jonathan/ISR
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: collapse the single-implementation backend abstraction, fix device resolution

Browse Source 
isr.py carried a plugin architecture — AudioBackend ABC, a _backend_classes
registry, _discover_backends, available_backends/get_backend, a
preferred_backend config knob — for exactly one implementation. It was
also the source of three real bugs:

* find_device() returned None before reaching the ALSA PCM-name fallback
  whenever `arecord -l` enumerated nothing, so the documented
  `device = shared_mic` (dsnoop from asound.conf) case failed outright on
  any host with no hardware card. The fallback is the whole point: those
  PCMs never appear in `arecord -l`.

* `device = monitor` could never resolve. ALSA has no generic monitor
  source and list_devices hardcoded is_monitor=False, so the keyword —
  advertised in config.example.ini as the "EASY" way to capture system
  audio — only ever produced "device not found". Removed; it is a
  PulseAudio idea that outlived the PulseAudio backend.

* list_all_devices() deduplicated by lowercased name "in case a device
  appears in multiple backends". With one backend what it actually did
  was delete the second of two identical USB cards. Dedup dropped.

find_device() now always returns a device, so _resolve_device's
unreachable "Available devices:" error is gone; an unknown spec is
passed to ALSA verbatim and logged. AudioDevice loses backend/
is_monitor plus channels/sample_rate — the latter were hardcoded 2/44100
and only ever printed by --list-devices, i.e. numbers with no bearing on
what gets recorded.

--list-devices now says outright that it lists hardware cards only, which
is where the shared_mic confusion started.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
admin
2026-07-29 20:28:30 +02:00
co-authored by Claude Opus 5
parent 4aa7f7253e
commit 1541110ffb
4 changed files with 155 additions and 281 deletions
Download Patch File Download Diff File Expand all files Collapse all files
isr.py
+117 -221
View File
@@ -50,113 +50,24 @@ FILENAME_FORMAT = '%Y%m%d_%H%M%S'
# =============================================================================
# Audio Device & Backend System
# Audio Device Discovery (ALSA)
# =============================================================================
@dataclass
class AudioDevice:
"""Represents an audio input device."""
id: str # Backend-specific identifier
name: str # Human-readable name
channels: int # Max input channels
sample_rate: int # Default sample rate
backend: str # Backend name ('alsa')
is_default: bool = False # Is system default
is_monitor: bool = False # Is a monitor/loopback source
"""An ALSA capture device.
Only the identity matters: capture format comes from config (sample_rate,
channels), not from the device, so this deliberately carries no
capability fields — reporting a card's channel count here would just be a
number that has no bearing on what actually gets recorded.
"""
id: str # ALSA device spec: 'hw:1,0', or a PCM name
name: str # Human-readable name from `arecord -l`
is_default: bool = False
def __str__(self):
flags = []
if self.is_default:
flags.append("DEFAULT")
if self.is_monitor:
flags.append("MONITOR")
flag_str = f" [{', '.join(flags)}]" if flags else ""
return f"{self.name}{flag_str} ({self.backend})"
class AudioBackend(ABC):
"""Abstract base for audio capture backends."""
name: str = "base"
@classmethod
@abstractmethod
def is_available(cls) -> bool:
"""Check if this backend can be used on the current system."""
pass
@classmethod
@abstractmethod
def list_devices(cls) -> List[AudioDevice]:
"""List all available input devices."""
pass
@abstractmethod
def open_stream(self, device: AudioDevice, sample_rate: int, channels: int,
callback: Callable[[bytes], None]) -> Any:
"""Open an audio capture stream. Returns a context manager."""
pass
class ALSABackend(AudioBackend):
"""ALSA backend using arecord (raw PCM output, no sound server required)."""
name = "alsa"
@classmethod
def is_available(cls) -> bool:
return shutil.which('arecord') is not None
@classmethod
def list_devices(cls) -> List[AudioDevice]:
devices = []
try:
result = subprocess.run(
['arecord', '-l'],
capture_output=True,
text=True,
timeout=5
)
if result.returncode != 0:
return devices
for line in result.stdout.split('\n'):
if not line.startswith('card '):
continue
try:
card_part, rest = line.split(':', 1)
card_num = card_part.replace('card', '').strip()
if ', device ' in rest:
dev_part = rest.split(', device ')[1]
dev_num = dev_part.split(':')[0].strip()
else:
dev_num = '0'
long_name = rest.split('[')[1].split(']')[0] if '[' in rest else rest.strip()
hw_id = f"hw:{card_num},{dev_num}"
devices.append(AudioDevice(
id=hw_id,
name=long_name,
channels=2,
sample_rate=44100,
backend=cls.name,
is_default=(card_num == '0' and dev_num == '0'),
is_monitor=False,
))
except (IndexError, ValueError):
continue
except Exception:
pass
return devices
def __init__(self, logger: logging.Logger):
self.logger = logger
def open_stream(self, device: AudioDevice, sample_rate: int, channels: int,
callback: Callable[[bytes], None]):
return ALSAStream(device, sample_rate, channels, callback, self.logger)
return f"{self.name} ({self.id})" + (" [DEFAULT]" if self.is_default else "")
class ALSAStream:
@@ -258,95 +169,94 @@ class ALSAStream:
class AudioSystem:
"""Manages audio backends and device discovery."""
"""Discovers ALSA capture devices and opens capture streams.
# All available backend classes
_backend_classes: List[type] = [ALSABackend]
ALSA (via `arecord`) is the only backend. This was once a plugin system
with a registry and an abstract base class; there has only ever been one
implementation, so it is a plain class now.
"""
def __init__(self, logger: logging.Logger):
self.logger = logger
self._backends: Dict[str, AudioBackend] = {}
self._discover_backends()
self.available = shutil.which('arecord') is not None
def _discover_backends(self):
"""Find and initialize available backends."""
for cls in self._backend_classes:
if cls.is_available():
self._backends[cls.name] = cls(self.logger)
self.logger.debug(f"Audio backend available: {cls.name}")
def list_devices(self) -> List[AudioDevice]:
"""Hardware capture devices, parsed out of `arecord -l`.
@property
def available_backends(self) -> List[str]:
return list(self._backends.keys())
Virtual PCMs defined in asound.conf (dsnoop, plug, ...) are NOT listed
by `arecord -l` and so never appear here — see find_device().
"""
devices = []
try:
result = subprocess.run(
['arecord', '-l'],
capture_output=True,
text=True,
timeout=5
)
if result.returncode != 0:
return devices
def get_backend(self, name: str) -> Optional[AudioBackend]:
return self._backends.get(name)
for line in result.stdout.split('\n'):
if not line.startswith('card '):
continue
try:
card_part, rest = line.split(':', 1)
card_num = card_part.replace('card', '').strip()
def list_all_devices(self) -> List[AudioDevice]:
"""List devices from all available backends."""
all_devices = []
seen_names = set()
if ', device ' in rest:
dev_part = rest.split(', device ')[1]
dev_num = dev_part.split(':')[0].strip()
else:
dev_num = '0'
for cls in self._backend_classes:
if cls.name in self._backends:
for dev in cls.list_devices():
# Deduplicate by name (same device may appear in multiple backends)
key = dev.name.lower()
if key not in seen_names:
all_devices.append(dev)
seen_names.add(key)
long_name = rest.split('[')[1].split(']')[0] if '[' in rest else rest.strip()
return all_devices
devices.append(AudioDevice(
id=f"hw:{card_num},{dev_num}",
name=long_name,
is_default=(card_num == '0' and dev_num == '0'),
))
except (IndexError, ValueError):
continue
except Exception:
pass
return devices
def find_device(self, spec: str, preferred_backend: str = None) -> Optional[AudioDevice]:
"""Find a device by ID, name, or pattern."""
devices = self.list_all_devices()
if not devices:
return None
# "default" - get the default device from preferred backend
if spec.lower() == 'default':
for dev in devices:
if dev.is_default:
if preferred_backend is None or dev.backend == preferred_backend:
return dev
return devices[0] if devices else None
# "monitor" - get first monitor source
if spec.lower() == 'monitor':
for dev in devices:
if dev.is_monitor:
if preferred_backend is None or dev.backend == preferred_backend:
return dev
return None
def find_device(self, spec: str) -> AudioDevice:
"""Resolve a config `device =` value to an ALSA device.
Never returns None. An unrecognised spec is passed to ALSA verbatim,
because a valid PCM name from asound.conf (`shared_mic`, `plughw:0`)
does not appear in `arecord -l` — and that must keep working even on a
host where no hardware card is enumerable at all, which is exactly the
case an earlier "no devices, give up" guard broke.
"""
devices = self.list_devices()
spec_lower = spec.lower()
# Try exact ID match first
for dev in devices:
if dev.id == spec:
return dev
if spec_lower == 'default':
for dev in devices:
if dev.is_default:
return dev
# 'default' is itself a valid ALSA PCM when no card is enumerable.
return devices[0] if devices else AudioDevice(id='default', name='default')
# Try exact name match
for dev in devices:
if dev.name.lower() == spec_lower:
return dev
exact_id = [d for d in devices if d.id == spec]
exact_name = [d for d in devices if d.name.lower() == spec_lower]
partial = [d for d in devices if spec_lower in d.name.lower()]
for candidates in (exact_id, exact_name, partial):
if candidates:
return candidates[0]
# Try partial name match
for dev in devices:
if spec_lower in dev.name.lower():
return dev
self.logger.info(
f"Device '{spec}' is not in `arecord -l` — using it as an ALSA PCM name")
return AudioDevice(id=spec, name=spec)
# No hardware device matched. Treat spec as a direct ALSA PCM name so
# virtual devices defined in asound.conf (dsnoop, plug, etc.) work without
# appearing in `arecord -l`.
if 'alsa' in self._backends:
return AudioDevice(
id=spec, name=spec, channels=2, sample_rate=44100, backend='alsa'
)
return None
def open_stream(self, device: AudioDevice, sample_rate: int, channels: int,
callback: Callable[[bytes], None]) -> 'ALSAStream':
"""Open a capture stream. Returns a context manager."""
return ALSAStream(device, sample_rate, channels, callback, self.logger)
class BaseRecorder(ABC):
@@ -675,7 +585,7 @@ class _AudioFileWriter:
class SoundcardRecorder(BaseRecorder):
"""Records from soundcard/audio devices using multiple backends."""
"""Records from a soundcard/audio device via ALSA."""
def __init__(self, name: str, config: Dict[str, Any], logger: logging.Logger,
audio_system: 'AudioSystem' = None,
@@ -685,9 +595,9 @@ class SoundcardRecorder(BaseRecorder):
# Initialize audio system if not provided
self.audio_system = audio_system or AudioSystem(logger)
if not self.audio_system.available_backends:
if not self.audio_system.available:
raise RuntimeError(
"No audio backends available.\n"
"arecord not found — soundcard capture is unavailable.\n"
"Install ALSA utilities: sudo apt install alsa-utils\n"
"(In Docker, also ensure /dev/snd is mapped into the container)"
)
@@ -696,7 +606,6 @@ class SoundcardRecorder(BaseRecorder):
self.device_spec = config.get('device', 'default')
self.sample_rate = config.get('sample_rate', 44100)
self.channels = config.get('channels', 2)
self.preferred_backend = config.get('backend', None)
# Validate format
if self.file_format not in ('wav', 'flac') and self.file_format != 'auto':
@@ -713,23 +622,19 @@ class SoundcardRecorder(BaseRecorder):
self.buffer_lock = threading.Lock()
def _resolve_device(self) -> AudioDevice:
"""Resolve device specification to AudioDevice."""
device = self.audio_system.find_device(self.device_spec, self.preferred_backend)
if device is None:
# List available devices in error
available = self.audio_system.list_all_devices()
device_list = "\n ".join(str(d) for d in available) if available else "None found"
raise ValueError(
f"[{self.name}] Device '{self.device_spec}' not found.\n"
f"Available devices:\n {device_list}"
)
"""Resolve the configured device spec to an AudioDevice.
find_device() always succeeds — an unknown spec becomes a literal ALSA
PCM name — so a typo surfaces as an arecord failure at capture time
rather than here. That is the price of supporting asound.conf PCMs,
which are invisible to `arecord -l`.
"""
device = self.audio_system.find_device(self.device_spec)
self.logger.info(f"[{self.name}] Using device: {device}")
return device
def _audio_callback(self, data: bytes):
"""Callback for audio data from backend."""
"""Callback for audio data from the capture thread."""
with self.buffer_lock:
self.audio_buffer.append(data)
@@ -767,13 +672,8 @@ class SoundcardRecorder(BaseRecorder):
retry_count = 0
next_split = self.get_next_split_time()
backend = self.audio_system.get_backend(self.device.backend)
if not backend:
self.logger.error(f"[{self.name}] Backend '{self.device.backend}' not available")
return
self.logger.info(f"[{self.name}] Starting recorder - Split every {self.split_duration} min")
self.logger.info(f"[{self.name}] Backend: {self.device.backend}, Device: {self.device.name}")
self.logger.info(f"[{self.name}] Device: {self.device}")
self.logger.info(f"[{self.name}] Sample rate: {self.sample_rate}, Channels: {self.channels}")
self.logger.info(f"[{self.name}] Next split: {next_split.strftime('%H:%M:%S')}")
@@ -782,7 +682,7 @@ class SoundcardRecorder(BaseRecorder):
if self.current_file is None:
self._open_output_file()
with backend.open_stream(
with self.audio_system.open_stream(
self.device, self.sample_rate, self.channels, self._audio_callback
) as stream:
self.logger.info(f"[{self.name}] Recording started")
@@ -862,8 +762,8 @@ class RecorderManager:
# Initialize shared audio system for soundcard recorders
self.audio_system = AudioSystem(self.logger)
if self.audio_system.available_backends:
self.logger.info(f"Audio backends: {', '.join(self.audio_system.available_backends)}")
if self.audio_system.available:
self.logger.info("Audio capture available: ALSA (arecord)")
# Parse source sections
for section in config.sections():
@@ -1005,7 +905,7 @@ class RecorderManager:
def list_audio_devices():
"""List available audio input devices from all backends."""
"""Print the ALSA capture devices usable as `device =` in config.ini."""
# Create a minimal logger for device listing
logger = logging.getLogger('isr-devices')
logger.setLevel(logging.WARNING) # Suppress debug output
@@ -1014,40 +914,36 @@ def list_audio_devices():
print(" ISR Audio Device Discovery")
print("=" * 70)
if not ALSABackend.is_available():
print("\n No audio backends available!")
audio_system = AudioSystem(logger)
if not audio_system.available:
print("\n arecord not found!")
print("\n Install ALSA utilities:")
print(" sudo apt install alsa-utils")
print()
return
print("\n Available Backends:")
print(" - ALSA (arecord)")
# Initialize audio system and list devices
audio_system = AudioSystem(logger)
devices = audio_system.list_all_devices()
if not devices:
print("\n No input devices found!")
print()
return
print("\n Input Devices:")
print(" " + "-" * 68)
for dev in devices:
flag_str = " [DEFAULT]" if dev.is_default else ""
print(f"\n {dev.name}{flag_str}")
print(f" ID: {dev.id} | Backend: {dev.backend}")
print(f" Channels: {dev.channels} | Sample Rate: {dev.sample_rate} Hz")
devices = audio_system.list_devices()
if devices:
print("\n Hardware Input Devices:")
print(" " + "-" * 68)
for dev in devices:
flag_str = " [DEFAULT]" if dev.is_default else ""
print(f"\n {dev.name}{flag_str}")
print(f" ID: {dev.id}")
else:
print("\n No hardware capture devices found.")
# This listing only ever shows hardware cards, so say so — a working
# asound.conf PCM will never appear here and users have gone looking.
print("\n" + "=" * 70)
print(" Configuration Examples:")
print("-" * 70)
print(" device = default # Use system default input")
print(" device = <name> # Match by partial name")
print(" device = <id> # Use exact ALSA ID, e.g. hw:0,0")
print(" device = <pcm> # Any ALSA PCM name from asound.conf, e.g. shared_mic")
print(" device = <pcm> # Any ALSA PCM name from asound.conf, e.g.")
print(" # shared_mic — these are never listed above,")
print(" # but work regardless")
print("=" * 70 + "\n")