Update app.py

app.py

@@ -1,6 +1,7 @@
 import os
 import sys
 import time
+import urllib.request
 from huggingface_hub import snapshot_download
 
 # --- 配置（可通过环境变量覆盖） ---
@@ -8,67 +9,104 @@ MODEL_REPO = os.getenv("MODEL_REPO", "mradermacher/Qwen3_Medical_GRPO-i1-GGUF")
 MODEL_FILE = os.getenv("MODEL_FILE", "Qwen3_Medical_GRPO.i1-Q4_K_M.gguf")
 MODEL_DIR = os.getenv("MODEL_DIR", "/models")
 MODEL_PATH = os.path.join(MODEL_DIR, MODEL_FILE)
-HF_TOKEN = os.getenv("HF_TOKEN", None)  # 如果模型是私有的，需要在 Spaces Secret 中设置这个值
-# 可选线程设置（不设置则默认 8）
+HF_TOKEN = os.getenv("HF_TOKEN", None)  # 如果模型是私有的，请在 Space Secrets 设置
 N_THREADS = int(os.getenv("N_THREADS", "8"))
 
-# --- 确保模型文件存在：若不存在，则从 Hugging Face Hub 下载 ---
 os.makedirs(MODEL_DIR, exist_ok=True)
 
+def download_via_http(url, dest_path, token=None, chunk_size=4*1024*1024):
+    """使用 urllib 分块下载，支持 token（Bearer）"""
+    print(f"Downloading via HTTP: {url} -> {dest_path}")
+    req = urllib.request.Request(url)
+    if token:
+        req.add_header("Authorization", f"Bearer {token}")
+    try:
+        with urllib.request.urlopen(req, timeout=120) as resp:
+            # 如果状态不是 200，抛错
+            if resp.status not in (200, 302, 301):
+                raise RuntimeError(f"HTTP download returned status {resp.status}")
+            # 写入临时文件，下载完成后重命名
+            tmp_dest = dest_path + ".part"
+            with open(tmp_dest, "wb") as fh:
+                while True:
+                    chunk = resp.read(chunk_size)
+                    if not chunk:
+                        break
+                    fh.write(chunk)
+            os.replace(tmp_dest, dest_path)
+            print("HTTP download finished.")
+    except Exception as e:
+        if os.path.exists(dest_path):
+            os.remove(dest_path)
+        raise
+
 def download_model_if_missing():
     if os.path.exists(MODEL_PATH):
         print(f"Model already exists at {MODEL_PATH}")
         return
 
-    print(f"Model not found at {MODEL_PATH}. Attempting to download from {MODEL_REPO} ...")
+    print(f"Model not found at {MODEL_PATH}. Trying snapshot_download from {MODEL_REPO} ...")
+    # 先尝试使用 huggingface_hub.snapshot_download（优先）
     try:
-        # snapshot_download 会把仓库内容下载到 MODEL_DIR；allow_patterns 只抓我们需要的文件
-        snapshot_download(
+        outdir = snapshot_download(
             repo_id=MODEL_REPO,
             repo_type="model",
             local_dir=MODEL_DIR,
             token=HF_TOKEN,
-            allow_patterns=[MODEL_FILE],
-            ignore_patterns=["*"]  # 先默认忽略所有，allow_patterns 会覆盖需要的
+            allow_patterns=[MODEL_FILE]  # 仅抓取我们需要的文件
         )
-    except Exception as e:
-        print("Error while trying to download the model:", e, file=sys.stderr)
-        print("If the model is private, make sure HF_TOKEN is set in Space Secrets and has read access.", file=sys.stderr)
-        raise
-
-    # 等待短时间让文件系统稳定（可选）
-    time.sleep(1)
-
-    if not os.path.exists(MODEL_PATH):
-        # 有时 snapshot_download 会把文件放在子目录，尝试在 MODEL_DIR 下搜索
+        # snapshot_download 有时会返回 download 目录；搜索目标文件
         found = None
-        for root, dirs, files in os.walk(MODEL_DIR):
+        for root, _, files in os.walk(outdir):
             if MODEL_FILE in files:
                 found = os.path.join(root, MODEL_FILE)
                 break
         if found:
-            print(f"Found model at {found}; moving to {MODEL_PATH}")
-            os.replace(found, MODEL_PATH)
+            # 如果找到了，把它移动到 MODEL_PATH（若已在正确位置则跳过）
+            if os.path.abspath(found) != os.path.abspath(MODEL_PATH):
+                print(f"Found model at {found}, moving to {MODEL_PATH}")
+                os.replace(found, MODEL_PATH)
+            print("snapshot_download succeeded.")
+            return
         else:
-            raise RuntimeError(f"Model download finished but {MODEL_PATH} still not found. Check repo contents.")
+            print("snapshot_download did not find the file (0 files). Will try direct HTTP download as fallback.")
+    except Exception as e:
+        print("snapshot_download failed / returned nothing:", e, file=sys.stderr)
+
+    # 备用：直接构造 resolve URL 并下载
+    direct_url = f"https://huggingface.co/{MODEL_REPO}/resolve/main/{MODEL_FILE}"
+    try:
+        download_via_http(direct_url, MODEL_PATH, token=HF_TOKEN)
+        return
+    except Exception as e:
+        print("Direct HTTP download failed:", e, file=sys.stderr)
+        # 最后再尝试在 MODEL_DIR 下搜索一遍（保险）
+        for root, _, files in os.walk(MODEL_DIR):
+            if MODEL_FILE in files:
+                found = os.path.join(root, MODEL_FILE)
+                print(f"Found model at {found} after fallback search; moving to {MODEL_PATH}")
+                os.replace(found, MODEL_PATH)
+                return
+        raise RuntimeError(f"Model download finished but {MODEL_PATH} still not found. Check repo contents and network.")
 
+# 执行下载
 download_model_if_missing()
+time.sleep(0.5)
 
-# --- 现在再导入并初始化 llama_cpp（确保模型已存在） ---
+# --- 导入并初始化 llama_cpp（确保模型存在） ---
 try:
     from llama_cpp import Llama
 except Exception as e:
-    print("Failed to import llama_cpp. Ensure the wheel you installed matches the runtime (musl vs glibc) and required libs are present.", file=sys.stderr)
+    print("Failed to import llama_cpp. Ensure the wheel matches the runtime and required system libs are present.", file=sys.stderr)
     raise
 
 if not os.path.exists(MODEL_PATH):
     raise RuntimeError(f"Model path does not exist after download: {MODEL_PATH}")
 
-# 初始化模型（给 N_THREADS 一个合理默认）
 n_threads = max(1, N_THREADS)
 llm = Llama(model_path=MODEL_PATH, n_ctx=4096, n_threads=n_threads)
 
-# --- system prompt 和 gradio 接口 ---
+# --- gradio 接口 ---
 import gradio as gr
 
 system_prompt = """You are given a problem.
@@ -82,7 +120,6 @@ def chat(user_input):
         response = llm(prompt, max_tokens=2048, temperature=0.7)
         return response["choices"][0]["text"]
     except Exception as e:
-        # 捕获运行时错误并返回友好提示（也会打印到容器日志）
         err_msg = f"Error while generating: {e}"
         print(err_msg, file=sys.stderr)
         return err_msg