Route Dailymotion directly when impersonation is available

Author: Michal Warda

deploy/pipeline/sam-cinematic-continuous.env

@@ -50,6 +50,9 @@ SAM_CONTINUOUS_PROVIDER_CIRCUIT_COOLDOWN_SECONDS=300
 SAM_CONTINUOUS_PROVIDER_CIRCUIT_MAX_COOLDOWN_SECONDS=3600
 SAM_CONTINUOUS_SOURCE_PROVIDER_WEIGHTS=dailymotion=12,bilibili=1,youtube=0.25,vimeo=1,tiktok=1,soundcloud=1,internet_archive=1
 SAM_YTDLP_CONCURRENT_FRAGMENTS=4
+# Dailymotion's CDN is faster and more reliable direct when curl-cffi is
+# installed; keep the proxy pool for providers that need IP rotation.
+SAM_YTDLP_DIRECT_PLATFORMS=dailymotion
 SAM_CONTINUOUS_SOURCE_SCAN_WORKERS=4
 SAM_CONTINUOUS_SCANNED_HIGH_WATER=64
 SAM_CONTINUOUS_SOURCE_EXTRACT_WORKERS=16

pipeline/src/sam_audio_pipeline/youtube_random.py

@@ -50,6 +50,11 @@
 YTDLP_CONCURRENT_FRAGMENTS = max(
     1, int(os.environ.get("SAM_YTDLP_CONCURRENT_FRAGMENTS", "4"))
 )
+YTDLP_DIRECT_PLATFORMS = frozenset(
+    value.strip().lower()
+    for value in os.environ.get("SAM_YTDLP_DIRECT_PLATFORMS", "").split(",")
+    if value.strip()
+)
 OUTPUT_SAMPLE_RATE = 48_000
 MIN_SOURCE_SAMPLE_RATE = 44_100
 MIN_SOURCE_BITRATE_KBPS = 120.0
@@ -551,6 +556,8 @@ def _protected_proxy_configs(path: Path) -> list[Path]:
 
 def _yt_dlp_proxy_args(source_platform: str, affinity_key: str = "") -> list[str]:
     """Select a protected all-provider proxy without credentials in argv."""
+    if source_platform.strip().lower() in YTDLP_DIRECT_PLATFORMS:
+        return []
     if YOUTUBE_PROXY_CONFIG is None:
         return []
     path = Path(YOUTUBE_PROXY_CONFIG)

pipeline/tests/test_youtube_random.py

@@ -117,6 +117,24 @@ def test_proxy_config_is_secret_backed_and_applies_to_all_providers(
     ]
 
 
+def test_direct_platform_bypasses_proxy_pool(tmp_path: Path, monkeypatch) -> None:
+    config = tmp_path / "proxy.conf"
+    config.write_text("--proxy http://user:secret@proxy.example:80\n")
+    config.chmod(0o600)
+    monkeypatch.setattr(youtube_random, "YOUTUBE_PROXY_CONFIG", config)
+    monkeypatch.setattr(
+        youtube_random,
+        "YTDLP_DIRECT_PLATFORMS",
+        frozenset({"dailymotion"}),
+    )
+
+    assert youtube_random._yt_dlp_proxy_args("dailymotion") == []
+    assert youtube_random._yt_dlp_proxy_args("youtube") == [
+        "--config-locations",
+        str(config),
+    ]
+
+
 def test_yt_dlp_transfer_args_use_bounded_fragment_concurrency(monkeypatch) -> None:
     monkeypatch.setattr(youtube_random, "YTDLP_CONCURRENT_FRAGMENTS", 4)