MAINT: Remove datalad

.circleci/config.yml

@@ -286,7 +286,6 @@ jobs:
           keys:
             - data-cache-eeg_matchingpennies-1
       - bash_env
-      - gitconfig  # email address is needed for datalad
       - run:
           name: Get eeg_matchingpennies
           command: |
@@ -306,7 +305,6 @@ jobs:
           keys:
             - data-cache-MNE-phantom-KIT-data-1
       - bash_env
-      - gitconfig  # email address is needed for datalad
       - run:
           name: Get MNE-phantom-KIT-data
           command: |

.circleci/setup_bash.sh

@@ -33,15 +33,13 @@ fi
 
 # Set up image
 sudo ln -s /usr/lib/x86_64-linux-gnu/libxcb-util.so.0 /usr/lib/x86_64-linux-gnu/libxcb-util.so.1
-wget -q -O- http://neuro.debian.net/lists/focal.us-tn.libre | sudo tee /etc/apt/sources.list.d/neurodebian.sources.list
-sudo apt-key adv --recv-keys --keyserver hkps://keyserver.ubuntu.com 0xA5D32F012649A5A9
 echo "export RUN_TESTS=\".circleci/run_dataset_and_copy_files.sh\"" >> "$BASH_ENV"
 echo "export DOWNLOAD_DATA=\"coverage run -m mne_bids_pipeline._download\"" >> "$BASH_ENV"
 
 # Similar CircleCI setup to mne-python (Xvfb, venv, minimal commands, env vars)
 wget -q https://raw.githubusercontent.com/mne-tools/mne-python/main/tools/setup_xvfb.sh
 bash setup_xvfb.sh
-sudo apt install -qq tcsh git-annex-standalone python3.10-venv python3-venv libxft2
+sudo apt install -qq tcsh python3.10-venv python3-venv libxft2
 python3.10 -m venv ~/python_env
 wget -q https://raw.githubusercontent.com/mne-tools/mne-python/main/tools/get_minimal_commands.sh
 source get_minimal_commands.sh

CONTRIBUTING.md

@@ -11,17 +11,11 @@ Once this is done, you should be able to run this in a terminal:
 
 `$ python -c "import mne; mne.sys_info()"`
 
-You can then install the following additional packages via `pip`. Note that
+You can then install the following additional package via `pip`. Note that
 the URL points to the bleeding edge version of `mne_bids`:
 
-`$ pip install datalad`
 `$ pip install https://github.com/mne-tools/mne-bids/zipball/main`
 
-To get the test data, you need to install `git-annex` on your system. If you
-installed MNE-Python via `conda`, you can simply call:
-
-`conda install -c conda-forge git-annex`
-
 Now, get the pipeline through git:
 
 `$ git clone https://github.com/mne-tools/mne-bids-pipeline.git`

Makefile

@@ -23,8 +23,6 @@ doc:
 
 check:
 	which python
-	git-annex version
-	datalad --version
 	openneuro-py --version
 	mri_convert --version
 	mne_bids --version

docs/source/examples/gen_examples.py

@@ -203,21 +203,14 @@ def _gen_demonstrated_funcs(example_config_path: Path) -> dict:
             f"{fname.name} :fontawesome-solid-square-poll-vertical:</a>\n\n"
         )
 
-    assert (
-        sum(key in options for key in ("openneuro", "git", "web", "datalad", "mne"))
-        == 1
-    )
+    assert sum(key in options for key in ("openneuro", "web", "mne")) == 1
     if "openneuro" in options:
         url = f'https://openneuro.org/datasets/{options["openneuro"]}'
-    elif "git" in options:
-        url = options["git"]
     elif "web" in options:
         url = options["web"]
-    elif "mne" in options:
-        url = f"https://mne.tools/dev/generated/mne.datasets.{options['mne']}.data_path.html"  # noqa: E501
     else:
-        assert "datalad" in options  # guaranteed above
-        url = ""
+        assert "mne" in options
+        url = f"https://mne.tools/dev/generated/mne.datasets.{options['mne']}.data_path.html"  # noqa: E501
 
     source_str = (
         f"## Dataset source\n\nThis dataset was acquired from " f"[{url}]({url})\n"

docs/source/v1.6.md.inc

@@ -35,6 +35,7 @@
 ### :medical_symbol: Code health
 
 - The package build backend has been switched from `setuptools` to `hatchling`. (#825 by @hoechenberger)
+- Removed dependencies on `datalad` and `git-annex` for testing (#867 by @larsoner)
 - Code formatting now uses `ruff format` instead of `black` (#834, #838 by @larsoner)
 - Code caching is now tested using GitHub Actions (#836 by @larsoner)
 - Steps in the documentation are now automatically parsed into flowcharts (#859 by @larsoner)

mne_bids_pipeline/_download.py

@@ -9,29 +9,6 @@
 DEFAULT_DATA_DIR = Path("~/mne_data").expanduser()
 
 
-def _download_via_datalad(*, ds_name: str, ds_path: Path):
-    import datalad.api as dl
-
-    print(f'datalad installing "{ds_name}"')
-    options = DATASET_OPTIONS[ds_name]
-    git_url = options["git"]
-    assert "exclude" not in options
-    assert "hash" not in options
-    dataset = dl.install(path=ds_path, source=git_url)
-
-    # XXX: git-annex bug:
-    # https://github.com/datalad/datalad/issues/3583
-    # if datalad fails, use "get" twice, or set `n_jobs=1`
-    if ds_name == "ds003104":
-        n_jobs = 16
-    else:
-        n_jobs = 1
-
-    for to_get in DATASET_OPTIONS[ds_name].get("include", []):
-        print(f'datalad get data "{to_get}" for "{ds_name}"')
-        dataset.get(to_get, jobs=n_jobs)
-
-
 def _download_via_openneuro(*, ds_name: str, ds_path: Path):
     import openneuro
 
@@ -88,20 +65,12 @@ def _download_via_mne(*, ds_name: str, ds_path: Path):
 def _download(*, ds_name: str, ds_path: Path):
     options = DATASET_OPTIONS[ds_name]
     openneuro_name = options.get("openneuro", "")
-    git_url = options.get("git", "")
-    osf_node = options.get("osf", "")
     web_url = options.get("web", "")
     mne_mod = options.get("mne", "")
-    assert (
-        sum(bool(x) for x in (openneuro_name, git_url, osf_node, web_url, mne_mod)) == 1
-    )
+    assert sum(bool(x) for x in (openneuro_name, web_url, mne_mod)) == 1
 
     if openneuro_name:
         download_func = _download_via_openneuro
-    elif git_url:
-        download_func = _download_via_datalad
-    elif osf_node:
-        raise RuntimeError("OSF downloads are currently not supported.")
     elif mne_mod:
         download_func = _download_via_mne
     else:

mne_bids_pipeline/tests/datasets.py

@@ -22,16 +22,6 @@ class DATASET_OPTIONS_T(TypedDict, total=False):
         "hash": "sha256:ddc94a7c9ba1922637f2770592dd51c019d341bf6bc8558e663e1979a4cb002f",  # noqa: E501
     },
     "eeg_matchingpennies": {
-        # This dataset started out on osf.io as dataset https://osf.io/cj2dr
-        # then moved to g-node.org. As of 2023/02/28 when we download it via
-        # datalad it's too (~200 kB/sec!) and times out at the end:
-        #
-        #   "git": "https://gin.g-node.org/sappelhoff/eeg_matchingpennies",
-        #   "web": "",
-        #   "include": ["sub-05"],
-        #
-        # So now we mirror this datalad-fetched git repo back on osf.io!
-        # original dataset: "osf": "cj2dr"
         "web": "https://osf.io/download/8rbfk?version=1",
         "hash": "sha256:06bfbe52c50b9343b6b8d2a5de3dd33e66ad9303f7f6bfbe6868c3c7c375fafd",  # noqa: E501
     },

pyproject.toml

@@ -57,7 +57,6 @@ tests = [
   "pytest-cov",
   "pooch",
   "psutil",
-  "datalad",
   "ruff",
   "mkdocs",
   "mkdocs-material >= 9.0.4",