fix: update virtualenv building so llamastack- prefix is not added, make notebook experience easier (#1225)

Make sure venv behaves like conda (no prefix is added to image_name) and
`--image-type venv` inside a notebook "just works" without any fiddling

.pre-commit-config.yaml

@@ -30,6 +30,7 @@ repos:
     rev: v0.9.4
     hooks:
     -   id: ruff
+        args: [ --fix ]
         exclude: ^llama_stack/strong_typing/.*$
     -   id: ruff-format
 

llama_stack/cli/stack/_build.py

@@ -37,6 +37,7 @@ from llama_stack.distribution.distribution import get_provider_registry
 from llama_stack.distribution.resolver import InvalidProviderError
 from llama_stack.distribution.utils.config_dirs import DISTRIBS_BASE_DIR
 from llama_stack.distribution.utils.dynamic import instantiate_class_type
+from llama_stack.distribution.utils.exec import in_notebook
 from llama_stack.providers.datatypes import Api
 
 TEMPLATES_PATH = Path(__file__).parent.parent.parent / "templates"
@@ -59,8 +60,16 @@ def run_stack_build_command(args: argparse.Namespace) -> None:
     if args.list_templates:
         return _run_template_list_cmd()
 
+    if args.image_type == "venv":
+        current_venv = os.environ.get("VIRTUAL_ENV")
+        image_name = args.image_name or current_venv
+        if not image_name and in_notebook():
+            image_name = "__system__"
+    elif args.image_type == "conda":
         current_conda_env = os.environ.get("CONDA_DEFAULT_ENV")
         image_name = args.image_name or current_conda_env
+    else:
+        image_name = args.image_name
 
     if args.template:
         available_templates = available_templates_specs()
@@ -256,6 +265,9 @@ def _run_stack_build_command_from_build_config(
     elif build_config.image_type == ImageType.conda.value:
         if not image_name:
             raise ValueError("Please specify an image name when building a conda image")
+    elif build_config.image_type == ImageType.venv.value:
+        if not image_name:
+            raise ValueError("Please specify an image name when building a venv image")
 
     if template_name:
         build_dir = DISTRIBS_BASE_DIR / template_name

llama_stack/distribution/build_venv.sh

@@ -16,6 +16,7 @@ TEST_PYPI_VERSION=${TEST_PYPI_VERSION:-}
 # Reference: https://github.com/astral-sh/uv/pull/1694
 UV_HTTP_TIMEOUT=${UV_HTTP_TIMEOUT:-500}
 UV_SYSTEM_PYTHON=${UV_SYSTEM_PYTHON:-}
+VIRTUAL_ENV=${VIRTUAL_ENV:-}
 
 if [ -n "$LLAMA_STACK_DIR" ]; then
   echo "Using llama-stack-dir=$LLAMA_STACK_DIR"
@@ -25,7 +26,7 @@ if [ -n "$LLAMA_MODELS_DIR" ]; then
 fi
 
 if [ "$#" -lt 3 ]; then
-  echo "Usage: $0 <distribution_type> <build_name> <pip_dependencies> [<special_pip_deps>]" >&2
+  echo "Usage: $0 <distribution_type> <env_name> <pip_dependencies> [<special_pip_deps>]" >&2
   echo "Example: $0 <distribution_type> mybuild ./my-stack-build.yaml 'numpy pandas scipy'" >&2
   exit 1
 fi
@@ -34,8 +35,7 @@ special_pip_deps="$3"
 
 set -euo pipefail
 
-build_name="$1"
-env_name="llamastack-$build_name"
+env_name="$1"
 pip_dependencies="$2"
 
 # Define color codes
@@ -75,8 +75,12 @@ run() {
   local pip_dependencies="$2"
   local special_pip_deps="$3"
   
-  if [ -n "$UV_SYSTEM_PYTHON" ]; then 
+  if [ -n "$UV_SYSTEM_PYTHON" ] || [ "$env_name" == "__system__" ]; then 
     echo "Installing dependencies in system Python environment"
+    # if env == __system__, ensure we set UV_SYSTEM_PYTHON
+    export UV_SYSTEM_PYTHON=1
+  elif [ "$VIRTUAL_ENV" == "$env_name" ]; then
+    echo "Virtual environment $env_name is already active"
   else
     echo "Using virtual environment $env_name"
     uv venv "$env_name"
@@ -90,6 +94,7 @@ run() {
     # shellcheck disable=SC2086
     # we are building a command line so word splitting is expected
     uv pip install --extra-index-url https://test.pypi.org/simple/ \
+      --index-strategy unsafe-best-match \
       llama-models=="$TEST_PYPI_VERSION" llama-stack=="$TEST_PYPI_VERSION" \
       $pip_dependencies
     if [ -n "$special_pip_deps" ]; then

llama_stack/distribution/library_client.py

@@ -41,6 +41,7 @@ from llama_stack.distribution.stack import (
     redact_sensitive_fields,
     replace_env_vars,
 )
+from llama_stack.distribution.utils.exec import in_notebook
 from llama_stack.providers.utils.telemetry.tracing import (
     end_trace,
     setup_logger,
@@ -52,19 +53,6 @@ logger = logging.getLogger(__name__)
 T = TypeVar("T")
 
 
-def in_notebook():
-    try:
-        from IPython import get_ipython
-
-        if "IPKernelApp" not in get_ipython().config:  # pragma: no cover
-            return False
-    except ImportError:
-        return False
-    except AttributeError:
-        return False
-    return True
-
-
 def convert_pydantic_to_json_value(value: Any) -> Any:
     if isinstance(value, Enum):
         return value.value

llama_stack/distribution/start_venv.sh

@@ -55,6 +55,7 @@ while [[ $# -gt 0 ]]; do
   esac
 done
 
+echo "Using virtual environment: $venv_path"
 # Activate virtual environment
 if [ ! -d "$venv_path" ]; then
   echo -e "${RED}Error: Virtual environment not found at $venv_path${NC}" >&2

llama_stack/distribution/utils/exec.py

@@ -22,6 +22,19 @@ def run_with_pty(command):
         return _run_with_pty_unix(command)
 
 
+def in_notebook():
+    try:
+        from IPython import get_ipython
+
+        if "IPKernelApp" not in get_ipython().config:  # pragma: no cover
+            return False
+    except ImportError:
+        return False
+    except AttributeError:
+        return False
+    return True
+
+
 # run a command in a pseudo-terminal, with interrupt handling,
 # useful when you want to run interactive things
 def _run_with_pty_unix(command):