, a_path
+            print(" # ", "matching", keep.name)
+            print(" rm", f'"{remove}"')
+            print()
+
+
+def delete_dupes():
+    rows = []
+
+    with connect_db() as con:
+        for row in rows:
+            file_path = Path(row["file_path"])
+            try:
+                file_path.unlink()
+                delete_video(con, file_path=row["file_path"])
+                con.commit()
+            except:
+                pass
+
+
+def save_suggested_names():
+    rows = []
+
+    with connect_db() as con:
+        for row in rows:
+            bandaid_path = Path(row["file_path"])
+            fiver_path = Path(row["fiver_path"])
+
+            if bandaid_path.stem == fiver_path.stem:
+                continue
+
+            if count_shared_words(bandaid_path.stem, fiver_path.stem) < 3:
+                print("unsimilar", row)
+                continue
+
+            if bandaid_path.stem.count("--") == 3:
+                better_stem = bandaid_path.stem
+            else:
+                better_stem = fiver_path.stem
+
+            better_stem = re.sub(r"(\d{4})(\d{2})(\d{2})", r"\1-\2-\3", better_stem)
+
+            if better_stem == fiver_path.stem:
+                continue
+
+            print(fiver_path.name, better_stem, sep="\n")
+
+            con.execute(
+                "update videos set suggested_filename = %s where file_path = %s",
+                [better_stem, fiver_path.name],
+            )
+            con.commit()
+
+
+def delete_remote_files():
+    rows = [
+        {"file_path": "/mnt/box/files/_raw/prt/5KPorn.24.10.02.Leya.Desantis.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/AllAnal.24.10.30.Aviana.Violet.And.Nicole.Aria.XXX.720p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/AllAnal.24.11.07.Khloe.Kingsley.And.Cherry.Kiss.XXX.720p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/AnalOnly.24.11.26.Aria.Sloane.And.Nicole.Doshi.XXX.720p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/AnalTherapyXXX.24.02.10.Willow.Ryder.Family.Competition.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/AngelsLove.23.07.29.Cherry.Candle.And.Stacy.Cruz.Hot.Juice.Is.Dripping.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BBCPie.20.05.31.Hazel.Moore.Morning.Wood.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BangBros18.24.06.16.Angel.Gostosa.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BangPOV.24.02.20.Justine.Jakobs.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Blacked.22.10.29.Kay.Lovely.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BlacksOnBlondes.22.11.29.Hazel.Moore.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BlacksOnBlondes.24.11.22.Kylie.Rocket.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrattySis.22.06.10.Angel.Gostosa.Everyone.Wants.To.Fuck.My.Stepsister.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrattySis.23.06.23.Angel.Gostosa.My.Stepsister.Takes.Charge.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrattySis.24.05.10.Jade.Maris.And.Vanessa.Marie.Sharing.Stepbros.Bed.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrattySis.24.05.31.Renee.Rose.Nobody.Wants.To.Fuck.A.Virgin.Stepbro.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrattySis.24.07.26.Jade.Maris.And.Sera.Ryder.I.Want.To.Bone.My.Stepsister.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrazzersExxtra.24.08.21.Destiny.Mira.Cyclist.Sweat.Leads.To.Shower.Sex.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrazzersExxtra.24.10.20.Baby.Gemini.Dripping.Wet.Hookup.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrazzersExxtra.24.11.04.Nichole.Saphir.Tattooed.Bombshells.Gaping.Anal.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrazzersExxtra.24.11.20.Katie.Kush.And.Kelly.Caprice.Free.Use.Maid.Service.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/BrazzersExxtra.24.11.20.Kira.Noir.Secret.Goth.Camgirl.Roomie.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/CherryPimps.Busted.24.08.14.Lacey.Jayne.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/CrazyCollegeGFs.24.02.20.Destiny.Mira.Bunk.Banging.The.Handyman.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Creampie-Angels.22.07.03.Nata.Ocean.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/CumSwappingSis.24.03.16.Jade.Maris.And.Liz.Jordan.Giving.Stepbro.Something.To.Do.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/DFXSoleMates.24.06.15.Angel.Gostosa.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/DatingMyStepson.24.10.03.Shalina.Devine.A.Game.Of.Dare.Always.Spices.Things.Up.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/DirtyAuditions.24.08.17.Kazumi.XXX.720p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/ElegantAngel.24.01.01.Hazel.Moore.A.Touch.Of.Class.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/ElegantAngel.24.11.29.Scarlett.Alexis.Stretch.My.Hole.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/EvilAngel.24.06.01.Chloe.Amour.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/EvilAngel.24.11.01.Princess.Alice.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/FilthyTaboo.24.10.19.Kelly.Caprice.Unsatisfied.MILF.Gets.What.She.Craves.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/GirlGirlXXX.24.01.16.Kylie.Rocket.And.Melissa.Stratton.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Hegre.24.10.22.Anna.L.Erotic.Goddess.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/HotwifeXXX.24.11.06.Willow.Ryder.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/InTheCrack.E1928.Lia.Lin.Provence.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/KarupsPC.20.04.01.Nata.Ocean.Horny.Housekeeper.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {
+            "file_path": "/mnt/box/files/_raw/prt/LegalPorno.23.11.10.Lady.Ana.Jureka.Del.Mar.Fisting.ATOGM.Double.Anal.Fist.Big.Gapes.Monster.ButtRose.Squirt.Creampie.Swallow.GIO2635.XXX.1080p.HEVC.x265.PRT.mkv"
+        },
+        {"file_path": "/mnt/box/files/_raw/prt/LegalPorno.24.06.19.Kristy.Black.GIO2815.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/LegalPorno.24.10.02.Leila.Botwin.AH007.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/LegendaryX.24.01.11.Melissa.Stratton.And.Nicole.Doshi.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/LegendaryX.24.02.01.Chantal.Danielle.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/LilSis.24.05.17.Jade.Maris.Let.The.Games.Begin.Part.1.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/LittleCaprice-Dreams.24.11.09.Tiffany.Tatum.And.Nata.Ocean.Nasstyx.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/MilfBody.24.06.14.Kelly.Caprice.Kellys.New.Routine.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/MomIsHorny.23.12.29.Justine.Jakobs.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/MomXXX.24.11.05.Kiara.Lord.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/MomsTeachSex.23.11.24.Justine.Jakobs.Getting.Down.To.Business.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/MrLuckyPOV.24.01.05.Gianna.Dior.And.Kazumi.Squirts.An.All.Star.Threesome.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/MrLuckyPOV.24.01.05.Gianna.Dior.And.Kazumi.Squirts.An.All.Star.Threesome.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/MySistersHotFriend.24.04.09.Willow.Ryder.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NewSensations.24.11.16.Ellie.Nova.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NubileFilms.23.10.17.Liz.Jordan.My.Roommate.Has.A.Crush.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NubileFilms.23.11.20.Hailey.Rose.And.Jade.Maris.Happy.Birthday.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NubileFilms.24.02.01.Liz.Jordan.February.2024.Fantasy.Of.The.Month.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NubileFilms.24.03.01.Jade.Maris.March.2024.Fantasy.Of.The.Month.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/NubileFilms.24.12.01.Chloe.Temple.December.2024.Fantasy.Of.The.Month.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Nubiles-Casting.24.10.28.Chanel.Camryn.Cast.Addison.Vodka.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/PascalsSubSluts.21.06.25.Rebel.Rhyder.Every.Inch.Every.Hole.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/Passion-HD.24.04.03.Aria.Valencia.Good.Help.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {
+            "file_path": "/mnt/box/files/_raw/prt/PissVids.23.05.20.Juis.Wild.Black.Pee.1on1.BBC.ATM.Balls.Deep.No.Pussy.Rough.Sex.Gapes.Pee.Drink.Cum.In.Mouth.Swallow.GL836.XXX.1080p.HEVC.x265.PRT.mp4"
+        },
+        {"file_path": "/mnt/box/files/_raw/prt/PornMegaLoad.24.07.18.Justine.Jakobs.Hardcore.40443.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/PornPlus.24.10.18.Hailey.Rose.Naturally.Gifted.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Private.23.11.18.Dolly.Dyson.Enjoys.An.Anal.Threesome.With.Mary.Popiense.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Private.24.11.18.Lia.Lin.Gets.Horny.By.The.Pool.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/RKPrime.24.10.28.Octokuro.Fucking.After.The.Club.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/RickysRoom.23.03.09.Destiny.Mira.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/SexSelector.23.03.17.Graycee.Baybee.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/StrapLez.23.08.17.Alexis.Crystal.And.Lilly.Bella.Everything.She.Wants.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/StrapLez.24.11.14.Cherry.Candle.And.Lilly.Mays.Pleasure.Control.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/StrapLez.24.11.28.Alexis.Crystal.And.Mia.Trejsi.Rhythm.Of.Desire.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/TeensLoveHugeCocks.23.04.28.Kylie.Rocket.Influencers.Delight.XXX.1080p.HEVC.x265.PRT.mkv"},
+        {"file_path": "/mnt/box/files/_raw/prt/ThePOVGod.24.10.25.Willow.Ryder.The.Call.Of.That.Booty.XXX.1080p.HEVC.x265.PRT.mp4"},
+        {"file_path": "/mnt/box/files/_raw/prt/Tiny4K.23.02.02.Liz.Jordan.Ice.Cream.Treat.XXX.1080p.HEVC.x265.PRT.mp4"},
+    ]
+    with connect_server("klein") as ssh, connect_db() as con:
+        for row in rows:
+            file_path = row["file_path"]
+            try:
+                print("deleting", file_path)
+                ssh.exec_command(f'rm -rf "{file_path}"')
+                delete_video(con, file_path=file_path)
+                con.commit()
+            except:
+                pass
+
+
+@contextlib.contextmanager
+def connect_server(host: str) -> typing.Generator[paramiko.SSHClient, typing.Any, None]:
+    config = paramiko.SSHConfig.from_path(str(Path(r"~/.ssh/config").expanduser()))
+    key = paramiko.RSAKey.from_private_key_file(filename=str(Path(r"~/.ssh/id_rsa").expanduser()))
+
+    target = config.lookup(host)
+
+    ssh_client = paramiko.SSHClient()
+    ssh_client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
+    ssh_client.connect(
+        hostname=target["hostname"],
+        port=target.as_int("port"),
+        username=target["user"],
+        pkey=key,
+    )
+
+    yield ssh_client
+
+    ssh_client.close()
+
+
+def collect_all():
+    root = Path("/Volumes/FATBLUE/_backup/_vidz/_all")
+    video_files = find_vids(root)
+
+    # re_good = re.compile(r"(?P<people>.+) - (?P<title>.+) --? (?P<studio>\S+) (?P<date_year>\d{4})(?P<date_month>\d{2})(?P<date_day>\d{2})")
+    # re_good = re.compile(r"(?P<people>.+) - (?P<title>.+) --? (?P<studio>\S+)")
+    # re_good = re.compile(r"(?P<people>.+) - (?P<title>.+) --? (?P<date_year>\d{4})(?P<date_month>\d{2})(?P<date_day>\d{2})")
+    re_good = re.compile(r"(?P<people>.+) - (?P<title>[^-]+) (?P<date_year>\d{4})(?P<date_month>\d{2})(?P<date_day>\d{2})")
+
+    for f in sorted(video_files):
+        m = re_good.search(f.stem)
+        if not m:
+            continue
+        params = m.groupdict()
+        title = params["title"]
+        if "  " not in title:
+            print(f.stem, title, sep="\t\t")
+        continue
+        # better_name = f"{m['people']} -- @{m['studio']} -- {m['title']} -- {m['date_year']}-{m['date_month']}-{m['date_day']}"
+        better_name = f"{m['people']} -- {m['title']} -- {m['date_year']}-{m['date_month']}-{m['date_day']}"
+        # better_name = f"{m['people']} -- @{m['studio']} -- {m['title']}"
+        print(f.stem, better_name, sep="\n")
+        better_path = f.with_stem(better_name)
+        f.rename(better_path)
+
+
+def clean_images():
+    from PIL import Image
+
+    for img in Path(r"/Users/abdus/Downloads/temp/Ploy Tigerstam -- OnlyFans Leak").glob("*.jpg"):
+        image = Image.open(img)
+        width, height = image.size
+
+        if width < 2000 or height < 2000:
+            img.unlink()
+
+
+def main():
+    dry = True
+    try:
+        # clean_images()
+        # collect_all()
+        # save_suggested_names()
+        # delete_remote_files()
+        # transfer()
+        # different_format(dry_run=dry)
+        # local_duped(dry_run=dry)
+        # reencoded_self(dry_run=dry)
+        # dupes()
+        # scan_videos()
+        # list_studios()
+        # lower_resolutions(dry_run=dry)
+        # large_unwanted()
+        unwanted(dry_run=dry)
+    finally:
+        pass
+        # clean_records()
+
+
+if __name__ == "__main__":
+    main()