Merge branch 'dev' into precommit_autofixing

Signed-off-by: Eric Kerfoot <17726042+ericspod@users.noreply.github.com>

Author: ericspod

.github/workflows/weekly-preview.yml

@@ -6,9 +6,18 @@ permissions:
 on:
   schedule:
   - cron: "0 2 * * 0"  # 02:00 of every Sunday
+  pull_request:
+    branches:
+      - dev
+
+env:
+  PYTHON_VER: '3.10'
+  PYTORCH_VER: '2.8.0'
+  PIP_EXTRA_INDEX_URL: "https://download.pytorch.org/whl/cpu"  # forces CPU PyTorch installation, should be faster
 
 jobs:
   static-checks:
+    if: github.event_name == 'schedule'  # only check on cron run, these checks are redundant in a PR
     runs-on: ubuntu-latest
     strategy:
       matrix:
@@ -25,10 +34,10 @@ jobs:
     - uses: actions/checkout@v7
       with:
         persist-credentials: false
-    - name: Set up Python 3.10
+    - name: Set up Python ${{ env.PYTHON_VER }}
       uses: actions/setup-python@v6
       with:
-        python-version: '3.10'
+        python-version: ${{ env.PYTHON_VER }}
         cache: 'pip'
     - name: Install dependencies
       run: |
@@ -40,22 +49,24 @@ jobs:
         $(pwd)/runtests.sh --build --clean
         $(pwd)/runtests.sh --build --${{ matrix.opt }}
 
-  packaging:
+  publish:
     if: github.repository == 'Project-MONAI/MONAI'
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v7
       with:
-        ref: dev
+        # get the ref for the PR branch or dev if this is a cron job
+        ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || 'dev' }}
         fetch-depth: 0
         persist-credentials: false
-    - name: Set up Python 3.10
+    - name: Set up Python ${{ env.PYTHON_VER }}
       uses: actions/setup-python@v6
       with:
-        python-version: '3.10'
-    - name: Install setuptools
+        python-version: ${{ env.PYTHON_VER }}
+        cache: 'pip'
+    - name: Install tools
       run: |
-        python -m pip install --user --upgrade setuptools wheel packaging
+        python -m pip install -U pip build
     - name: Build distribution
       run: |
         export HEAD_COMMIT_ID=$(git rev-parse HEAD)
@@ -72,9 +83,15 @@ jobs:
         git tag "1.7.dev${YEAR_WEEK}"
         git log -1
         git tag --list
-        python setup.py sdist bdist_wheel
-
+        python -m build
+        ls -lh dist
+    - name: Test Installation
+      run: |
+        pip install dist/*.whl
+        pip list
+        (cd "$(mktemp -d)" && python -c 'import monai; print(monai.__version__)')
     - name: Publish to PyPI
+      if: github.event_name == 'schedule'  # only publish on cron run
       uses: pypa/gh-action-pypi-publish@release/v1
       with:
         user: __token__

docs/requirements.txt

@@ -20,8 +20,8 @@ sphinxcontrib-serializinghtml
 sphinx-autodoc-typehints==1.11.1
 pandas
 einops
-transformers>=4.53.0
-mlflow>=2.12.2,<3.13
+transformers>=5.5.0
+mlflow>=3.15.2
 clearml>=1.10.0rc0
 tensorboardX
 imagecodecs; platform_system == "Linux" or platform_system == "Darwin"

docs/source/utils.rst

@@ -80,3 +80,8 @@ Ordering
 --------
 .. automodule:: monai.utils.ordering
   :members:
+
+Safe Evaluation
+---------------
+.. automodule:: monai.utils.safeeval
+  :members:

monai/bundle/config_parser.py

@@ -162,7 +162,14 @@ def __getattr__(self, key: str) -> Any:
         try:
             return self._chain(key)
         except KeyError:
-            return getattr(self._value, key)
+            pass
+        if isinstance(self._value, dict) and key in self._value:
+            # the chained id is absent from the resolver (for example when this proxy is
+            # backed by a `$@ref`, whose children have no ids of their own), but the key
+            # does exist in the container: resolve it like `__getitem__` does, so dot- and
+            # bracket-notation agree and config keys keep precedence over dict methods.
+            return self._value[key]
+        return getattr(self._value, key)
 
     def __getitem__(self, key: str | int) -> Any:
         try:

monai/bundle/scripts.py

@@ -25,6 +25,7 @@
 from textwrap import dedent
 from typing import Any
 
+import numpy as np
 import torch
 from torch.cuda import is_available
 
@@ -51,13 +52,13 @@
     min_version,
     optional_import,
     pprint_edges,
+    safe_eval,
 )
 
 validate, _ = optional_import("jsonschema", name="validate")
 ValidationError, _ = optional_import("jsonschema.exceptions", name="ValidationError")
 Checkpoint, has_ignite = optional_import("ignite.handlers", IgniteInfo.OPT_IMPORT_VERSION, min_version, "Checkpoint")
 requests, has_requests = optional_import("requests")
-onnx, _ = optional_import("onnx")
 huggingface_hub, _ = optional_import("huggingface_hub")
 
 logger = get_logger(module_name=__name__)
@@ -158,10 +159,12 @@ def _get_fake_spatial_shape(shape: Sequence[str | int], p: int = 1, n: int = 1,
             if i == "*":
                 ret.append(any)
             else:
-                for c in _get_var_names(i):
-                    if c not in ["p", "n"]:
-                        raise ValueError(f"only support variables 'p' and 'n' so far, but got: {c}.")
-                ret.append(eval(i, {"p": p, "n": n}))
+                bad_names = set(c for c in _get_var_names(i) if c not in {"p", "n"})
+                if bad_names:
+                    raise ValueError(f"Only variables `p` and `n` currently supported. Invalid names: {bad_names}")
+
+                # evaluate using Numpy types to prevent slow Python DoS attacks
+                ret.append(int(safe_eval(i, {"p": np.int32(p), "n": np.int32(n)}, rewrite_np=True)))
         else:
             raise ValueError(f"spatial shape items must be int or string, but got: {type(i)} {i}.")
     return tuple(ret)
@@ -994,7 +997,8 @@ def run(
             common parameters shown below will be added and can be passed through the `override` parameter of this method.
 
             - ``"output_dir"``: the path to save mlflow tracking outputs locally, default to "<bundle root>/eval".
-            - ``"tracking_uri"``: uri to save mlflow tracking outputs, default to "/output_dir/mlruns".
+            - ``"tracking_uri"``: uri to save mlflow tracking outputs, default to a local SQLite database
+              at "<output_dir>/mlruns.db" with run artifacts kept under "<output_dir>/mlruns".
             - ``"experiment_name"``: experiment name for this run, default to "monai_experiment".
             - ``"run_name"``: the name of current run.
             - ``"save_execute_config"``: whether to save the executed config files. It can be `False`, `/path/to/artifacts`
@@ -1433,6 +1437,7 @@ def onnx_export(
     converter_kwargs_.update({"inputs": inputs_, "use_trace": use_trace_})
 
     def save_onnx(onnx_obj: Any, filename_prefix_or_stream: str, **kwargs: Any) -> None:
+        onnx, _ = optional_import("onnx")
         onnx.save(onnx_obj, filename_prefix_or_stream)
 
     _export(

monai/bundle/utils.py

@@ -116,8 +116,10 @@
     "configs": {
         # if no "output_dir" in the bundle config, default to "<bundle root>/eval"
         "output_dir": "$@bundle_root + '/eval'",
-        # use URI to support linux, mac and windows os
-        "tracking_uri": "$monai.utils.path_to_uri(@output_dir) + '/mlruns'",
+        # MLflow 3.13+ rejects the filesystem (file store) tracking backend, so default tracking
+        # to a local SQLite database. The handler keeps run artifacts under "<output_dir>/mlruns"
+        # (next to the db). A URI is used so the path is valid on linux, mac and windows os.
+        "tracking_uri": "$monai.utils.path_to_sqlite_uri(@output_dir + '/mlruns.db')",
         "experiment_name": "monai_experiment",
         "run_name": None,
         # may fill it at runtime

monai/bundle/workflows.py

@@ -15,6 +15,7 @@
 import os
 import sys
 import time
+import warnings
 from abc import ABC, abstractmethod
 from collections.abc import Sequence
 from copy import copy
@@ -34,6 +35,23 @@
 logger = get_logger(module_name=__name__)
 
 
+def _warn_logging_file_execution(logging_file: str) -> None:
+    """
+    Warn that ``logging_file`` is about to be executed by `logging.config.fileConfig`.
+
+    Called immediately before every `fileConfig` invocation in this module, so the warning is only
+    raised when the file is really executed -- not when it is missing or logging is disabled.
+    """
+    warnings.warn(
+        f"applying logging config {logging_file}: `logging.config.fileConfig` passes the `class=` and "
+        "`args=` fields of the INI's handler and formatter sections to Python `eval()`, so this file "
+        "runs as code. A bundle ships its own `configs/logging.conf` and it is applied by default, "
+        "before any of the bundle's config is parsed. Only proceed if this file is from a source you "
+        "trust (see https://github.com/Project-MONAI/MONAI/security/advisories/GHSA-wvpx-5qmp-46g3).",
+        stacklevel=3,
+    )
+
+
 class BundleWorkflow(ABC):
     """
     Base class for the workflow specification in bundle, it can be a training, evaluation or inference workflow.
@@ -55,6 +73,10 @@ class BundleWorkflow(ABC):
         meta_file: filepath of the metadata file, if this is a list of file paths, their contents will be merged in order.
         logging_file: config file for `logging` module in the program. for more details:
             https://docs.python.org/3/library/logging.config.html#logging.config.fileConfig.
+            Security note: `fileConfig` passes the INI's `class=` and `args=` fields to Python
+            `eval()`, so this file runs as code and applying it raises a warning -- once per call
+            site, as Python's default warning filter suppresses repeats
+            (see https://github.com/Project-MONAI/MONAI/security/advisories/GHSA-wvpx-5qmp-46g3).
 
     """
 
@@ -72,6 +94,7 @@ def __init__(
             if not os.path.isfile(logging_file):
                 raise FileNotFoundError(f"Cannot find the logging config file: {logging_file}.")
             logger.info(f"Setting logging properties based on config: {logging_file}.")
+            _warn_logging_file_execution(logging_file)
             fileConfig(logging_file, disable_existing_loggers=False)
 
         if meta_file is not None:
@@ -273,6 +296,10 @@ class PythonicWorkflow(BundleWorkflow):
         meta_file: filepath of the metadata file, if this is a list of file paths, their contents will be merged in order.
         logging_file: config file for `logging` module in the program. for more details:
             https://docs.python.org/3/library/logging.config.html#logging.config.fileConfig.
+            Security note: `fileConfig` passes the INI's `class=` and `args=` fields to Python
+            `eval()`, so this file runs as code and applying it raises a warning -- once per call
+            site, as Python's default warning filter suppresses repeats
+            (see https://github.com/Project-MONAI/MONAI/security/advisories/GHSA-wvpx-5qmp-46g3).
 
     """
 
@@ -375,6 +402,10 @@ class ConfigWorkflow(BundleWorkflow):
             https://docs.python.org/3/library/logging.config.html#logging.config.fileConfig.
             If None, default to "configs/logging.conf", which is commonly used for bundles in MONAI model zoo.
             If False, the logging logic for the bundle will not be modified.
+            Security note: `fileConfig` passes the INI's `class=` and `args=` fields to Python
+            `eval()`, so this file runs as code and applying it raises a warning -- once per call
+            site, as Python's default warning filter suppresses repeats
+            (see https://github.com/Project-MONAI/MONAI/security/advisories/GHSA-wvpx-5qmp-46g3).
         init_id: ID name of the expected config expression to initialize before running, default to "initialize".
             allow a config to have no `initialize` logic and the ID.
         run_id: ID name of the expected config expression to run, default to "run".
@@ -444,6 +475,7 @@ def __init__(
                 else:
                     raise FileNotFoundError(f"Cannot find the logging config file: {logging_file}.")
             else:
+                _warn_logging_file_execution(str(logging_file))
                 fileConfig(str(logging_file), disable_existing_loggers=False)
                 logger.info(f"Setting logging properties based on config: {logging_file}.")