change: add image_uri_config for vLLM

sagemaker-core/src/sagemaker/core/image_uri_config/sglang.json

@@ -0,0 +1,62 @@
+{
+    "inference": {
+        "processors": [
+            "gpu"
+        ],
+        "version_aliases": {
+            "0.5": "0.5.18"
+        },
+        "versions": {
+            "0.5.18": {
+                "py_versions": [
+                    "py312"
+                ],
+                "registries": {
+                    "af-south-1": "626614931356",
+                    "ap-east-1": "871362719292",
+                    "ap-east-2": "975050140332",
+                    "ap-northeast-1": "763104351884",
+                    "ap-northeast-2": "763104351884",
+                    "ap-northeast-3": "364406365360",
+                    "ap-south-1": "763104351884",
+                    "ap-south-2": "772153158452",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ap-southeast-5": "550225433462",
+                    "ap-southeast-6": "633930458069",
+                    "ap-southeast-7": "590183813437",
+                    "ca-central-1": "763104351884",
+                    "ca-west-1": "204538143572",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-central-2": "380420809688",
+                    "eu-north-1": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "eu-south-2": "503227376785",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "il-central-1": "780543022126",
+                    "me-central-1": "914824155844",
+                    "me-south-1": "217643126080",
+                    "mx-central-1": "637423239942",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "tag_prefix": "0.5.18",
+                "repository": "sglang",
+                "container_version": {
+                    "gpu": "cu130-ubuntu24.04-sagemaker"
+                }
+            }
+        }
+    }
+}

sagemaker-core/src/sagemaker/core/image_uri_config/vllm.json

@@ -0,0 +1,62 @@
+{
+    "inference": {
+        "processors": [
+            "gpu"
+        ],
+        "version_aliases": {
+            "0.28": "0.28.0"
+        },
+        "versions": {
+            "0.28.0": {
+                "py_versions": [
+                    "py312"
+                ],
+                "registries": {
+                    "af-south-1": "626614931356",
+                    "ap-east-1": "871362719292",
+                    "ap-east-2": "975050140332",
+                    "ap-northeast-1": "763104351884",
+                    "ap-northeast-2": "763104351884",
+                    "ap-northeast-3": "364406365360",
+                    "ap-south-1": "763104351884",
+                    "ap-south-2": "772153158452",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ap-southeast-5": "550225433462",
+                    "ap-southeast-6": "633930458069",
+                    "ap-southeast-7": "590183813437",
+                    "ca-central-1": "763104351884",
+                    "ca-west-1": "204538143572",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-central-2": "380420809688",
+                    "eu-north-1": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "eu-south-2": "503227376785",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "il-central-1": "780543022126",
+                    "me-central-1": "914824155844",
+                    "me-south-1": "217643126080",
+                    "mx-central-1": "637423239942",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "tag_prefix": "0.28.0",
+                "repository": "vllm",
+                "container_version": {
+                    "gpu": "cu130-ubuntu24.04-sagemaker"
+                }
+            }
+        }
+    }
+}

sagemaker-core/tests/unit/image_uris/test_sglang.py

@@ -0,0 +1,94 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+from __future__ import absolute_import
+
+import pytest
+
+from sagemaker.core import image_uris
+from sagemaker.core.common_utils import ALTERNATE_DOMAINS
+
+# SGLang images are GPU-only; a GPU instance type selects the "gpu" processor.
+INSTANCE_TYPE = "ml.g5.2xlarge"
+DEFAULT_DOMAIN = "amazonaws.com"
+
+# Regions whose ECR host suffix is stable across botocore versions (commercial,
+# China, GovCloud). Exact-host assertions are limited to these; every other region
+# is still covered by the account/region/repository/tag checks in test_sglang_uris.
+FULL_URI_REGIONS = ["us-east-1", "us-west-2", "eu-west-1", "cn-north-1", "us-gov-west-1"]
+
+
+@pytest.mark.parametrize("load_config", ["sglang.json"], indirect=True)
+def test_sglang_uris(load_config):
+    """Every (version, region) resolves to the sglang repo with the expected account + tag."""
+    config = load_config
+    assert config["inference"]["processors"] == ["gpu"]
+    versions = config["inference"]["versions"]
+    for version, version_config in versions.items():
+        py_version = version_config["py_versions"][0]
+        container_version = version_config["container_version"]["gpu"]
+        expected_tag = f"{version}-gpu-{py_version}-{container_version}"
+        for region, account in version_config["registries"].items():
+            uri = image_uris.retrieve(
+                framework="sglang",
+                region=region,
+                version=version,
+                image_scope="inference",
+                instance_type=INSTANCE_TYPE,
+            )
+            # account (registry), region, repository and tag are config-controlled;
+            # the domain suffix is resolved by botocore and asserted separately below.
+            assert uri.startswith(f"{account}.dkr.ecr.{region}."), uri
+            assert uri.endswith(f"/sglang:{expected_tag}"), uri
+
+
+@pytest.mark.parametrize("load_config", ["sglang.json"], indirect=True)
+def test_sglang_full_uri_for_representative_regions(load_config):
+    """Exact URI (including domain) for representative commercial/China/GovCloud regions."""
+    config = load_config
+    versions = config["inference"]["versions"]
+    for version, version_config in versions.items():
+        py_version = version_config["py_versions"][0]
+        container_version = version_config["container_version"]["gpu"]
+        for region in FULL_URI_REGIONS:
+            if region not in version_config["registries"]:
+                continue
+            account = version_config["registries"][region]
+            domain = ALTERNATE_DOMAINS.get(region, DEFAULT_DOMAIN)
+            expected = (
+                f"{account}.dkr.ecr.{region}.{domain}"
+                f"/sglang:{version}-gpu-{py_version}-{container_version}"
+            )
+            uri = image_uris.retrieve(
+                framework="sglang",
+                region=region,
+                version=version,
+                image_scope="inference",
+                instance_type=INSTANCE_TYPE,
+            )
+            assert uri == expected
+
+
+@pytest.mark.parametrize("load_config", ["sglang.json"], indirect=True)
+def test_sglang_version_aliases_resolve_to_newest_patch(load_config):
+    """Each minor alias resolves to its newest patch version."""
+    config = load_config
+    aliases = config["inference"]["version_aliases"]
+    for alias, target_version in aliases.items():
+        uri = image_uris.retrieve(
+            framework="sglang",
+            region="us-west-2",
+            version=alias,
+            image_scope="inference",
+            instance_type=INSTANCE_TYPE,
+        )
+        assert f"/sglang:{target_version}-gpu-" in uri, uri

sagemaker-core/tests/unit/image_uris/test_vllm.py

@@ -0,0 +1,95 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+#     http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+from __future__ import absolute_import
+
+import pytest
+
+from sagemaker.core import image_uris
+from sagemaker.core.common_utils import ALTERNATE_DOMAINS
+
+# vLLM images are GPU-only; a GPU instance type selects the "gpu" processor.
+INSTANCE_TYPE = "ml.g5.2xlarge"
+DEFAULT_DOMAIN = "amazonaws.com"
+
+# Regions whose ECR host suffix is stable across botocore versions (commercial,
+# China, GovCloud). Exact-host assertions are limited to these; every other region
+# is still covered by the account/region/repository/tag checks in test_vllm_uris,
+# which avoids depending on botocore endpoint data for newer ISO partitions.
+FULL_URI_REGIONS = ["us-east-1", "us-west-2", "eu-west-1", "cn-north-1", "us-gov-west-1"]
+
+
+@pytest.mark.parametrize("load_config", ["vllm.json"], indirect=True)
+def test_vllm_uris(load_config):
+    """Every (version, region) resolves to the vllm repo with the expected account + tag."""
+    config = load_config
+    assert config["inference"]["processors"] == ["gpu"]
+    versions = config["inference"]["versions"]
+    for version, version_config in versions.items():
+        py_version = version_config["py_versions"][0]
+        container_version = version_config["container_version"]["gpu"]
+        expected_tag = f"{version}-gpu-{py_version}-{container_version}"
+        for region, account in version_config["registries"].items():
+            uri = image_uris.retrieve(
+                framework="vllm",
+                region=region,
+                version=version,
+                image_scope="inference",
+                instance_type=INSTANCE_TYPE,
+            )
+            # account (registry), region, repository and tag are config-controlled;
+            # the domain suffix is resolved by botocore and asserted separately below.
+            assert uri.startswith(f"{account}.dkr.ecr.{region}."), uri
+            assert uri.endswith(f"/vllm:{expected_tag}"), uri
+
+
+@pytest.mark.parametrize("load_config", ["vllm.json"], indirect=True)
+def test_vllm_full_uri_for_representative_regions(load_config):
+    """Exact URI (including domain) for representative commercial/China/GovCloud regions."""
+    config = load_config
+    versions = config["inference"]["versions"]
+    for version, version_config in versions.items():
+        py_version = version_config["py_versions"][0]
+        container_version = version_config["container_version"]["gpu"]
+        for region in FULL_URI_REGIONS:
+            if region not in version_config["registries"]:
+                continue
+            account = version_config["registries"][region]
+            domain = ALTERNATE_DOMAINS.get(region, DEFAULT_DOMAIN)
+            expected = (
+                f"{account}.dkr.ecr.{region}.{domain}"
+                f"/vllm:{version}-gpu-{py_version}-{container_version}"
+            )
+            uri = image_uris.retrieve(
+                framework="vllm",
+                region=region,
+                version=version,
+                image_scope="inference",
+                instance_type=INSTANCE_TYPE,
+            )
+            assert uri == expected
+
+
+@pytest.mark.parametrize("load_config", ["vllm.json"], indirect=True)
+def test_vllm_version_aliases_resolve_to_newest_patch(load_config):
+    """Each minor alias resolves to its newest patch version."""
+    config = load_config
+    aliases = config["inference"]["version_aliases"]
+    for alias, target_version in aliases.items():
+        uri = image_uris.retrieve(
+            framework="vllm",
+            region="us-west-2",
+            version=alias,
+            image_scope="inference",
+            instance_type=INSTANCE_TYPE,
+        )
+        assert f"/vllm:{target_version}-gpu-" in uri, uri