Add B2Share support #88 (#89)

Co-authored-by: Jonathan de Bruin <[email protected]>

Author: micafer

datahugger/config.py

@@ -1,4 +1,5 @@
 from datahugger.services import ArXivDataset
+from datahugger.services import B2shareDataset
 from datahugger.services import DataDryadDataset
 from datahugger.services import DataEuropaDataset
 from datahugger.services import DataOneDataset
@@ -118,6 +119,7 @@
     "trolling.uit.no": DataverseDataset,
     "www.sodha.be": DataverseDataset,
     "www.uni-hildesheim.de": DataverseDataset,
+    "b2share.eudat.eu": B2shareDataset,
     "data.europa.eu": DataEuropaDataset,
 }
 

datahugger/services.py

@@ -320,7 +320,7 @@ class MendeleyDataset(DatasetDownloader):
 class OSFDataset(DatasetDownloader):
     """Downloader for OSF repository."""
 
-    REGEXP_ID = r"osf\.io\/(?P<record_id>.*)/"
+    REGEXP_ID = r"osf\.io\/(?P<record_id>[^\/]*)\/{0,1}"
 
     # the base entry point of the REST API
     API_URL = "https://api.osf.io/v2/nodes/"
@@ -425,3 +425,23 @@ class SeaNoeDataset(DatasetDownloader):
     ATTR_SIZE_JSONPATH = "size"
     ATTR_HASH_JSONPATH = "checksum"
     ATTR_HASH_TYPE_VALUE = "sha256"
+
+
+class B2shareDataset(DatasetDownloader):
+    """Downloader for B2Share repository."""
+
+    REGEXP_ID = r"b2share\.eudat\.eu\/records\/(?P<record_id>[0-9a-z]+)"
+
+    # the base entry point of the REST API
+    API_URL = "https://b2share.eudat.eu/api/"
+
+    # the files and metadata about the dataset
+    API_URL_META = "{api_url}records/{record_id}"
+    META_FILES_JSONPATH = "files[*]"
+
+    # paths to file attributes
+    ATTR_NAME_JSONPATH = "key"
+    ATTR_FILE_LINK_JSONPATH = "ePIC_PID"
+    ATTR_SIZE_JSONPATH = "size"
+    ATTR_HASH_JSONPATH = "checksum"
+    ATTR_HASH_TYPE_VALUE = "md5"

tests/test_repositories.toml

@@ -114,6 +114,10 @@ files = "AA_age.tab"
 location = "https://github.com/j535d165/cbsodata"
 files = "cbsodata-main/README.md"
 
+[[b2share]]
+location = "https://b2share.eudat.eu/records/db2ef5890fa44c7a85af366a50de73b9"
+files = "2024-02-13.sav"
+
 [[dataeuropa]]
 location = "https://data.europa.eu/data/datasets/65e092e4009f18f050b14216"
 files = "consolidation-wattzhub-schema-irve-statique-20240220-152202.csv"