Commit graph

165 commits

Author SHA1 Message Date
jtrzupek
c154deab37 feat(sources): 0-5★ ranking on Sites (freshness/metadata/plays) + playback telemetry
Rates each source on three axes the user asked for:
- freshness: how recently/often new content arrives (newest age + 7d volume)
- richness: metadata coverage (thumbnail/tags/performers/description/studio/duration)
- plays: does it actually play — from real playback telemetry when available,
  else a proxy from the resolve mechanism. 0★ = offline (gates the overall stars,
  so a fresh+rich source that doesn't play still ranks bottom — the hqfap/4k69 case)

Backend:
- playback_events: fire-and-forget telemetry POST from the app per playback attempt
  (origin + success/error + time-to-first-frame), append-only, 30d retention
- source_stats: per-origin computed scores, refreshed by a scheduler job (6h);
  /sources joins it and sorts by stars
- models + local migration 0025; new GOON_SCHED_SOURCE_STATS_HOURS setting

Mobile:
- Sites rows show ★ rating; tap the stars for a breakdown (axes + metadata %, plus
  whether "plays" is measured or estimated)
- PlayerScreen reports playback success/failure per source (native path only —
  symmetric, conservative); origin threaded through Scene/Movie play callsites

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-22 10:00:59 +02:00
jtrzupek
f34a75f4c6 feat(ingest): disable hqfap/4k69 (broken playback), latestpornvideo → browse
- hqfap + 4k69: both ingested fresh but playback is dead (hqfap serves a fixed
  ~3MB "server down" stub for every scene; 4k69 resolves no playable URL).
  Removed from ALL_BROWSE_SCRAPERS so no new dead sources get ingested; existing
  live playback_sources marked dead in prod (scenes drop out of has_playback /
  Sites). Extractors kept in registry for easy re-enable if the hosts recover.
- latestpornvideo: was a performer-search scraper, so it never picked up the
  site's "latest" feed — users saw a stale set. Converted to a browse scraper
  reading /page/N/ (studio+date from title/thumb, category tags; performers via
  canonical merge). Moved DIRECT → BROWSE list.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-22 09:34:47 +02:00
jtrzupek
4afebacad8 feat(mobile): movies — performer filter + 3-column grid
Two Movies-list reports. (1) 1044cd34 'do movies have a metadata base for performers/categories/studio/year': yes — 90% have year, 92% studio, 81% performers, 93% tags, and the filter already covered studio/genre/year. Added the missing dimension: a performer search-and-select in MovieFiltersSheet (backend listMovies + api.ts already accepted performer_ids; only the UI was missing). (2) 0200956f 'use the space better': Movies grid goes 2 -> 3 columns (poster card is flex:1, scales fine) so ~50% more films per screen.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 23:21:22 +02:00
jtrzupek
960bc75be4 fix(hqfap): reject 3MB video_down.mp4 stub (placeholder, not real video)
hqfap migrated its JSON-LD contentUrl (and the *.workers.dev mirror) to /upload/videos/video_down.mp4, which serves a FIXED ~3.04MB file for EVERY scene regardless of declared length (verified 5/5 scenes at 14-47min all = 3.04MB, 2026-06-21). It is a placeholder/'server down' clip, not the content — the browser's own player streamed the same stub via MediaSource. We were handing users that 3MB stub (reports c382d441/ef10b946). Now reject the video_down.mp4 contentUrl and return no source, so scenes fall through to other sources or show no playback instead of a fake clip. Real older scenes (cdnde.com / okcdn.ru direct mp4) still resolve. This also makes the proxy-fallback question moot — there is no source to proxy.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 23:01:19 +02:00
jtrzupek
249ad49430 fix(mobile): double-tap seek no longer pops the center pause control
Report dc4e91fb: double-tapping to skip ±15s also called setControlsVisible(true), throwing the full controls (big center pause button) on screen for 3.5s. Seek already has its own ±15s hint overlay, so the controls pop was redundant — removed it. Single-tap still toggles controls.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 22:27:08 +02:00
jtrzupek
b643b2cb77 fix(movies): dedup playback sources by target (cross-mirror dupes)
Movie detail showed ~100 playback links (report 41ca1fa4) because the 3 dooplay mirrors (mangoporn/pandamovies/streamporn) each record the SAME hoster embed as a separate row (e.g. luluvid/e/X from all three). Dedup by real target (embed_url/stream_url/page_url) after the priority sort, keeping the highest-priority copy — one verified movie drops 101 -> 58 unique.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-21 22:26:50 +02:00
jtrzupek
78d26c4bc6 feat(mobile): strip .com/.org clutter from site names
User-report 18105d14: drop the TLD suffix from Sites list + SiteScenes header (hqporner.com -> hqporner, fpo.xxx -> fpo). Logos skipped (needs a per-site logo source) — TLD strip is the quick win.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-20 16:38:45 +02:00
jtrzupek
ac84da92a4 feat(siska): convert to browse scraper, re-enable (search broken site-side)
siska's ?s= search ignores the query (returns latest regardless), so the performer-driven search scraper always yielded 0 and was disabled. Rewrote SiskaScraper as a latest-browse scraper (BaseBrowseScraper, /page/<n>/) and moved it to ALL_BROWSE_SCRAPERS. The listing tile carries everything (no detail fetch): title, duration (MM:SS span), thumbnail (img data-src), performer + studio (img alt 'Performer - Title - Studio'), category (thumbnail path). Playback unchanged: fresh videos embed playmogo + luluvid, resolved phone-side via _embed_iframe. Verified ingest: 26 seen / 11 new / 15 updated / 0 errors — and 15 updated means siska scenes match existing canonical scenes, adding playback coverage rather than orphans. Now covered by the browse ingest-watchdog (48h) and the 6h browse-latest + deep-crawl jobs. Old self-player videos (player.siska.video -> cfglobalcdn, ~2018) are dead and age out.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-20 16:25:11 +02:00
jtrzupek
0b6f663528 investigate(siska): keep disabled — site search is broken (ignores query)
Revisited siska re-enable (user fa4083a2). Findings: (1) fresh siska videos (videoID 227xxx) embed playmogo + luluvid and ARE phone-resolvable; updated siska.py scene regex + extractor path to the current video.php?videoID= format (old /<slug>/ format is gone). (2) BUT siska's ?s=<query> search is broken site-side — it returns the latest videos regardless of query (angela white == riley reid == homepage), so as a performer-driven BaseSearchScraper it always yields 0 (title token filter rejects everything). Reviving siska would require converting it to a browse/latest scraper (changes ingest character) — left as a decision. Old self-player videos (player.siska.video -> cfglobalcdn) are dead. Scraper stays disabled.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-20 16:15:02 +02:00
jtrzupek
8b216018a2 feat(mobile): Hidden content screen — blacklist tags/performers/studios
User-report 86a9ec72 ('remove all gay scenes from randomly popping up'): there was no UI to hide a tag, nor to view/undo the blacklist — even though the 'Hide performer' alert promised 'undo from Settings -> Blacklist' (a screen that never existed). New BlacklistScreen: search-and-add any tag to hide (e.g. a category), plus manage/unhide all blacklisted tags/performers/studios. Reached via Settings -> Content -> Hidden content. Backend already drops blacklisted-entity scenes from every /scenes (device-scoped); this just exposes it.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-20 15:53:09 +02:00
jtrzupek
b0e15935c6 fix(mobile): stop full scene-list refetch on back-navigation (perf)
Returning to the Scenes list from a scene caused a full reload + phone load spike (report 5df48551). Cause: invalidateQueries(['scenes']) in SceneDetail/Player/Performer/Studio handlers — including the silent auto-enrich-thumbnail that fires on opening any thumbnail-less scene — forces react-query to refetch EVERY loaded page of the infinite list. Added refetchType:'none' to all ['scenes'] invalidations: marks stale without refetching the active list, which refreshes on pull-to-refresh / filter change instead. Scene detail (['scene', id]) still updates immediately.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-20 14:14:07 +02:00
jtrzupek
c524b43fa3 fix(mobile): drop background ANR noise from Sentry (beforeSend)
Android Background ANRs captured via AppExitInfo (GOON-1D) are OS-side noise: the OS freezes a backgrounded app and reports it as not-responding, with zero JS/app frames and nothing to fix. beforeSend now drops events that are ANRs (ApplicationNotResponding) AND backgrounded (contexts.app.in_foreground === false). Foreground ANRs are kept (those can be real jank).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-19 20:07:32 +02:00
jtrzupek
f014a901de feat(scheduler): periodic title+duration dedup (missing-merge tube dupes)
Missing-merge duplicates (same performer + identical normalized title + identical duration-to-the-second) that bulk_dedup misses — tube re-scrapes and cross-tube re-ingests like porn00 pulling a video already present from xnxx (reports 28fe8181/32df33b1). Extracted the proven merge_exact_title_duration logic into app/scheduler/title_duration_dedup.py (script now a thin wrapper), wired a 12h scheduler job (playback-only = what users actually see, GOON_SCHED_TITLE_DEDUP_HOURS). Signal is near-certain (two different videos don't share byte-identical title AND exact duration); no shared performer = not merged (over-match guard). Verified: job registers (jobs=14), backlog currently 0 after the one-shot global merge.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-19 11:20:48 +02:00
jtrzupek
476cbb8d16 fix(ingest): race-safe scene_tags insert (ON CONFLICT) — GOON-M
scene_resolver._sync_tags used check-then-insert (select existing -> add if None), which races under concurrent ingest of the same scene: two runs both see existing=None, both add, flush -> IntegrityError pk_scene_tags (Sentry GOON-M, 4 events). Switched to pg_insert(...).on_conflict_do_nothing(index_elements=[scene_id, tag_id]) + in-batch dedup, identical to movie_resolver._sync_tags.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-19 11:09:06 +02:00
jtrzupek
567a8fb3b5 fix(mobile): scene-list scroll perf + native phone-side fpoxxx resolver
(1) Scroll jank/device load on long scene lists (report 5b7ca1e1): SceneTile is now React.memo'd so typing in search no longer re-renders every mounted tile, and sceneGridProps bounds the render window (windowSize 7 etc.) — required because removeClippedSubviews stays false to avoid thumbnail blanking. Applies to all scene grids. (2) fpoxxx played an ad instead of the video via the WebView fallback (reports f79beefb/cfa207c7). fpoxxx is KVS with an IP-bound + session-bound get_file token (cross-IP 403 confirmed), so it must resolve phone-side: new fpoxxxResolver fetches the page + follows get_file on the device (KVS real_url port for the function/0 case), wired into SceneDetailScreen like sxyprn/eporner. Verified from a residential IP: get_file -> CDN returns 206 video/mp4.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-19 11:02:21 +02:00
jtrzupek
e4cb94bc59 feat(scheduler): hetzner bandwidth monitor + search-tube watchdog coverage
Two observability additions to the worker scheduler (intertwined in the same files): (1) ingest-watchdog now also covers performer-driven search scrapers (ALL_DIRECT_SCRAPERS) with a separate 7d threshold, not just browse tubes at 48h — several search tubes (perverzija, fpoxxx, porndish, ...) had frozen silently for weeks. (2) New Hetzner Cloud bandwidth monitor (app/scheduler/hetzner_monitor.py): polls outgoing_traffic vs included_traffic and fires a Sentry message at info/warning/error % thresholds with a per-level fingerprint. The config fields existed for ages but the monitor was never implemented. No-op until HETZNER_API_TOKEN + HETZNER_SERVER_ID are set in .env (verified: returns {enabled: False}, job registers as 'hetzner-monitor every 6h', jobs=13).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-18 09:18:59 +02:00
jtrzupek
b1a530611f fix(latestpornvideo): revive search via /actor/ listing + metadata
Old regex matched junk (/wp-json etc.), not scenes (scenes are /<post_id>/).
Frozen since 06-13. Rewrote search() to scrape the /actor/<slug>/ listing
and parse <article> cards: scene URL, title, performers + tags from the
class (actors-*/tag-*/category-*, dropping performer-name fragment tags),
thumbnail. Studio + release date parsed from the "<Studio>-YYYY-MM-DD"
thumbnail filename, with a title-prefix "<Studio> YY MM DD" fallback.
Multi-performer works; no duration in listing; playback unchanged (hoster).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-16 23:20:02 +02:00
jtrzupek
e77deef667 fix(mypornerleak): revive search via /actor/ listing + metadata
Content moved to the w8.mypornerleak.com (wN) load-balancer subdomain, so
the old bare-domain scene regex matched nothing (frozen since 05-07).
Rewrote search() to scrape the canonical /actor/<slug>/ listing: scene
URL (wN host normalized to canonical for stable dedup), title, duration,
performers and category-tags from the <article> class (actors-*/category-*),
thumbnail. No studio (OnlyFans/amateur leaks have none). Multi-performer
works; playback unchanged (hoster, phone-side).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-16 23:16:02 +02:00
jtrzupek
5b67aeeeaf fix(sxyland): revive search via /actor/ pages + rich metadata
sxyland dropped the /<numeric_id>/<slug>/ scene URL format for /<slug>/,
so the old regex matched nothing (frozen since 06-07). Rewrote search()
to use the performer page /actor/<slug>/ and fetch each scene for full
metadata: all performers (with co-stars, from /actor/ links), tags
(scoped to the scene's tags-list, not the sidebar), duration + upload
date (itemprop), studio from the title prefix (BraZZers/MilfCoach/... ,
guarded so a performer-name prefix isn't mistaken for a studio). Junk
nav pages (Terms of Use etc.) are dropped via a no-duration-and-no-tags
guard. Verified: clean studio/performers/tags in DB, 0 errors.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-16 23:11:44 +02:00
jtrzupek
e0e69189a8 fix(sxyprn): revive search via performer pages + rich metadata
sxyprn ingest was frozen since 05-07: the old ?type=videos&query= endpoint
returns trending (not performer-filtered), so the strict token filter
correctly dropped everything -> 0 ingest. Real "search" is the performer
page /<First-Last>.html. Rewrote search() to scrape those cards: clean
performer (the query, avoids sxyprn's Dallas/Rae name fragmentation),
studio (channel subcat), tags (#hashtags), duration, thumbnail. Token
filter now runs on the card title so only genuine matches attach the
performer. Verified: Lana Rhoades/Riley Reid/Angela White return results,
metadata persists in DB (studio e.g. Vixen, 10-31 tags/scene), playback
mp4 206.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-16 22:58:52 +02:00
jtrzupek
00f4779abe feat(mobile): column toggle, duration filter, saved searches, screen protection (mobilism feedback)
Batch from user feedback: (1) Grid columns 1/2/3 setting (PreferencesContext, persisted) across all scene grids — default 2 was too small on phones. (2) Min-duration filter chips (5/10/20/30+ min) to hide ad-clips. (3) Saved-search chips + Save button (backed by /saved-searches). (4) Re-enabled screen-capture protection (Recents hide + screenshot block) for distributed users — verified active on emulator (screencap returns 0 bytes). (5) 'Checking for updates' gate before the PIN screen so a background OTA restart no longer causes a double PIN prompt. Changelog entry added. Published OTA runtime 1.1 (a9620b12).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-16 13:52:27 +02:00
jtrzupek
bcee5851e9 feat(api): per-device saved searches (keyword favorites)
User-report (mobilism): scenes are often poorly titled, so saved keyword queries are a useful extra retrieval strategy. New saved_searches table (device-scoped via X-Device-Id, unique per device+query, 50/device cap) + GET/POST/DELETE /saved-searches. Migration 0024. Verified CRUD on prod: add trims+dedups idempotently, empty rejected 422, delete idempotent.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-16 13:52:18 +02:00
jtrzupek
0424cb9138 feat(scheduler): per-origin ingest freshness watchdog -> Sentry
The global source monitor can't catch a single stalled tube because every tube scraper shares one Source row (tube-scraper), so an aggregate run still reports success while one origin freezes (freshporno browsing the rotating KVS homepage root, report 14f3a655). New watchdog checks max(created_at) per active browse-scraper origin (tube:<sitetag>); if a tube with history hasn't produced a new scene in > max_age_hours it fires a Sentry message with a stable per-origin fingerprint (age in extras, not the title, so it stays one grouped issue). Runs every 6h, 48h threshold, both env-tunable (GOON_SCHED_INGEST_WATCHDOG_HOURS / GOON_INGEST_WATCHDOG_MAX_AGE_HOURS). Verified: 0 stale at 48h post-fix, detects neporn at a strict 12h threshold.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-15 10:26:25 +02:00
jtrzupek
4b71689a95 fix(scrapers): freshporno browse from /latest-updates/ not homepage root
The homepage root / is a KVS page with cache-control: no-store and a fresh PHPSESSID per request; the server rotates its featured block and on a cold session can serve an old set instead of the newest scenes. Result: browse-latest skipped everything for 3 days (root served 20 May content), no new freshporno scenes since 12 Jun (user report). Switch _listing_url to the explicit date-sorted /latest-updates/ feed (pagination /latest-updates/N/), which is not subject to that rotation.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-15 09:59:40 +02:00
jtrzupek
3714afa22f fix(mobile): capture site/origin text params in bug-report auto-context
SiteScenes passes the tube as origin/name (strings), not UUIDs, so the existing UUID-only auto-context loop dropped them. Reports like 'ingest of this site has been stuck 2 days' (14f3a655) arrived without any site identifier. Add a second loop for known string identity params (origin/name/sitetag/tag/q), length-capped, so per-site/per-performer reports become actionable.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-15 09:35:58 +02:00
jtrzupek
8b4783771f feat(scheduler): periodic thumb-asset dedup (hdporn.gg/fullmovies.xxx)
The one-off cleanup merged ~13.5k same-video-different-title dupes, but they regrow as
these sibling tubes re-ingest under new titles. Wire the asset-id+duration merge into
the scheduler (every 12h, GOON_SCHED_THUMB_DEDUP_HOURS, 0=off) so it stays clean.

Shared logic lives in app/scheduler/thumb_dedup.py (run_thumb_asset_dedup); the one-shot
script now imports it. Same tight signature as the cleanup: family hosts only + identical
duration (the bare asset-id number is reused across unrelated CDNs, so cross-host/diff-
duration grouping is excluded). Reports 205b17d9 / 5a2944cb.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 14:56:45 +02:00
jtrzupek
b5d9473898 feat(scripts): merge tube dupes by thumbnail asset-id (hdporn.gg/fullmovies.xxx family)
These sibling platforms share one video-id space and ingest the same video under
different titles, which bulk_dedup misses (different titles, no phash). Match by the
asset-id in the thumbnail path (/<bucket>000/<id>/) on img.hdporn.gg|fullmovies.xxx plus
identical duration, and merge. Hard host restriction + duration guard: the bare number
is reused for unrelated videos on other CDNs (verified via dry-run), so cross-host or
different-duration grouping is excluded. Run scoped (studio id) or global; dry-run by
default. Reports 205b17d9 / 5a2944cb. Ran on Parasited: 43 pairs merged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 14:18:44 +02:00
jtrzupek
b66dd99eba fix(mobile): show Refresh thumbnail when the hero image actually fails to load
The button keyed on thumbnail_url presence, but a URL can be present yet broken (hqfap
404 → blank hero, no button — report ef0c6a5a). Tie it to the hero Image load state
(onLoad ok / onError broken / no url none) and show Refresh only when the image is
broken or missing. Reconciles 26c114ed (hidden for good previews) with ef0c6a5a.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 14:18:44 +02:00
jtrzupek
81d617efc2 fix(extractors): 4k69 direct okcdn extraction (replaces WebView fallback)
Reverse-engineered the migrated 4k69 player: jwplayer now serves OK.ru CDN (okcdn.ru)
mp4s. The static page (SSR behind Cloudflare, fetched via proxy) carries "file"+"label"
pairs for every quality. okcdn's srcIp param is NOT enforced (cross-IP test 2026-06-14:
206 video/mp4 from a residential IP != srcIp), so the URL plays from any IP. Parse the
okcdn sources server-side and return them mobile_direct_ok — the phone plays the direct
video, no WebView, no VAST preroll, no age-gate, zero VPS proxy. Skips 4K/2K. Reverts
the brief _vps_blocked_fallback routing (WebView grabbed the preroll ad, not content).
Verified on emulator: native player streams the actual scene (report 5de3fbc5).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 11:39:36 +02:00
jtrzupek
2a9445fe4a feat(mobile): auto-accept age-gate modal in WebView fallback
4k69 (and similar) show an "Are you 18 or above? Yes/No" modal that blocks the jwplayer
from initialising, so the WebView fallback never extracts a stream. Click the age-gate
accept button by id (#pop_up_18_yes and id*=18_yes/age_yes variants) on the same loop as
the consent/play-poster auto-clickers. Verified on emulator: 4k69 age-gate clears and the
player initialises (ExoPlayer hands off). A VAST preroll is still grabbed instead of the
okcdn content for 4k69 specifically (report 5de3fbc5 stays open) - separate ad-filter work.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 11:31:12 +02:00
jtrzupek
08410fddd1 fix(mobile): show Refresh thumbnail only when preview missing or broken
The Refresh thumbnail button appeared on every scene, which is noise for the majority
that already have a good preview (report 26c114ed). Show it only when no source has a
usable thumbnail or the only thumbnails are rotting (sxyprn/trafficdeposit), which is
exactly when a manual refresh helps (the original d3376a71 case).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 11:17:18 +02:00
jtrzupek
29da1fbaa6 fix(extractors): route 4k69 to WebView fallback after player migration
4k69 swapped its player from get_file (4kporno.xxx) to jwplayer + okcdn.ru, whose token
carries srcIp= (IP-bound); the site is also behind Cloudflare (VPS fetch only via proxy).
The native get_file extractor matched nothing and returned None, surfacing as a "host
problem" error even though the video plays fine (report 5de3fbc5). Switch 4k69com to
_vps_blocked_fallback: the on-device WebView (residential IP) clears Cloudflare, the
okcdn token binds to the phone IP, and INJECTED_JS hands the jwplayer source to ExoPlayer.
fourk69.extract stays in the module in case the site reverts to get_file.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-14 11:17:18 +02:00
jtrzupek
9269b02a4c feat(mobile): source-code link in Settings + Refresh thumbnail button
- AppLockSettings: a "Source code" row linking the public OSS repo (report 4c5066b8) -
  a trust signal for a sideloaded FOSS app (audit / self-host / contribute).
- SceneDetail: a "Refresh thumbnail" button (force) for scenes whose preview is broken
  or stale (report d3376a71).
- changelog: new What's New entry for this batch.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-13 19:04:11 +02:00
jtrzupek
e512665d26 feat(scenes): force-refresh thumbnail via enrich-thumbnail ?force
enrich-thumbnail was fill-only (skipped scenes that already had a thumbnail), so a
broken or stale preview (rotting sxyprn/trafficdeposit) could not be refreshed. Add a
force flag that re-fetches the source page and overwrites the existing thumbnail.
Backs the new "Refresh thumbnail" button (report d3376a71).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-13 19:04:10 +02:00
jtrzupek
32919d6a6c feat(extractors): detect deleted porntrex videos and mark dead
Porntrex soft-deletes: a removed video returns HTTP 200 with a "this video was deleted"
message instead of a player, so extract returned [] (transient) and the source was never
marked dead, leaving users on a permanently broken link (report 75dbf53e). Match the
deletion message and raise HosterDead so resolve marks the source dead.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-13 19:04:10 +02:00
jtrzupek
9d4384cef3 fix(ingest): cap code/director to column length (GOON-J)
Some sources (sexlikereal) build a giant `code`/`director` from a multi-performer
compilation title, overflowing scenes.code varchar(128) -> StringDataRightTruncation,
and the scene silently dropped from ingest. Cap both at the column limit in
_create_canonical and the fill path; code/director are stored metadata, not match keys,
so truncation is safe.

Fixes GOON-J

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-13 19:04:10 +02:00
jtrzupek
86b3e88f08 fix(mobile): remount Favorites lists on numColumns change (GOON-11)
The new Scenes tab uses a 2-column FlatList while Performers/Studios/Movies are
1-column. Switching tabs reused the same FlatList instance, so numColumns changed on
the fly and RN threw "Changing numColumns on the fly is not supported" (5 users).
Give the Scenes list a distinct key ("fav-scenes") from the shared single-column key
("fav-list") so React remounts a fresh FlatList across the 1<->2 boundary.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-12 16:19:19 +02:00
jtrzupek
e618087eae feat(mobile): "What's new" popup after OTA updates
After an OTA bundle is applied, show a one-time popup listing recent changes. The
changelog ships in the bundle (mobile/src/changelog.ts), so it is always in sync with
the code that just arrived. WhatsNewModal compares the newest entry id against the last
one seen (SecureStore); shows unseen entries, marks seen on dismiss, and stays quiet
until the next update adds an entry. First run shows only the newest entry (no history
dump). Mounted over the navigator when signed in.

Each OTA publish should prepend a new entry at the top of CHANGELOG.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-12 11:41:54 +02:00
jtrzupek
a00acdddfb feat(mobile): "Your messages" inbox on bug FAB + geo-block playback hint
Bug FAB now has two tabs: "Report a bug" (existing) and "Your messages", which lists
this device's reports with any admin reply in a highlighted box. A badge dot on the FAB
shows unread replies; opening the tab marks them seen. Polls every 90s and on open.

PlayerScreen: when the WebView fallback (residential IP) cannot extract a stream within
25s and there is no 404/410, show a one-time hint that the source may be blocked in the
user's region or by their ISP (try another source or a VPN) - so a geo/network block on
the user's side does not read as a broken app.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-12 11:35:44 +02:00
jtrzupek
d1f2f035b0 feat(bug-reports): two-way replies (device-scoped) + admin reply endpoint
Reports were anonymous and one-way. Tie each report to the submitting device
(X-Device-Id), add an admin response back-channel, and let the app fetch replies for
its own device:
- migration 0023: bug_reports gains device_id, response, responded_at, response_seen.
- create_bug_report captures device_id.
- GET /bug-reports/mine (device-scoped) returns this device's reports + unseen count.
- POST /bug-reports/mine/seen clears the unseen flag.
- POST /bug-reports/{id}/reply sets the admin response (authored during triage).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-12 11:35:44 +02:00
jtrzupek
aebacc0389 feat(mobile): Favorites "Scenes" tab to view saved scenes
Users could heart individual scenes from SceneDetail, but the Favorites screen only
had Performers/Studios/Movies tabs, so saved scenes were invisible (bug report: got a
bunch of scenes saved but no way to see them). Add a Scenes tab (now the default)
listing favorited scenes as tiles via GET /scene-favorites, long-press to remove.
Adds client.listSceneFavorites().

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-12 10:06:12 +02:00
jtrzupek
1654d78d59 fix(ingest): strip NUL bytes from raw payloads before Postgres write
A source (TPDB) returned a performer alias containing a literal U+0000 ("Ramon..").
Postgres cannot store  in JSONB or text, so the external_records JSONB insert in
_upsert_external_record failed with UntranslatableCharacter and the scene never ingested
(GOON-Z). Recursively strip NUL from the raw payload (-> external_records.raw) and, when
present, also re-validate the RawScene/RawMovie so normalize -> typed text columns get
clean data too. Gated by a cheap _has_nul scan so clean records (the overwhelming
majority) pay no extra cost.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 19:48:22 +02:00
jtrzupek
16eb633bde feat(mobile): phone-side resolvers for IP-bound tubes (sxyprn, eporner, voe)
These CDNs bind their signed video URL to the IP that fetched the page, so a
server-side resolve hands the phone a URL bound to the server IP -- the device then
gets a placeholder/403 and falls back through the proxy, streaming the whole video
through the server. Resolve on the device instead (token binds to the phone IP) so
playback goes direct with zero proxy bandwidth.

Ports of the existing backend extractors:
- sxyprnResolver.ts: data-vnfo + boo/ssut51 transform
- epornerResolver.ts: vid+hash -> /xhr/video mp4 sources
- voeResolver.ts: mirror redirect + 7-step payload decoder

Wired into SceneDetailScreen.onPress (sxyprn/eporner) and MovieDetailScreen.playVoe (voe).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 16:14:25 +02:00
jtrzupek
aa05ce2647 feat(playback): direct-HLS manifest passthrough + proxy stream drop handling
Time-bound HLS hosters whose manifest URL lacks a .m3u8 extension (e.g. pornhat's
"...mp4,?..." path) were mis-detected by ExoPlayer as progressive MP4 and failed,
forcing a full proxy fallback that streamed the whole video through the server. Serve
such manifests via /proxy/hls/<token>/play.m3u8 with child URLs left absolute on the
CDN, so the device fetches variant+segments directly and only the ~1KB manifest is
proxied. Routed only for mobile_direct_ok (time-bound) HLS without a .m3u8 path.

Also swallow httpx.TransportError in the stream proxy body generator: an upstream CDN
closing the connection mid-stream is benign (client just retries a range) and should
not surface as an unhandled error.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 16:14:25 +02:00
jtrzupek
072f2608b3 chore: gitignore marketing-shots/ and one-off _*.py scripts
Keep local-only marketing material and throwaway backfill scripts out of
the public repo (same rationale as the existing screenshots/ entry).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 19:28:22 +02:00
jtrzupek
a9f0f94321 feat(sxyprn): mark dead posts during thumbnail refresh sweep
resolve_post() now distinguishes "Post Not Found" (mark dead_at — the
link wouldn't play anyway) from a live page with no fresh poster (leave
untouched), on top of the existing thumbnail refresh. Batched into
refresh_batch() with refreshed/dead/untouched counters.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 19:20:28 +02:00
jtrzupek
956a0feb22 docs: correct Bright Data proxy type (ISP, flat-rate not per-GB)
It is an ISP proxy (static ISP IPs, flat billing), not residential —
so HTML-ingest bandwidth is free and the full deep-crawl is fine.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 19:18:40 +02:00
jtrzupek
21bc8bf1fe feat(superporn): browse scraper via Bright Data residential proxy
superporn hard-blocks the VPS IP with Cloudflare 403 on every TLS
impersonation, so HTML ingest routes through Bright Data residential
(BRIGHTDATA_PROXY_URL, parsed in config). First scraper to use a proxy:
optional _proxy on the browse base, threaded into browser_get.

JSON-LD VideoObject (title/desc/uploadDate/thumb/duration) + pornstar
and category chips; superporn double-encodes HTML entities so titles
are unescaped twice. Thumbnails fetch fine from the VPS (no proxy).

Playback stays off-proxy: the <source> mp4 token is IP-bound to the
fetcher, so resolve is phone-side via WebView (extractor superporncom
-> _vps_blocked_fallback), same as porndoe.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 18:47:45 +02:00
jtrzupek
80fd83cb4e feat(tubes): add 4k69 + neporn browse scrapers, shared PlayTube base
4k69.com (~65k scenes): same PlayTube CMS as hqfap - common logic moved
to _playtube.py (sitemap catalog, JSON-LD, pills). Studio classified by
matching category pills against the studios index page. Streams are
get_file (fullmovies family) returned unresolved with mobile_direct,
2160p skipped.

neporn.com: KVS engine, latest-updates listing, JSON-LD + video:duration
meta, performers from models links with flashvars video_tags fallback
for fresh uploads. Resolve via _kvs; final URL portable cross-IP.

superporn.com rejected: Cloudflare 403 from VPS on all TLS impersonations.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 18:15:13 +02:00
jtrzupek
6de986b9a7 feat(hqfap): browse scraper + native mp4 extractor (~120k scenes)
PlayTube CMS. Sitemap-based pagination (listing has no GET paging),
JSON-LD VideoObject metadata, pornstar/category pills, " Clips"
categories mapped to studio. Direct mp4 (cdnde.com/okcdn.ru), tokens
time-bound and portable cross-IP, so mobile plays direct.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-10 17:51:04 +02:00