instance-repo 1.3.2.dev0__py3-none-any.whl → 1.3.2.dev1__py3-none-any.whl

instance_repo/__init__.py

@@ -18,7 +18,7 @@ from .errors import (
     TransportError,
 )
 
-__version__ = "1.3.2.dev0"
+__version__ = "1.3.2.dev1"
 
 __all__ = [
     "Repo", "load_instances",

instance_repo/clients/splits.py

@@ -261,29 +261,26 @@ class SplitsClient:
                          page: int, page_size: int) -> PagedResult:
         """OSS 回退的 splits.list — 处理 version_scope 和 status 过滤。
 
-        version_scope="unversioned" → OSS 回退天然有版本结构，返回空。
-        version_scope="exact" → 只返回精确匹配的 version。
-        version_scope="all"（或空）+ version 非空 → 精确匹配 version。
+        注意：回退场景（dataset 未注册）下 ``version_scope`` 是 MongoDB 语义，
+        OSS 探测无从区分 versioned/unversioned——**忽略 scope 并返回全部探测结果**
+        （G4：否则 ap CLI 无 --version 的 split list 恒空）。``version`` 非空时
+        仍精确匹配该 version。
         """
         from ..store.oss_discovery import (
             parse_dataset_level, build_synthetic_split,
-            _in_memory_paged, _empty_paged_result, _warn_if_truncated,
+            _in_memory_paged, _warn_if_truncated,
         )
-        if (version_scope or "").strip() == "unversioned":
-            return _empty_paged_result(int(page), int(page_size))
         info = parse_dataset_level(
             self._content, dataset, oss_prefix=self._oss_prefix)
         _warn_if_truncated(info, dataset, 500)
         splits: list[DatasetSplit] = []
         for v in info.versions:
-            if version and (version_scope or "all") != "all" and v.version != version:
-                continue
-            elif version and v.version != version:
+            if version and v.version != version:
                 continue
             if v.has_default_instances:
                 splits.append(build_synthetic_split(
                     dataset, v.version, "default",
-                    instance_count=v.instance_count))
+                    instance_count=v.default_instance_count))
             for s in v.splits:
                 splits.append(build_synthetic_split(
                     dataset, v.version, s))
@@ -306,7 +303,7 @@ class SplitsClient:
             if split_name == "default" and v.has_default_instances:
                 return build_synthetic_split(
                     dataset, v.version, "default",
-                    instance_count=v.instance_count)
+                    instance_count=v.default_instance_count)
             if split_name in v.splits:
                 return build_synthetic_split(
                     dataset, v.version, split_name)

instance_repo/clients/versions.py

@@ -235,11 +235,15 @@ class VersionsClient:
     def list_paged(self, dataset: str, *, status: str | None = None,
                    environment: str | None = None,
                    metadata_model: str | None = None,
-                   page: int = 1, page_size: int = 20) -> "PagedResult":
+                   page: int = 1, page_size: int = 20,
+                   fallback_to_oss: bool = True) -> "PagedResult":
         """版本分页列表（单页，``GET /datasets/versions``）。
 
         服务端参数集仅 status + environment/metadata_model + 分页（**无 q 模糊
         搜索**——找数据集用 datasets.list_paged(q=)，两段式检索）。
+
+        ``fallback_to_oss``: dataset 未注册（92001）时自动回退到 OSS 目录探测
+        （与 :meth:`list` 一致），在内存中按 status 过滤 + 分页。
         """
         from ..models import PagedResult
         if not (dataset or "").strip():
@@ -254,7 +258,14 @@ class VersionsClient:
             q += f"&status={quote(s, safe='')}"
         q += (f"&page={int(page)}&page_size={int(page_size)}"
               f"{self._version_extra_q(environment, metadata_model)}")
-        data, pagination = self._t.request_with_meta("GET", f"{self._BASE}?{q}")
+        try:
+            data, pagination = self._t.request_with_meta("GET", f"{self._BASE}?{q}")
+        except TransportError as e:
+            if (e.biz_code == 92001 and fallback_to_oss
+                    and self._content is not None):
+                return self._discover_versions_paged(
+                    dataset, status, int(page), int(page_size))
+            raise
         rows = data.get("versions", data) if isinstance(data, dict) else data
         rows = rows if isinstance(rows, list) else []
         items = [DatasetVersion(**_pick(r)) for r in rows if isinstance(r, dict)]
@@ -891,6 +902,15 @@ class VersionsClient:
             raise DatasetNotFound(f"dataset {dataset!r} not found in OSS")
         return [build_synthetic_version(dataset, v) for v in info.versions]
 
+    def _discover_versions_paged(self, dataset: str, status: str | None,
+                                 page: int, page_size: int):
+        """OSS 回退的 list_paged：全量探测后按 status 过滤 + 内存分页。"""
+        from ..store.oss_discovery import _in_memory_paged
+        items = self._discover_versions(dataset)
+        if (status or "").strip():
+            items = [v for v in items if v.status == status.strip()]
+        return _in_memory_paged(items, int(page), int(page_size))
+
     def _discover_version(self, dataset: str, version: str):
         from ..store.oss_discovery import (
             parse_dataset_level, build_synthetic_version, _warn_if_truncated)

instance_repo/models.py

@@ -21,7 +21,9 @@ RUN_TYPE_VALUES = ("train", "eval")
 STORAGE_TYPE_VALUES = ("oss",)
 # permission 是签发 OSS 凭据时的**权限封顶**（apiserver credential_handler 的 permission
 # 字段）：只能把本次令牌收窄到不高于调用方真实角色，绝不提权。ACR 分支不支持该字段。
-PERMISSION_VALUES = ("admin", "writer", "reader")
+# "list" 是 list-only 模式（OSS 回退发现专用）：仅授 ListObjects + GetBucketInfo，
+# 跳过 DatasetRBAC 检查（未注册 dataset 无 RBAC 记录），无任何对象级读写动作。
+PERMISSION_VALUES = ("admin", "writer", "reader", "list")
 
 
 def canonical_json(obj: Any) -> str:

instance_repo/store/oss_discovery.py

@@ -59,6 +59,8 @@ class OssVersionInfo:
     splits: list[str]
     has_default_instances: bool
     instance_count: int
+    # default split（3 层布局）的直接 .json 数，与 instance_count（含命名 split）区分
+    default_instance_count: int = 0
 
 
 @dataclass
@@ -279,9 +281,11 @@ def parse_dataset_level(
                 version_dir, permission="list")
             v_json_keys = _direct_json_keys(v_all_keys, version_dir)
             has_default_instances = len(v_json_keys) > 0
+            default_instance_count = len(v_json_keys)
             instance_count = len(v_json_keys)
         except Exception:  # noqa: BLE001
             has_default_instances = False
+            default_instance_count = 0
             instance_count = 0
 
         # Count instances in each named split directory.
@@ -303,6 +307,7 @@ def parse_dataset_level(
             splits=splits,
             has_default_instances=has_default_instances,
             instance_count=instance_count,
+            default_instance_count=default_instance_count,
         ))
 
     return OssDatasetLevelInfo(
@@ -528,12 +533,15 @@ def build_synthetic_instance(
     key = P.metadata_key(
         dataset, version, instance_id, split, prefix=oss_prefix)
 
+    # 存在性判定用 LIST（permission="list" 仅需 ListObjects）而非 HEAD——
+    # list-only STS 无 oss:GetObject，object_exists 的 HEAD 会在 STS 签发或
+    # OSS 侧被拒，导致已存在的实例恒被误判为不存在。
     try:
-        exists = content_store.object_exists(key)
+        keys = content_store.list_objects(key, permission="list")
     except Exception:  # noqa: BLE001
         return None
 
-    if not exists:
+    if key not in keys:
         return None
 
     inst = DatasetInstance(

instance_repo/transport.py

@@ -333,11 +333,13 @@ class Transport:
         降到 ~1 次。dataset_id / dataset_name 至少传一个（细粒度授权场景），二者可并存，
         服务端校验一致性。
 
-        ``permission``（可选，admin/writer/reader）是**权限封顶**：apiserver 取
+        ``permission``（可选，admin/writer/reader/list）是**权限封顶**：apiserver 取
         ``min(调用方真实角色, permission)``，**只能收窄不能提权**——传 admin 不会把
         reader 变成 admin。用途是最小权限原则（只读任务显式请求 reader，即便自己是
         admin），以及避免把可写令牌交给下游只读组件。admin 与 writer 在 STS 策略层等价，
-        真正有区别的是 reader（只给 GetObject）。ACR 分支不支持该字段（镜像令牌是实例级的）。
+        真正有区别的是 reader（只给 GetObject）。``list`` 是 list-only 模式（OSS 回退
+        发现专用）：仅授 ListObjects + GetBucketInfo，跳过 DatasetRBAC 检查，用于未
+        注册 dataset 的目录结构探测。ACR 分支不支持该字段（镜像令牌是实例级的）。
         """
         # 空串与 None 一律视为"不封顶"（与 Go 的 Permission == "" 语义对齐）。
         if permission and permission not in PERMISSION_VALUES:

{instance_repo-1.3.2.dev0.dist-info → instance_repo-1.3.2.dev1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: instance-repo
-Version: 1.3.2.dev0
+Version: 1.3.2.dev1
 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

{instance_repo-1.3.2.dev0.dist-info → instance_repo-1.3.2.dev1.dist-info}/RECORD

@@ -1,4 +1,4 @@
-instance_repo/__init__.py,sha256=gbklD8A9cq8YL2H0n3_MB0XyZJ5FSrIC7c6mfbNAYA0,1422
+instance_repo/__init__.py,sha256=8ybL0ZVRO_IDt_dQeYOOUIvnJMx0OEPjNrT0Z_v38n4,1422
 instance_repo/_routing.py,sha256=XbdKPF-Zh2sEkr4UyLC9HuiRcTLP-ssvhp6SCDC98YM,8326
 instance_repo/_site_defaults.py,sha256=ck7aHHTup8ibBo0l5MSR03XpTtxJ11hYo7oM2DzWs4A,2383
 instance_repo/artifacts.py,sha256=OOcSI1ehB_dw1ZNScjyhzghPqOdL534kQp_iI2Hz7IU,10659
@@ -12,13 +12,13 @@ instance_repo/errors.py,sha256=2W6VS1fpkfkBsKPaNKCq1iJenAg60_G2BPjzWt0Ta3Y,5621
 instance_repo/image.py,sha256=tUTRyeNNiAzd2ULbmjhNj436558j_IPwpTDXTld1KVY,3822
 instance_repo/layout.py,sha256=8o4dVD4-76hTAvRSp7UZFNvXG5AxzEqMCVgrVXNfdok,11943
 instance_repo/loader.py,sha256=lWjhw0MjEIIxbZ5w6cjyaDIWkNVFDhjjNdu1lc5zRuo,474
-instance_repo/models.py,sha256=bPKh3EEgpF-mqZruuvK5fIT9qWU5E7XxrePY8aKAgL4,35902
+instance_repo/models.py,sha256=OHBslrA2c5-c-QR_-008hk1ryxQKezNlCZG7JWt_f1E,36111
 instance_repo/paths.py,sha256=EL0d613T0mnabiYCtZQBC50rZT3mWKUq-3TepPbTRec,8438
 instance_repo/release.py,sha256=Up96Qeh2GRH26BSAHVEjDFQnFCaEA-vIvSme3a8FQFo,13882
 instance_repo/repo.py,sha256=5D7jYLeaDSZLKnf548YlfXT9tUPNgdcXT2j31wzShpA,40305
 instance_repo/retry.py,sha256=8y-AX9qmnLqGwqRzVJ5PKqZt2fWyqUL2t2P0suSBAA4,3438
 instance_repo/tasktoml.py,sha256=b-tWfNiOHN9D-ixCbPpCLnZaaU9Q6LRGG0Wl4YN8r10,5152
-instance_repo/transport.py,sha256=Hhw89xxQ7YCJaxmr35he0t9IALeYGvIqcx1jUwzKSzo,21420
+instance_repo/transport.py,sha256=B5ZxixaHuqyi9BzjTRCKlBdkrrG0Dft6O2FhL7inceY,21616
 instance_repo/validate.py,sha256=NMR-vFBIU_FQ0WKdBZuQ-ua4jtkAzvWwEjnAHyAFwDs,11877
 instance_repo/clients/__init__.py,sha256=5V7Pe0b6nnMiwIhACA4ctQowovrL4iHTiHKKcjmsBQo,488
 instance_repo/clients/benchmarks.py,sha256=RzHY8VzDP1o2jT4M8gxxbcr3amuNr4mlSblZhhJqG_w,2778
@@ -28,17 +28,17 @@ instance_repo/clients/images.py,sha256=8i0FXxAAL2mOUUTrVgDdWSYM2imJ5nOEXzpVRBa-G
 instance_repo/clients/instances.py,sha256=IOAKl3MAlW3UcftX8b5cGIEdC5_1qfouaVSE_0-OAxE,58125
 instance_repo/clients/reports.py,sha256=4Z8bpcSi-mmfkQLKpMklelLI4mgHcnjfb0725bE2nkI,3001
 instance_repo/clients/scaffold.py,sha256=VLicNFnqKi2iYB-gjOkQ9wa1unNOJXO7UxBZlz7El5A,4998
-instance_repo/clients/splits.py,sha256=h4WBveww1z-a7A8W3FJa6CBGfuH2M5eLVmviu3okbsU,27366
-instance_repo/clients/versions.py,sha256=5aJOvONQyn_eP0D2xrEv0p4IgFSjJqGP_2SjPTIveF0,51631
+instance_repo/clients/splits.py,sha256=9hUZAiLCv0gZpxbEPvxtsJZ0qeGxSxaYl21ARF3sl58,27211
+instance_repo/clients/versions.py,sha256=jeLwV0K5h4lpeeEMPQMnv-BpzfRcm0jiqcGIa0Nl2RI,52658
 instance_repo/clients/workflows.py,sha256=6ftl1tuyEGx8BtMLSjfExP9_VcQhK4veo7ex1jDOdlc,3686
 instance_repo/store/__init__.py,sha256=fGwePNjKuojqx1yagSaY-svONE5pjaEz2BZmG0ISUZ0,674
 instance_repo/store/acr.py,sha256=OD7LH3GZY4HrrbEht86Mf3NfW_lh-8MhCTOGyX6gLT4,19468
 instance_repo/store/base.py,sha256=7feHdDXDvSkq60wXYvF6p8TK9KRj7YTLlESdgZ5MQHo,2622
 instance_repo/store/oss.py,sha256=4jHGPO7ceTx7xC6MqzNK26eHRTmvIe_rLrV36oPDfkk,30718
-instance_repo/store/oss_discovery.py,sha256=S9V1FhzxALkSVBbvS5eXu2HbjmyHHSeW-1X8Xr4QrWI,19472
+instance_repo/store/oss_discovery.py,sha256=2nwJwtdWO6Pa5FK0MfOFZlFhSav8jzFrh7uNHA_oLqg,20032
 instance_repo/store/registry.py,sha256=1R2GqXsuDZ8wc3IGpncF9yL6IHk-FeiwUlZN7v2ndSc,10837
-instance_repo-1.3.2.dev0.dist-info/METADATA,sha256=c0EhI853YMKbYrN6Qh-zdHsie_R-PC6feSmw5kRLpyQ,14227
-instance_repo-1.3.2.dev0.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
-instance_repo-1.3.2.dev0.dist-info/entry_points.txt,sha256=tn6jgwYChSx7eAqw9YNed_JFYYgoRtE7WkFn8xqa8Rw,53
-instance_repo-1.3.2.dev0.dist-info/top_level.txt,sha256=mblK0Bkwwp6jCZioS6q3sSuFFs8NR9DKTJ0DJpIRDmY,14
-instance_repo-1.3.2.dev0.dist-info/RECORD,,
+instance_repo-1.3.2.dev1.dist-info/METADATA,sha256=gFmfOf-KacPoB2hUdJeAqRfvNOfR6T4Mu61IEMsNf7s,14227
+instance_repo-1.3.2.dev1.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
+instance_repo-1.3.2.dev1.dist-info/entry_points.txt,sha256=tn6jgwYChSx7eAqw9YNed_JFYYgoRtE7WkFn8xqa8Rw,53
+instance_repo-1.3.2.dev1.dist-info/top_level.txt,sha256=mblK0Bkwwp6jCZioS6q3sSuFFs8NR9DKTJ0DJpIRDmY,14
+instance_repo-1.3.2.dev1.dist-info/RECORD,,