From b5c2dc7cd2f68367a5a212bf556adf6c548eca04 Mon Sep 17 00:00:00 2001
From: Zioywishing <Zioywishing@users.noreply.github.com>
Date: Sun, 31 May 2026 13:55:40 +0800
Subject: [PATCH] fix(qqbot): skip voice detection for file uploads
 (content_type='file')

Minimal fix: add 'if ct == "file": return False' before extension
matching. The original fallback logic is preserved for empty/unknown
content_types. Only the bug case (file uploads with audio extensions)
is fixed.

Removed the over-engineered _looks_like_voice helper.
---
 gateway/platforms/qqbot/adapter.py | 40 ++++++++++--------------------
 tests/gateway/test_qqbot.py        | 10 +-------
 2 files changed, 14 insertions(+), 36 deletions(-)

diff --git a/gateway/platforms/qqbot/adapter.py b/gateway/platforms/qqbot/adapter.py
index 44f6aef97fe..0cab4bb794e 100644
--- a/gateway/platforms/qqbot/adapter.py
+++ b/gateway/platforms/qqbot/adapter.py
@@ -151,15 +151,6 @@ def _coerce_list(value: Any) -> List[str]:
 # ---------------------------------------------------------------------------
 
 
-def _looks_like_voice(filename: str) -> bool:
-    """Return True if *filename* has a known voice/audio extension."""
-    fn = filename.strip().lower()
-    return any(
-        fn.endswith(ext)
-        for ext in (".silk", ".amr", ".mp3", ".wav", ".ogg", ".m4a", ".aac", ".speex", ".flac")
-    )
-
-
 class QQAdapter(BasePlatformAdapter):
     """QQ Bot adapter backed by the official QQ Bot WebSocket Gateway + REST API."""
 
@@ -1816,27 +1807,22 @@ class QQAdapter(BasePlatformAdapter):
 
     @staticmethod
     def _is_voice_content_type(content_type: str, filename: str) -> bool:
-        """Check if an attachment is a voice/audio message.
-
-        The QQ Bot API explicitly sets ``content_type="voice"`` for voice
-        messages and ``content_type="file"`` for file uploads.  When
-        ``content_type`` is a known non-voice type (``"file"``, ``"image/*"``,
-        ``"video/*"``), we trust it unconditionally and **skip** extension
-        sniffing — files sent via QQ's file-transfer feature can have audio
-        extensions (e.g. ``.wav``, ``.mp3``) but must not be routed through
-        the STT pipeline.
-
-        When ``content_type`` is empty or unrecognised we fall back to
-        filename-extension matching as a defensive measure.
-        """
+        """Check if an attachment is a voice/audio message."""
         ct = content_type.strip().lower()
+        fn = filename.strip().lower()
         if ct == "voice" or ct.startswith("audio/"):
             return True
-        # Explicit non-voice types — never treat as voice regardless of extension.
-        if ct in ("file", "") or ct.startswith("image/") or ct.startswith("video/"):
-            return ct == "" and _looks_like_voice(filename)
-        # Unknown content_type — defensive extension fallback.
-        return _looks_like_voice(filename)
+        # QQ file uploads have content_type="file" — never treat as voice,
+        # even if the filename has an audio extension.
+        if ct == "file":
+            return False
+        _VOICE_EXTENSIONS = (
+            ".silk", ".amr", ".mp3", ".wav", ".ogg",
+            ".m4a", ".aac", ".speex", ".flac",
+        )
+        if any(fn.endswith(ext) for ext in _VOICE_EXTENSIONS):
+            return True
+        return False
 
     def _qq_media_headers(self) -> Dict[str, str]:
         """Return Authorization headers for QQ multimedia CDN downloads.
diff --git a/tests/gateway/test_qqbot.py b/tests/gateway/test_qqbot.py
index 50b09d74235..10d558c8260 100644
--- a/tests/gateway/test_qqbot.py
+++ b/tests/gateway/test_qqbot.py
@@ -151,20 +151,12 @@ class TestIsVoiceContentType:
         assert self._fn("", "recording.amr") is True
 
     def test_file_upload_with_audio_extension(self):
-        """File upload with audio extension must NOT be treated as voice."""
+        """content_type='file' is never voice, even with audio extension."""
         assert self._fn("file", "song.mp3") is False
         assert self._fn("file", "audio-30251.instrumental..wav") is False
-
-    def test_file_upload_never_voice(self):
-        """content_type='file' is never voice, regardless of extension."""
         assert self._fn("file", "recording.silk") is False
         assert self._fn("file", "voice.amr") is False
 
-    def test_unknown_content_type_extension_fallback(self):
-        """Unknown content_type falls back to extension matching."""
-        assert self._fn("unknown/type", "voice.ogg") is True
-        assert self._fn("unknown/type", "data.json") is False
-
 
 # ---------------------------------------------------------------------------
 # Voice attachment SSRF protection