feat: ckan harvester supports named schema

ckanext/harvest_basket/harvesters/ckan_harvester.py

@@ -1,6 +1,5 @@
 import json
 import logging
-from urllib.parse import urljoin
 
 import ckan.plugins.toolkit as tk
 from ckan import model
@@ -9,6 +8,7 @@
 from ckanext.harvest.harvesters.ckanharvester import SearchError
 
 from ckanext.harvest_basket.harvesters.base_harvester import BasketBasicHarvester
+from ckanext.transmute.utils import get_schema
 
 
 log = logging.getLogger(__name__)
@@ -20,9 +20,14 @@ class CustomCKANHarvester(CKANHarvester, BasketBasicHarvester):
     def import_stage(self, harvest_object):
         package_dict = json.loads(harvest_object.content)
         self._set_config(harvest_object.source.config)
-        self.transmute_data(package_dict, self.config.get("tsm_schema"))
-        harvest_object.content = json.dumps(package_dict)
 
+        schema = self.config.get("tsm_schema")
+        if not schema and (schema_name := self.config.get("tsm_named_schema")):
+            schema = get_schema(schema_name)
+
+        self.transmute_data(package_dict, schema)
+
+        harvest_object.content = json.dumps(package_dict)
         super().import_stage(harvest_object)
 
     def _search_for_datasets(self, remote_ckan_base_url, fq_terms=None):

ckanext/harvest_basket/harvesters/csw.py

@@ -2,9 +2,6 @@
 
 import logging
 from lxml import etree
-
-from ckan import model
-import ckan.plugins as p
 from ckanext.spatial.harvesters import CSWHarvester
 from ckanext.spatial.lib.csw_client import CswService, CswError, PropertyIsEqualTo
 from ckanext.transmute.utils import get_schema

ckanext/harvest_basket/harvesters/dcat.py

@@ -9,7 +9,7 @@
 
 class BasketDcatJsonHarvester(DCATJSONHarvester, BasketBasicHarvester):
     SRC_ID = "DCAT"
-    n = 0
+
     def info(self):
         return {
             "name": "basket_dcat_json",
@@ -31,7 +31,3 @@ def modify_package_dict(self, package_dict, dcat_dict, harvest_object):
         self.transmute_data(package_dict, schema)
 
         return package_dict
-
-    def _get_guids_and_datasets(self, content):
-        for idx, item in enumerate(super()._get_guids_and_datasets(content)):
-            yield item