DroneMind/voice_drone/core/mic_device_select.py

"""启动时列出 arecord -l 与 PyAudio 输入设备，并把 ALSA card/device 映射到 PyAudio 索引供交互选择。"""

from __future__ import annotations

import re
import subprocess
from typing import List, Optional, Tuple, Any

from voice_drone.logging_ import get_logger

logger = get_logger("mic_device_select")


def run_arecord_l() -> str:
    try:
        r = subprocess.run(
            ["arecord", "-l"],
            capture_output=True,
            text=True,
            timeout=5,
            check=False,
        )
        out = (r.stdout or "").rstrip()
        err = (r.stderr or "").strip()
        body = out + (f"\n{err}" if err else "")
        return body.strip() if body.strip() else "(arecord 无输出)"
    except FileNotFoundError:
        return "(未找到 arecord，可安装 alsa-utils)"
    except Exception as e:  # noqa: BLE001
        return f"(执行 arecord -l 失败: {e})"


def parse_arecord_capture_lines(text: str) -> List[Tuple[int, int, str]]:
    rows: List[Tuple[int, int, str]] = []
    for line in text.splitlines():
        m = re.search(r"card\s+(\d+):.+?,\s*device\s+(\d+):", line, re.IGNORECASE)
        if m:
            rows.append((int(m.group(1)), int(m.group(2)), line.strip()))
    return rows


def _pyaudio_input_devices() -> List[Tuple[int, Any]]:
    from voice_drone.core.portaudio_env import fix_ld_path_for_portaudio

    fix_ld_path_for_portaudio()
    import pyaudio

    pa = pyaudio.PyAudio()
    out: List[Tuple[int, Any]] = []
    try:
        for i in range(pa.get_device_count()):
            try:
                inf = pa.get_device_info_by_index(i)
                if int(inf.get("maxInputChannels", 0)) <= 0:
                    continue
                out.append((i, inf))
            except Exception:
                continue
        return out
    finally:
        pa.terminate()


def match_alsa_hw_to_pyaudio_index(
    card: int,
    dev: int,
    pa_items: List[Tuple[int, Any]],
) -> Optional[int]:
    want1 = f"(hw:{card},{dev})"
    want2 = f"(hw:{card}, {dev})"
    for idx, inf in pa_items:
        name = str(inf.get("name", ""))
        if want1 in name or want2 in name:
            return idx
    return None


def print_mic_device_menu() -> List[int]:
    """
    打印 arecord + PyAudio + 映射表。
    返回本菜单中列出的 PyAudio 索引列表（顺序与 [1]、[2]… 一致）。
    """
    alsa_text = run_arecord_l()
    pa_items = _pyaudio_input_devices()

    print("\n" + "=" * 72, flush=True)
    print("录音设备（先 ALSA，再 PortAudio；请记下要用的 PyAudio 索引）", flush=True)
    print("=" * 72, flush=True)
    print("\n--- arecord -l（系统硬件视角）---\n", flush=True)
    print(alsa_text, flush=True)

    print("\n--- PyAudio 可录音设备（maxInputChannels > 0）---\n", flush=True)
    ordered_indices: List[int] = []
    if not pa_items:
        print("（无）\n", flush=True)
    for rank, (idx, inf) in enumerate(pa_items, start=1):
        ordered_indices.append(idx)
        mic = int(inf.get("maxInputChannels", 0))
        outc = int(inf.get("maxOutputChannels", 0))
        name = str(inf.get("name", "?"))
        print(
            f"  [{rank}]  PyAudio_index={idx}  in={mic} out={outc}  {name}",
            flush=True,
        )

    alsa_rows = parse_arecord_capture_lines(alsa_text)
    print(
        "\n--- 映射：arecord 的 card / device → PyAudio 索引（匹配设备名中的 hw:X,Y）---\n",
        flush=True,
    )
    if not alsa_rows:
        print("  （未解析到 card/device 行，请直接用上一表的 PyAudio_index）", flush=True)
    for card, dev, line in alsa_rows:
        pidx = match_alsa_hw_to_pyaudio_index(card, dev, pa_items)
        short = line if len(line) <= 76 else line[:73] + "..."
        if pidx is not None:
            print(
                f"  card {card}, device {dev}  →  PyAudio 索引 {pidx}\n      {short}",
                flush=True,
            )
        else:
            print(
                f"  card {card}, device {dev}  →  （无 in>0 设备名含 hw:{card},{dev}）\n      {short}",
                flush=True,
            )

    print(
        "\n说明：程序只会用「一个」PyAudio 索引打开麦克风；"
        "HDMI 等若 in=0 不会出现在可录音列表。\n"
        + "=" * 72
        + "\n",
        flush=True,
    )
    return ordered_indices


def prompt_for_input_device_index() -> int:
    """交互式选择，返回写入 audio.input_device_index 的 PyAudio 索引。"""
    ordered = print_mic_device_menu()
    if not ordered:
        print("错误：没有发现可录音的 PyAudio 设备。", flush=True)
        raise SystemExit(2)

    valid_set = set(ordered)
    print(
        "请输入菜单序号 [1-"
        f"{len(ordered)}]（推荐），或直接输入 PyAudio_index 数字；q 退出。",
        flush=True,
    )
    while True:
        try:
            raw = input("录音设备> ").strip()
        except EOFError:
            raise SystemExit(1) from None
        if not raw:
            continue
        if raw.lower() in ("q", "quit", "exit"):
            raise SystemExit(0)
        if not raw.isdigit():
            print("请输入正整数或 q。", flush=True)
            continue
        n = int(raw)
        if 1 <= n <= len(ordered):
            chosen = ordered[n - 1]
            print(f"已选择：菜单 [{n}] → PyAudio 索引 {chosen}\n", flush=True)
            logger.info("交互选择录音设备 PyAudio index=%s", chosen)
            return chosen
        if n in valid_set:
            print(f"已选择：PyAudio 索引 {n}\n", flush=True)
            logger.info("交互选择录音设备 PyAudio index=%s", n)
            return n
        print(
            f"无效：{n} 不在可选列表。可选序号为 1～{len(ordered)}，"
            f"或索引之一 {sorted(valid_set)}。",
            flush=True,
        )


def apply_input_device_index_only(index: int) -> None:
    """写入运行时配置：仅用索引选设备，其余 yaml 中的 hw/名称匹配不再参与 logic。"""
    from voice_drone.core.configuration import SYSTEM_AUDIO_CONFIG

    SYSTEM_AUDIO_CONFIG["input_device_index"] = int(index)
    SYSTEM_AUDIO_CONFIG["input_hw_card_device"] = None
    SYSTEM_AUDIO_CONFIG["input_device_name_match"] = None
    SYSTEM_AUDIO_CONFIG["input_strict_selection"] = False