AI story writer enhancements, text to video and voice generation, subscription management, and more.

backend/api/story_writer/utils/hd_video.py

@@ -1,13 +1,11 @@
 from __future__ import annotations
 
-from typing import Any, Dict, Optional
+from typing import Any, Dict
 
 from fastapi import HTTPException
 from loguru import logger
 from uuid import uuid4
 
-from .media_utils import load_story_image_bytes
-
 
 def generate_hd_video_payload(request: Any, user_id: str) -> Dict[str, Any]:
     """Handles synchronous HD video generation."""
@@ -57,8 +55,8 @@ def generate_hd_video_payload(request: Any, user_id: str) -> Dict[str, Any]:
 
 def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any]:
     """
-    Handles per-scene HD video generation including prompt enhancement,
-    subscription validation, and optional image conditioning.
+    Handles per-scene HD video generation including prompt enhancement
+    and subscription validation.
     """
     from services.database import get_db as get_db_validation
     from services.onboarding.api_key_manager import APIKeyManager
@@ -71,7 +69,6 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
     scene_number = request.scene_number
     logger.info(f"[StoryWriter] Generating HD video for scene {scene_number} for user {user_id}")
 
-    # Step 1: Validate API key
     hf_token = APIKeyManager().get_api_key("hf_token")
     if not hf_token:
         logger.error("[StoryWriter] Pre-flight: HF token not configured - blocking video generation")
@@ -83,7 +80,6 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
             },
         )
 
-    # Step 2: Subscription limits
     db_validation = next(get_db_validation())
     try:
         pricing_service = PricingService(db_validation)
@@ -93,7 +89,6 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
     finally:
         db_validation.close()
 
-    # Stage 1: Prompt enhancement
     enhanced_prompt = enhance_scene_prompt_for_video(
         current_scene=request.scene_data,
         story_context=request.story_context,
@@ -102,15 +97,6 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
     )
     logger.info(f"[StoryWriter] Generated enhanced prompt ({len(enhanced_prompt)} chars) for scene {scene_number}")
 
-    # Stage 2: Optional image reference
-    scene_image_bytes: Optional[bytes] = None
-    if getattr(request, "scene_image_url", None):
-        scene_image_bytes = load_story_image_bytes(request.scene_image_url)
-        if scene_image_bytes:
-            logger.info(f"[StoryWriter] Using scene image reference for scene {scene_number}")
-        else:
-            logger.warning(f"[StoryWriter] Scene image could not be loaded for scene {scene_number}, falling back to text-only video")
-
     kwargs: Dict[str, Any] = {}
     if getattr(request, "model", None):
         kwargs["model"] = request.model
@@ -129,7 +115,6 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
         prompt=enhanced_prompt,
         provider=getattr(request, "provider", None) or "huggingface",
         user_id=user_id,
-        input_image_bytes=scene_image_bytes,
         **kwargs,
     )
 
@@ -151,4 +136,3 @@ def generate_hd_video_scene_payload(request: Any, user_id: str) -> Dict[str, Any
         "model": getattr(request, "model", None) or "tencent/HunyuanVideo",
     }
 
-

backend/api/story_writer/utils/media_utils.py

@@ -11,6 +11,8 @@ from loguru import logger
 BASE_DIR = Path(__file__).resolve().parents[3]  # backend/
 STORY_IMAGES_DIR = (BASE_DIR / "story_images").resolve()
 STORY_IMAGES_DIR.mkdir(parents=True, exist_ok=True)
+STORY_AUDIO_DIR = (BASE_DIR / "story_audio").resolve()
+STORY_AUDIO_DIR.mkdir(parents=True, exist_ok=True)
 
 
 def load_story_image_bytes(image_url: str) -> Optional[bytes]:
@@ -48,6 +50,41 @@ def load_story_image_bytes(image_url: str) -> Optional[bytes]:
         return None
 
 
+def load_story_audio_bytes(audio_url: str) -> Optional[bytes]:
+    """
+    Resolve an authenticated story audio URL (e.g., /api/story/audio/<file>) to raw bytes.
+    Returns None if the file cannot be located.
+    """
+    if not audio_url:
+        return None
+
+    try:
+        parsed = urlparse(audio_url)
+        path = parsed.path if parsed.scheme else audio_url
+        prefix = "/api/story/audio/"
+        if prefix not in path:
+            logger.warning(f"[StoryWriter] Unsupported audio URL for video reference: {audio_url}")
+            return None
+
+        filename = path.split(prefix, 1)[1].split("?", 1)[0].strip()
+        if not filename:
+            return None
+
+        file_path = (STORY_AUDIO_DIR / filename).resolve()
+        if not str(file_path).startswith(str(STORY_AUDIO_DIR)):
+            logger.error(f"[StoryWriter] Attempted path traversal when resolving audio: {audio_url}")
+            return None
+
+        if not file_path.exists():
+            logger.warning(f"[StoryWriter] Referenced scene audio not found on disk: {file_path}")
+            return None
+
+        return file_path.read_bytes()
+    except Exception as exc:
+        logger.error(f"[StoryWriter] Failed to load reference audio for video gen: {exc}")
+        return None
+
+
 def resolve_media_file(base_dir: Path, filename: str) -> Path:
     """
     Returns a safe resolved path for a media file stored under base_dir.
@@ -62,8 +99,50 @@ def resolve_media_file(base_dir: Path, filename: str) -> Path:
         raise HTTPException(status_code=status.HTTP_403_FORBIDDEN, detail="Access denied")
 
     if not resolved.exists():
+        alternate = _find_alternate_media_file(base_dir, filename)
+        if alternate:
+            logger.warning(
+                "[StoryWriter] Requested media file '%s' missing; serving closest match '%s'",
+                filename,
+                alternate.name,
+            )
+            return alternate
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=f"File not found: {filename}")
 
     return resolved
 
 
+def _find_alternate_media_file(base_dir: Path, filename: str) -> Optional[Path]:
+    """
+    Attempt to find the most recent media file that matches the original name prefix.
+
+    This helps when files are regenerated with new UUID/hash suffixes but the frontend still
+    references an older filename.
+    """
+    try:
+        base_dir = base_dir.resolve()
+    except Exception:
+        return None
+
+    stem = Path(filename).stem
+    suffix = Path(filename).suffix
+
+    if not suffix or "_" not in stem:
+        return None
+
+    prefix = stem.rsplit("_", 1)[0]
+    pattern = f"{prefix}_*{suffix}"
+
+    try:
+        candidates = sorted(
+            (p for p in base_dir.glob(pattern) if p.is_file()),
+            key=lambda p: p.stat().st_mtime,
+            reverse=True,
+        )
+    except Exception as exc:
+        logger.debug(f"[StoryWriter] Failed to search alternate media files for {filename}: {exc}")
+        return None
+
+    return candidates[0] if candidates else None
+
+