WIP checkpoint: US-1: Asset Import Pipeline \u2014 glTF upload endpoint with metadata extraction

backend/app/routers/assets.py

@@ -1,5 +1,6 @@
 """Asset management router."""
 
+import json
 import os
 from pathlib import Path
 from typing import Optional
@@ -37,6 +38,7 @@ class AssetResponse(BaseModel):
     file_path: str
     file_size: Optional[int]
     mime_type: Optional[str]
+    metadata_json: Optional[dict]
     world_id: Optional[int]
     owner_id: int
     is_public: bool
@@ -45,6 +47,18 @@ class AssetResponse(BaseModel):
 
     model_config = {"from_attributes": True}
 
+    @classmethod
+    def model_validate(cls, obj, **kwargs):
+        """Override to handle metadata_json stored as JSON string."""
+        if isinstance(obj, dict):
+            obj = dict(obj)
+            if "metadata_json" in obj and isinstance(obj["metadata_json"], str):
+                try:
+                    obj["metadata_json"] = json.loads(obj["metadata_json"])
+                except (json.JSONDecodeError, TypeError):
+                    obj["metadata_json"] = None
+        return super().model_validate(obj, **kwargs)
+
 
 class AssetUpdate(BaseModel):
     """Schema for updating an asset."""
@@ -64,6 +78,144 @@ def _sanitize_filename(filename: str) -> str:
     return safe or "file"
 
 
+def _extract_gltf_metadata(content: bytes) -> Optional[dict]:
+    """Extract metadata from a glTF/GLB file.
+
+    Returns a dict with mesh_count, vertex_count, has_animation, texture_refs
+    if the file is a valid glTF/GLB, or None otherwise.
+
+    Uses only the standard library (struct, json) so no extra dependencies
+    are required.
+    """
+    if not content:
+        return None
+
+    # --- GLB detection ---
+    # GLB magic: 0x676C5442 ("glTF") followed by version 2 and payload length.
+    if len(content) >= 12:
+        magic = int.from_bytes(content[0:4], "little")
+        if magic == 0x46546C67:  # "glTF"
+            version = int.from_bytes(content[4:8], "little")
+            if version == 2:
+                return _parse_glb(content)
+
+    # --- .gltf JSON detection ---
+    # Try parsing as JSON; if it has a top-level "scene" or "scenes" key,
+    # it's likely a .gltf file.
+    try:
+        data = json.loads(content)
+    except (json.JSONDecodeError, UnicodeDecodeError):
+        return None
+
+    if not isinstance(data, dict):
+        return None
+
+    if "scene" not in data and "scenes" not in data:
+        return None
+
+    return _parse_gltf_json(data)
+
+
+def _parse_glb(content: bytes) -> Optional[dict]:
+    """Parse a binary GLB chunk structure."""
+    # Header: magic(4) + version(4) + length(4) = 12 bytes
+    # Then chunks: length(4) + type(4) + data...
+    offset = 12
+    json_start = None
+    json_length = 0
+    binary_start = None
+    binary_length = 0
+
+    while offset < len(content):
+        if offset + 8 > len(content):
+            break
+        chunk_length = int.from_bytes(content[offset:offset + 4], "little")
+        chunk_type = int.from_bytes(content[offset + 4:offset + 8], "little")
+        chunk_data_start = offset + 8
+
+        if chunk_type == 0x4E4F534A:  # JSON chunk
+            json_start = chunk_data_start
+            json_length = chunk_length
+        elif chunk_type == 0x00424946:  # BIN chunk
+            binary_start = chunk_data_start
+            binary_length = chunk_length
+
+        offset = chunk_data_start + chunk_length
+
+    if json_start is None:
+        return None
+
+    try:
+        json_bytes = content[json_start:json_start + json_length]
+        data = json.loads(json_bytes)
+    except (json.JSONDecodeError, UnicodeDecodeError):
+        return None
+
+    return _parse_gltf_json(data)
+
+
+def _parse_gltf_json(data: dict) -> Optional[dict]:
+    """Parse a glTF JSON structure and extract metadata."""
+    meshes = data.get("meshes") or []
+    animations = data.get("animations") or []
+    textures = data.get("textures") or []
+    images = data.get("images") or []
+    accessors = data.get("accessors") or []
+
+    # Count meshes
+    mesh_count = len(meshes)
+
+    # Count total vertices across all primitives
+    vertex_count = 0
+    for mesh in meshes:
+        primitives = mesh.get("primitives") or []
+        for primitive in primitives:
+            indices = primitive.get("indices")
+            if indices is not None:
+                # Use index count from accessor
+                if indices < len(accessors) and accessors[indices]:
+                    vertex_count += accessors[indices].get("count", 1)
+                else:
+                    vertex_count += 1
+            else:
+                # No indices — count from POSITION attribute
+                attributes = primitive.get("attributes") or {}
+                pos_accessor = attributes.get("POSITION")
+                if pos_accessor is not None:
+                    if pos_accessor < len(accessors) and accessors[pos_accessor]:
+                        vertex_count += accessors[pos_accessor].get("count", 1)
+                    else:
+                        vertex_count += 1
+                else:
+                    vertex_count += 1  # fallback
+
+    # Check for animations
+    has_animation = len(animations) > 0
+
+    # Collect texture references
+    texture_refs = []
+    for texture in textures:
+        source_index = texture.get("source")
+        ref = {}
+        if source_index is not None and source_index < len(images):
+            image = images[source_index]
+            if isinstance(image, dict):
+                if image.get("name"):
+                    ref["name"] = image["name"]
+                if image.get("uri"):
+                    ref["uri"] = image["uri"]
+                if image.get("mimeType"):
+                    ref["mime_type"] = image["mimeType"]
+        texture_refs.append(ref)
+
+    return {
+        "mesh_count": mesh_count,
+        "vertex_count": vertex_count,
+        "has_animation": has_animation,
+        "texture_refs": texture_refs,
+    }
+
+
 # --- Endpoints ---
 
 @router.post("/", response_model=AssetResponse, status_code=201)
@@ -204,6 +356,13 @@ async def upload_asset_file(
     asset.mime_type = file.content_type
     asset.file_size = file_size
 
+    # Extract glTF metadata for glTF/GLB files.
+    filename_lower = (file.filename or "").lower()
+    if filename_lower.endswith(".glb") or filename_lower.endswith(".gltf"):
+        metadata = _extract_gltf_metadata(content)
+        if metadata is not None:
+            asset.metadata_json = json.dumps(metadata)
+
     await db.commit()
     await db.refresh(asset)
     return asset