Fix handling of DOIs that start with doi: (#59)

* Fix handling of DOIs that start with `doi:`

* Run tests on Python 3.9; remove schedule

.github/workflows/run-linter.yml

@@ -11,8 +11,8 @@ on:
   create:
     branches: [main]
     tags: ['**']
-  schedule:
-    - cron: "0 4 * * *"
+  # schedule:
+  #   - cron: "0 4 * * *"
 
 jobs:
   lint:

.github/workflows/run-tests.yml

@@ -11,8 +11,8 @@ on:
   create:
     branches: [main]
     tags: ['**']
-  schedule:
-    - cron: "0 4 * * *"
+  # schedule:
+  #   - cron: "0 4 * * *"
 
 jobs:
   test:
@@ -25,7 +25,7 @@ jobs:
       fail-fast: false
       matrix:
         # python-version: [3.7, 3.8, 3.9]
-        python-version: [3.7]
+        python-version: [3.9]
 
     steps:
     - uses: actions/checkout@v2

CHANGES.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 2022.1.0
+
+- Fix handling of DOIs that start with `doi:`, as found e.g. in `ds002778`.
+
 ## 2021.10.1
 
 - New release for PyPI.

openneuro/download.py

@@ -490,16 +490,22 @@ def _get_local_tag(
                            '"DatasetDOI" field. Are you sure this is the '
                            'correct directory?')
 
+    local_doi = local_json['DatasetDOI']
+    if local_doi.startswith('doi:'):
+        # Remove the "protocol" prefix
+        local_doi = local_doi[4:]
+
     expected_doi_start = f'10.18112/openneuro.{dataset_id}.v'
-    if not local_json['DatasetDOI'].startswith(expected_doi_start):
+
+    if not local_doi.startswith(expected_doi_start):
         raise RuntimeError(f'The existing dataset in the target directory '
                            f'appears to be different from the one you '
                            f'requested to download. "DatasetDOI" field in '
                            f'local "dataset_description.json": '
                            f'{local_json["DatasetDOI"]}. '
                            f'Requested dataset: {dataset_id}')
 
-    local_version = (local_json['DatasetDOI']
+    local_version = (local_doi
                      .replace(f'10.18112/openneuro.{dataset_id}.v', ''))
     return local_version
 

openneuro/tests/test_download.py

@@ -87,3 +87,34 @@ def test_ds000248(tmp_path: Path):
         include=['participants.tsv'],
         target_dir=tmp_path
     )
+
+
+def test_doi_handling(tmp_path: Path):
+    """Test that we can handle DOIs that start with 'doi:`."""
+    dataset = 'ds000248'
+    download(
+        dataset=dataset,
+        include=['participants.tsv'],
+        target_dir=tmp_path
+    )
+
+    # Now inject a `doi:` prefix into the DOI
+    dataset_description_path = tmp_path / 'dataset_description.json'
+    dataset_description = json.loads(
+        dataset_description_path.read_text(encoding='utf-8')
+    )
+    assert not dataset_description['DatasetDOI'].startswith('doi:')
+    dataset_description['DatasetDOI'] = (
+        'doi:' + dataset_description['DatasetDOI']
+    )
+    dataset_description_path.write_text(
+        data=json.dumps(dataset_description, indent=2),
+        encoding='utf-8'
+    )
+
+    # Try to download again
+    download(
+        dataset=dataset,
+        include=['participants.tsv'],
+        target_dir=tmp_path
+    )

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = openneuro-py
-version = 2021.10.1
+version = 2022.1.0
 author = Richard Höchenberger <[email protected]>
 author_email = [email protected]
 url = https://github.com/hoechenberger/openneuro-py