Merge pull request #105 from LUMC/release1.3.0

Release 1.3.0

.travis.yml

@@ -2,7 +2,7 @@ language: python
 install:
   - pip install tox
 dist: xenial
-python: 3.5  # Use the oldest supported version of python as default.
+python: 3.6  # Use the oldest supported version of python as default.
 script:
     - tox -e $TOX_ENV
 matrix:
@@ -11,30 +11,25 @@ matrix:
     # Use default python3 version here.
     - env: TOX_ENV=docs
     - env: TOX_ENV=lint
-
     # UNIT TESTS
     # On most recent versions of python.
-    - python: 3.6
-      env: TOX_ENV=py36
-    - python: 3.7
-      env: TOX_ENV=py37
-    - python: 3.8
-      env: TOX_ENV=py38
-    - python: 3.5
-      env: TOX_ENV=py35
+    - env: TOX_ENV=py36
       install:
         - pip install codecov
         - pip install tox
       after_success:
         - codecov -v  # -v to make sure coverage upload works.
+    - python: 3.7
+      env: TOX_ENV=py37
+    - python: 3.8
+      env: TOX_ENV=py38
 
     # FUNCTIONAL TESTS
-    - python: 3.6
-      env: TOX_ENV=snakemake
-    - python: 3.7
-      env: TOX_ENV=snakemake
+    # Use default python here.
+    - env: TOX_ENV=snakemake
+    - env: TOX_ENV=miniwdl
     - env: TOX_ENV=cromwell
-      before_install:
+      install:
         # Install conda
         - export MINICONDA=${HOME}/miniconda
         - export PATH=${MINICONDA}/bin:${PATH}
@@ -44,5 +39,6 @@ matrix:
         - conda config --add channels defaults
         - conda config --add channels bioconda
         - conda config --add channels conda-forge
-        - conda create -n my_env cromwell tox  # Install tox for good integration within the conda env.
+        # TODO: Upgrade cromwell once https://github.com/broadinstitute/cromwell/pull/5456 is fixed.
+        - conda create -n my_env cromwell=48 tox  # Install tox for good integration within the conda env.
         - source activate my_env

HISTORY.rst

@@ -7,6 +7,42 @@ Changelog
 .. This document is user facing. Please word the changes in such a way
 .. that users understand how the changes affect the new version.
 
+version 1.3.0
+---------------------------
+Python 3.6 and pytest 5.4.0.0 are now minimum requirements for pytest-workflow.
+This was necessary for fixing the deprecation warning issue and the issue with
+the subdirectory evaluation. This also gave the opportunity to simplify the
+source code using new python 3.6 syntax.
+
++ Using the ``name`` keyword argument in workflow marks will be deprecated
+  from 1.4.0 onwards. A warning will be given if this is used. For example:
+  ``pytest.mark.workflow(name="my_workflow")``. Use the name as argument
+  instead: ``pytest.mark.workflow("my_workflow")``.
++ Allow running custom tests on multiple workflows. You can now use
+  ``pytest.mark.workflow("worflow name 1", "workflow name 2", ...)``.
+  (`Issue #75 <https://github.com/LUMC/pytest-workflow/issues/75>`_)
++ Add a miniwdl example to the documentation.
++ Added a ``--symlink`` flag to the CLI that changes the copying behavior.
+  Instead of copying, it creates a similar directory structure where all files
+  are linked to with symbolic links. (`Issue #96
+  <https://github.com/LUMC/pytest-workflow/issues/98>`_)
++ Refactored the code base. Python 3.6's f-strings and type annotation were
+  used consistently throughout the project. Some code was rewritten to be more
+  concise and readable.
++ Improved speed for searching string content in files. This was achieved by
+  removing intermediate functions and simplifying the search function.
++ Improved speed for calculating md5sums by increasing the read buffer size
+  from 8k to 64k.
++ Solve issue where pytest would display a lot of deprecation warnings when
+  running pytest-workflow. (`Issue #98
+  <https://github.com/LUMC/pytest-workflow/issues/98>`_)
++ Fix issues with later versions of Cromwell and Snakemake in CI testing.
++ Add correct subdirectory evaluation to fix issue where ``/parent-dir/child``
+  was evaluated as a subdirectory of ``/parent`` due to starting with the same
+  string. (`Issue #95 <https://github.com/LUMC/pytest-workflow/issues/95>`_)
++ Fix error in cromwell example which did not allow it to remove folders
+  correctly.
+
 version 1.2.3
 ---------------------------
 + Added missing ``help`` section for ``--tag`` on the CLI.

docs/examples.rst

@@ -55,9 +55,9 @@ created by the same user that runs the test (docker containers run as root by
 default). This will ensure that files can be deleted by pytest-workflow
 afterwards.
 
-The following yaml file tests a Cromwell pipeline. In this case Cromwell is
-installed via conda. The conda installation adds a wrapper to Cromwell so it
-can be used as a command, instead of having to use the jar.
+The following yaml file tests a WDL pipeline run with Cromwell. In this case
+Cromwell is installed via conda. The conda installation adds a wrapper to
+Cromwell so it can be used as a command, instead of having to use the jar.
 
 .. code-block:: yaml
 
@@ -70,3 +70,42 @@ can be used as a command, instead of having to use the jar.
       contains:
         - "WorkflowSucceededState"
 
+WDL with miniwdl example
+------------------------
+
+For miniwdl please consult the `runner reference
+<https://miniwdl.readthedocs.io/en/stable/runner_reference.html>`_ for more
+information on the localization of output files as well as options to modify
+the running of miniwdl from the environment.
+
+Miniwdl will localize all the output files to an ``output_links`` directory
+inside the test output directory. If you have a workflow with the output:
+
+.. code-block::
+
+        output {
+            File moo_file = moo_task.out
+            Array[File] stats = moo_task.stats_files
+        }
+
+Inside the ``output_links`` directory the directories ``moo_file`` and
+``stats`` will be created. Inside these directories will be the produced files.
+
+The following yaml file tests a WDL pipeline run with miniwdl.
+
+.. code-block:: yaml
+
+  - name: My pipeline
+    command: miniwdl run -i inputs.json -d test-output/ moo.wdl
+    files:
+      - path: test-output/output_links/moo_file/moo.txt.gz
+        md5sum: 173fd8023240a8016033b33f42db14a2
+      - path: test-output/output_links/stats/number_of_moos_per_cow.tsv
+        contains:
+          - 42
+      - path: test-output/output_links/stats/joy_invoking_moos.tsv
+        must_not_contain:
+          - 0
+
+Please note that the trailing slash in ``-d test-output/`` is important. It
+will ensure the files end up in the ``test-output`` directory.

docs/running_pytest_workflow.rst

@@ -26,8 +26,8 @@ Specific pytest options for pytest workflow
     :func: __pytest_workflow_cli
     :prog: pytest
 
-Temporary directory cleanup
----------------------------
+Temporary directory cleanup and creation
+----------------------------------------
 
 The temporary directories are cleaned up after the tests are completed.
 If you wish to inspect the output of a failing
@@ -51,6 +51,16 @@ use ``--basetemp <dir>`` to change pytest's base temp directory.
   DO NOT use ``--basetemp`` on directories where none of the
   contents should be deleted.
 
+The temporary directories created are copies of pytest's root directory, the
+directory from which it runs the tests. If you have lots of tests, and if you
+have a large repository, this may take a lot of disk space. To alleviate this
+you can use the ``--symlink`` flag which will create the same directory layout
+but instead symlinks the files instead of copying them. This is *slower* for
+lots of small files, and it carries with it the risk that the tests may alter
+files from your work directory. If there are a lot of large files and files are
+used read-only in tests, then it will use a lot less disk space and be faster
+as well.
+
 Running multiple workflows simultaneously
 -----------------------------------------
 

docs/writing_tests.rst

@@ -86,23 +86,34 @@ workflow.
     import pathlib
     import pytest
 
-    @pytest.mark.workflow(name='files containing numbers')
+    @pytest.mark.workflow('files containing numbers')
     def test_div_by_three(workflow_dir):
         number_file = pathlib.Path(workflow_dir, "123.txt")
         number_file_content = number_file.read_text()
         assert int(number_file_content) % 3 == 0
 
-The ``@pytest.mark.workflow(name='files containing numbers')`` marks the test
-as belonging to a workflow named ``files containing numbers``. The mark can
-also be written without the explicit ``name`` key as
-``@pytest.mark.workflow('files containing nummbers')``. This test will only run
-if the workflow 'files containing numbers' has run.
+The ``@pytest.mark.workflow('files containing numbers')`` marks the test
+as belonging to a workflow named ``files containing numbers``. This test will
+only run if the workflow 'files containing numbers' has run.
+
+Multiple workflows can use the same custom test like this:
+
+.. code-block:: python
+
+    import pathlib
+    import pytest
+
+    @pytest.mark.workflow('my_workflow', 'another_workflow',
+                          'yet_another_workflow')
+    def test_ensure_long_logs_are_written(workflow_dir):
+        log = pathlib.Path(workflow_dir, "log.out")
+        assert len(log.readtext()) > 10000
 
 ``workflow_dir`` is a fixture. It does not work without a
 ``pytest.mark.workflow('workflow_name')`` mark.  This is a
 `pathlib.Path <https://docs.python.org/3/library/pathlib.html>`_ object that
-points to the folder where the named workflow was executed. This allows writing of
-advanced python tests for each file produced by the workflow.
+points to the folder where the named workflow was executed. This allows writing
+of advanced python tests for each file produced by the workflow.
 
 .. note::
 

requirements.txt

@@ -1,3 +1,3 @@
 pyyaml
-pytest>=4
+pytest>=5.4.0
 jsonschema

setup.py

@@ -21,7 +21,7 @@
 
 setup(
     name="pytest-workflow",
-    version="1.2.3",
+    version="1.3.0",
     description="A pytest plugin for configuring workflow/pipeline tests "
                 "using YAML files",
     author="Leiden University Medical Center",
@@ -40,7 +40,6 @@
     classifiers=[
         "Programming Language :: Python :: 3 :: Only",
         "Programming Language :: Python :: 3",
-        "Programming Language :: Python :: 3.5",
         "Programming Language :: Python :: 3.6",
         "Programming Language :: Python :: 3.7",
         "Programming Language :: Python :: 3.8",
@@ -49,9 +48,10 @@
         "GNU Affero General Public License v3 or later (AGPLv3+)",
         "Framework :: Pytest",
     ],
-    python_requires=">=3.5",  # Because we use type annotation.
+    # Because we use the resolve(strict=False) feature from pathlib.
+    python_requires=">=3.6",
     install_requires=[
-        "pytest>=4",
+        "pytest>=5.4.0",  # To use from_parent node instantiation.
         "pyyaml",
         "jsonschema"
     ],

src/pytest_workflow/__init__.py

@@ -13,25 +13,3 @@
 #
 # You should have received a copy of the GNU Affero General Public License
 # along with pytest-workflow.  If not, see <https://www.gnu.org/licenses/
-
-import re
-import shutil
-from pathlib import Path
-from typing import List
-
-
-# This function was created to ensure the same conversion is used throughout
-# pytest-workflow.
-def replace_whitespace(string: str, replace_with: str = '_') -> str:
-    """
-    Replaces all whitespace with the string in replace_with.
-    :param string: input string
-    :param replace_with: Replace whitespace with this string. Default: '_'
-    :return: The string with whitespace converted.
-    """
-    return re.sub(r'\s+', replace_with, string)
-
-
-def rm_dirs(directories: List[Path]):
-    for directory in directories:
-        shutil.rmtree(str(directory))