From 9058cf061d481fb0ca9678c3777555d495115597 Mon Sep 17 00:00:00 2001
From: Robert Wojciechowski <hello@robertvoy.com>
Date: Sat, 28 Feb 2026 00:02:31 +0000
Subject: [PATCH] fix: sync audio media refs for worker prompt validation

---
 api/orchestration/media_sync.py | 15 ++++++++-------
 tests/api/test_media_sync.py    | 14 +++++++++++++-
 2 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/api/orchestration/media_sync.py b/api/orchestration/media_sync.py
index 24223e1..877d4ed 100644
--- a/api/orchestration/media_sync.py
+++ b/api/orchestration/media_sync.py
@@ -12,12 +12,13 @@ from ...utils.trace_logger import trace_debug, trace_info
 
 
 LIKELY_FILENAME_RE = re.compile(
-    r"\.(ckpt|safetensors|pt|pth|bin|yaml|json|png|jpg|jpeg|webp|gif|bmp|latent|txt|vae|lora|embedding)"
+    r"\.(ckpt|safetensors|pt|pth|bin|yaml|json|png|jpg|jpeg|webp|gif|bmp|mp4|avi|mov|mkv|webm|"
+    r"wav|mp3|flac|m4a|aac|ogg|opus|aiff|aif|wma|latent|txt|vae|lora|embedding)"
     r"(\s*\[\w+\])?$",
     re.IGNORECASE,
 )
-IMAGE_OR_VIDEO_RE = re.compile(
-    r"\.(png|jpg|jpeg|webp|gif|bmp|mp4|avi|mov|mkv|webm)(\s*\[\w+\])?$",
+MEDIA_FILE_RE = re.compile(
+    r"\.(png|jpg|jpeg|webp|gif|bmp|mp4|avi|mov|mkv|webm|wav|mp3|flac|m4a|aac|ogg|opus|aiff|aif|wma)(\s*\[\w+\])?$",
     re.IGNORECASE,
 )
 
@@ -39,7 +40,7 @@ def convert_paths_for_platform(obj, target_separator):
                     return trimmed
 
                 # Keep relative media-style paths in forward-slash form (Comfy-style annotated paths).
-                if not has_drive and not is_absolute and not has_protocol and IMAGE_OR_VIDEO_RE.search(trimmed):
+                if not has_drive and not is_absolute and not has_protocol and MEDIA_FILE_RE.search(trimmed):
                     return re.sub(r"[\\]+", "/", trimmed)
 
                 if target_separator == "\\":
@@ -56,17 +57,17 @@ def convert_paths_for_platform(obj, target_separator):
 
 
 def _find_media_references(prompt_obj):
-    """Find media file references in image/video inputs used by worker prompts."""
+    """Find media file references in image/video/audio inputs used by worker prompts."""
     media_refs = set()
     for node in prompt_obj.values():
         if not isinstance(node, dict):
             continue
         inputs = node.get("inputs", {})
-        for key in ("image", "video"):
+        for key in ("image", "video", "audio"):
             value = inputs.get(key)
             if isinstance(value, str):
                 cleaned = re.sub(r"\s*\[\w+\]$", "", value).strip().replace("\\", "/")
-                if IMAGE_OR_VIDEO_RE.search(cleaned):
+                if MEDIA_FILE_RE.search(cleaned):
                     media_refs.add(cleaned)
     return sorted(media_refs)
 
diff --git a/tests/api/test_media_sync.py b/tests/api/test_media_sync.py
index f402f37..99cabf1 100644
--- a/tests/api/test_media_sync.py
+++ b/tests/api/test_media_sync.py
@@ -99,11 +99,16 @@ class ConvertPathsForPlatformTests(unittest.TestCase):
         self.assertNotIn("/", result["ckpt_name"])
 
     def test_relative_media_paths_always_stay_forward_slash(self):
-        """Relative image/video paths (Comfy annotated style) must not be backslash-ified."""
+        """Relative image/video/audio paths (Comfy annotated style) must not be backslash-ified."""
         obj = {"image": "subfolder/my_photo.png"}
         result = ms.convert_paths_for_platform(obj, "\\")
         self.assertEqual(result["image"], "subfolder/my_photo.png")
 
+    def test_relative_audio_paths_stay_forward_slash(self):
+        obj = {"audio": "subfolder/my_track.wav"}
+        result = ms.convert_paths_for_platform(obj, "\\")
+        self.assertEqual(result["audio"], "subfolder/my_track.wav")
+
     def test_annotated_relative_media_path_stays_forward_slash(self):
         obj = {"image": "input/frame.jpg [abc123]"}
         result = ms.convert_paths_for_platform(obj, "\\")
@@ -169,6 +174,11 @@ class FindMediaReferencesTests(unittest.TestCase):
         refs = ms._find_media_references(prompt)
         self.assertIn("clip.mp4", refs)
 
+    def test_finds_audio_input(self):
+        prompt = {"1": {"class_type": "LoadAudio", "inputs": {"audio": "track.wav"}}}
+        refs = ms._find_media_references(prompt)
+        self.assertIn("track.wav", refs)
+
     def test_strips_annotation_suffix(self):
         prompt = {"1": {"class_type": "LoadImage", "inputs": {"image": "photo.jpg [abc123]"}}}
         refs = ms._find_media_references(prompt)
@@ -219,10 +229,12 @@ class FindMediaReferencesTests(unittest.TestCase):
         prompt = {
             "1": {"class_type": "LoadImage", "inputs": {"image": "frame.png"}},
             "2": {"class_type": "LoadVideo", "inputs": {"video": "clip.mp4"}},
+            "3": {"class_type": "LoadAudio", "inputs": {"audio": "track.wav"}},
         }
         refs = ms._find_media_references(prompt)
         self.assertIn("frame.png", refs)
         self.assertIn("clip.mp4", refs)
+        self.assertIn("track.wav", refs)
 
 
 if __name__ == "__main__":