Stille-Suche: Folge erst in Python laden, ffmpeg liest die lokale Datei

Die statische ffmpeg-Build stürzt im Container beim Auflösen von Hostnamen ab (SIGSEGV).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

podcasts/silence.py

@@ -7,17 +7,22 @@ episode (PodcastEpisode.silences); the browser then jumps over them while it
 plays (podcasts-web.js).
 """
 import logging
+import os
 import re
 import subprocess
+import tempfile
 import threading
 
+import requests
+
 from django.db import close_old_connections
 
 log = logging.getLogger(__name__)
 
 NOISE = '-35dB'        # quieter than this counts as silence
 MIN_SILENCE = 0.4      # seconds; shorter pauses stay (they belong to speech)
-TIMEOUT = 900          # seconds for one episode (download and decode)
+TIMEOUT = 900          # seconds for decoding one episode
+MAX_BYTES = 800 * 1024 * 1024   # a larger file isn't an episode to analyse
 MAX_RUNNING = 2        # episodes analysed at the same time
 
 _START = re.compile(r'silence_start: (-?[\d.]+)')
@@ -47,18 +52,36 @@ def parse(stderr):
 
 
 def analyze(url):
-    """The pauses of the audio at `url` (ffmpeg reads it straight from there)."""
+    """The pauses of the audio at `url`. Downloaded here first: the static ffmpeg
+    build can't look up host names (it crashes doing so), it reads a local file."""
     import imageio_ffmpeg
-    proc = subprocess.run(
+    fd, path = tempfile.mkstemp(suffix='.audio')
-        [imageio_ffmpeg.get_ffmpeg_exe(), '-nostdin', '-hide_banner', '-nostats',
+    try:
-         '-user_agent', 'diora podcast silence', '-vn', '-sn', '-dn', '-i', url,
+        with os.fdopen(fd, 'wb') as out, requests.get(
-         '-af', f'silencedetect=noise={NOISE}:d={MIN_SILENCE}', '-f', 'null', '-'],
+                url, stream=True, timeout=(10, 60), headers={'User-Agent': 'diora podcast silence'}) as r:
-        capture_output=True, text=True, timeout=TIMEOUT,
+            if r.status_code >= 400:
-    )
+                raise RuntimeError(f'HTTP {r.status_code}')
-    if proc.returncode != 0:
+            size = 0
-        tail = (proc.stderr or '').strip().splitlines()[-1:] or ['ffmpeg failed']
+            for chunk in r.iter_content(1 << 20):
-        raise RuntimeError(tail[0][:200])
+                size += len(chunk)
-    return parse(proc.stderr)
+                if size > MAX_BYTES:
+                    raise RuntimeError('file too large')
+                out.write(chunk)
+        proc = subprocess.run(
+            [imageio_ffmpeg.get_ffmpeg_exe(), '-nostdin', '-hide_banner', '-nostats',
+             '-vn', '-sn', '-dn', '-i', path,
+             '-af', f'silencedetect=noise={NOISE}:d={MIN_SILENCE}', '-f', 'null', '-'],
+            capture_output=True, text=True, timeout=TIMEOUT,
+        )
+        if proc.returncode != 0:
+            tail = (proc.stderr or '').strip().splitlines()[-1:] or [f'ffmpeg failed ({proc.returncode})']
+            raise RuntimeError(tail[0][:200])
+        return parse(proc.stderr)
+    finally:
+        try:
+            os.remove(path)
+        except OSError:
+            pass
 
 
 def ensure(episode_id):