Audyt duplikatow po phashu miniatur wykazal dwa rozne zjawiska, ktore wygladaly
jak jedno.
1. Zdegenerowane phashe. 73 wartosci wystepowaly przy >=10 scenach kazda, lacznie
przy 4375 scenach; rekordzistka byla dzielona przez 892 sceny o 892 roznych
tytulach i 1886 roznych performerach. To zaslepki i czarne klatki, nie odciski
scen. find_by_phash_within bierze najblizsza wartosc z calej tabeli, wiec taka
zaslepka zawsze wygrywala z prawdziwym duplikatem (dist 0). Do tej pory bronila
nas bramka dur_prox i nic sie nie skleilo, ale to zabezpieczenie drugiej linii.
Blacklista jest tabela, nie jednorazowym DELETE, bo sam DELETE nic nie daje:
zaslepka wraca przy kolejnym ingescie. Trzeba pamietac, ze wartosc jest
bezuzyteczna. Job co 24h dopisuje nowe i czysci odciski. Po czyszczeniu zero
grup >=10, najwieksza pozostala ma 9.
2. Realne duplikaty. 4647 scalonych. Przyczyna byla jedna: 97 procent par ma
perverzije po dokladnie jednej stronie, bo pisze tytuly z prefiksem studia i
performera, a reszta tubow daje goly tytul. Wbrew mojej pierwszej diagnozie NIE
trzeba tu ruszac scoringu tytulu (token_set_ratio i tak radzi sobie z prefiksem,
a sciezka phash idzie przed composite): 80 procent par to dlug sprzed 60+ dni, a
biezacy wyciek to okolo 1 dziennie. Nadmiarowe wiersze: 18452 na 9449.
merge_phash_exact_dupes.py dostal wykluczenie blacklisty. Bez tego byl grozny:
sam klaster 892 scen dawal ~397 tys. par do rozwazenia.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Pelne sceny, nie klipy: mediana ~33 min, zero ponizej 10 min, ~140-240 dziennie,
listing chronologiczny mimo naglowka Featured.
Extractor sklada URL sam, bez tokenu i bez wygasania: id w iframe jest zapisane
wspak, a koncowka to maska bitowa jakosci (1/2/4/8 = 360/480/720/1080). Id bywa
numeryczne albo szesnastkowe, a to drugie to ~25% katalogu. Blokada CDN jest na
naglowku (bez UA 403, bez Referera 404), NIE na fingerprincie TLS, wiec telefon
przechodzi -> mobile_direct_ok. Zweryfikowane 206 + faststart do 1080p.
Metadane cienkie: brak studia, mediana 3 tagi, tytuly przepisane pod SEO. Za to
data jako unix timestamp i dlugosc co do sekundy. Bramka na obsade, bo sceny bez
performera scalily sie 0/23 w dry-runie.
Phash miniatury liczony celowo i to on robi robote: dry-run resolvera bez niego
dawal 31% scalen, pilot z nim dal 65% (seen 62, new 22, updated 40, errors 0).
85 z 89 performerow ma ref tpdb/stashdb.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Metadane ma najlepsze z tej serii: kanal = realne studio na 97% scen, 96%
performerow z refem tpdb/stashdb, dokladna dlugosc w sekundach, swiezosc w
godzinach, 120 scen na stronie listingu.
Nie wlaczamy go jednak do ALL_BROWSE_SCRAPERS, bo jednostka tresci jest zla:
to klipy 5-10 minut, nie sceny. Mediana 571 s na stronie 30 i 459 s na 200,
zero filmow >=20 min w calym katalogu. Pilot na 240 scenach dal 1 canonical
match, bo dlugosc i tytul, czyli sygnaly resolvera, sa przepisane pod tube.
Przy ~340 uploadach dziennie to ~10k nierozwiazywalnych wierszy miesiecznie.
Kod zostaje kompletny i zweryfikowany: POST /api/video-info przez Bright Data
(CF blokuje sam POST z VPS), HLS playerData.src zamiast mp4 sources (te maja
ip= requestera w tokenie), master konczy sie .mp4 wiec lapie go istniejacy
hls_needs_passthrough. Gdyby serwis zaczal wrzucac pelne sceny, wystarczy
dopisac klase z powrotem do rejestru.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Sitemapa i /watch/*.html są za Cloudflare, paginacja zwraca stronę 1, a load-more
chce logowania, więc katalog buduje się z homepage (~115 scen), a detale lecą przez
?link1=watch&id=. Bramka na obsadę przepuszcza 62/115; z tego 97% performerów ma
ref tpdb/stashdb, a 60% scen dopina się do tego, co już mamy.
Extractor czyta tablicę sources JWPlayera, nie JSON-LD contentUrl (ten jest zawsze
144p). Cap 720p, bo okcdn dławi połączenie do ~2,1 Mbps. Stream przenośny cross-IP
(206 bez Referera, faststart) → mobile_direct_ok.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
pornmike.com puts everything in JSON-LD @graph -> ItemPage.mainEntity
(name, duration, uploadDate, actor[], genre[], keywords, description,
thumbnail), so the parser reads one JSON blob instead of scraping markup.
Cast is clean: the whole scene page carries exactly as many /pornstar/
links as there are actors, none of the sidebar pollution that got xxxfiles
rejected. 82% of sampled performers already carry a tpdb/stashdb ref and
19-20 of 20 channels are studios we already know.
Ingest is gated on a non-empty actor[]: 23% of the catalog has no cast and
those scenes could neither be attributed nor deduped. Tags and categories
come only from JSON-LD (keywords + genre), never from the HTML, which
carries 38 /category/ and 22 /tag/ nav and sidebar links per page.
Pagination is ?p=N only: the /N/ form 404s and ?page=N is silently ignored,
returning page 1.
Playback is the simplest in the portfolio: a plain <source> mp4 on twincdn
with no token, no query string and no expiry. Verified 206 on a Range
request from the VPS and from another machine in another country, both
without a Referer, so it is neither hotlink-guarded nor IP-bound and the
phone streams it directly.
Two honest caveats recorded in the module docstring: these are 5-12 minute
clips (median ~487s against 1800-2400s for the tubes we accepted), so for
the ~20% of the catalog from Tushy/Blacked Raw/Milfy/Anilos they will sit
as a short shadow next to full canonical scenes; and uploadDate is the
tube's import date, not the studio release, so pages past the second are
marked backfill.
Pilot ingest of 3 pages: 127 seen, 49 merged into existing scenes, 66 new,
0 errors; on page one the gate passed 42 of 62 links with zero scenes
missing cast, duration or studio.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
fullvideosporn.com is sextu.com rebranded, not a clone of fullmovies.xxx
(different engine, different catalog, 0/140 title overlap with our
fullmoviesxxx corpus).
The site is only worth ingesting behind a gate: 65-75% of its catalog has
no performer at all, and those scenes also carry SEO-spun titles, so they
would never match canonical and would land as empty orphans. So we ingest
only scenes with at least one performer. That keeps ~25-35% of the catalog
(verified: 19 of 60 on page one) where the signal is good, since 88-89% of
the performer names in the research sample already resolve to a canonical
performer in our DB.
Three site-specific traps, all handled:
- Titles come from the player's vit:"..." field, not og:title/h1, which are
sometimes an AI SEO rewrite rather than the real scene title.
- Cast is read only from the <h3>Porn-stars:</h3> section; the page carries
~22 videos.php?q= links overall but only 1-2 real performers, the same
pollution that got xxxfiles rejected. Porn Site / Porn Categories are
separate h3 blocks and are parsed per-section so they don't bleed.
- Every fetch passes a cookie gate: a fresh session gets HTTP 429 plus a
small JS challenge, so we read the cookie out of it and retry on the same
session. Hence the custom crawl_page instead of the base browser_get.
The TXXX video_url decoder moved out of vjav into _txxx.py since both tubes
share the engine; vjav keeps an alias and was re-verified after the move.
Playback resolves videofile.php -> decode -> get_file -> 302 -> znvcdn, and
the final CDN URL is portable cross-IP so the phone streams it directly.
Note for future debugging: the VPS itself gets 429 from that CDN because of
datacenter IP reputation, so playback health-checks run from the VPS will
be falsely negative.
Pilot ingest of 2 pages: 39 seen, 11 merged into existing scenes, 28 new,
0 errors; 0/19 without cast or duration on the sampled page.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
pornbusy.com exposes every field as a discrete itemprop node (title,
duration, uploadDate, thumbnail, description) plus a dedicated
<div id="video-actors"> cast block with no sidebar pollution. Measured
orphan risk is low: 80% of a 100-scene sample have a performer that
already carries a tpdb/stashdb ref, and 21% of titles strongly match
scenes we already hold. No studio field exists on the site.
Homepage pagination is broken (pages 1/2/3 return an identical set), so
the listing is driven off sitemap_index.xml -> 10 post-sitemaps sorted by
lastmod, chunked in crawl_page the same way the PlayTube base does it.
Playback dispatches on embedURL and covers ~69% of the catalog: the
seekplayer family (already handled by the engine after the earlier host
regex widening) and zpi.cx (the embedURL is the file itself), plus loadvid
at ~41%, which needed new plumbing.
loadvid hands back the CONTENT of an m3u8 over POST /videos/resolve-token
(CSRF + videoToken from the embed page) and has no manifest URL at all:
GET on that endpoint is 405 and the guessable .m3u8 paths are 404. So
make_token grew an optional `producer` marker and /proxy/hls calls the
producer instead of GETting a URL. Segments in the returned manifest are
absolute and fully portable (verified: 206 on a Range request with no
headers, no referer, no token), so the phone still pulls them straight
from the CDN and only the manifest travels through the VPS.
upload18 (~12%) and the tail are deliberately left unresolved: their token
embeds the fetcher's /24, so resolving server-side would force the whole
video through the VPS.
Verified: 19 scenes/page, 19/19 with duration, 17/19 with cast, 19/19 with
tags (names read from the title attribute since an icon element precedes
the anchor text), playback 8/8 via loadvid, and /proxy/hls returning a
116KB manifest with 869 absolute segments. Pilot ingest of 3 pages: 59
seen, 35 merged into existing scenes, 24 new, 0 errors.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
galaxyporn.net carries paysite rips with a dedicated <div id="video-actors">
cast block (no sidebar pollution), studio + date in the title prefix, healthy
/page/N/ pagination. Measured orphan risk is low: 92% of a 75-performer
sample already have a tpdb/stashdb ref in our DB and 79% of titles match
scenes we already hold.
seekplayer_engine host regex widened to cover seekplays|4meplayer|ezplayer|
seeks|upn and the pro|cloud|one TLDs. This is the same engine (identical
AES key/IV and /api/v1/video endpoint), just newer domains; the whitelist is
additive so existing hosts keep matching (verified, including that
upns.evil.com is still rejected). Unlocks all four galaxyporn player hosts
and the same family on other sites.
Two things the extractor had to get right:
- Call seekplayer_engine directly rather than via extract_stream_from_hoster:
the wrapper verifies the resulting URL and the hotlink-guarded HLS 403s, so
it discarded correctly decoded streams.
- Referer must be the PLAYER origin, not galaxyporn.net. With the site referer
the manifest 403s; with the player origin it returns 200. Verified end to
end: manifest 200 -> variant 200 (570 segments) -> segment 206.
Duration is absent from the HTML, JSON-LD and player payload, and a NULL
duration would silently hide scenes behind the min_duration_sec filter (the
porntrex incident), so it is computed from the player's thumbnail.vtt last
cue. The player API throttles bursts, hence the retry/backoff: that took
missing durations from 9/21 down to 3/21. The value runs ~2% short of the
true length (one thumbnail interval), which beats having none.
Pilot: 21 scenes/page with studio 21/21 and clean cast (max 3), ingest of
2 pages = 42 seen, 34 new, 8 merged, 0 errors.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
youperv.com (DataLife Engine) carries paysite rips titled
"Studio - Performer - Title" (71% of a 132-title sample), ~60-70 new
scenes/day. Browse the homepage + /page/N/ (19 scenes/page, no overlap);
scene URL is /<category>/<id>-<slug>.html.
Cast is scoped to the fmeta block (up to the Related section): the whole
page carries 19-26 /xfsearch/pornstar/ links but only 1-2 are the actual
cast, so without scoping this would repeat the page-wide pollution that
got xxxfiles rejected. Studio comes from the title prefix, guarded so a
performer name is never mistaken for a studio. Duration, ISO release date,
per-scene tags and thumbnail all come from the same block.
Playback is a plain <source> mp4 (files.klubnichka-hd.com) with no token
or expiry, but the CDN hotlink-guards on Referer: bare Range gets 403,
Range + Referer + browser UA gets 206 cross-IP from the VPS. So the
extractor returns it with referer + mobile_direct_ok and the phone streams
straight from the CDN, no WebView and no proxy. Path is percent-encoded
because the filenames contain spaces.
Deep-crawl capped at 2000 pages: beyond that (<=09.2023) the catalog turns
into generic amateur uploads with no performers and dead CDN files.
Verified: 19 scenes/page with studio+cast+duration+date, max 2 performers
per scene (pollution guard holds), 0/19 missing duration, playback 206
video/mp4. Pilot ingest 3 pages: 57 seen, 39 attached to existing canonical
scenes, 18 new, 0 errors.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
1) Ekran gasl w trakcie ogladania (bug-reports fcbaa933 2026-07-18, c40fb79f
2026-07-23, dwoch userow). Android wygasza po timeoucie, bo dotyk nie pada, a samo
granie wideo nie trzyma wake-locka. expo-keep-awake NIE jest w top-level
node_modules (siedzi w node_modules/expo/node_modules), wiec metro nie rozwiaze
importu, ale natywny modul ExpoKeepAwake JEST w APK (expo go autolinkuje) -
siegamy po niego przez requireNativeModule z expo-modules-core (top-level).
Dzieki temu leci OTA, bez nowego APK. Wake-lock tylko gdy realnie gra (przy pauzie
ekran gasnie normalnie); WebView-player trzyma przez caly czas zycia ekranu, bo
tam nie ma sygnalu play/pause.
2) Miniaturki youporn (bug-report bebf8c92): youporn przeszedl dla czesci katalogu
na imgproxy z PODPISANYM URL-em (?hash=...&validto=<ts>), ktory WYGASA - zapisany
w bazie po czasie zwraca 410 (8666 z 62930 zrodel). Statyczne .../original/N.jpg
dzialaja bezterminowo. Nowy endpoint /proxy/youporn-thumb/<video_id> resolwuje
swiezy poster przy serwowaniu (jak sxyprn-thumb), z html.unescape bo og:image ma
& i psulo to podpis (403). Podmiana TYLKO dla podpisanych - statycznych nie
ruszamy, zero dodatkowych fetchy.
Zweryfikowane: 8/8 wczesniej-410 teraz 200 (57-142KB), statyczne bez regresji.
hqporner byl TYLKO w ALL_DIRECT_SCRAPERS (search po nazwisku performera), wiec
swieze sceny wchodzily wylacznie gdy worker trafil na pasujacego performera →
0 nowych scen w 7 dni (audit 2026-07-26, user: 'nie widze nowych scen z hqporner'),
mimo ze tube publikuje codziennie.
HQPornerBrowseScraper: listing / + ?p=N (~50 scen/strone), detail /hdporn/<id>-<slug>.
Metadane z detail page: tytul z <title> (Title Case; <h1> jest lowercase), dlugosc z
meta description ('Video duration is 46min 58sec'), performerki z /actress/ (display
name, nie slug), tagi z /category/, miniaturka fastporndelivery + phash (z Refererem,
bo CDN bez niego 403).
external_id identyczny jak w HQPornerScraper.search (hqpornercom:<url>) → obie sciezki
trafiaja w te sama scene zamiast robic duplikat. Tagi jakosciowe (1080p-porn, 4k-porn)
odfiltrowane, bo to atrybut pliku a nie tag tresci.
Zweryfikowane na zywo: 50 scen z page 1, dur 2818s/1138s/1805s, performerzy+tagi+phash.
1) eporner/phone-resolve: gdy strona parsuje sie OK, ale CDN jest nieosiagalny z
sieci usera (audit: siec zglaszajacego nulluje vid-*.eporner.com -> 0.0.0.0, a
www.eporner.com dziala), phone re-resolve zwracal TEN SAM URL, a lancuch fallback
nie ratowal: dla phone-resolved linkow fallbackProxyUrl === url (resolver ustawia
stream_url = direct_url), wiec krok proxy byl no-opem i szlo prosto do WebView =
strona hostera z reklamami ('zaladowala sie reklama zamiast wideo'). Teraz po
nieudanym phone-resolve bierzemy PRAWDZIWE proxy z backendu (VPS pobiera swoim
DNS-em). Tylko na sciezce bledu. Eporner globalnie zdrowy (89-100%, ~100 urzadzen
dziennie), wiec to fix dla userow z blokowanym CDN, nie regresja globalna.
2) porntrex zmienil markup kafelka: class="duration">MM:SS< ->
<div class="durations"><i class="fa fa-clock-o"></i> MM:SS</div>. Stary wzorzec
przestal matchowac -> duration_sec=NULL dla ~95% swiezych scen (6479/6815 w 7 dni),
a mobile domyslnie filtruje min_duration_sec=60, wiec te sceny byly NIEWIDOCZNE
('nie widze nowych scen z porntrex'). Wzorzec obsluguje oba warianty + H:MM:SS.
Zweryfikowane na zywym listingu: 84/85 kafelkow, 20:45 -> 1245s.
1) _build_scene_out (detal) nie stosował _sxyprn_thumb_url jak lista → detal
pokazywał martwy trafficdeposit URL, refresh nie pomagał (sxyprn IP-bound). Teraz
detal też podmienia na żywy /proxy/sxyprn-thumb/<id>. Backend-only.
2) Mute WebView-scrape: attribute-autoplay (<video autoplay>) omijał override play()
i grał unmuted. Dokładam proaktywne ciszenie na play/loadeddata/volumechange +
poll 40ms do gestu (property-lock trzyma między tickami). Web Audio i tak gated
autoplay-policy. Raport usera: dalej głos przed ExoPlayerem.
Ten sam tag z różnych tubów tworzył osobne wpisy (Blowjob/Blow Job, Big Tits/
Bigtits, deep-throat/deepthroat, creampie-1/creampie, mojibake CJK 'Japanese'…)
bo resolve_tag brał slug/slugify(name) verbatim. ~11k nadmiarowych tagów.
- resolve_tag: przed utworzeniem nowego tagu szuka istniejącego kanonicznego po
alnum-kluczu nazwy (_resolve_by_altkey) i reużywa go → warianty nie regenerują.
- Indeks funkcyjny ix_tags_name_altkey (migracja 0028) pod ten lookup.
- scripts/merge_altkey_tags.py: jednorazowy backfill istniejących duplikatów
(kanoniczny = max scene_count, scene_tags/movie_tags/blacklisted_tags przepięte
z dedupem na PK, dropy skasowane, taxonomy_counts odświeżone).
Kategorie/studia podszywające się pod performera (over-attribution z luźnego
tube-searcha: 'Big Tits', 'Brazzers', 'Ball Licking') zaśmiecały ranking similar
— klastrował po nich zamiast po treści. Odsiew: slug performera pokrywa się z
tagiem I brak kanonicznego refa (tpdb/stashdb). Prawdziwi performerzy z tym samym
tagiem mają kanoniczny ref → zostają (Jax Slayher/Angela White OK). Scena z samymi
śmieciowymi performerami spada na tag-matching (i pusto gdy ma tylko mega-tag).
Backend GET /scenes/{id}/similar: ranking po WSPÓLNYCH performerach (waga 3) i
tagach (waga 1). Performerzy selektywni → mała pula kandydatów (cap 1500); tagi
liczone tylko w tej puli. Fill z tagów (bounded 400, HAVING>=2 wspólne) tylko gdy
performerów mało/brak (anonimowe tube sceny) — inaczej szeroki tag skanowałby
setki tysięcy scen. Filtr widoczności jak lista: grywalne + brama JAV + blacklist
+ nie-stub + nie-self. Ranking/sort w Pythonie nad ograniczonymi zbiorami.
Zmierzone: 210-944ms wg liczby performerów seeda.
Mobile: SimilarRail (poziomy) na dole SceneDetail, lazy przez react-query, znika
gdy brak dopasowań, tap → nav.push kolejnego SceneDetail.
freshporno gra natywnie (zweryfikowane na realnym telefonie, PlaybackState=
PLAYING) ale udane odtworzenia nie raportują się jako success event → sam błąd
w telemetrii → 0% success → health=0 → fałszywe 'offline' i demote.
_overall_stars: gdy health==0 ale źródło jest ewidentnie żywe (freshness>=4 i
richness>=4), floor 2★ 'flaky' zamiast 0★ 'offline'. Martwe źródła mają niski
freshness więc gate dalej działa.
The JAV vertical gate excluded JAV origins whenever jav=false, including when the user
browsed a specific JAV site (Sites -> javflix/vjav/supjav -> SiteScenes passes
origin=... without jav=true), so those pages showed "No videos" (3 reports 07-19).
Only apply the exclusion to the main feed (origin is None). Browsing a specific origin
shows it regardless; the JAV tab (jav=true) still shows only JAV. Verified: javflix/vjav/
supjav now return scenes via origin filter, main feed still gates JAV out.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
No-accounts app, so favorites/saved-searches/hidden-list/watch-progress live
per-device (keyed by X-Device-Id). Users asked how to move them to a new phone.
Backend: POST /me/import/{source_device} copies all device-scoped tables from a
source device to the caller (dedup via NOT EXISTS, new UUIDs for id-PK tables like
saved_searches). Copy not move, so the old phone keeps everything; idempotent. The
device_id is a random UUID so knowing it is the authorization (fine for a keyless app).
Mobile: Settings -> "Backup & sync" shows this device's sync code (copy button) and a
Restore field to paste the other device's code; invalidates favorites queries after.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
supjav is an embed-aggregator (voe/streamtape/turbovid/fc2stream). It was playing
through the WebView fallback because none of its hosters had a native phone-side
resolver wired into the scene player. voe.sx does have one (voeResolver.ts, already
used for movies) — it just was not wired for scenes.
- PlayerScreen: wire isVoeUrl/resolveVoePage into the hoster routing (like
dood/filemoon), so a voe embed resolves on the phone to a direct m3u8
(token bound to the phone /16, plays direct, no WebView, no proxy).
- supjav extractor: return voe first (sorted by URL), and stop putting the raw
hostname in `quality` (the digit in "fc2stream" was parsed as quality 2 and won
_pick_best) — use a clean digit-free server label instead.
- SceneDetailScreen: for hoster-only scenes, open the first embed (the extractor
puts the natively-resolvable one first) rather than _pick_best's choice.
Verified: supjav's voe embed resolves to a real HLS manifest (200,
application/vnd.apple.mpegurl, #EXTM3U). WebView stays only as last-resort fallback.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fourth JAV vertical source (origin tube:supjav, gated to JAV tab via JAV_ORIGINS).
Browse: CF-blocks datacenter IPs, so listing goes through the Bright Data ISP proxy
with retry (the proxy rotates IPs and CF only lets some through). Homepage-latest only
(~66 newest); deeper /page/N/ stays CF-403 even via proxy. Metadata parsed from the
listing cards (title, thumbnail, JAV code, year/month) rather than per-post details,
which are flaky and slow through the proxy.
Stream: supjav hides the real hoster behind a per-server data-link (hex). base.js loads
lk1.supremejav.com/supjav.php?l=<data-link>, which reverses the hex string and fetches
?c=<reversed>, 302-ing to the hoster (TV->turbovid, FST->fc2stream, ST->streamtape,
VOE->voe). The extractor reproduces that: fetch detail via proxy, reverse each data-link,
resolve through lk1 (reachable direct from the server), return the hosters as type=hoster
so the phone resolves them (dood/filemoon native, the rest via the WebView fallback on the
residential IP). lk1 needs no proxy; only the detail fetch does.
Backend-only, no mobile change.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Second JAV vertical source (origin tube:vjav, gated to JAV tab via JAV_ORIGINS).
Browse: SPA has no server-rendered listing, so id-walk over sitemap_vids
(newest = highest video id) feeding the rich JSON metadata API
(/api/json/video/1/<floor>/<id>/<id>.json): title, duration, post_date,
channel->studio, models->performers, categories+tags.
Stream: videofile.php returns video_url in two obfuscation layers, Cyrillic
homoglyphs (M/C/A/E) over a custom base64 alphabet (comma->slash, tilde->pad,
dash->plus). Decoded get_file is absolute (old shared-txxx videos) or relative
(new vjav-infra, prepend host); adding f=video.m3u8 yields a portable, time-bound
ahcdn HLS (referer=none whitelisted, not IP-bound). Returned as m3u8 +
mobile_direct_ok so playback routes it through /proxy/hls: manifest passthrough,
segments direct from the phone (verified 206 cross-IP, 0 VPS video bandwidth).
Backend-only, no mobile change (JAV tab + jav param already shipped).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
JAV is a distinct vertical (user decision): Asian codes/titles that do not dedup
against the western catalog, so they must not flood the main feed.
Backend (scenes.py): JAV_ORIGINS = {tube:javflix, tube:javguru, tube:vjav,
tube:supjav}; list_scenes gains a `jav` param. Default (jav=false) excludes any
scene with a live JAV-origin source; jav=true returns only those. The cached
default-count and _is_pure_default also exclude JAV so the main feed count matches.
JavflixScraper is now registered in ALL_BROWSE_SCRAPERS (scheduled ingest lands in
the JAV section, gated). Scraper hardened: requires a real server button
(class="myLink") so static pages (Terms/FAQ) are skipped, and unescapes HTML
entities in the title.
Mobile: a "JAV" top tab reuses ScenesScreen with { jav: true } (route param ->
listScenes jav=true). The 60s minimum-duration default is disabled in the JAV tab
because javflix does not expose duration (NULL >= 60 would hide the whole section).
Verified on prod: 16 javflix scenes appear only in the JAV feed and are excluded
from the 2.29M main feed; playback resolves to voe/doodstream/emturbovid.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First site of the JAV vertical (user request, separate section). javflix.cc is a
WordPress JAV aggregator, VPS-reachable and server-rendered. Scraper parses itemprop
metadata (title, thumbnail, uploadDate; no duration) and the JAV code (BKD-368 style)
as a tag. Playback: server buttons are <a class="myLink" href="<embed>"> (streamtape,
voe, doodstream, emturbovid) whose href sits in the raw HTML; the generic _embed_iframe
extractor already resolves those via its anchor-hoster pattern, so javflix registers
under sitetag "javflix" with a thin wrapper that drops the players.mp4 placeholder.
Added emturbovid to the anchor-hoster host list.
Verified end-to-end on prod: scrape a listing -> RawScene with metadata + code tag,
resolve -> emturbovid/voe/doodstream hoster sources (all Goon-playable).
NOT registered in ALL_BROWSE_SCRAPERS yet. JAV is a separate vertical (user decision)
that must be gated out of the main scenes feed before ingest so it does not flood the
western catalog. Next: feed gating (exclude JAV origins by default) + a mobile JAV tab
+ on-device playback test, then enable scheduled ingest. jav.guru / vjav / supjav
follow the same pattern.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
extract_stream_from_hoster fell through to yt-dlp for DoodStream clones (playmogo,
doply, myvidplay, dood.*), which yt-dlp does not support -> "ERROR: Unsupported URL"
spam dominating the movie-resolve logs, plus wasted seconds per resolve before the
caller returned the embed as type='hoster' anyway. Now detects the dood-clone family
up front and returns None immediately (same pattern as the filemoon SPA handling), so
the embed goes straight to the mobile doodstream.ts resolver (pass_md5 with the user's
residential IP, which clears the invisible Turnstile). Verified: playmogo/myvidplay/
doply URLs now return None with no yt-dlp attempt.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Scenes feed repeated the same videos on scroll (report 304c9258): OFFSET
pagination + continuous ingest means new scenes shift the offset, so later pages
re-include earlier rows. Backend now appends Scene.id as a unique final tiebreaker
to every sort (deterministic order), and the mobile Scenes list dedups by scene id
when concatenating pages so a drifted overlap can never show a duplicate.
- Player: holding to fast-forward at 2x now mutes the sped-up (chipmunk) audio for
the duration of the hold and restores the user's real mute preference on release
(report 35bbf428).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
hqfap was hard-removed 2026-06-25 when its entire CDN library started serving a
fixed /upload/videos/video_down.mp4 "server down" stub for every scene. The site
is back and now serves real video from a new CDN (d*.vstor.top/whlvid/...). Probe
of the live sitemap: 5/6 sampled scenes return a real mp4 (0 stubs), and the CDN
is portable cross-IP (HTTP 206 from a residential IP and from the VPS with the
hqfap referer).
Restores the PlayTube CMS browse scraper (HQFapScraper, sitetag hqfapcom) and the
JSON-LD contentUrl extractor, re-registered in both registries. The extractor now
sets mobile_direct_ok=True (vstor.top is not matched by _TIME_BOUND_CDN_RE, but is
verified portable) so mobile plays direct with zero proxy. The video_down.mp4 stub
guard is kept defensively. 4k69 stays removed (not re-checked).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two bug-report clusters about missing thumbnails:
1. watchporn scenes had no thumbnail on the list, only appearing after opening
SceneDetail (which auto-enriches og:image). Coverage was 1.3% (463/36,915).
The browse scraper never captured a thumbnail. KVS stores the poster at a fixed
derivable path (contents/videos_screenshots/<id//1000*1000>/<id>/preview.jpg,
verified 8/8 loading). Scraper now sets thumbnail_url (og:image, else derived);
backfilled 36,687 existing rows -> 100% coverage.
2. SceneDetail showed no thumb where the list showed one: the mobile detail picks
the first source with a thumbnail_url (origin ASC often puts sxyprncom first),
and sxyprn/trafficdeposit stored thumbs rot to 404. The list already swaps those
for a live resolver (/proxy/sxyprn-thumb/), but the detail builder did not. It
now applies the same live-resolver swap and nulls other rotting thumbs so the
detail lands on a working image.
Both backend-only, no OTA needed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Generic movie titles ("Monster Tits", "Pirates") map to many different TPDB
films with identical titles (different studios/casts, all title-score 1.0). The
old matcher searched per_page=10 and ranked by title only, so for a generic title
the correct film was often not even in the top 10, and among same-title
candidates it picked arbitrarily. Result: a ~95% no_match rate and silent
misattribution (e.g. "Monster Tits" by Venom Digital Media would get Galaxy
Productions' TPDB entry).
_best_match now:
- searches per_page=40 (the right film for a generic title is often past top 10),
- ranks title-gate survivors by a composite of title + studio similarity +
performer overlap (our studio/cast from the primary source disambiguate which
same-title film it is),
- guards against misattribution: if we have a studio/cast signal and there is
more than one near-identical-title candidate but the winner shares neither
studio nor cast, return no_match instead of attaching a wrong same-title film.
Verified on prod data: a no_match-with-studio sample now matches 18/18 with the
correct studio (fixing Galaxy to Venom, Exquisite to Rodney Moore, and a no_match
to Cherry Boxxx), and an already-enriched sample keeps 16/18 identical picks with
the 2 differences being the same studio (benign TPDB duplicate). No wrong-studio
regressions observed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The zload CDN URL (remote_control.php) is a valid faststart progressive mp4 that
ExoPlayer plays fine when fresh: 206, no IP/referer binding, ~2h token TTL,
verified from a residential IP. Real-device failures (player_error/gone
telemetry) come from the static URL going stale: a rotated or dead CDN node, or
an expired token.
Instead of falling back to the tube page in a WebView (removed, it papered over
the real issue), PlayerScreen now re-resolves the scene fresh via the backend on
initial-load error for backend-native KVS tubes (watchporn). That yields a live
node plus a fresh token and stays on the native direct stream (0 VPS bandwidth).
For these origins a CDN 'gone' (404) means stale URL, not deleted, so we
re-resolve on gone too; a genuinely deleted post raises HosterDead (410) and
falls through to the normal fallback chain.
Also: add playback_events.error_detail (raw ExoPlayer message) to pin down the
exact failure of tubes we cannot reproduce on the emulator. Keep FLAG_SECURE on
release builds only (!__DEV__) so debug builds stay screenshottable for local UI
verification.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
From user bug reports (8f8c10c0 freeomovie, c156e4b7/8db71220 watchporn scene):
freeomovie (app/connectors/freeomovie.py):
- Parse the detail-page poster (<img class="rmbd">); RawMovie was built with no
poster_url, so title-only orphans (no canonical match) rendered blank cards. Re-ingest
backfilled existing blanks 11 -> 1.
- _host_label now returns the registrable domain (parts[-2]) not the subdomain, so
video.player4me.xyz labels as 'player4me' (was garbage 'freeomovie:video').
- Skip player4me (JS-SPA, no known resolver); pruned 28 dead freeomovie:video sources.
- myvidplay.com added to mobile DOOD_HOSTS (doodstream.ts): it 301s to playmogo.com
(doodcdn clone), so it now routes to the native dood resolver instead of a dead WebView.
watchporn: the reported 4 duplicate/broken quality rows were the API process serving a
STALE extractor registry (never restarted after the re-enable deploy) -> _embed_iframe
scraped 2 tokenless get_file (403) + 2 preview trailers (404). The running api now serves
the native extractor (2 clean playing links, verified). Hardened _embed_iframe to skip
preview/videos_screenshots clips and tokenless get_file so this class can't recur.
Also: DonateScreen em-dashes -> commas/semicolon (no-em-dash rule).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
freeomovie.to (WordPress bestia theme, not dooplay) publishes fresh
full-length DVD titles daily. New standalone BaseMovieConnector scoped to
/category/full-movie/ (homepage mixes scene-clips that would orphan as
movies). Parses the listing thumi items for clean titles, and the detail
var TABS array for playback hosters (voe/luluvid/vidhide resolve VPS-side;
myvidplay is a DoodStream clone, phone-side), skipping dead streamtape/
mxdrop. Genre tags come from articleSection (the part before "XXX Movies");
release_year is deliberately NOT set (datePublished is the post/upload date,
not the film's production year, which would poison year-scoring on canonical
match) and performers/studio are skipped (articleSection has no clean
delimiter -> junk-performer risk) -- the movie rides in as a title-trigram
mirror and TPDB enrichment supplies cast/studio/tags authoritatively.
Verified pilot: 10 movies -> 9 attached to existing canonical movies
(mirror playback), 1 new, 0 errors.
Follow-up (mobile, not required for playability since voe/luluvid cover it):
add 'myvidplay.com' to DOOD_HOSTS in doodstream.ts + OTA for a 4th hoster.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
watchporn.to was disabled 2026-05-12 (all iframes were DoodStream/CAPTCHA
-> black-screen WebView). The site has since been rebuilt as a KVS tube:
the DoodStream hosters are gone, scenes now expose flashvars get_file
direct mp4. Rewrote the dead ?s= search scraper as a /latest-updates/
browse scraper (og:title, JSON-LD duration/uploadDate, /models/ performers,
/tags/ tags, /categories/ studio) and added a KVS get_file extractor
(porntrex-style same-session 302 resolve; token not IP-bound, plays from
VPS). Moved to ALL_BROWSE_SCRAPERS, extractor swapped from _embed_iframe.
Verified: 35 fresh scenes/page (today), multi-performer, playback 1080p mp4 206.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
streamporn.nl went stale (0 new since 2026-05-23); streamporn.vip is the
live sibling on the same PsyPlay theme with a fresher catalog. Same
DooplayConnector works out of the box (verified: full title/studio/
performers/tags/poster + 11 playback sources per movie). New source name
'streampornvip' so origins don't mix with the dead .nl rows. Swapped into
get_movie_connectors in place of the dead .nl.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Enrich existing movies (from paradisehill/dooplay, which mostly lack cast)
with metadata from TPDB's /movies API: cast, categories (tags), studio,
director + a canonical TPDB UUID for dedup. Chosen over IAFD after a
source-comparison research pass — IAFD has strong cast/studio but ZERO
categories, while TPDB /movies has ~11 tags/movie, cast, studio, director,
a canonical UUID (+ sparse phash), is already an integrated API (no
scraping/anti-bot), and covers ~75-85% of our western-DVD-feature catalog.
Enrichment only ever augments EXISTING movies and never creates new ones
(TPDB has no playback, so a standalone TPDB movie would be unplayable).
Writes to movie_performers / movie_tags / movie.studio_id, which the movies
API + mobile detail already render, so no schema/API/UI change is needed.
- connectors/tpdb.py: search_movies() + fetch_movie() + _parse_movie()
reusing the existing _parse_studio/_parse_performer/_parse_tag.
- enrich/tpdb_movies.py: match our movie to a TPDB /movies result by
token_sort_ratio on normalized titles (sort, not set, to reject the
short-title-subset trap "Fantasies" -> "Tara's Fetish Fantasies") with a
+/-2yr guard; then attach cast/tags/studio/director. Incoming performers
deduped by external_id to avoid the performer_external_refs PK clash.
- resolve/movie_merge.py: merge_movies() mirror of scene_merge; two of our
movies mapping to the same TPDB UUID are the same film -> merge.
- scheduler: _job_tpdb_movie_enrich every 6h, batch 200, prioritizing
playable movies missing cast/studio.
Verified on a 150-movie batch: 119 enriched, 4 deduped, 26 no-match,
0 errors; matched titles/studios spot-checked correct (Big Butts Drive Me
Nuts 4 -> 33 tags, Seinfeld #2 -> 10 cast/17 tags, German BB Video titles
-> categories+studio).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-up cleanup from the ultra-review (behaviour-preserving, verified equivalent):
- scenes.py: extract live_playback_exists / blacklist_clauses / stub_exclusion_clause;
list_scenes and favorites now share ONE definition of "visible scene" (was a hand-kept
copy in favorites -> the count-vs-list drift class). Verified identical: helper vs old
inline both count 1,853,327 scenes.
- favorites.py: replace the two copy-pasted count blocks with one _new_counts(kind=...)
that counts in SQL (count(*) FILTER over the windowed subquery) instead of streaming up
to N*200 rows to Python; joins the favorite table for per-row last_seen. Deployed
_new_counts verified == hand SQL (studios 364). Dropped now-unused imports.
- mobile: extract lib/newScenes.ts (isNewScene / sortNewFirst); SceneTile + Performer/
StudioScenes use it (was triplicated, already drifted once in the deleted FavoriteSceneRow).
- SceneDetail: fold tube:pornxpph into the phoneResolver map (was a verbatim-duplicated block).
- deep_crawl: persist the cursor after every completed page (was once at run end), so a
mid-page hard-kill past the soft budget can't lose progress (GOON-V hardening).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Addresses the ultra-review findings on this branch:
Player (PlayerScreen.tsx): the new recoveryPending mirrored the fallback-chain guards
by hand and could deadlock into a permanent "Reconnecting" spinner with no way to Mark
broken — for gone (410) sources on IP-bound tubes (re-resolve bails before setting
reResolveDone) and for any post-load error on those tubes (re-resolve is initial-load
only). Derive one reResolveApplicable flag (IP-bound AND initial-load AND not-gone) and
use it for both the chain gate and the spinner, so gone/post-load errors fall through to
proxy/WebView or the terminal error card. Seek-recovery now falls through to the chain
when player.replace() throws instead of returning.
Quick-play (SceneDetail): the autoplay route param persisted and autoPlay={i===0} re-fired
when the source list reordered (e.g. after Mark broken drops the dead source), bouncing the
user into the player. Consume it once via onAutoPlayConsumed -> nav.setParams({autoplay:false}).
Backfill semantics: performer-driven direct-scraper "backward fill" now tags scenes
backfill=True (search-by-name pulls the whole old catalog); merge coalesces backfill
(keep AND drop) so a fresh scene merged into a dead dup keeps NEW; deep-crawl only tags
backfill on a tube's FIRST sweep (swept_once) so re-sweep catalog growth stays genuine;
pilot script tags backfill.
Perf/migration: migration 0026 is now idempotent (IF NOT EXISTS; prod got the column via
manual ALTER) and adds ix_scene_performers_performer_id (favorites count filtered
performer_id with no index); index also created on prod.
Cleanup: deleted dead FavoriteSceneRow (unused import in two screens, stale isNew without
the backfill guard); removed em-dashes from all lines this branch added (user CLAUDE.md
rule), including the user-facing changelog / Settings / player-overlay strings.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Browse scrapers backfilling old catalogs stamp the tube's import/post date as
release_date, so old content (e.g. 83 MissaX classics via perverzija, ~3600/3 days
across eporner/youporn/etc.) fake-ranked as newest and flooded the favorites "+N".
NULLing the dates was a non-starter — the stub filter would hide 251k performer-less
scenes. Instead: a Scene.backfill flag marks bulk catalog imports; they stay visible but
never count as "new".
- scenes.backfill column (+ index, migration 0026); resolve_scene/_process_scene thread it.
- deep_crawl tags scenes from pages beyond the "latest" threshold (>2) as backfill;
latest pages + TPDB/StashDB delta stay genuine. Cursor reset re-sweeps page 1 so real
new content is always caught fresh.
- favorites +N (performers + studios) excludes backfill within the top-200 window.
- SceneOut exposes `backfill`; mobile NEW badge + NEW-first re-sort skip it (badge==count).
- Retroactive: tagged 439k existing scenes in bulk (>10 non-canonical / studio / day)
clusters. Device check: favorite-studios +N 11626 (naive) -> 236.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The "+N new" badge counted every live-playback scene with created_at > last_seen_at,
but the performer/studio list the user opens applies the device blacklist AND only shows
the first page (per_page=200, sort release_date desc) — so a freshly-ingested scene with
an old release date, or a blacklisted one, inflated "+N" while being invisible ("+6, no
new scenes"). Count now mirrors the list: same visibility filters (live playback +
blacklist + stub-exclusion for studios) and the same top-200 window under release_date
desc. On the main device this dropped the naive total 1663 -> 1406 (phantom off-page
scenes), and blacklisted content no longer inflates counts.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The NEW badge keys off scene.created_at. merge_scenes kept the survivor's created_at,
but the dedup caller may pick the freshly re-ingested mirror as keep_id — so deduplicated
old content got a recent created_at and showed up as NEW (report f17799b3). Coalesce to
min(keep, drop) so a merged scene keeps its first-seen date.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
run_deep_crawl picks one tube/run and crawls 60 pages under _job_deep_crawl's 3600s
hard timeout. A detail-fetch scraper on a slow patch (per-scene page fetch, e.g. via
proxy) could exceed it → the run is killed mid-page, the cursor is never saved (orphan
thread), and that tube makes zero progress — recurring Sentry GOON-V. Added a 3000s
in-run budget that breaks after a completed page, saves the cursor, and returns cleanly;
the next run continues. budget_hit surfaced in the summary log to spot the slow tube.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
freshporno (5★ fresh+rich, verified working — 206/507MB) was labeled OFFLINE off 10
playback attempts that all failed in one unlucky window (a CDN-node blip; it resolves
fine now). 10 was too thin a sample to zero out a known-good source's stars. Raised
the telemetry-trust threshold 10→25; below it we fall back to the proxy/heuristic
health instead of declaring offline (user-report cb526949).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A single request with 194 studio_slugs + 23 tag filters (each tag = a correlated
EXISTS) plus an ILIKE search built a query heavy enough that the OOM killer killed the
Postgres backend, triggering a full crash-recovery (~1s prod-wide outage, all in-flight
connections dropped). Any user could do this with a big enough filter. Cap studios to
50, tags to 15, performers to 15 (far above any real UI usage) and return 422 instead
of executing — bounding query complexity regardless of the planner's choice.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Re-check 2026-06-25 across the full id range confirmed both PlayTube tubes
serve only the fixed `/upload/videos/video_down.mp4` "server down" stub, never
a real file: hqfap 0/80 real (79 stub, 1 none), 4k69 0/40 real (38 stub, 2
none). Both were disabled 2026-06-22; CDN never came back, so removing entirely
(mirrors the pornhub/redtube/0dayxx/pornditt/pornhat removals).
Removed the extractor registry entries (hqfapcom, 4k69com) + module files and
the browse scrapers + imports. Prod DB data deleted separately (28,398
solo-orphan scenes + 46,196 playback_sources). `_playtube.py` kept: superporn
and neporn still use its JSON-LD helpers.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_embed_iframe returned xtremestream's player endpoint (player/xs1.php?data=) labeled
type=mp4, but it's an IP-bound JS player page (403 cross-IP), not a real file — the
native player loaded it forever ("perverzija nie działa" / "loading w nieskończoność").
Added xtremestream.* to _IP_BOUND_CDN_RE so Stage 1 skips it and falls through to the
hoster fallback: the phone WebView loads the index.php player with its residential IP
and the stream plays in-session.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Browse over /best/<YYYY-MM>/<page> (SSR; xnxx has no clean /new/ and its homepage is
JS-rendered) for a latest-feed freshness signal next to the performer-driven search
scraper. JSON-LD VideoObject only — xnxx detail (unlike its xvideos twin) doesn't
expose /models/ or /tags/ in SSR, so performers/tags come via canonical merge + the
search scraper. Title is html.unescaped (JSON-LD ships ,/! entities).
xhamster and sxyprn intentionally left search-only: xhamster Cloudflare-blocks the
VPS on listing pages (1KB challenge), sxyprn has no clean SSR listing (IP-bound) —
a flaky browse scraper would be worse than the working search + 168h watchdog.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Browse over /browse/time/?page=<n> (SSR) for guaranteed latest-feed freshness next to
the existing performer-driven search scraper. JSON-LD VideoObject only (title /
duration / uploadDate / thumbnail) — deliberately NOT scraping performers/tags from
the detail page: JSON-LD has no actor field and the /pornstar//category links are
sidebar-polluted with no scene-scoped container, so a naive regex attached the same
2 pornstars to every scene. Performers/tags come via canonical merge + the search
scraper instead.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Both were search-only — fresh only as long as the performer queue cycles and the
site search keeps working. Added browse scrapers next to the existing search ones
(xvideos/eporner pattern: search keeps performer back-catalog coverage, browse
guarantees latest-feed freshness → watchdog 48h instead of 168h):
- porntrex: KVS /latest-updates/<n>/ (title + thumb + phash)
- mypornerleak: WP REST /wp-json/wp/v2/posts?_embed=1 (title + date + studio from
category + performers from the actors taxonomy)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Watchdog flagged porndish as frozen (search ?s= stopped yielding new scenes
2026-05-07, 1151h). It's WordPress and the VPS can reach it, so converted to a browse
scraper over the WP REST API (/wp-json/wp/v2/posts?_embed=1), same pattern as
perverzija: title, date, featured thumbnail, studio (category — FreeUseFantasy /
I Have A Wife / … paysite content) and tags. Performers via canonical merge. Playback
unchanged (embed iframe → phone-side). 60 fresh scenes on first crawl.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Search (?s=) started returning 429 and the homepage is JS-rendered (no post links in
raw HTML), so the old search scraper got 0 (frozen since 2026-05-07). perverzija is
WordPress and the VPS can reach it (200, not CF-blocked), so converted to a browse
scraper over the WP REST API (/wp-json/wp/v2/posts?_embed=1): one structured call per
page gives title, date, featured thumbnail, studio (category — DadCrush/FamilyStrokes/
… TeamSkeet-family paysite re-ups) and genre tags. Performers via canonical merge
(stars taxonomy isn't REST-exposed; title carries names). Playback unchanged (embed
iframe → phone-side). 15 fresh + 45 refreshed on first crawl.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
