Merge pull request #5 from Australian-Imaging-Service/github-download

added retrieve from github function

medimages4tests/mri/neuro/bold.py

@@ -0,0 +1,21 @@
+from pathlib import Path
+from medimages4tests import base_cache_dir
+from medimages4tests.utils import retrieve_from_openneuro, OpenneuroSpec
+
+
+cache_dir = base_cache_dir / "mri" / "neuro" / "bold"
+
+
+SAMPLES = {
+    "ds002014-01": OpenneuroSpec(
+        dataset="ds002014",
+        tag="1.0.1",
+        path="sub-01/func/sub-01_task-languageproduction_run-01_bold",
+    )
+}
+
+
+def get_image(out_dir: Path = None, sample: str = "ds002014-01"):
+    if out_dir is None:
+        out_dir = cache_dir / sample
+    return retrieve_from_openneuro(SAMPLES[sample], out_dir)

medimages4tests/mri/neuro/dwi.py

@@ -0,0 +1,23 @@
+from pathlib import Path
+from medimages4tests import base_cache_dir
+from medimages4tests.utils import retrieve_from_openneuro, OpenneuroSpec
+
+
+cache_dir = base_cache_dir / "mri" / "neuro" / "t1w"
+
+
+SAMPLES = {
+    "ds004024-CON031": OpenneuroSpec(
+        dataset="ds004024",
+        tag="1.0.1",
+        path="sub-CON031/ses-mri/dwi/sub-CON031_ses-mri_dwi",
+    )
+}
+
+
+def get_image(out_dir: Path = None, sample: str = "ds004024-CON031"):
+    if out_dir is None:
+        out_dir = cache_dir / sample
+    return retrieve_from_openneuro(
+        SAMPLES[sample], out_dir, suffixes=(".nii.gz", ".json", ".bvec", ".bval")
+    )

medimages4tests/mri/neuro/t1w.py

@@ -1,3 +1,4 @@
+from pathlib import Path
 from medimages4tests import base_cache_dir
 from medimages4tests.utils import retrieve_from_openneuro, OpenneuroSpec
 
@@ -10,9 +11,26 @@
         dataset="ds004130",
         tag="1.0.0",
         path="sub-ON01016/anat/sub-ON01016_acq-fspgr_run-01_T1w",
-    )
+    ),
+    "ds002014-01": OpenneuroSpec(
+        dataset="ds002014",
+        tag="1.0.1",
+        path="sub-01/anat/sub-01_T1w",
+    ),
+    "ds001743-01": OpenneuroSpec(
+        dataset="ds001743",
+        tag="1.0.1",
+        path="sub-01/anat/sub-01_T1w",
+    ),
+    "ds004024-CON031": OpenneuroSpec(
+        dataset="ds004024",
+        tag="1.0.1",
+        path="sub-CON031/ses-mri/dwi/sub-CON031_ses-mri_T1w",
+    ),
 }
 
 
-def get_image(sample="ds004130-ON01016"):
-    return retrieve_from_openneuro(SAMPLES[sample], cache_dir / sample)
+def get_image(out_dir: Path = None, sample: str = "ds004130-ON01016"):
+    if out_dir is None:
+        out_dir = cache_dir / sample
+    return retrieve_from_openneuro(SAMPLES[sample], out_dir)

medimages4tests/utils.py

@@ -1,13 +1,17 @@
 from tempfile import mkdtemp
+import requests
+import tarfile
 import shutil
+import os
+import typing as ty
 from pathlib import Path
 import openneuro
 import attrs
+from . import base_cache_dir
 
 
 @attrs.define
 class OpenneuroSpec:
-
     dataset: str
     tag: str
     path: Path = attrs.field(converter=Path)
@@ -16,6 +20,7 @@ class OpenneuroSpec:
 def retrieve_from_openneuro(
     sample, cache_path, suffixes=(".nii.gz", ".json"), force_download=False
 ):
+    """Retrieves an image from the OpenNeuro repository"""
     if not cache_path.parent.exists():
         cache_path.parent.mkdir(parents=True)
     out_path = cache_path.with_suffix(suffixes[0])
@@ -32,3 +37,66 @@ def retrieve_from_openneuro(
                 (tmpdir / sample.path).with_suffix(ext), cache_path.with_suffix(ext)
             )
     return out_path
+
+
+def retrieve_from_github(
+    org: str,
+    repo: str,
+    path: str,
+    tag: str = "main",
+    compressed: bool = True,
+    cache_dir: ty.Union[Path, str, None] = None,
+) -> Path:
+    """Retrieves a sample file from a path within a GitHub repository
+
+    Parameters
+    ----------
+    org: str
+        the Github organisation
+    repo : str
+        the name of the git repository within the Github organisation
+    path : str
+        the path to the file relative to the repository
+    tag : str, optional
+        the git tag (version) to use, "main" by default
+    compressed : bool, optional
+        whether the file within the git repo has been archived with tar/gzip and
+        needs to be uncompressed before use, True by default
+    cache_dir : Path | str, optional
+        the directory in which to download and cache the requested file, by default uses
+        "~/.medimages/cache/github"
+    """
+    if cache_dir is None:
+        cache_dir = base_cache_dir / "github"
+    else:
+        cache_dir = Path(cache_dir).expanduser()
+    cache_path = (cache_dir / repo / tag).joinpath(*path.split("/"))
+    if cache_path.exists():
+        return cache_path
+    if not cache_path.parent.exists():
+        cache_path.parent.mkdir(parents=True)
+    url = f"https://raw.githubusercontent.com/{repo}/{tag}/{path}"
+    if compressed:
+        url += ".tar.gz"
+    response = requests.get(url)
+    if response.status_code != "200":
+        raise ValueError(f"Did not find a file to download at '{url}'")
+    if compressed:
+        tmp_dir = Path(mkdtemp())
+        download_path = tmp_dir / url.split("/")[-1]
+    else:
+        download_path = cache_path
+    with open(download_path, "wb") as f:
+        f.write(response.content)
+    if compressed:
+        extract_dir = tmp_dir / "extracted"
+        extract_dir.mkdir()
+        with tarfile.open(download_path) as tfile:
+            tfile.extractall(path=extract_dir)
+        dir_contents = list(extract_dir.iterdir())
+        if len(dir_contents) > 1:
+            raise ValueError(
+                f"Contents or tar file at {url} contain more than one file/sub-dir ({dir_contents})"
+            )
+        os.rename(dir_contents[0], cache_path)
+    return cache_path

tests/test_github.py

@@ -0,0 +1,15 @@
+import pytest
+import nibabel as nb
+import numpy as np
+from medimages4tests.utils import retrieve_from_github
+
+
+@pytest.mark.xfail
+def test_github_retrieve():
+
+    nifti_fpath = retrieve_from_github(
+        org="nipype", repo="pydra-fsl-testdata", path="melodic_ica"
+    )
+    nifti = nifti = nb.load(nifti_fpath)
+
+    assert np.array_equal(nifti.header["dim"][:4], [3, 204, 256, 256])