fix: escape registry credentials and validate REST options

Author: KAJdev

README.md

@@ -297,7 +297,7 @@ pod = runpod.create_pod("test", "runpod/stack", instance_id="cpu3c-2-4")
 runpod.stop_pod(pod["id"])
 
 # resume the pod
-runpod.resume_pod(pod["id"], 1)
+runpod.resume_pod(pod["id"])
 
 # terminate the pod
 runpod.terminate_pod(pod["id"])
@@ -314,6 +314,8 @@ runpod.terminate_pod(pod["id"])
   requires `network_volume_id`.
 - For GPU pods, `min_memory_in_gb` and `min_vcpu_count` specify minimum host RAM
   and vCPUs **per GPU**, not GPU VRAM or totals for the pod.
+- CPU `instance_id` must use `<cpu-flavor>-<vcpu-count>-<memory>`, with positive
+  integer vCPU and memory values, for example `cpu3c-4-8`.
 - REST v2 cannot require a public-IP-capable host. `support_public_ip=False` is
   the default and does not disable public networking. `True` raises `ValueError`
   rather than silently ignoring that requirement.
@@ -326,6 +328,8 @@ runpod.terminate_pod(pod["id"])
 - `create_endpoint` accepts only `QUEUE_DELAY` and `REQUEST_COUNT` scaling.
   `idle_timeout` applies to `QUEUE_DELAY` (default: 5 seconds); explicitly setting
   it with `REQUEST_COUNT` raises `ValueError`.
+- `resume_pod(pod_id)` keeps the existing GPU allocation. REST v2 does not support
+  resizing on resume, so providing `gpu_count` raises `ValueError`.
 
 ## 📁 | Directory
 

docs/api/queries.md

@@ -14,7 +14,7 @@ for gpu in gpus:
 ### get_gpus output
 
 ```python
-{
+[{
     "id": "NVIDIA GeForce RTX 4090",
     "name": "RTX 4090",
     "pool": "ADA_24",
@@ -24,7 +24,7 @@ for gpu in gpus:
     "community": True,
     "price": {"secure": 0.44, "community": 0.31, "serverless": 1.1},
     "maxCount": {"secure": 8, "community": 4},
-}
+}]
 ```
 
 ## get_gpu

examples/rest_wrapper.py

@@ -28,7 +28,7 @@
 print("Waiting for pod to be stopped...")
 time.sleep(10)
 
-pod = runpod.resume_pod(pod["id"], 1)
+pod = runpod.resume_pod(pod["id"])
 print(pod)
 
 print("Waiting for pod to be resumed...")

runpod/api/ctl_commands.py

@@ -2,6 +2,7 @@
 
 # pylint: disable=too-many-arguments,too-many-locals
 
+import re
 from typing import Any, Iterable, Optional
 from urllib.parse import quote
 
@@ -31,21 +32,17 @@ def _environment(env: Optional[dict]) -> dict[str, str]:
     return {str(key): str(value) for key, value in (env or {}).items()}
 
 
-def _cpu_config(instance_id: Optional[str], min_vcpu_count: int) -> dict[str, Any]:
+def _cpu_config(instance_id: Optional[str]) -> dict[str, Any]:
     if not instance_id:
         raise ValueError("instance_id must be provided for CPU pods")
 
-    parts = instance_id.split("-")
-    vcpu_count = max(2, min_vcpu_count)
-    if len(parts) > 1:
-        try:
-            vcpu_count = int(parts[1])
-        except ValueError as exc:
-            raise ValueError(
-                "instance_id must use the format <cpu-flavor>-<vcpu-count>-<memory>"
-            ) from exc
+    match = re.fullmatch(r"([^-\s]+)-([1-9][0-9]*)-([1-9][0-9]*)", instance_id)
+    if match is None:
+        raise ValueError(
+            "instance_id must use the format <cpu-flavor>-<vcpu-count>-<memory>"
+        )
 
-    return {"id": parts[0], "vcpuCount": vcpu_count}
+    return {"id": match[1], "vcpuCount": int(match[2])}
 
 
 def get_user(api_key: Optional[str] = None) -> dict:
@@ -212,7 +209,7 @@ def create_pod(
             gpu["allowedCudaVersions"] = _split_values(allowed_cuda_versions)
         body["gpu"] = gpu
     else:
-        body["cpu"] = _cpu_config(instance_id, min_vcpu_count)
+        body["cpu"] = _cpu_config(instance_id)
 
     return run_rest_request("POST", "/v2/pods", json=body)
 
@@ -226,9 +223,10 @@ def stop_pod(pod_id: str) -> dict:
     )
 
 
-def resume_pod(pod_id: str, gpu_count: int) -> dict:
-    """Start a stopped pod."""
-    _ = gpu_count
+def resume_pod(pod_id: str, gpu_count: Optional[int] = None) -> dict:
+    """Start a stopped pod without changing its GPU allocation."""
+    if gpu_count is not None:
+        raise ValueError("REST API v2 does not support gpu_count when resuming a pod")
     return run_rest_request(
         "POST",
         f"/v2/pods/{_path_segment(pod_id)}/action",

runpod/api/mutations/container_register_auth.py

@@ -1,5 +1,7 @@
 """GraphQL operations for container registry credentials."""
 
+import json
+
 
 def update_container_registry_auth(
     registry_auth_id: str, username: str, password: str
@@ -11,7 +13,7 @@ def update_container_registry_auth(
         "password": password,
     }
     input_str = ", ".join(
-        f'{key}: "{value}"' for key, value in input_dict.items()
+        f"{key}: {json.dumps(value)}" for key, value in input_dict.items()
     )
 
     return f"""

tests/test_api/test_ctl_commands.py

@@ -264,26 +264,17 @@ def allocate(method, path, *, json):
     assert pod["machineId"] == "at-boundary"
 
 
-def test_create_cpu_pod_translates_instance_id():
-    with patch(
-        "runpod.api.ctl_commands.run_rest_request", return_value={"id": "pod"}
-    ) as request:
-        ctl_commands.create_pod("cpu-pod", "python:3.11", instance_id="cpu3c-4-8")
-
-    assert request.call_args.kwargs["json"]["cpu"] == {
-        "id": "cpu3c",
-        "vcpuCount": 4,
-    }
-
-
 def test_create_cpu_pod_requires_instance_id():
     with pytest.raises(ValueError, match="instance_id"):
         ctl_commands.create_pod("cpu-pod", "python:3.11")
 
 
-def test_create_cpu_pod_validates_instance_id():
+@pytest.mark.parametrize(
+    "instance_id", ["cpu3c-invalid", "cpu3c", "cpu3c-4-not-a-number"]
+)
+def test_create_cpu_pod_validates_instance_id(template_backend, instance_id):
     with pytest.raises(ValueError, match="format"):
-        ctl_commands.create_pod("cpu-pod", "python:3.11", instance_id="cpu3c-invalid")
+        ctl_commands.create_pod("cpu-pod", "python:3.11", instance_id=instance_id)
 
 
 def test_create_pod_validates_image_and_cloud():
@@ -310,19 +301,12 @@ def test_create_gpu_pod_rejects_unsupported_constraints(template_backend, kwargs
         ctl_commands.create_pod("pod", "image", gpu_type_id="NVIDIA A100", **kwargs)
 
 
-def test_stop_and_resume_pod_use_actions():
-    with patch(
-        "runpod.api.ctl_commands.run_rest_request", return_value={"id": "pod"}
-    ) as request:
-        assert ctl_commands.stop_pod("pod") == {"id": "pod"}
-        assert ctl_commands.resume_pod("pod", 8) == {"id": "pod"}
-
-    assert request.call_args_list[0].args == (
-        "POST",
-        "/v2/pods/pod/action",
-    )
-    assert request.call_args_list[0].kwargs == {"json": {"action": "stop"}}
-    assert request.call_args_list[1].kwargs == {"json": {"action": "start"}}
+def test_resume_pod_rejects_gpu_count():
+    with (
+        patch("runpod.api.ctl_commands.run_rest_request", return_value={"id": "pod"}),
+        pytest.raises(ValueError, match="gpu_count"),
+    ):
+        ctl_commands.resume_pod("pod", 8)
 
 
 def test_terminate_pod_deletes_resource():
@@ -461,21 +445,6 @@ def test_create_container_registry_auth_uses_rest():
     )
 
 
-def test_update_container_registry_auth_uses_graphql():
-    with patch(
-        "runpod.api.ctl_commands.run_graphql_query",
-        return_value={"data": {"updateRegistryAuth": {"id": "registry"}}},
-    ) as request:
-        result = ctl_commands.update_container_registry_auth(
-            "registry", "user", "password"
-        )
-
-    assert result == {"id": "registry"}
-    mutation = request.call_args.args[0]
-    assert "mutation UpdateRegistryAuth" in mutation
-    assert 'id: "registry"' in mutation
-
-
 def test_delete_container_registry_auth_uses_rest():
     with patch(
         "runpod.api.ctl_commands.run_rest_request", return_value=None