instance-repo 1.0.9.dev3__tar.gz → 1.0.9.dev5__tar.gz

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: instance-repo
-Version: 1.0.9.dev3
+Version: 1.0.9.dev5
 Summary: InstanceRepo SDK — 评测 Instance 统一存储客户端（SDK-only；控制面由 apiserver 承载）
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
@@ -149,10 +149,9 @@ keys = repo.list_user_data(uid)
 irepo --help
 # 全局参数：--api-env / --api-base / --cluster / --storage-env / --metadata-model / --profile
 # 无 --api-key：凭据只从 AP_API_KEY（旧别名 INSTANCEREPO_TOKEN）读取，避免进 shell history
-# 20 个子命令：validate / push / push-many / deliver / pull / list / get
+# 18 个子命令：validate / push / push-many / deliver / pull / list / get
 #              publish / grant / feedback / report / whoami / user-data
 #              repo-config / create / claim / version / image
-#              split（list/get/set-run-type）/ dataset（list/get/update/access/revoke/benchmarks/workflows）
 irepo image ref --dataset alibaba/mybench --instance-id inst-1 --split test
 irepo image pull '<acr>/<ns>/mybench:test-inst-1' ./out.tar
 ```

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/README.md

@@ -136,10 +136,9 @@ keys = repo.list_user_data(uid)
 irepo --help
 # 全局参数：--api-env / --api-base / --cluster / --storage-env / --metadata-model / --profile
 # 无 --api-key：凭据只从 AP_API_KEY（旧别名 INSTANCEREPO_TOKEN）读取，避免进 shell history
-# 20 个子命令：validate / push / push-many / deliver / pull / list / get
+# 18 个子命令：validate / push / push-many / deliver / pull / list / get
 #              publish / grant / feedback / report / whoami / user-data
 #              repo-config / create / claim / version / image
-#              split（list/get/set-run-type）/ dataset（list/get/update/access/revoke/benchmarks/workflows）
 irepo image ref --dataset alibaba/mybench --instance-id inst-1 --split test
 irepo image pull '<acr>/<ns>/mybench:test-inst-1' ./out.tar
 ```

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/__init__.py

@@ -18,7 +18,7 @@ from .errors import (
     TransportError,
 )
 
-__version__ = "1.0.9.dev3"
+__version__ = "1.0.9.dev5"
 
 __all__ = [
     "Repo", "load_instances",

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/_routing.py

@@ -45,6 +45,42 @@ def _split_scheme(endpoint: str) -> tuple[str, str]:
     return "", endpoint
 
 
+def _probe_private_endpoint(endpoint: str, timeout: float = 3.0) -> bool:
+    """Quick TCP connect probe to decide whether a private endpoint is reachable.
+
+    Performs a raw TCP connect (no TLS handshake) to the endpoint host/port.
+    Returns True if the connect succeeds, False on any error.  The default
+    timeout is 3 seconds — short enough to save 60+ seconds of waiting when
+    the private endpoint is unreachable, yet long enough for a normal TCP
+    handshake on a reachable host.
+    """
+    _, host = _split_scheme(endpoint)
+    # Determine port: default 443 for https, 80 for http.
+    scheme, _ = _split_scheme(endpoint)
+    default_port = 443 if scheme == "https://" else 80
+    # Parse host:port — handle IPv4 with optional port.
+    if ":" in host:
+        parts = host.rsplit(":", 1)
+        hostname = parts[0]
+        try:
+            port = int(parts[1])
+        except (ValueError, IndexError):
+            hostname = host
+            port = default_port
+    else:
+        hostname = host
+        port = default_port
+    # Strip any trailing path from hostname.
+    if "/" in hostname:
+        hostname = hostname.split("/", 1)[0]
+    try:
+        conn = socket.create_connection((hostname, port), timeout=timeout)
+        conn.close()
+        return True
+    except Exception:  # noqa: BLE001 - any failure means unreachable
+        return False
+
+
 def _plan_oss_route(endpoint: str) -> _EndpointRoute:
     value = (endpoint or "").strip()
     scheme, host = _split_scheme(value)

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/cli.py

@@ -870,17 +870,19 @@ def main(argv=None, *, repo: Repo | None = None, out=None) -> int:
             # 1.0.10：取版本完整详情（splits 明细 + manifest + 时间戳）。
             # split_first 分支时服务端聚合 legacy 投影。
             dv = r.versions.get(dataset, version, environment=args.environment)
-            # versions.get 返回 DatasetVersion dataclass（字段访问，非 dict 取值）。
-            print(f"version\t{dv.version}\tstatus={dv.status}"
-                  f"\tstorage_path={dv.storage_path or ''}", file=out)
-            print(f"published_at={dv.published_at or ''}"
-                  f"\tlast_ingested_at={dv.last_ingested_at or ''}", file=out)
-            print(f"dataset_version_id={dv.dataset_version_id or ''}", file=out)
-            for s in dv.splits:
-                print(f"  split\t{s.get('name', '')}\tstatus={s.get('status', '')}"
-                      f"\toss_prefix={s.get('oss_prefix', '')}"
+            print(f"version\t{dv.get('version','')}\tstatus={dv.get('status','')}"
+                  f"\tstorage_path={dv.get('storage_path','')}", file=out)
+            print(f"published_at={dv.get('published_at') or ''}"
+                  f"\tlast_ingested_at={dv.get('last_ingested_at') or ''}",
+                  file=out)
+            print(f"dataset_version_id={dv.get('dataset_version_id','')}",
+                  file=out)
+            splits = dv.get("splits") or []
+            for s in splits:
+                print(f"  split\t{s.get('name','')}\tstatus={s.get('status','')}"
+                      f"\toss_prefix={s.get('oss_prefix','')}"
                       f"\tinstance_count={s.get('instance_count') or 0}", file=out)
-            manifest = dv.manifest or {}
+            manifest = dv.get("manifest") or {}
             print(f"manifest\tinstance_count={manifest.get('instance_count') or 0}"
                   f"\tfile_count={manifest.get('file_count') or 0}"
                   f"\tsize_bytes={manifest.get('size_bytes') or 0}", file=out)

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/clients/splits.py

@@ -224,14 +224,11 @@ class SplitsClient:
 
     def update(self, dataset: str, split: str | None = None, *,
                version: str | None = None, run_type: str = "",
-               dataset_split_id: str | None = None,
-               environment: str | None = None) -> DatasetSplit:
+               dataset_split_id: str | None = None) -> DatasetSplit:
         """更新 split 记录（``PATCH /datasets/splits/detail``）。
 
         服务端 body 仅接受 ``run_type``（**必填**："train"/"eval"，空串=清除）；
         仅 online 环境（split 记录 pre 环境的写经 versions.update 走集合路由）。
-        ``environment`` 缺省由 profile 派生（同 get/list）；profile 派生为 pre
-        而目标 split 记录在 online 集合时，须显式传 ``environment="online"``。
 
         寻址二选一：``dataset_split_id``（优先，split-first 精确记录 id）或
         ``dataset``+``split``（+可选 ``version``）name 三元组。
@@ -251,8 +248,6 @@ class SplitsClient:
             raise SchemaError(
                 f"invalid run_type {rt!r}; must be one of "
                 f"{', '.join(RUN_TYPE_VALUES)} (empty string to clear)")
-        if (env := self._env(environment)):
-            q += f"&environment={quote(env, safe='')}"
         data = self._t.patch(f"{self._BASE}/detail?{q}", {"run_type": rt})
         return DatasetSplit.from_dict(data) if isinstance(data, dict) else DatasetSplit()
 

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/clients/versions.py

@@ -209,8 +209,9 @@ class VersionsClient:
                 out[iid] = img
         return out
 
-    def list(self, dataset: str) -> list[DatasetVersion]:
-        extra = self._version_extra_q()
+    def list(self, dataset: str, *, environment: str | None = None,
+             metadata_model: str | None = None) -> list[DatasetVersion]:
+        extra = self._version_extra_q(environment, metadata_model)
         data = self._t.get(
             f"{self._BASE}?dataset_name={quote(dataset, safe='')}{extra}")
         rows = data.get("versions", data) if isinstance(data, dict) else data
@@ -264,8 +265,10 @@ class VersionsClient:
         data = self._t.get(f"{self._BASE}/detail?{q}")
         return DatasetVersion.from_dict_full(data) if isinstance(data, dict) else DatasetVersion(version=version, dataset=dataset)
 
-    def status(self, dataset: str, version: str) -> str:
-        extra = self._version_extra_q()
+    def status(self, dataset: str, version: str, *,
+               environment: str | None = None,
+               metadata_model: str | None = None) -> str:
+        extra = self._version_extra_q(environment, metadata_model)
         data = self._t.get(
             f"{self._BASE}/detail?dataset_name={quote(dataset, safe='')}"
             f"&version={quote(version, safe='')}{extra}")
@@ -382,7 +385,9 @@ class VersionsClient:
     def update(self, dataset: str, version: str, *, status: str | None = None,
                splits: list[str] | None = None, run_type: str | None = None,
                split_run_types=None,
-               manifest: dict | None = None) -> None:
+               manifest: dict | None = None,
+               environment: str | None = None,
+               metadata_model: str | None = None) -> None:
         """通用版本更新（``PATCH /apis/v1/datasets/versions/detail``）。
 
         只发送非 None 字段。apiserver 侧 ``run_type`` / ``split_run_types`` / ``splits``
@@ -398,6 +403,8 @@ class VersionsClient:
           manifest        : 版本 manifest 整体替换（``{instance_count, file_count,
                             size_bytes, validation_status, validation_message}``
                             子集均可）——服务端支持但此前 SDK 未暴露（1.0.10 补）。
+          environment     : 显式 environment 覆盖 profile 派生值（如发布后操作 online）。
+          metadata_model  : 显式 metadata_model 覆盖 profile 派生值。
         """
         norm_srt = _normalize_split_run_types(split_run_types) if split_run_types else None
         # 三选一互斥（对齐 apiserver：run_type / split_run_types / splits 两两互斥）。
@@ -427,10 +434,13 @@ class VersionsClient:
             body["manifest"] = manifest
         if not body:
             raise SchemaError("update requires at least one field to change")
-        self._patch_detail(dataset, version, body)
+        self._patch_detail(dataset, version, body,
+                           environment=environment, metadata_model=metadata_model)
 
     def set_run_type(self, dataset: str, version: str, *,
-                     run_type: str | None = None, split_run_types=None) -> None:
+                     run_type: str | None = None, split_run_types=None,
+                     environment: str | None = None,
+                     metadata_model: str | None = None) -> None:
         """便捷入口：仅给 version/split 打 run_type 标签（train/eval）。
 
         version 级与 split 级二选一：传 ``run_type`` 打整个版本（无 logical split 的版本），
@@ -440,14 +450,20 @@ class VersionsClient:
             raise SchemaError(
                 "set_run_type requires run_type or split_run_types")
         self.update(dataset, version, run_type=run_type,
-                    split_run_types=split_run_types)
+                    split_run_types=split_run_types,
+                    environment=environment, metadata_model=metadata_model)
 
-    def deprecate(self, dataset: str, version: str) -> None:
+    def deprecate(self, dataset: str, version: str, *,
+                  environment: str | None = None,
+                  metadata_model: str | None = None) -> None:
         """把版本置为 deprecated（``update(status="deprecated")`` 的薄封装）。"""
-        self.update(dataset, version, status="deprecated")
+        self.update(dataset, version, status="deprecated",
+                    environment=environment, metadata_model=metadata_model)
 
-    def _patch_detail(self, dataset: str, version: str, body: dict) -> None:
-        extra = self._version_extra_q()
+    def _patch_detail(self, dataset: str, version: str, body: dict, *,
+                      environment: str | None = None,
+                      metadata_model: str | None = None) -> None:
+        extra = self._version_extra_q(environment, metadata_model)
         self._t.patch(
             f"{self._BASE}/detail?dataset_name={quote(dataset, safe='')}"
             f"&version={quote(version, safe='')}{extra}",

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/models.py

@@ -124,7 +124,7 @@ class DatasetSeries:
     available_envs: list[str] = field(default_factory=list)
     # 服务端 `DatasetSeriesStatus`（"" 正常 / "deprecated"）；列表项
     # `DatasetListItem.Status` 带 `omitempty`，正常态不出现。
-    status: str = ""
+    # 与 class 开头 status 字段合并到此位置。
 
     def validate(self) -> None:
         if self.visibility not in VISIBILITY_VALUES:
@@ -139,12 +139,11 @@ class DatasetSeries:
         """create 请求体：剔除服务端派生/服务端分配的字段。
 
         teacher_direction / teacher_type 由 apiserver 从 benchmark 派生（传了也被忽略），
-        dataset_id 由服务端分配；dataset_l1 / dataset_l2 由 dataset_name 推导。status
-        是服务端 DatasetSeriesStatus（deprecated 打标），create 不接受。都不发送，
+        dataset_id 由服务端分配；dataset_l1 / dataset_l2 由 dataset_name 推导。都不发送，
         避免让调用方误以为这些值能由客户端决定。
         """
         drop = {"teacher_direction", "teacher_type", "dataset_id",
-                "dataset_l1", "dataset_l2", "status"}
+                "dataset_l1", "dataset_l2"}
         return {k: v for k, v in self.to_dict().items() if k not in drop}
 
 

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/retry.py

@@ -22,6 +22,9 @@ _DEFAULT_SLEEP = time.sleep
 #: 可重试的 4xx：429=限流（退避后可成功）、408=请求超时（服务端主动断，重发安全）。
 RETRIABLE_4XX = frozenset({408, 429})
 
+#: Tengine 等反向代理偶发返回的重定向状态码；SDK 应视为瞬时故障并重试。
+REDIRECT_STATUSES = frozenset({301, 302, 307, 308})
+
 
 def _status_of(exc: BaseException) -> int | None:
     """尽力从异常里取 HTTP 状态码：TransportError.status / oss2.*.status / .code。"""
@@ -46,7 +49,8 @@ def is_transient(exc: BaseException) -> bool:
     status = _status_of(exc)
     if status is not None:
         return (status < 0 or 500 <= status <= 599
-                or status in RETRIABLE_4XX)
+                or status in RETRIABLE_4XX
+                or status in REDIRECT_STATUSES)
     return False
 
 

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/store/oss.py

@@ -15,6 +15,7 @@ from .._routing import (
     _OperationIntent,
     _is_reachability_error,
     _plan_oss_route,
+    _probe_private_endpoint,
 )
 from ..content import sha256_file, verify_file_digest
 from ..errors import DigestMismatch, PrivateNetworkRequired, TransportError
@@ -157,6 +158,9 @@ class OssContentStore(BaseContentStore):
         self._retry_attempts = retry_attempts
         self._cache_buckets = cache_buckets
         self._local = threading.local()
+        # Injectable TCP probe for private endpoint reachability.  Defaults to
+        # the real raw-TCP-connect probe; tests may replace it with a stub.
+        self._probe_private_endpoint = _probe_private_endpoint
 
     @property
     def bucket_name(self) -> str:
@@ -251,6 +255,25 @@ class OssContentStore(BaseContentStore):
 
         private_endpoint = route.private_endpoint
         private_failure: BaseException | None = None
+
+        # Quick TCP probe to decide whether to attempt the private endpoint.
+        # If the route is standard (has a known public fallback) and the probe
+        # fails, skip the private attempt entirely to save 60+ seconds of
+        # waiting for a full timeout.  Downloads are NOT affected — they
+        # always try private only (no public fallback).
+        if route.standard and route.public_endpoint is not None:
+            if not self._probe_private_endpoint(private_endpoint):
+                private_failure = ConnectionRefusedError(
+                    f"TCP probe failed for {private_endpoint}")
+                try:
+                    bucket = self._open_bucket(
+                        sts, prefix, route.public_endpoint,
+                        private_download=False)
+                    return self._do(lambda: operation(bucket))
+                except Exception as public_exc:  # noqa: BLE001
+                    public_exc.private_route_failure = private_failure
+                    raise
+
         try:
             bucket = self._open_bucket(
                 sts, prefix, private_endpoint, private_download=False)

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo/transport.py

@@ -196,7 +196,7 @@ class Transport:
             raise SchemaError(
                 "apiserver 根地址未配置；请通过 AP_BASE_URL 提供"
                 "（旧别名 INSTANCEREPO_API_BASE 仍兼容；SDK 不内置任何域名，见部署文档）")
-        retry = (method == "GET") if idempotent is None else idempotent
+        retry = True if idempotent is None else idempotent
         if not retry:
             return self._request_once(method, path, body)
         return retry_transient(

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/instance_repo.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: instance-repo
-Version: 1.0.9.dev3
+Version: 1.0.9.dev5
 Summary: InstanceRepo SDK — 评测 Instance 统一存储客户端（SDK-only；控制面由 apiserver 承载）
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
@@ -149,10 +149,9 @@ keys = repo.list_user_data(uid)
 irepo --help
 # 全局参数：--api-env / --api-base / --cluster / --storage-env / --metadata-model / --profile
 # 无 --api-key：凭据只从 AP_API_KEY（旧别名 INSTANCEREPO_TOKEN）读取，避免进 shell history
-# 20 个子命令：validate / push / push-many / deliver / pull / list / get
+# 18 个子命令：validate / push / push-many / deliver / pull / list / get
 #              publish / grant / feedback / report / whoami / user-data
 #              repo-config / create / claim / version / image
-#              split（list/get/set-run-type）/ dataset（list/get/update/access/revoke/benchmarks/workflows）
 irepo image ref --dataset alibaba/mybench --instance-id inst-1 --split test
 irepo image pull '<acr>/<ns>/mybench:test-inst-1' ./out.tar
 ```

{instance_repo-1.0.9.dev3 → instance_repo-1.0.9.dev5}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "instance-repo"
-version = "1.0.9.dev3"
+version = "1.0.9.dev5"
 description = "InstanceRepo SDK — 评测 Instance 统一存储客户端（SDK-only；控制面由 apiserver 承载）"
 requires-python = ">=3.10"
 readme = "README.md"