Add data.europa.eu support (#87)

datahugger/base.py

@@ -11,7 +11,7 @@
 from urllib.parse import urlparse
 
 import requests
-from jsonpath_ng import parse
+from jsonpath_ng.ext import parse
 from scitree import scitree
 from tqdm import tqdm
 

datahugger/config.py

@@ -1,5 +1,6 @@
 from datahugger.services import ArXivDataset
 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
@@ -117,6 +118,7 @@
     "trolling.uit.no": DataverseDataset,
     "www.sodha.be": DataverseDataset,
     "www.uni-hildesheim.de": DataverseDataset,
+    "data.europa.eu": DataEuropaDataset,
 }
 
 # regexp lookup

datahugger/services.py

@@ -390,6 +390,23 @@ def _get_attr_hash_type(self, record):
         return self._get_attr_attr(record, self.ATTR_HASH_JSONPATH).split(":")[0]
 
 
+class DataEuropaDataset(DatasetDownloader):
+    """Downloader for European data repository."""
+
+    REGEXP_ID = r"data\.europa\.eu\/data\/datasets\/(?P<record_id>.+)"
+
+    # the base entry point of the REST API
+    API_URL = "https://data.europa.eu/api/hub/repo/"
+
+    API_URL_META = "{api_url}datasets/{record_id}"
+    META_FILES_JSONPATH = '$.@graph[?(@.@type == "dcat:Distribution")]'
+
+    # paths to file attributes
+    ATTR_FILE_LINK_JSONPATH = "'dcat:accessURL'.@id"
+    ATTR_NAME_JSONPATH = "'dct:title'"
+    ATTR_SIZE_JSONPATH = "'dcat:byteSize'.@value"
+
+
 class SeaNoeDataset(DatasetDownloader):
     """Downloader for SeaNoe publication."""
 

tests/test_repositories.toml

@@ -110,6 +110,10 @@ files = "AA_age.tab"
 location = "https://github.com/j535d165/cbsodata"
 files = "cbsodata-main/README.md"
 
+[[dataeuropa]]
+location = "https://data.europa.eu/data/datasets/65e092e4009f18f050b14216"
+files = "consolidation-wattzhub-schema-irve-dynamic-20240918-033000.csv"
+
 [[seanoe]]
 location = "https://doi.org/10.17882/101042"
 files = "111609.xlsx"