From 3e8a2219810c37b4432f3cbe4d523f49bc030eec Mon Sep 17 00:00:00 2001
From: jtrzupek <jtrzupek@gmail.com>
Date: Tue, 9 Jun 2026 09:35:58 +0200
Subject: [PATCH] feat(extractors): native HLS for xhamster; hqporner flyflv
 player

xhamster: move from WebView fallback to server-side native HLS. The scene page
is fetchable server-side and the xhcdn master m3u8 (variants + segments) is
time-bound, not IP-bound (verified cross-IP), so mobile plays the HLS direct
with zero proxy bandwidth. New tubes/xhamster.py pulls the master m3u8 from
SSR HTML and returns type='m3u8' mobile_direct; registry remaps xhamstercom
off _vps_blocked_fallback.

hqporner: add flyflv to the player-iframe host whitelist. hqporner rotated
some players to flyflv.com; the CDN host was already whitelisted but the iframe
host was not, so those scenes returned no stream.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 app/extractors/__init__.py       | 12 +++--
 app/extractors/tubes/hqporner.py |  2 +-
 app/extractors/tubes/xhamster.py | 87 ++++++++++++++++++++++++++++++++
 3 files changed, 96 insertions(+), 5 deletions(-)
 create mode 100644 app/extractors/tubes/xhamster.py

diff --git a/app/extractors/__init__.py b/app/extractors/__init__.py
index 7ee6df4..f7604dc 100644
--- a/app/extractors/__init__.py
+++ b/app/extractors/__init__.py
@@ -39,6 +39,7 @@ from app.extractors.tubes import (
     pornhat,
     porntrex,
     sxyprn,
+    xhamster,
     yespornvip,
 )
 
@@ -84,10 +85,6 @@ _REGISTRY: dict[str, Callable[[str], list[StreamSource] | None]] = {
     # flashvars `video_url` → `get_file` 302 → CDN time-bound signed URL
     # (`expires`+`md5`, NIE IP-bound) → mobile gra direct, zero VPS bandwidth.
     "porntrexcom": porntrex.extract,
-    # VPS-blocked tubes — KVS / Cloudflare blokuje Hetzner IP, ale działają z residential
-    # IP (potwierdzone Chrome DevTools MCP 2026-05-15). Mobile WebView + INJECTED_JS
-    # (PlayerScreen.tsx:805) skanuje <video>.src + XHR — łapie URL po decode-ie player JS.
-    "xhamstercom": _vps_blocked_fallback.extract,
     # pornditt — KVS jak yespornvip (function/0 + license). VPS dociera → resolve
     # server-side (decode + follow 302 → portable twa.tgprn.com CDN). Wcześniej WebView
     # fallback łapał VAST preroll (trafostatic) zamiast contentu. Patrz pornditt.py/_kvs.py.
@@ -114,6 +111,13 @@ _REGISTRY: dict[str, Callable[[str], list[StreamSource] | None]] = {
     # xxxfreewatch — DELISTED 2026-05-18. 790 solo-orphan scen, 0% match, CF-walled z VPS.
     "latestleaksco": _embed_iframe.extract,
     "mypornerleakcom": _embed_iframe.extract,
+    # xhamster — 2026-06-08 PRZEPIĘTE z _vps_blocked_fallback na natywny server-side HLS.
+    # Re-test (DevTools + cross-IP): VPS pobiera scene page bez CF challenge, master m3u8
+    # w SSR HTML, manifest+segmenty time-bound (portable, nie IP-bound). Mobile gra HLS
+    # direct, multi-quality, zero VPS proxy/WebView/reklam. Patrz tubes/xhamster.py.
+    # ~155k solo-scen upgrade z WebView-z-reklamami na natywne. Wcześniej WebView fallback
+    # ładował ad-heavy stronę z phone IP (działało, ale gorszy UX + preroll VAST).
+    "xhamstercom": xhamster.extract,
     # PornHat — dedicated extractor: tylko `<source>` z player area (skip sidebar
     # trailer URLs `_preview*.mp4`), dedupe po filename. Get_file 302 → CDN, proxy
     # follow_redirects=True wymagane (fix w stream_proxy.py).
diff --git a/app/extractors/tubes/hqporner.py b/app/extractors/tubes/hqporner.py
index 49fb227..23e6903 100644
--- a/app/extractors/tubes/hqporner.py
+++ b/app/extractors/tubes/hqporner.py
@@ -48,7 +48,7 @@ _PLAYER_IFRAME_RE = re.compile(r'<iframe[^>]+src=["\']([^"\']+)', re.IGNORECASE)
 # smartpop, popcash, reebr) → reklama. Brak match = fail safe (return None),
 # nie próbujemy go odpalić jako hostera bo to ad-redirect → pop-under.
 _VIDEO_IFRAME_HOST_RE = re.compile(
-    r"//(?:[a-z0-9-]+\.)?(?:mydaddy|hqwo|hqporner)\.[a-z]{2,4}/",
+    r"//(?:[a-z0-9-]+\.)?(?:mydaddy|hqwo|hqporner|flyflv)\.[a-z]{2,4}/",
     re.IGNORECASE,
 )
 
diff --git a/app/extractors/tubes/xhamster.py b/app/extractors/tubes/xhamster.py
new file mode 100644
index 0000000..3ed8331
--- /dev/null
+++ b/app/extractors/tubes/xhamster.py
@@ -0,0 +1,87 @@
+"""xhamster.com — natywny server-side HLS extractor.
+
+2026-06-08: re-test Chrome DevTools + cross-IP NAPRAWIA założenie z `_vps_blocked_fallback`.
+Wcześniej `xhamstercom` szedł przez WebView fallback (założenie: Cloudflare blokuje Hetzner
+IP). Re-test pokazał:
+  1. VPS pobiera scene page (HTTP 200, BEZ Cloudflare challenge — blok się zdjął).
+  2. Master HLS URL jest w SSR HTML plain: `video-nss.xhcdn.com/<token>,<expiry>/media=hls4/
+     multi=.../...m3u8`. `<expiry>` to UNIX ts → token TIME-BOUND, nie IP-bound.
+  3. Cross-IP test (VPS Hetzner): master m3u8 → 200, wariant playlist → 200, segment .m4s
+     → 206 video/mp4. Cały łańcuch PORTABLE — mobile gra HLS direct z residential IP,
+     zero VPS proxy bandwidth.
+
+Dlatego resolvujemy SERVER-SIDE jak porntrex/freshporno: fetch page (curl_cffi chrome) →
+wyłuskaj master m3u8 → oddaj jako type='m3u8' mobile_direct. ExoPlayer robi adaptive
+multi-quality z jednego master URL.
+
+NB `sources.standard.av1/h264` w HTML to ZASZYFROWANE hex-bloby (player deszyfruje w JS),
+bezużyteczne server-side — dlatego bierzemy HLS, nie mp4.
+"""
+from __future__ import annotations
+
+import logging
+import re
+
+from app.extractors._fetch import _DEFAULT_IMPERSONATE, _DEFAULT_UA, _HAS_CURL_CFFI, fetch_tube_html
+from app.extractors._models import HosterDead, StreamSource
+
+log = logging.getLogger(__name__)
+
+_BASE = "https://xhamster.com"
+
+# Master HLS na xhcdn (video-nss.xhcdn.com / fallback inne sub-domeny). JSON w HTML
+# escape'uje slashe (`https:\/\/...`), więc unescape przed matchowaniem.
+_M3U8_RE = re.compile(r"https://[a-z0-9.\-]*xhcdn\.com/[^\"'\\ ]+?\.m3u8", re.IGNORECASE)
+
+# Markery skasowanej sceny (strona istnieje, ale bez wideo) → HosterDead.
+_DEAD_MARKERS = (
+    "this video has been deleted",
+    "this video was deleted",
+    "video is no longer available",
+    "has been removed",
+)
+
+
+def extract(page_url: str, *, timeout: float = 60.0) -> list[StreamSource] | None:
+    html = ""
+    if _HAS_CURL_CFFI:
+        from curl_cffi import requests as _cf_requests
+
+        session = _cf_requests.Session(impersonate=_DEFAULT_IMPERSONATE)
+        try:
+            resp = session.get(
+                page_url,
+                headers={"User-Agent": _DEFAULT_UA, "Accept": "text/html,application/xhtml+xml"},
+                timeout=timeout,
+                allow_redirects=True,
+            )
+            html = resp.text if resp.status_code < 400 else ""
+        except Exception as e:
+            log.info("xhamster: page fetch failed %s: %s", page_url, e)
+            html = ""
+    if not html:
+        # fetch_tube_html podnosi TubePageError dla 404/410 (caller → dead_at).
+        html = fetch_tube_html(page_url, timeout=timeout)
+
+    # JSON-escaped slashe → plain, żeby regex złapał master URL.
+    unescaped = html.replace("\\/", "/")
+    m = _M3U8_RE.search(unescaped)
+    if not m:
+        low = unescaped.lower()
+        if any(marker in low for marker in _DEAD_MARKERS):
+            raise HosterDead(f"xhamster: scene deleted {page_url}")
+        log.info("xhamster: no HLS master URL on %s", page_url)
+        return None
+
+    master = m.group(0)
+    return [
+        StreamSource(
+            link=master,
+            type="m3u8",
+            quality=None,  # HLS master = adaptive multi-quality (ExoPlayer wybiera)
+            referer=_BASE + "/",
+            # Master + warianty + segmenty są time-bound (nie IP/cookie-bound),
+            # zweryfikowane cross-IP 2026-06-08 → mobile gra direct, zero VPS proxy.
+            raw={"mobile_direct_ok": True},
+        )
+    ]