fix: prevent repeated model GPU submissions

modelhub_submmit_api/modelhub_client.py

@@ -21,6 +21,23 @@ class ModelHubAPIError(RuntimeError):
         self.payload = payload
 
 
+def parse_model_submission_precheck(payload: Any) -> dict[str, Any]:
+    """Validate the community lookup response instead of failing open."""
+    if not isinstance(payload, dict):
+        raise ModelHubAPIError("Community model precheck returned a non-object response", payload=payload)
+    data = payload.get("data")
+    if not isinstance(data, dict) or "verifyResult" not in data:
+        raise ModelHubAPIError("Community model precheck response is incomplete", payload=payload)
+    verify_result = data.get("verifyResult") or {}
+    if not isinstance(verify_result, dict):
+        raise ModelHubAPIError("Community model precheck verifyResult is invalid", payload=payload)
+
+    return {
+        "isInDB": data.get("isInDB") is True,
+        "processedGpus": set(str(gpu) for gpu in verify_result),
+    }
+
+
 class ModelHubClient:
     def __init__(
         self,
@@ -42,9 +59,13 @@ class ModelHubClient:
             retries=retries,
         )
 
-    def search_by_model_id(self, model_id: str) -> dict[str, Any]:
+    def search_by_model_id(self, model_id: str, *, force_refresh: bool = False) -> dict[str, Any]:
+        del force_refresh
         return self._request("GET", "/api/computility/models/search-by-model-id", query={"modelId": model_id})
 
+    def model_submission_precheck(self, model_id: str, *, force_refresh: bool = False) -> dict[str, Any]:
+        return parse_model_submission_precheck(self.search_by_model_id(model_id, force_refresh=force_refresh))
+
     def is_model_processed_for_gpu(self, model_id: str, target_gpu: str) -> bool:
         gpu_result = self.get_verify_result_map(model_id).get(target_gpu)
         if gpu_result is None:
@@ -56,7 +77,7 @@ class ModelHubClient:
         return ((payload.get("data") or {}).get("verifyResult") or {})
 
     def processed_gpus_for_model(self, model_id: str) -> set[str]:
-        return set(self.get_verify_result_map(model_id).keys())
+        return set(self.model_submission_precheck(model_id)["processedGpus"])
 
     def list_tasks_page(
         self,
@@ -256,6 +277,16 @@ DUPLICATE_SUBMISSION_MARKERS = (
     "already in progress",
 )
 
+MODEL_UNIQUENESS_ERROR_MARKERS = (
+    "模型唯一性检查",
+    "唯一性检查没有通过",
+    "模型已存在",
+    "model uniqueness",
+    "uniqueness check",
+    "duplicate model",
+    "model already exists",
+)
+
 DEFAULT_CAPACITY_STATE_PATH = Path(".modelhub_state/account_capacity.json")
 
 
@@ -275,6 +306,13 @@ def is_duplicate_submission_error(error: ModelHubAPIError) -> bool:
     return any(marker in message for marker in DUPLICATE_SUBMISSION_MARKERS)
 
 
+def is_model_uniqueness_error(error: ModelHubAPIError) -> bool:
+    message = str(error).strip().lower()
+    if isinstance(error.payload, dict):
+        message = f"{message} {error.payload.get('message') or ''}".lower()
+    return any(marker in message for marker in MODEL_UNIQUENESS_ERROR_MARKERS)
+
+
 class ModelHubClientPool:
     def __init__(
         self,
@@ -385,13 +423,6 @@ class ModelHubClientPool:
         except Exception:
             return max(0, max_count - 1)
 
-    def _safe_search_by_model_id(self, client: ModelHubClient, model_id: str) -> dict[str, Any]:
-        try:
-            payload = client.search_by_model_id(model_id)
-            return payload if isinstance(payload, dict) else {}
-        except Exception:
-            return {}
-
     def _safe_list_tasks(self, client: ModelHubClient, kwargs: dict[str, Any]) -> list[dict[str, Any]]:
         try:
             return client.list_tasks(**kwargs)
@@ -546,29 +577,33 @@ class ModelHubClientPool:
                 if now - cached_at >= self._verify_cache_ttl:
                     self._verify_cache.pop(model_id, None)
 
-    def search_by_model_id(self, model_id: str) -> dict[str, Any]:
+    def search_by_model_id(self, model_id: str, *, force_refresh: bool = False) -> dict[str, Any]:
         # Reuse recent model verification results across short poll cycles.
-        with self._state_lock:
-            cached = self._read_from_cache(model_id)
-        if cached is not None:
-            return cached
+        if not force_refresh:
+            with self._state_lock:
+                cached = self._read_from_cache(model_id)
+            if cached is not None:
+                return cached
 
         # Only query ONE client (the reader) instead of fanning out to all clients.
         # verifyResult is model-specific platform data, not account-specific.
-        response = self._safe_search_by_model_id(self._reader, model_id)
+        response = self._reader.search_by_model_id(model_id)
         if not isinstance(response, dict):
-            response = {"code": 0, "data": {"verifyResult": {}}}
+            raise ModelHubAPIError("Community model precheck returned a non-object response", payload=response)
 
         with self._state_lock:
             self._write_to_cache(model_id, response)
         return response
 
+    def model_submission_precheck(self, model_id: str, *, force_refresh: bool = False) -> dict[str, Any]:
+        return parse_model_submission_precheck(self.search_by_model_id(model_id, force_refresh=force_refresh))
+
     def get_verify_result_map(self, model_id: str) -> dict[str, Any]:
         payload = self.search_by_model_id(model_id)
         return ((payload.get("data") or {}).get("verifyResult") or {})
 
     def processed_gpus_for_model(self, model_id: str) -> set[str]:
-        return set(self.get_verify_result_map(model_id).keys())
+        return set(self.model_submission_precheck(model_id)["processedGpus"])
 
     def _reserve_account(self, excluded: set[int]) -> tuple[int, int] | None:
         with self._state_lock: