From 6ca501b0e0397cbbf583e0eba05dfcc790544e1c Mon Sep 17 00:00:00 2001
From: Miguel Caballer Fernandez <micafer1@upv.es>
Date: Sun, 29 Sep 2024 09:42:15 +0200
Subject: [PATCH] Add support to SeaNoe (#85)

---
 datahugger/config.py         |  2 ++
 datahugger/services.py       | 20 ++++++++++++++++++++
 tests/test_repositories.toml |  4 ++++
 3 files changed, 26 insertions(+)

diff --git a/datahugger/config.py b/datahugger/config.py
index 3176c87..2c5e90f 100644
--- a/datahugger/config.py
+++ b/datahugger/config.py
@@ -10,6 +10,7 @@
 from datahugger.services import MendeleyDataset
 from datahugger.services import OSFDataset
 from datahugger.services import PangaeaDataset
+from datahugger.services import SeaNoeDataset
 from datahugger.services import ZenodoDataset
 
 # fast lookup
@@ -112,6 +113,7 @@
     "researchdata.ntu.edu.sg": DataverseDataset,
     "rin.lipi.go.id": DataverseDataset,
     "ssri.is": DataverseDataset,
+    "www.seanoe.org": SeaNoeDataset,
     "trolling.uit.no": DataverseDataset,
     "www.sodha.be": DataverseDataset,
     "www.uni-hildesheim.de": DataverseDataset,
diff --git a/datahugger/services.py b/datahugger/services.py
index a0075e7..3a0ce60 100644
--- a/datahugger/services.py
+++ b/datahugger/services.py
@@ -388,3 +388,23 @@ def _get_attr_hash(self, record):
 
     def _get_attr_hash_type(self, record):
         return self._get_attr_attr(record, self.ATTR_HASH_JSONPATH).split(":")[0]
+
+
+class SeaNoeDataset(DatasetDownloader):
+    """Downloader for SeaNoe publication."""
+
+    REGEXP_ID = r"https://www.seanoe\.org/data/[0-9]+/(?P<record_id>.*)/"
+
+    # the base entry point of the REST API
+    API_URL = "https://www.seanoe.org/api/"
+
+    # the files and metadata about the dataset
+    API_URL_META = "{api_url}find-by-id/{record_id}"
+    META_FILES_JSONPATH = "files[*]"
+
+    # paths to file attributes
+    ATTR_NAME_JSONPATH = "fileName"
+    ATTR_FILE_LINK_JSONPATH = "fileUrl"
+    ATTR_SIZE_JSONPATH = "size"
+    ATTR_HASH_JSONPATH = "checksum"
+    ATTR_HASH_TYPE_VALUE = "sha256"
diff --git a/tests/test_repositories.toml b/tests/test_repositories.toml
index 1758b46..ce22f08 100644
--- a/tests/test_repositories.toml
+++ b/tests/test_repositories.toml
@@ -109,3 +109,7 @@ files = "AA_age.tab"
 [[github]]
 location = "https://github.com/j535d165/cbsodata"
 files = "cbsodata-main/README.md"
+
+[[seanoe]]
+location = "https://doi.org/10.17882/101042"
+files = "111609.xlsx"