fix: convert qq_official voice messages to wav

astrbot/core/platform/sources/qqofficial/qqofficial_message_event.py

@@ -6,6 +6,8 @@
 from typing import cast
 
 import aiofiles
+import aiofiles.os
+import aiofiles.ospath
 import botpy
 import botpy.errors
 import botpy.message
@@ -22,6 +24,7 @@
 from astrbot.api.platform import AstrBotMessage, PlatformMetadata
 from astrbot.core.utils.astrbot_path import get_astrbot_temp_path
 from astrbot.core.utils.io import download_image_by_url, file_to_base64
+from astrbot.core.utils.media_utils import convert_audio_to_wav
 from astrbot.core.utils.tencent_record_helper import wav_to_tencent_silk
 
 
@@ -608,7 +611,19 @@ async def _parse_to_qqofficial(message: MessageChain):
                         temp_dir,
                         f"qqofficial_{uuid.uuid4()}.silk",
                     )
+                    converted_record_wav_path = None
                     try:
+                        if not await QQOfficialMessageEvent._is_wav_audio_file(
+                            record_wav_path
+                        ):
+                            converted_record_wav_path = os.path.join(
+                                temp_dir,
+                                f"qqofficial_{uuid.uuid4()}.wav",
+                            )
+                            record_wav_path = await convert_audio_to_wav(
+                                record_wav_path,
+                                converted_record_wav_path,
+                            )
                         duration = await wav_to_tencent_silk(
                             record_wav_path,
                             record_tecent_silk_path,
@@ -621,6 +636,16 @@ async def _parse_to_qqofficial(message: MessageChain):
                     except Exception as e:
                         logger.error(f"处理语音时出错: {e}")
                         record_file_path = None
+                    finally:
+                        if converted_record_wav_path and await aiofiles.ospath.exists(
+                            converted_record_wav_path
+                        ):
+                            try:
+                                await aiofiles.os.remove(converted_record_wav_path)
+                            except OSError as e:
+                                logger.warning(
+                                    f"[QQOfficial] failed to remove converted audio file: {e}"
+                                )
             elif isinstance(i, Video) and not video_file_source:
                 if i.file.startswith("file:///"):
                     video_file_source = i.file[8:]
@@ -648,3 +673,12 @@ async def _parse_to_qqofficial(message: MessageChain):
             file_source,
             file_name,
         )
+
+    @staticmethod
+    async def _is_wav_audio_file(file_path: str) -> bool:
+        try:
+            async with aiofiles.open(file_path, "rb") as f:
+                header = await f.read(12)
+        except OSError:
+            return False
+        return len(header) >= 12 and header[:4] == b"RIFF" and header[8:12] == b"WAVE"

tests/test_qqofficial_message_event.py

@@ -0,0 +1,97 @@
+from pathlib import Path
+
+import pytest
+
+from astrbot.core.message.components import Record
+from astrbot.core.message.message_event_result import MessageChain
+from astrbot.core.platform.sources.qqofficial.qqofficial_message_event import (
+    QQOfficialMessageEvent,
+)
+
+
+def _wav_header() -> bytes:
+    return b"RIFF\x00\x00\x00\x00WAVEfmt "
+
+
+@pytest.mark.asyncio
+async def test_parse_to_qqofficial_converts_non_wav_record_before_silk(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
+):
+    source_path = tmp_path / "voice.mp3"
+    source_path.write_bytes(b"ID3fake-mp3")
+    converted_paths: list[Path] = []
+
+    async def fake_convert_audio_to_wav(audio_path: str, output_path: str | None = None):
+        assert audio_path == str(source_path)
+        assert output_path is not None
+        converted_path = Path(output_path)
+        converted_path.write_bytes(_wav_header())
+        converted_paths.append(converted_path)
+        return output_path
+
+    async def fake_wav_to_tencent_silk(wav_path: str, silk_path: str):
+        assert converted_paths
+        assert wav_path == str(converted_paths[0])
+        Path(silk_path).write_bytes(b"fake-silk")
+        return 1200
+
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.get_astrbot_temp_path",
+        lambda: str(tmp_path),
+    )
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.convert_audio_to_wav",
+        fake_convert_audio_to_wav,
+    )
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.wav_to_tencent_silk",
+        fake_wav_to_tencent_silk,
+    )
+
+    result = await QQOfficialMessageEvent._parse_to_qqofficial(
+        MessageChain([Record.fromFileSystem(str(source_path))])
+    )
+
+    assert converted_paths
+    assert not converted_paths[0].exists()
+    assert result[3] is not None
+    assert str(result[3]).endswith(".silk")
+
+
+@pytest.mark.asyncio
+async def test_parse_to_qqofficial_skips_conversion_for_wav_record(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
+):
+    source_path = tmp_path / "voice.wav"
+    source_path.write_bytes(_wav_header())
+    conversions: list[tuple[str, str | None]] = []
+
+    async def fake_convert_audio_to_wav(audio_path: str, output_path: str | None = None):
+        conversions.append((audio_path, output_path))
+        return output_path or audio_path
+
+    async def fake_wav_to_tencent_silk(wav_path: str, silk_path: str):
+        assert wav_path == str(source_path)
+        Path(silk_path).write_bytes(b"fake-silk")
+        return 800
+
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.get_astrbot_temp_path",
+        lambda: str(tmp_path),
+    )
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.convert_audio_to_wav",
+        fake_convert_audio_to_wav,
+    )
+    monkeypatch.setattr(
+        "astrbot.core.platform.sources.qqofficial.qqofficial_message_event.wav_to_tencent_silk",
+        fake_wav_to_tencent_silk,
+    )
+
+    result = await QQOfficialMessageEvent._parse_to_qqofficial(
+        MessageChain([Record.fromFileSystem(str(source_path))])
+    )
+
+    assert conversions == []
+    assert result[3] is not None
+    assert str(result[3]).endswith(".silk")