fix(avatar): align knowledge upload limit with production

digital-avatar-app/backend/routers/knowledge.py

@@ -17,7 +17,8 @@ UPLOAD_DIR = os.path.abspath(os.getenv("UPLOAD_DIR", os.path.join(BASE_DIR, "upl
 os.makedirs(UPLOAD_DIR, exist_ok=True)
 
 ALLOWED_EXT = {".md", ".txt", ".pdf", ".doc", ".docx", ".xlsx"}
-MAX_UPLOAD_BYTES = 20 * 1024 * 1024
+MAX_UPLOAD_BYTES = 100 * 1024 * 1024
+UPLOAD_CHUNK_BYTES = 1024 * 1024
 
 
 class QAIn(BaseModel):
@@ -80,17 +81,25 @@ async def upload_doc(avatar_id: str, file: UploadFile = File(...), authorization
     os.makedirs(avatar_dir, exist_ok=True)
     stored = f"{uuid.uuid4().hex}{ext}"
     path = os.path.join(avatar_dir, stored)
-    content = await file.read()
-    if len(content) > MAX_UPLOAD_BYTES:
-        return fail("文件不能超过 20MB", code=400)
-    with open(path, "wb") as f:
-        f.write(content)
+    file_size = 0
+    try:
+        # Stream large files to disk so a 100MB upload does not occupy 100MB RAM.
+        with open(path, "wb") as f:
+            while chunk := await file.read(UPLOAD_CHUNK_BYTES):
+                file_size += len(chunk)
+                if file_size > MAX_UPLOAD_BYTES:
+                    raise ValueError("文件不能超过 100MB")
+                f.write(chunk)
+    except ValueError as exc:
+        if os.path.exists(path):
+            os.remove(path)
+        return fail(str(exc), code=400)
     doc = KnowledgeDoc(
         id=uuid.uuid4().hex,
         avatar_id=avatar_id,
         filename=file.filename,
         file_type=ext.lstrip("."),
-        file_size=len(content),
+        file_size=file_size,
         file_url=f"/api/files/{avatar_id}/{stored}",
         status="parsing",
     )