Merge pull request #4 from micafer/b2share

Add B2Share support J535D165#88
micafer · Sep 24, 2024 · b4a9c04 · b4a9c04
2 parents 62698ef + 9be54fd
commit b4a9c04
Show file tree

Hide file tree

Showing 4 changed files with 31 additions and 7 deletions.
diff --git a/datahugger/base.py b/datahugger/base.py
@@ -159,12 +159,8 @@ def download_file(
             if self.progress:
                 print(f"{_format_filename(file_name)}: SKIPPED")
             return
-
-        if (
-            self.filter_files and
-            not re.match(self.filter_files, file_name)
-            ):
-
+
+        if self.filter_files and not re.match(self.filter_files, file_name):
             logging.info(f"Skipping file by filter {file_link}")
             if self.progress:
                 print(f"{_format_filename(file_name)}: SKIPPED")

diff --git a/datahugger/config.py b/datahugger/config.py
@@ -1,10 +1,11 @@
 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
 from datahugger.services import DataverseDataset
 from datahugger.services import DjehutyDataset
 from datahugger.services import DSpaceDataset
-from datahugger.services import DataEuropaDataset
 from datahugger.services import FigShareDataset
 from datahugger.services import GitHubDataset
 from datahugger.services import HuggingFaceDataset
@@ -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,
 }
 

diff --git a/datahugger/services.py b/datahugger/services.py
@@ -390,6 +390,28 @@ def _get_attr_hash_type(self, record):
         return self._get_attr_attr(record, self.ATTR_HASH_JSONPATH).split(":")[0]
 
 
+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_SIZE_JSONPATH = "size"
+    ATTR_HASH_JSONPATH = "checksum"
+    ATTR_HASH_TYPE_VALUE = "md5"
+
+    def _get_attr_link(self, record, base_url=None):
+        return f"{base_url}/files/{self._params['record_id']}/{record['key']}"
+
+
 class SeaNoeDataset(DatasetDownloader):
     """Downloader for SeaNoe publication."""
 

diff --git a/tests/test_repositories.toml b/tests/test_repositories.toml
@@ -110,6 +110,10 @@ 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"
+
 [[seanoe]]
 location = "https://doi.org/10.17882/101042"
 files = "111609.xlsx"