[FIX] Package pipeline-catalog as data files in CLI installation (#379

)

* move pipeline-catalog submodule into bagel package

* update path to processing pipeline index

* add pipeline-catalog to package installation

* move tests out of bagel module and update paths

* move submodules for test data into tests module

* update dockerignore

* update test data paths

* do not install CLI in editable mode in CI tests

* install minimal dependencies in Docker image

.dockerignore

@@ -38,6 +38,5 @@ CITATION.cff
 LICENSE
 README.md
 
-# submodules used for testing
-bids-examples
-neurobagel_examples
+# testing
+tests

.github/workflows/test.yml

@@ -23,7 +23,7 @@ jobs:
       run: |
         python -m pip install --upgrade pip
         pip install -r dev_requirements.txt
-        pip install --no-deps -e .
+        pip install --no-deps .
     - name: Lint with flake8
       run: |
         # stop the build if there are Python syntax errors or undefined names

.gitmodules

@@ -1,9 +1,9 @@
 [submodule "bids-examples"]
-	path = bids-examples
+	path = tests/bids-examples
 	url = https://github.com/bids-standard/bids-examples.git
 [submodule "neurobagel_examples"]
-	path = neurobagel_examples
+	path = tests/neurobagel_examples
 	url = https://github.com/neurobagel/neurobagel_examples.git
 [submodule "pipeline-catalog"]
-	path = pipeline-catalog
+	path = bagel/pipeline-catalog
 	url = https://github.com/nipoppy/pipeline-catalog.git

Dockerfile

@@ -8,7 +8,7 @@ COPY . /app/src
 # 1. install the environment from our lockfile
 RUN pip install -r /app/src/requirements.txt
 # 2. install the CLI script without touching the dependencies again
-RUN pip install --no-cache-dir --no-deps /app/src[all]
+RUN pip install --no-cache-dir --no-deps /app/src
 
 ENTRYPOINT [ "bagel" ]
 CMD ["--help"]

bagel/mappings.py

@@ -38,8 +38,9 @@
     "assessment_tool": NB.pf + ":Assessment",
 }
 
+# TODO: Use importlib.resources.files(bagel) to get the path to the pipeline-catalog instead?
 PROCESSING_PIPELINE_PATH = (
-    Path(__file__).parents[1] / "pipeline-catalog" / "processing"
+    Path(__file__).parent / "pipeline-catalog" / "processing"
 )
 
 

setup.cfg

@@ -29,6 +29,10 @@ zip_safe = False
 packages = find:
 include_package_data = True
 
+[options.package_data]
+bagel = 
+    pipeline-catalog/**/*
+
 [options.extras_require]
 dev =
     flake8

bagel/tests/conftest.py → tests/conftest.py

@@ -13,7 +13,7 @@ def runner():
 
 @pytest.fixture(scope="session")
 def bids_path(tmp_path_factory):
-    return Path(__file__).absolute().parent.parent.parent / "bids-examples"
+    return Path(__file__).absolute().parent / "bids-examples"
 
 
 @pytest.fixture(scope="session")
@@ -28,9 +28,7 @@ def test_data():
 
 @pytest.fixture(scope="session")
 def neurobagel_examples_path(tmp_path_factory):
-    return (
-        Path(__file__).absolute().parent.parent.parent / "neurobagel_examples"
-    )
+    return Path(__file__).absolute().parent / "neurobagel_examples"
 
 
 @pytest.fixture(scope="session")

bagel/tests/test_cli_bids.py → tests/test_cli_bids.py

@@ -133,6 +133,7 @@ def test_imaging_sessions_have_expected_metadata(
 def test_bids_data_with_sessions_have_correct_paths(
     runner,
     test_data_upload_path,
+    bids_synthetic,
     default_pheno_bids_output_path,
     load_test_json,
 ):
@@ -147,7 +148,7 @@ def test_bids_data_with_sessions_have_correct_paths(
             "--jsonld-path",
             test_data_upload_path / "example_synthetic.jsonld",
             "--bids-dir",
-            Path(__file__).parent / "../../bids-examples/synthetic",
+            bids_synthetic,
             "--output",
             default_pheno_bids_output_path,
         ],

bagel/tests/test_utility.py → tests/test_utility.py

@@ -514,15 +514,16 @@ def test_create_acquisitions(bids_path, bids_dir, acquisitions, bids_session):
     "bids_sub_id, session",
     [("01", "01"), ("02", "02"), ("03", "01")],
 )
-def test_get_session_path_when_session_exists(bids_sub_id, session):
+def test_get_session_path_when_session_exists(
+    bids_sub_id, session, bids_synthetic
+):
     """
     Test that given a subject and session ID (i.e. when BIDS session layer exists for dataset),
     get_session_path() returns a path to the subject's session directory.
     """
-    bids_dir = Path(__file__).parent / "../../bids-examples/synthetic"
     session_path = butil.get_session_path(
-        layout=BIDSLayout(bids_dir, validate=True),
-        bids_dir=bids_dir,
+        layout=BIDSLayout(bids_synthetic, validate=True),
+        bids_dir=bids_synthetic,
         bids_sub_id=bids_sub_id,
         session=session,
     )
@@ -534,12 +535,12 @@ def test_get_session_path_when_session_exists(bids_sub_id, session):
 
 
 @pytest.mark.parametrize("bids_sub_id", ["01", "03", "05"])
-def test_get_session_path_when_session_missing(bids_sub_id):
+def test_get_session_path_when_session_missing(bids_sub_id, bids_path):
     """
     Test that given only a subject ID (i.e., when BIDS session layer is missing for dataset),
     get_session_path() returns the path to the subject directory.
     """
-    bids_dir = Path(__file__).parent / "../../bids-examples/ds001"
+    bids_dir = bids_path / "ds001"
     session_path = butil.get_session_path(
         layout=BIDSLayout(bids_dir, validate=True),
         bids_dir=bids_dir,