From 41eb58945ee1d77ca515bdef5f44b96f3f074458 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 29 Jul 2024 16:07:32 +0200
Subject: [PATCH 01/19] fix: Update parsing of the dataset of the license and
 rights fields

---
 ckanext/dcatapchharvest/dcat_helpers.py | 89 ++++++++++++++++++++-----
 ckanext/dcatapchharvest/profiles.py     | 37 +++++-----
 2 files changed, 91 insertions(+), 35 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 2e1d800..9312cd5 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -4,7 +4,7 @@
 from urlparse import urlparse
 from ckantoolkit import config
 from rdflib import URIRef, Graph
-from rdflib.namespace import Namespace, RDF, SKOS
+from rdflib.namespace import Namespace, RDF, SKOS, FOAF
 import xml.etree.ElementTree as ET
 import logging
 
@@ -17,6 +17,7 @@
 
 SKOSXL = Namespace("http://www.w3.org/2008/05/skos-xl#")
 RDFS = Namespace("http://www.w3.org/2000/01/rdf-schema#")
+FOAF = Namespace("http://xmlns.com/foaf/0.1/")
 
 frequency_namespaces = {
   "skos": SKOS,
@@ -38,6 +39,7 @@
   "skosxl": SKOSXL,
   "rdf": RDF,
   "rdfs": RDFS,
+  "foaf": FOAF,
 }
 
 theme_namespaces = {
@@ -167,38 +169,95 @@ def get_frequency_values():
     return frequency_mapping
 
 
-def get_license_uri_by_name(vocabulary_name):
-    license_vocabulary = get_license_values()
-    for key, value in license_vocabulary.items():
+def get_license_ref_uri_by_name(vocabulary_name):
+    _, license_ref_literal_vocabulary, _ = get_license_values()
+    for key, value in license_ref_literal_vocabulary.items():
         if unicode(vocabulary_name) == unicode(value):
             return key
     return None
 
 
-def get_license_name_by_uri(vocabulary_uri):
-    license_vocabulary = get_license_values()
-    for key, value in license_vocabulary.items():
+def get_license_ref_uri_by_homepage_uri(vocabulary_name):
+    _, _, license_homepage_ref_vocabulary = get_license_values()
+    for key, value in license_homepage_ref_vocabulary.items():
+        if unicode(vocabulary_name) == unicode(key):
+            return value
+    return None
+
+
+def get_license_name_by_ref_uri(vocabulary_uri):
+    _, license_ref_literal_vocabulary, _ = get_license_values()
+    for key, value in license_ref_literal_vocabulary.items():
+        if unicode(vocabulary_uri) == unicode(key):
+            return unicode(value)
+    return None
+
+
+def get_license_name_by_homepage_uri(vocabulary_uri):
+    license_homepages_literal_vocabulary, _, _ = get_license_values()
+    for key, value in license_homepages_literal_vocabulary.items():
         if unicode(vocabulary_uri) == unicode(key):
             return unicode(value)
     return None
 
 
+def get_license_homepage_uri_by_name(vocabulary_name):
+    license_homepages_literal_vocabulary, _, _ = get_license_values()
+    for key, value in license_homepages_literal_vocabulary.items():
+        if unicode(vocabulary_name) == unicode(value):
+            return key
+    return None
+
+
+def get_license_homepage_uri_by_uri(vocabulary_uri):
+    _, _, license_homepage_ref_vocabulary = get_license_values()
+    license_homepages = list(license_homepage_ref_vocabulary.keys())
+    if vocabulary_uri in license_homepages:
+        return unicode(vocabulary_uri)
+    else:
+        for key, value in license_homepage_ref_vocabulary.items():
+            if unicode(vocabulary_uri) == unicode(value):
+                return unicode(key)
+    return
+
+
 def get_license_values():
     g = Graph()
-    license_mapping = {}
+    license_ref_literal_mapping = {}
+    license_homepages_literal_mapping = {}
+    license_homepage_ref_mapping = {}
+
     for prefix, namespace in license_namespaces.items():
         g.bind(prefix, namespace)
     file = os.path.join(__location__, 'license.ttl')
     g.parse(file, format='turtle')
     for ogdch_license_ref in g.subjects(predicate=RDF.type,
                                         object=SKOS.Concept):
-        license_mapping[ogdch_license_ref] = None
-        for license_pref_label in g.objects(subject=ogdch_license_ref,
-                                            predicate=SKOSXL.prefLabel):
-            for license_literal in g.objects(subject=license_pref_label,
-                                             predicate=SKOSXL.literalForm):
-                license_mapping[ogdch_license_ref] = license_literal
-    return license_mapping
+        license_homepage = None
+        for homepage in g.objects(subject=ogdch_license_ref,
+                                  predicate=FOAF.homepage):
+            license_homepage = homepage
+            break  # Assume one homepage per concept
+
+        license_literal = None
+        try:
+            for license_pref_label in g.objects(subject=ogdch_license_ref,
+                                                predicate=SKOSXL.prefLabel):
+                for literal in g.objects(subject=license_pref_label,
+                                         predicate=SKOSXL.literalForm):
+                    license_literal = literal
+                    break  # Assume one literal per concept
+
+            license_homepages_literal_mapping[license_homepage] = license_literal  # noqa
+            license_ref_literal_mapping[ogdch_license_ref] = license_literal
+            license_homepage_ref_mapping[license_homepage] = ogdch_license_ref
+
+        except Exception as e:
+            raise ValueError("SKOSXL.prefLabel is missing in the RDF-file: %s"
+                             % e)
+
+    return (license_homepages_literal_mapping, license_ref_literal_mapping,
+            license_homepage_ref_mapping)
 
 
 def get_theme_mapping():
diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 16accca..b64234f 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -277,23 +277,13 @@ def _get_iana_media_type(self, subject):
             if media_type_key in valid_media_types:
                 return media_type_key
 
-    def _license_rights_name(self, subject, predicate):
-        for node in self.g.objects(subject, predicate):
-            # DCAT-AP CH v1: the license as a literal (should be
-            # the code for one of the DCAT-AP CH licenses)
-            if isinstance(node, Literal):
-                return unicode(node)
-            if isinstance(node, URIRef):
-                return dh.get_license_name_by_uri(node)
-        return None
-
-    def _license_rights_uri(self, subject, predicate):
+    def _license_rights_homepage_uri(self, subject, predicate):
         for node in self.g.objects(subject, predicate):
             # DCAT-AP CH v2 compatible license has to be a URI.
             if isinstance(node, Literal):
-                return dh.get_license_uri_by_name(node)
+                return dh.get_license_homepage_uri_by_name(node)
             if isinstance(node, URIRef):
-                return node
+                return dh.get_license_homepage_uri_by_uri(node)
         return None
 
     def _keywords(self, subject):
@@ -633,21 +623,28 @@ def parse_dataset(self, dataset_dict, dataset_ref):  # noqa
                 if value:
                     resource_dict[key] = value
 
-            #  Rights & License save name
-            rights = self._license_rights_name(distribution, DCT.rights)
-            license = self._license_rights_name(distribution, DCT.license)
+            #  Rights & License save homepage uri
+            rights = self._license_rights_homepage_uri(distribution, DCT.rights)
+            license = self._license_rights_homepage_uri(distribution, DCT.license)
+
             if rights is None and license is not None:
                 resource_dict['license'] = license
                 resource_dict['rights'] = license
-            if rights is not None and license is None:
-                resource_dict['license'] = rights
+            elif rights is not None and license is None:
                 resource_dict['rights'] = rights
-            if license is not None and rights is not None:
+                if 'cc' not in rights:
+                    resource_dict['license'] = rights
+                else:
+                    resource_dict['license'] = None
+            elif license is not None and rights is not None:
                 resource_dict['license'] = license
                 resource_dict['rights'] = rights
-                if 'cc' in rights:
+                if 'cc' in license and 'cc' not in rights:
                     resource_dict['license'] = rights
                     resource_dict['rights'] = license
+            else:
+                resource_dict['license'] = None
+                resource_dict['rights'] = None
 
             # Format & Media type
             resource_dict['format'] = \

From c5263bfe500e76afc1d565b815ad3fc197b27781 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 29 Jul 2024 16:40:27 +0200
Subject: [PATCH 02/19] fix: Update the dataset to graph logic

---
 ckanext/dcatapchharvest/dcat_helpers.py |  2 +-
 ckanext/dcatapchharvest/profiles.py     | 18 ++++++++----------
 2 files changed, 9 insertions(+), 11 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 9312cd5..bc57608 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -13,11 +13,11 @@
 DCT = Namespace("http://purl.org/dc/terms/")
 EUTHEMES = \
     Namespace("http://publications.europa.eu/resource/authority/data-theme/")
+FOAF = Namespace("http://xmlns.com/foaf/0.1/") # noqa
 HYDRA = Namespace('http://www.w3.org/ns/hydra/core#')
 
 SKOSXL = Namespace("http://www.w3.org/2008/05/skos-xl#")
 RDFS = Namespace("http://www.w3.org/2000/01/rdf-schema#")
-FOAF = Namespace("http://xmlns.com/foaf/0.1/")
 
 frequency_namespaces = {
   "skos": SKOS,
diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index b64234f..040d561 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -1035,7 +1035,7 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
     def _rights_and_license_to_graph(self, resource_dict, distribution):
         g = self.g
         if resource_dict.get('rights'):
-            rights_uri = dh.get_license_uri_by_name(
+            rights_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('rights')
             )
             if rights_uri is not None:
@@ -1043,13 +1043,12 @@ def _rights_and_license_to_graph(self, resource_dict, distribution):
                 g.add((rights_ref, RDF.type, DCT.RightsStatement))
                 g.add((distribution, DCT.rights, rights_ref))
             if rights_uri is None:
-                rights_name = dh.get_license_name_by_uri(
+                rights_name = dh.get_license_name_by_homepage_uri(
                     resource_dict.get('rights')
                     )
                 if rights_name is not None:
-                    resource_rights_ref = URIRef(
-                        resource_dict.get('rights')
-                        )
+                    resource_rights_ref = dh.get_license_ref_uri_by_name(
+                        rights_name)
                     g.add((
                         resource_rights_ref,
                         RDF.type,
@@ -1058,7 +1057,7 @@ def _rights_and_license_to_graph(self, resource_dict, distribution):
                     g.add((distribution, DCT.rights, resource_rights_ref))
 
         if resource_dict.get('license'):
-            license_uri = dh.get_license_uri_by_name(
+            license_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('license')
             )
             if license_uri is not None:
@@ -1066,13 +1065,12 @@ def _rights_and_license_to_graph(self, resource_dict, distribution):
                 g.add((license_ref, RDF.type, DCT.LicenseDocument))
                 g.add((distribution, DCT.license, license_ref))
             if license_uri is None:
-                license_name = dh.get_license_name_by_uri(
+                license_name = dh.get_license_name_by_homepage_uri(
                     resource_dict.get('license')
                     )
                 if license_name is not None:
-                    resource_license_ref = URIRef(
-                        resource_dict.get('license')
-                        )
+                    resource_license_ref = dh.get_license_ref_uri_by_name(
+                        license_name)
                     g.add((
                         resource_license_ref,
                         RDF.type,

From 0a114cec3fbbb3d062367e129b094118709ce0e9 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 29 Jul 2024 20:08:33 +0200
Subject: [PATCH 03/19] fix: Update tests tu use uri and not str as a value for
 license and rights

---
 ckanext/dcatapchharvest/dcat_helpers.py              |  2 +-
 ckanext/dcatapchharvest/profiles.py                  |  8 ++++++--
 .../dcatapchharvest/tests/test_dcatap_ch_parse.py    | 10 +++++-----
 .../tests/test_dcatap_ch_serialize.py                | 12 ++++++------
 4 files changed, 18 insertions(+), 14 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index bc57608..62292b2 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -13,7 +13,7 @@
 DCT = Namespace("http://purl.org/dc/terms/")
 EUTHEMES = \
     Namespace("http://publications.europa.eu/resource/authority/data-theme/")
-FOAF = Namespace("http://xmlns.com/foaf/0.1/") # noqa
+FOAF = Namespace("http://xmlns.com/foaf/0.1/")
 HYDRA = Namespace('http://www.w3.org/ns/hydra/core#')
 
 SKOSXL = Namespace("http://www.w3.org/2008/05/skos-xl#")
diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 040d561..6f030c5 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -624,8 +624,12 @@ def parse_dataset(self, dataset_dict, dataset_ref):  # noqa
                     resource_dict[key] = value
 
             #  Rights & License save homepage uri
-            rights = self._license_rights_homepage_uri(distribution, DCT.rights)
-            license = self._license_rights_homepage_uri(distribution, DCT.license)
+            rights = self._license_rights_homepage_uri(
+                distribution, DCT.rights
+            )
+            license = self._license_rights_homepage_uri(
+                distribution, DCT.license
+            )
 
             if rights is None and license is not None:
                 resource_dict['license'] = license
diff --git a/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py b/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
index 26de93d..d729b38 100644
--- a/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
+++ b/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
@@ -32,8 +32,8 @@ def test_rights_license(self):
              # Resources
             eq_(len(dataset['resources']), 1)
             resource = dataset['resources'][0]
-            eq_(resource['rights'], u'NonCommercialAllowed-CommercialAllowed-ReferenceRequired')
-            eq_(resource['license'], u'NonCommercialAllowed-CommercialWithPermission-ReferenceRequired')
+            eq_(unicode(resource['rights']), u'https://opendata.swiss/en/terms-of-use/#terms_by')
+            eq_(unicode(resource['license']), u'https://opendata.swiss/en/terms-of-use/#terms_by_ask')
 
     def test_dataset_all_fields(self):
 
@@ -146,8 +146,8 @@ def test_dataset_all_fields(self):
         eq_(resource['format'], u'html')
         eq_(resource['media_type'], u'text/html')
         eq_(resource['identifier'], u'346265-fr@bundesamt-fur-statistik-bfs')
-        eq_(resource['rights'], u'NonCommercialAllowed-CommercialAllowed-ReferenceRequired')
-        eq_(resource['license'], u'Creative Commons Zero 1.0 Universal (CC0 1.0)')
+        eq_(resource['license'], u'https://opendata.swiss/en/terms-of-use/#terms_by')
+        eq_(resource['rights'], u'http://www.opendefinition.org/licenses/cc-zero')
         eq_(resource['language'], [u'fr'])
         eq_(resource['issued'], u'1900-12-31T00:00:00')
         eq_(resource['temporal_resolution'], u'P1D')
@@ -402,7 +402,7 @@ def test_multiple_rights_statements(self):
         dataset = [d for d in p.datasets()][0]
         resource = dataset["resources"][0]
 
-        eq_(resource['rights'], u"NonCommercialAllowed-CommercialWithPermission-ReferenceRequired")
+        eq_(unicode(resource['rights']), u"https://opendata.swiss/en/terms-of-use/#terms_by_ask")
 
     def test_eu_themes_mapping(self):
         contents = self._get_file_contents('catalog-themes.xml')
diff --git a/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py b/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
index f686931..e8cc47e 100644
--- a/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
+++ b/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
@@ -113,22 +113,22 @@ def test_graph_from_dataset(self):
             if resource_dict.get('rights') == 'Creative Commons Zero 1.0 Universal (CC0 1.0)':
                 assert self._triple(g, distribution, DCT.rights, URIRef("https://creativecommons.org/publicdomain/zero/1.0/"))
 
-            if resource_dict.get('license') == 'NonCommercialAllowed-CommercialAllowed-ReferenceNotRequired':
+            if resource_dict.get('license') == 'https://opendata.swiss/terms-of-use/#terms_open':
                 assert self._triple(g, distribution, DCT.license, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_open"))
 
             # 28e75e40-e1a1-497b-a1b9-8c1834d60201
-            if resource_dict.get('rights') == "http://dcat-ap.ch/vocabulary/licenses/terms_by":
+            if resource_dict.get('rights') == "https://opendata.swiss/terms-of-use#terms_by":
                 assert self._triple(g, distribution, DCT.rights, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_by"))
 
-            if resource_dict.get('license') == "NonCommercialAllowed-CommercialAllowed-ReferenceRequired":
+            if resource_dict.get('license') == "https://opendata.swiss/terms-of-use#terms_by":
                 assert self._triple(g, distribution, DCT.license, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_by"))
 
             # 0cfce6ba-28f4-4229-b733-f6492c650395
-            if resource_dict.get('rights') == "http://dcat-ap.ch/vocabulary/licenses/terms_by_ask":
+            if resource_dict.get('rights') == "https://opendata.swiss/terms-of-use#terms_by_ask":
                 assert self._triple(g, distribution, DCT.rights, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_by_ask"))
 
-            if resource_dict.get('license') == "https://creativecommons.org/licenses/by/4.0/":
-                assert self._triple(g, distribution, DCT.license, URIRef("https://creativecommons.org/licenses/by/4.0/"))
+            if resource_dict.get('rights') == "	http://www.opendefinition.org/licenses/cc-by/":
+                assert self._triple(g, distribution, DCT.rights, URIRef("https://creativecommons.org/licenses/by/4.0/"))
 
             if resource_dict.get('format') == "CSV":
                 assert self._triple(g, distribution, DCT['format'], URIRef("http://publications.europa.eu/resource/authority/file-type/CSV"))

From 14fd57f79d5d9ca679f6b1e1e55aeab5cca26ddc Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 29 Jul 2024 20:13:18 +0200
Subject: [PATCH 04/19] fix: Remove unsuded Namespace

---
 ckanext/dcatapchharvest/dcat_helpers.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 62292b2..992b758 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -4,7 +4,7 @@
 from urlparse import urlparse
 from ckantoolkit import config
 from rdflib import URIRef, Graph
-from rdflib.namespace import Namespace, RDF, SKOS, FOAF
+from rdflib.namespace import Namespace, RDF, SKOS
 import xml.etree.ElementTree as ET
 import logging
 

From c6b7d567b0478588c3ed07aceb78aec4ec94a8ee Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Tue, 30 Jul 2024 10:17:11 +0200
Subject: [PATCH 05/19] fix: License and Rights could be just a homepage URI in
 our dataset

---
 ckanext/dcatapchharvest/profiles.py | 49 ++++++-----------------------
 1 file changed, 10 insertions(+), 39 deletions(-)

diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 6f030c5..4aa79d9 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -279,7 +279,7 @@ def _get_iana_media_type(self, subject):
 
     def _license_rights_homepage_uri(self, subject, predicate):
         for node in self.g.objects(subject, predicate):
-            # DCAT-AP CH v2 compatible license has to be a URI.
+            # Rights and license has to be a homepage URI
             if isinstance(node, Literal):
                 return dh.get_license_homepage_uri_by_name(node)
             if isinstance(node, URIRef):
@@ -1039,50 +1039,21 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
     def _rights_and_license_to_graph(self, resource_dict, distribution):
         g = self.g
         if resource_dict.get('rights'):
-            rights_uri = dh.get_license_ref_uri_by_homepage_uri(
+            rights_ref_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('rights')
             )
-            if rights_uri is not None:
-                rights_ref = URIRef(rights_uri)
-                g.add((rights_ref, RDF.type, DCT.RightsStatement))
-                g.add((distribution, DCT.rights, rights_ref))
-            if rights_uri is None:
-                rights_name = dh.get_license_name_by_homepage_uri(
-                    resource_dict.get('rights')
-                    )
-                if rights_name is not None:
-                    resource_rights_ref = dh.get_license_ref_uri_by_name(
-                        rights_name)
-                    g.add((
-                        resource_rights_ref,
-                        RDF.type,
-                        DCT.RightsStatement)
-                        )
-                    g.add((distribution, DCT.rights, resource_rights_ref))
+            rights_ref = URIRef(rights_ref_uri)
+            g.add((rights_ref, RDF.type, DCT.RightsStatement))
+            g.add((distribution, DCT.rights, rights_ref))
 
         if resource_dict.get('license'):
-            license_uri = dh.get_license_ref_uri_by_homepage_uri(
+            license_ref_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('license')
             )
-            if license_uri is not None:
-                license_ref = URIRef(license_uri)
-                g.add((license_ref, RDF.type, DCT.LicenseDocument))
-                g.add((distribution, DCT.license, license_ref))
-            if license_uri is None:
-                license_name = dh.get_license_name_by_homepage_uri(
-                    resource_dict.get('license')
-                    )
-                if license_name is not None:
-                    resource_license_ref = dh.get_license_ref_uri_by_name(
-                        license_name)
-                    g.add((
-                        resource_license_ref,
-                        RDF.type,
-                        DCT.LicenseDocument)
-                        )
-                    g.add(
-                        (distribution, DCT.license, resource_license_ref)
-                        )
+            license_ref = URIRef(license_ref_uri)
+            g.add((license_ref, RDF.type, DCT.LicenseDocument))
+            g.add((distribution, DCT.license, license_ref))
+
 
     def _format_and_media_type_to_graph(self, resource_dict, distribution):
         g = self.g

From d54a75a74ad18ef2b9fa94355165633e1e5922f8 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Tue, 30 Jul 2024 10:19:18 +0200
Subject: [PATCH 06/19] style: Remove blank line

---
 ckanext/dcatapchharvest/profiles.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 4aa79d9..458de82 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -1054,7 +1054,6 @@ def _rights_and_license_to_graph(self, resource_dict, distribution):
             g.add((license_ref, RDF.type, DCT.LicenseDocument))
             g.add((distribution, DCT.license, license_ref))
 
-
     def _format_and_media_type_to_graph(self, resource_dict, distribution):
         g = self.g
         # Export format value if it matches EU vocabulary

From eec20dd4f296effc46b5452bf3e735ecce9d83b0 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Tue, 30 Jul 2024 10:44:13 +0200
Subject: [PATCH 07/19] fix: Rollback to the previous function look, before we
 migrate all licenses to uri

---
 ckanext/dcatapchharvest/profiles.py | 48 +++++++++++++++++++++++------
 1 file changed, 39 insertions(+), 9 deletions(-)

diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 458de82..6f030c5 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -279,7 +279,7 @@ def _get_iana_media_type(self, subject):
 
     def _license_rights_homepage_uri(self, subject, predicate):
         for node in self.g.objects(subject, predicate):
-            # Rights and license has to be a homepage URI
+            # DCAT-AP CH v2 compatible license has to be a URI.
             if isinstance(node, Literal):
                 return dh.get_license_homepage_uri_by_name(node)
             if isinstance(node, URIRef):
@@ -1039,20 +1039,50 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
     def _rights_and_license_to_graph(self, resource_dict, distribution):
         g = self.g
         if resource_dict.get('rights'):
-            rights_ref_uri = dh.get_license_ref_uri_by_homepage_uri(
+            rights_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('rights')
             )
-            rights_ref = URIRef(rights_ref_uri)
-            g.add((rights_ref, RDF.type, DCT.RightsStatement))
-            g.add((distribution, DCT.rights, rights_ref))
+            if rights_uri is not None:
+                rights_ref = URIRef(rights_uri)
+                g.add((rights_ref, RDF.type, DCT.RightsStatement))
+                g.add((distribution, DCT.rights, rights_ref))
+            if rights_uri is None:
+                rights_name = dh.get_license_name_by_homepage_uri(
+                    resource_dict.get('rights')
+                    )
+                if rights_name is not None:
+                    resource_rights_ref = dh.get_license_ref_uri_by_name(
+                        rights_name)
+                    g.add((
+                        resource_rights_ref,
+                        RDF.type,
+                        DCT.RightsStatement)
+                        )
+                    g.add((distribution, DCT.rights, resource_rights_ref))
 
         if resource_dict.get('license'):
-            license_ref_uri = dh.get_license_ref_uri_by_homepage_uri(
+            license_uri = dh.get_license_ref_uri_by_homepage_uri(
                 resource_dict.get('license')
             )
-            license_ref = URIRef(license_ref_uri)
-            g.add((license_ref, RDF.type, DCT.LicenseDocument))
-            g.add((distribution, DCT.license, license_ref))
+            if license_uri is not None:
+                license_ref = URIRef(license_uri)
+                g.add((license_ref, RDF.type, DCT.LicenseDocument))
+                g.add((distribution, DCT.license, license_ref))
+            if license_uri is None:
+                license_name = dh.get_license_name_by_homepage_uri(
+                    resource_dict.get('license')
+                    )
+                if license_name is not None:
+                    resource_license_ref = dh.get_license_ref_uri_by_name(
+                        license_name)
+                    g.add((
+                        resource_license_ref,
+                        RDF.type,
+                        DCT.LicenseDocument)
+                        )
+                    g.add(
+                        (distribution, DCT.license, resource_license_ref)
+                        )
 
     def _format_and_media_type_to_graph(self, resource_dict, distribution):
         g = self.g

From b8643298bae509c7744435f525e2c9ea823a402b Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Tue, 30 Jul 2024 10:45:02 +0200
Subject: [PATCH 08/19] fix: Update license and right values for the tests

---
 ckanext/dcatapchharvest/tests/fixtures/dataset.json       | 6 +++---
 ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/ckanext/dcatapchharvest/tests/fixtures/dataset.json b/ckanext/dcatapchharvest/tests/fixtures/dataset.json
index cf863af..7bc5cbc 100644
--- a/ckanext/dcatapchharvest/tests/fixtures/dataset.json
+++ b/ckanext/dcatapchharvest/tests/fixtures/dataset.json
@@ -121,8 +121,8 @@
         "https://example.com/documentation-resource-1",
         "https://example.com/documentation-resource-2"
       ],
-      "rights": "Creative Commons Zero 1.0 Universell (CC0 1.0)",
-      "license": "NonCommercialAllowed-CommercialAllowed-ReferenceNotRequired",
+      "rights": "http://www.opendefinition.org/licenses/cc-zero",
+      "license": "https://opendata.swiss/terms-of-use/#terms_open",
       "format": "CSV",
       "issued": "2015-06-26T15:21:09.034694",
       "modified": "2015-06-30T15:21:09.000000"
@@ -135,7 +135,7 @@
         "https://example.com/documentation-resource-2"
       ],
       "rights": "http://dcat-ap.ch/vocabulary/licenses/terms_by",
-      "license": "NonCommercialAllowed-CommercialAllowed-ReferenceRequired",
+      "license": "https://opendata.swiss/en/terms-of-use/#terms_by",
       "format": "HTML"
     },
     {
diff --git a/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py b/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
index e8cc47e..333936b 100644
--- a/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
+++ b/ckanext/dcatapchharvest/tests/test_dcatap_ch_serialize.py
@@ -110,10 +110,10 @@ def test_graph_from_dataset(self):
                 assert self._triple(g, distribution, DCAT.accessService, URIRef(link))
 
             # e2c50e70-67ad-4f86-bb1b-3f93867eadaa
-            if resource_dict.get('rights') == 'Creative Commons Zero 1.0 Universal (CC0 1.0)':
+            if resource_dict.get('rights') == "http://www.opendefinition.org/licenses/cc-zero":
                 assert self._triple(g, distribution, DCT.rights, URIRef("https://creativecommons.org/publicdomain/zero/1.0/"))
 
-            if resource_dict.get('license') == 'https://opendata.swiss/terms-of-use/#terms_open':
+            if resource_dict.get('license') == "https://opendata.swiss/terms-of-use/#terms_open":
                 assert self._triple(g, distribution, DCT.license, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_open"))
 
             # 28e75e40-e1a1-497b-a1b9-8c1834d60201
@@ -127,7 +127,7 @@ def test_graph_from_dataset(self):
             if resource_dict.get('rights') == "https://opendata.swiss/terms-of-use#terms_by_ask":
                 assert self._triple(g, distribution, DCT.rights, URIRef("http://dcat-ap.ch/vocabulary/licenses/terms_by_ask"))
 
-            if resource_dict.get('rights') == "	http://www.opendefinition.org/licenses/cc-by/":
+            if resource_dict.get('rights') == "http://www.opendefinition.org/licenses/cc-by/":
                 assert self._triple(g, distribution, DCT.rights, URIRef("https://creativecommons.org/licenses/by/4.0/"))
 
             if resource_dict.get('format') == "CSV":

From 71e0e7e90971a1412f38db32d33bf441a7765621 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 09:49:20 +0200
Subject: [PATCH 09/19] feat: Add additional check if cc-license in dct:license

---
 ckanext/dcatapchharvest/profiles.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 6f030c5..103791d 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -646,6 +646,8 @@ def parse_dataset(self, dataset_dict, dataset_ref):  # noqa
                 if 'cc' in license and 'cc' not in rights:
                     resource_dict['license'] = rights
                     resource_dict['rights'] = license
+                elif 'cc' in license and 'cc' in rights:
+                    resource_dict['license'] = None
             else:
                 resource_dict['license'] = None
                 resource_dict['rights'] = None

From 67b9ae83614131d1c01e32c5a786798349beffd5 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 09:57:58 +0200
Subject: [PATCH 10/19] fix: Update the license_literal mapping

---
 ckanext/dcatapchharvest/dcat_helpers.py | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 992b758..444d594 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -243,11 +243,10 @@ def get_license_values():
         try:
             for license_pref_label in g.objects(subject=ogdch_license_ref,
                                                 predicate=SKOSXL.prefLabel):
-                for literal in g.objects(subject=license_pref_label,
-                                         predicate=SKOSXL.literalForm):
-                    license_literal = literal
+                license_literal = next(g.objects(subject=license_pref_label,
+                                                 predicate=SKOSXL.literalForm))
+                if license_literal is not None:
                     break  # Assume one literal per concept
-
             license_homepages_literal_mapping[license_homepage] = license_literal  # noqa
             license_ref_literal_mapping[ogdch_license_ref] = license_literal
             license_homepage_ref_mapping[license_homepage] = ogdch_license_ref

From 46f7d842a1ec77461df35e41f05aa89fca0e5d51 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 10:24:20 +0200
Subject: [PATCH 11/19] fix: Restracture _rights_license to graph function

---
 ckanext/dcatapchharvest/profiles.py | 75 ++++++++++++-----------------
 1 file changed, 31 insertions(+), 44 deletions(-)

diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 103791d..6087d26 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -1038,53 +1038,40 @@ def graph_from_dataset(self, dataset_dict, dataset_ref):  # noqa
                 g.add((distribution, DCAT.byteSize,
                        Literal(resource_dict['byte_size'])))
 
+    def _get_rights_and_license_uri(self, resource_dict, property='license'):
+        if property not in ['license', 'rights']:
+            raise ValueError("Property must be 'license' or 'rights'")
+
+        homepage_uri = resource_dict.get(property)
+        if not homepage_uri:
+            return None
+
+        uri = dh.get_license_ref_uri_by_homepage_uri(homepage_uri)
+        if uri is not None:
+            return URIRef(uri)
+
+        name = dh.get_license_name_by_homepage_uri(homepage_uri)
+        if name is not None:
+            uri = dh.get_license_ref_uri_by_name(name)
+            if uri is not None:
+                return URIRef(uri)
+
+        return None
+
     def _rights_and_license_to_graph(self, resource_dict, distribution):
         g = self.g
-        if resource_dict.get('rights'):
-            rights_uri = dh.get_license_ref_uri_by_homepage_uri(
-                resource_dict.get('rights')
-            )
-            if rights_uri is not None:
-                rights_ref = URIRef(rights_uri)
-                g.add((rights_ref, RDF.type, DCT.RightsStatement))
-                g.add((distribution, DCT.rights, rights_ref))
-            if rights_uri is None:
-                rights_name = dh.get_license_name_by_homepage_uri(
-                    resource_dict.get('rights')
-                    )
-                if rights_name is not None:
-                    resource_rights_ref = dh.get_license_ref_uri_by_name(
-                        rights_name)
-                    g.add((
-                        resource_rights_ref,
-                        RDF.type,
-                        DCT.RightsStatement)
-                        )
-                    g.add((distribution, DCT.rights, resource_rights_ref))
 
-        if resource_dict.get('license'):
-            license_uri = dh.get_license_ref_uri_by_homepage_uri(
-                resource_dict.get('license')
-            )
-            if license_uri is not None:
-                license_ref = URIRef(license_uri)
-                g.add((license_ref, RDF.type, DCT.LicenseDocument))
-                g.add((distribution, DCT.license, license_ref))
-            if license_uri is None:
-                license_name = dh.get_license_name_by_homepage_uri(
-                    resource_dict.get('license')
-                    )
-                if license_name is not None:
-                    resource_license_ref = dh.get_license_ref_uri_by_name(
-                        license_name)
-                    g.add((
-                        resource_license_ref,
-                        RDF.type,
-                        DCT.LicenseDocument)
-                        )
-                    g.add(
-                        (distribution, DCT.license, resource_license_ref)
-                        )
+        rights_uri_ref = self._get_rights_and_license_uri(resource_dict,
+                                                          'rights')
+        if rights_uri_ref is not None:
+            g.add((rights_uri_ref, RDF.type, DCT.RightsStatement))
+            g.add((distribution, DCT.rights, rights_uri_ref))
+
+        license_uri_ref = self._get_rights_and_license_uri(resource_dict,
+                                                           'license')
+        if license_uri_ref is not None:
+            g.add((license_uri_ref, RDF.type, DCT.LicenseDocument))
+            g.add((distribution, DCT.license, license_uri_ref))
 
     def _format_and_media_type_to_graph(self, resource_dict, distribution):
         g = self.g

From df83d15aa38f4c30dfb1978deafa32818c0f689c Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 10:35:50 +0200
Subject: [PATCH 12/19] fix: Set keys and values to unicode when setting up the
 mappings

---
 ckanext/dcatapchharvest/dcat_helpers.py | 32 ++++++++++++++-----------
 1 file changed, 18 insertions(+), 14 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 444d594..a99cbc1 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -172,7 +172,7 @@ def get_frequency_values():
 def get_license_ref_uri_by_name(vocabulary_name):
     _, license_ref_literal_vocabulary, _ = get_license_values()
     for key, value in license_ref_literal_vocabulary.items():
-        if unicode(vocabulary_name) == unicode(value):
+        if vocabulary_name == value:
             return key
     return None
 
@@ -180,7 +180,7 @@ def get_license_ref_uri_by_name(vocabulary_name):
 def get_license_ref_uri_by_homepage_uri(vocabulary_name):
     _, _, license_homepage_ref_vocabulary = get_license_values()
     for key, value in license_homepage_ref_vocabulary.items():
-        if unicode(vocabulary_name) == unicode(key):
+        if vocabulary_name == key:
             return value
     return None
 
@@ -188,23 +188,23 @@ def get_license_ref_uri_by_homepage_uri(vocabulary_name):
 def get_license_name_by_ref_uri(vocabulary_uri):
     _, license_ref_literal_vocabulary, _ = get_license_values()
     for key, value in license_ref_literal_vocabulary.items():
-        if unicode(vocabulary_uri) == unicode(key):
-            return unicode(value)
+        if vocabulary_uri == key:
+            return value
     return None
 
 
 def get_license_name_by_homepage_uri(vocabulary_uri):
     license_homepages_literal_vocabulary, _, _ = get_license_values()
     for key, value in license_homepages_literal_vocabulary.items():
-        if unicode(vocabulary_uri) == unicode(key):
-            return unicode(value)
+        if vocabulary_uri == key:
+            return value
     return None
 
 
 def get_license_homepage_uri_by_name(vocabulary_name):
     license_homepages_literal_vocabulary, _, _ = get_license_values()
     for key, value in license_homepages_literal_vocabulary.items():
-        if unicode(vocabulary_name) == unicode(value):
+        if vocabulary_name == value:
             return key
     return None
 
@@ -213,12 +213,12 @@ def get_license_homepage_uri_by_uri(vocabulary_uri):
     _, _, license_homepage_ref_vocabulary = get_license_values()
     license_homepages = list(license_homepage_ref_vocabulary.keys())
     if vocabulary_uri in license_homepages:
-        return unicode(vocabulary_uri)
+        return vocabulary_uri
     else:
         for key, value in license_homepage_ref_vocabulary.items():
-            if unicode(vocabulary_uri) == unicode(value):
-                return unicode(key)
-    return
+            if vocabulary_uri == value:
+                return key
+    return None
 
 
 def get_license_values():
@@ -247,9 +247,13 @@ def get_license_values():
                                                  predicate=SKOSXL.literalForm))
                 if license_literal is not None:
                     break  # Assume one literal per concept
-            license_homepages_literal_mapping[license_homepage] = license_literal  # noqa
-            license_ref_literal_mapping[ogdch_license_ref] = license_literal
-            license_homepage_ref_mapping[license_homepage] = ogdch_license_ref
+
+            license_homepages_literal_mapping[unicode(license_homepage)] = \
+                unicode(license_literal)
+            license_ref_literal_mapping[unicode(ogdch_license_ref)] = \
+                unicode(license_literal)
+            license_homepage_ref_mapping[unicode(license_homepage)] = \
+                unicode(ogdch_license_ref)
 
         except Exception as e:
             raise ValueError("SKOSXL.prefLabel is missing in the RDF-file: %s"

From 4c2ac834c5bd0255d0597982c1fc6c06a3dc70ba Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 12:43:50 +0200
Subject: [PATCH 13/19] fix: Make vocabulary_uri to unicode for the comparison
 in helper functions

---
 ckanext/dcatapchharvest/dcat_helpers.py | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index a99cbc1..8e4b263 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -172,7 +172,7 @@ def get_frequency_values():
 def get_license_ref_uri_by_name(vocabulary_name):
     _, license_ref_literal_vocabulary, _ = get_license_values()
     for key, value in license_ref_literal_vocabulary.items():
-        if vocabulary_name == value:
+        if unicode(vocabulary_name) == value:
             return key
     return None
 
@@ -180,7 +180,7 @@ def get_license_ref_uri_by_name(vocabulary_name):
 def get_license_ref_uri_by_homepage_uri(vocabulary_name):
     _, _, license_homepage_ref_vocabulary = get_license_values()
     for key, value in license_homepage_ref_vocabulary.items():
-        if vocabulary_name == key:
+        if unicode(vocabulary_name) == key:
             return value
     return None
 
@@ -188,7 +188,7 @@ def get_license_ref_uri_by_homepage_uri(vocabulary_name):
 def get_license_name_by_ref_uri(vocabulary_uri):
     _, license_ref_literal_vocabulary, _ = get_license_values()
     for key, value in license_ref_literal_vocabulary.items():
-        if vocabulary_uri == key:
+        if unicode(vocabulary_uri) == key:
             return value
     return None
 
@@ -196,7 +196,7 @@ def get_license_name_by_ref_uri(vocabulary_uri):
 def get_license_name_by_homepage_uri(vocabulary_uri):
     license_homepages_literal_vocabulary, _, _ = get_license_values()
     for key, value in license_homepages_literal_vocabulary.items():
-        if vocabulary_uri == key:
+        if unicode(vocabulary_uri) == key:
             return value
     return None
 
@@ -204,7 +204,7 @@ def get_license_name_by_homepage_uri(vocabulary_uri):
 def get_license_homepage_uri_by_name(vocabulary_name):
     license_homepages_literal_vocabulary, _, _ = get_license_values()
     for key, value in license_homepages_literal_vocabulary.items():
-        if vocabulary_name == value:
+        if unicode(vocabulary_name) == value:
             return key
     return None
 
@@ -212,11 +212,11 @@ def get_license_homepage_uri_by_name(vocabulary_name):
 def get_license_homepage_uri_by_uri(vocabulary_uri):
     _, _, license_homepage_ref_vocabulary = get_license_values()
     license_homepages = list(license_homepage_ref_vocabulary.keys())
-    if vocabulary_uri in license_homepages:
-        return vocabulary_uri
+    if unicode(vocabulary_uri) in license_homepages:
+        return unicode(vocabulary_uri)
     else:
         for key, value in license_homepage_ref_vocabulary.items():
-            if vocabulary_uri == value:
+            if unicode(vocabulary_uri) == value:
                 return key
     return None
 

From ab73216844bbefe91a0536c36a513699913e1fec Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 13:34:15 +0200
Subject: [PATCH 14/19] refactor: Add LicenseHandler class to handle cashing of
 license vocabulary

---
 ckanext/dcatapchharvest/dcat_helpers.py | 149 ++++++++++++++++--------
 ckanext/dcatapchharvest/profiles.py     |  13 +--
 2 files changed, 107 insertions(+), 55 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 8e4b263..c5785d4 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -169,56 +169,109 @@ def get_frequency_values():
     return frequency_mapping
 
 
-def get_license_ref_uri_by_name(vocabulary_name):
-    _, license_ref_literal_vocabulary, _ = get_license_values()
-    for key, value in license_ref_literal_vocabulary.items():
-        if unicode(vocabulary_name) == value:
-            return key
-    return None
-
-
-def get_license_ref_uri_by_homepage_uri(vocabulary_name):
-    _, _, license_homepage_ref_vocabulary = get_license_values()
-    for key, value in license_homepage_ref_vocabulary.items():
-        if unicode(vocabulary_name) == key:
-            return value
-    return None
-
-
-def get_license_name_by_ref_uri(vocabulary_uri):
-    _, license_ref_literal_vocabulary, _ = get_license_values()
-    for key, value in license_ref_literal_vocabulary.items():
-        if unicode(vocabulary_uri) == key:
-            return value
-    return None
-
-
-def get_license_name_by_homepage_uri(vocabulary_uri):
-    license_homepages_literal_vocabulary, _, _ = get_license_values()
-    for key, value in license_homepages_literal_vocabulary.items():
-        if unicode(vocabulary_uri) == key:
-            return value
-    return None
-
-
-def get_license_homepage_uri_by_name(vocabulary_name):
-    license_homepages_literal_vocabulary, _, _ = get_license_values()
-    for key, value in license_homepages_literal_vocabulary.items():
-        if unicode(vocabulary_name) == value:
-            return key
-    return None
-
-
-def get_license_homepage_uri_by_uri(vocabulary_uri):
-    _, _, license_homepage_ref_vocabulary = get_license_values()
-    license_homepages = list(license_homepage_ref_vocabulary.keys())
-    if unicode(vocabulary_uri) in license_homepages:
-        return unicode(vocabulary_uri)
-    else:
+class LicenseHandler:
+    def __init__(self):
+        self._license_cache = None
+
+    def _get_license_values(self):
+        if self._license_cache is None:
+            try:
+                g = Graph()
+                license_ref_literal_mapping = {}
+                license_homepages_literal_mapping = {}
+                license_homepage_ref_mapping = {}
+
+                for prefix, namespace in license_namespaces.items():
+                    g.bind(prefix, namespace)
+                file = os.path.join(__location__, 'license.ttl')
+                g.parse(file, format='turtle')
+                for ogdch_license_ref in g.subjects(predicate=RDF.type,
+                                                    object=SKOS.Concept):
+                    license_homepage = None
+                    for homepage in g.objects(subject=ogdch_license_ref,
+                                              predicate=FOAF.homepage):
+                        license_homepage = homepage
+                        break  # Assume one homepage per concept
+
+                    license_literal = None
+                    try:
+                        for license_pref_label in g.objects(
+                                subject=ogdch_license_ref,
+                                predicate=SKOSXL.prefLabel):
+                            license_literal = next(
+                                g.objects(subject=license_pref_label,
+                                          predicate=SKOSXL.literalForm))
+                            if license_literal is not None:
+                                break  # Assume one literal per concept
+
+                        license_homepages_literal_mapping[
+                            unicode(license_homepage)] = \
+                            unicode(license_literal)
+                        license_ref_literal_mapping[
+                            unicode(ogdch_license_ref)] = \
+                            unicode(license_literal)
+                        license_homepage_ref_mapping[
+                            unicode(license_homepage)] = \
+                            unicode(ogdch_license_ref)
+
+                    except Exception as e:
+                        raise ValueError(
+                            "SKOSXL.prefLabel is missing in the RDF-file: %s"
+                            % e)
+
+                self._license_cache = (license_homepages_literal_mapping,
+                                       license_ref_literal_mapping,
+                                       license_homepage_ref_mapping)
+            except Exception as e:
+                raise RuntimeError("Failed to load license values: %s"
+                                   % e)
+        return self._license_cache
+
+    def get_license_ref_uri_by_name(self, vocabulary_name):
+        _, license_ref_literal_vocabulary, _ = self._get_license_values()
+        for key, value in license_ref_literal_vocabulary.items():
+            if unicode(vocabulary_name) == value:
+                return key
+        return None
+
+    def get_license_ref_uri_by_homepage_uri(self, vocabulary_name):
+        _, _, license_homepage_ref_vocabulary = self._get_license_values()
         for key, value in license_homepage_ref_vocabulary.items():
-            if unicode(vocabulary_uri) == value:
+            if unicode(vocabulary_name) == key:
+                return value
+        return None
+
+    def get_license_name_by_ref_uri(self, vocabulary_uri):
+        _, license_ref_literal_vocabulary, _ = self._get_license_values()
+        for key, value in license_ref_literal_vocabulary.items():
+            if unicode(vocabulary_uri) == key:
+                return value
+        return None
+
+    def get_license_name_by_homepage_uri(self, vocabulary_uri):
+        license_homepages_literal_vocabulary, _, _ = self._get_license_values()
+        for key, value in license_homepages_literal_vocabulary.items():
+            if unicode(vocabulary_uri) == key:
+                return value
+        return None
+
+    def get_license_homepage_uri_by_name(self, vocabulary_name):
+        license_homepages_literal_vocabulary, _, _ = self._get_license_values()
+        for key, value in license_homepages_literal_vocabulary.items():
+            if unicode(vocabulary_name) == value:
                 return key
-    return None
+        return None
+
+    def get_license_homepage_uri_by_uri(self, vocabulary_uri):
+        _, _, license_homepage_ref_vocabulary = self._get_license_values()
+        license_homepages = list(license_homepage_ref_vocabulary.keys())
+        if unicode(vocabulary_uri) in license_homepages:
+            return unicode(vocabulary_uri)
+        else:
+            for key, value in license_homepage_ref_vocabulary.items():
+                if unicode(vocabulary_uri) == value:
+                    return key
+        return None
 
 
 def get_license_values():
diff --git a/ckanext/dcatapchharvest/profiles.py b/ckanext/dcatapchharvest/profiles.py
index 6087d26..335e494 100644
--- a/ckanext/dcatapchharvest/profiles.py
+++ b/ckanext/dcatapchharvest/profiles.py
@@ -13,9 +13,8 @@
 from ckanext.dcat.profiles import CleanedURIRef, RDFProfile, SchemaOrgProfile
 
 log = logging.getLogger(__name__)
-
+license_handler = dh.LicenseHandler()
 valid_frequencies = dh.get_frequency_values()
-valid_licenses = dh.get_license_values()
 eu_theme_mapping = dh.get_theme_mapping()
 valid_formats = dh.get_format_values()
 valid_media_types = dh.get_iana_media_type_values()
@@ -281,9 +280,9 @@ def _license_rights_homepage_uri(self, subject, predicate):
         for node in self.g.objects(subject, predicate):
             # DCAT-AP CH v2 compatible license has to be a URI.
             if isinstance(node, Literal):
-                return dh.get_license_homepage_uri_by_name(node)
+                return license_handler.get_license_homepage_uri_by_name(node)
             if isinstance(node, URIRef):
-                return dh.get_license_homepage_uri_by_uri(node)
+                return license_handler.get_license_homepage_uri_by_uri(node)
         return None
 
     def _keywords(self, subject):
@@ -1046,13 +1045,13 @@ def _get_rights_and_license_uri(self, resource_dict, property='license'):
         if not homepage_uri:
             return None
 
-        uri = dh.get_license_ref_uri_by_homepage_uri(homepage_uri)
+        uri = license_handler.get_license_ref_uri_by_homepage_uri(homepage_uri)
         if uri is not None:
             return URIRef(uri)
 
-        name = dh.get_license_name_by_homepage_uri(homepage_uri)
+        name = license_handler.get_license_name_by_homepage_uri(homepage_uri)
         if name is not None:
-            uri = dh.get_license_ref_uri_by_name(name)
+            uri = license_handler.get_license_ref_uri_by_name(name)
             if uri is not None:
                 return URIRef(uri)
 

From 8638c5848354d60e270f6fc2a2b15641f8f3b136 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 14:07:07 +0200
Subject: [PATCH 15/19] refactor: Devide license mapping function to smaller
 stages

---
 ckanext/dcatapchharvest/dcat_helpers.py | 142 +++++++++++++-----------
 1 file changed, 76 insertions(+), 66 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index c5785d4..17c279c 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -173,52 +173,63 @@ class LicenseHandler:
     def __init__(self):
         self._license_cache = None
 
+    def _bind_namespaces(self, graph):
+        for prefix, namespace in license_namespaces.items():
+            graph.bind(prefix, namespace)
+
+    def _parse_graph(self, graph):
+        file = os.path.join(__location__, 'license.ttl')
+        graph.parse(file, format='turtle')
+
+    def _get_license_homepage(self, graph, license_ref):
+        for homepage in graph.objects(subject=license_ref,
+                                      predicate=FOAF.homepage):
+            return homepage
+        return None
+
+    def _get_license_literal(self, graph, license_ref):
+        for license_pref_label in graph.objects(subject=license_ref,
+                                                predicate=SKOSXL.prefLabel):
+            try:
+                return next(graph.objects(subject=license_pref_label,
+                                          predicate=SKOSXL.literalForm))
+            except StopIteration:
+                continue
+        return None
+
+    def _process_graph(self, graph):
+        license_ref_literal_mapping = {}
+        license_homepages_literal_mapping = {}
+        license_homepage_ref_mapping = {}
+
+        for ogdch_license_ref in graph.subjects(predicate=RDF.type,
+                                                object=SKOS.Concept):
+            license_homepage = self._get_license_homepage(graph,
+                                                          ogdch_license_ref)
+            license_literal = self._get_license_literal(graph,
+                                                        ogdch_license_ref)
+
+            license_homepages_literal_mapping[unicode(license_homepage)] = \
+                unicode(license_literal)
+            license_ref_literal_mapping[unicode(ogdch_license_ref)] = \
+                unicode(license_literal)
+            license_homepage_ref_mapping[unicode(license_homepage)] = \
+                unicode(ogdch_license_ref)
+
+        return (license_homepages_literal_mapping,
+                license_ref_literal_mapping, license_homepage_ref_mapping)
+
+
     def _get_license_values(self):
         if self._license_cache is None:
             try:
                 g = Graph()
-                license_ref_literal_mapping = {}
-                license_homepages_literal_mapping = {}
-                license_homepage_ref_mapping = {}
-
-                for prefix, namespace in license_namespaces.items():
-                    g.bind(prefix, namespace)
-                file = os.path.join(__location__, 'license.ttl')
-                g.parse(file, format='turtle')
-                for ogdch_license_ref in g.subjects(predicate=RDF.type,
-                                                    object=SKOS.Concept):
-                    license_homepage = None
-                    for homepage in g.objects(subject=ogdch_license_ref,
-                                              predicate=FOAF.homepage):
-                        license_homepage = homepage
-                        break  # Assume one homepage per concept
-
-                    license_literal = None
-                    try:
-                        for license_pref_label in g.objects(
-                                subject=ogdch_license_ref,
-                                predicate=SKOSXL.prefLabel):
-                            license_literal = next(
-                                g.objects(subject=license_pref_label,
-                                          predicate=SKOSXL.literalForm))
-                            if license_literal is not None:
-                                break  # Assume one literal per concept
-
-                        license_homepages_literal_mapping[
-                            unicode(license_homepage)] = \
-                            unicode(license_literal)
-                        license_ref_literal_mapping[
-                            unicode(ogdch_license_ref)] = \
-                            unicode(license_literal)
-                        license_homepage_ref_mapping[
-                            unicode(license_homepage)] = \
-                            unicode(ogdch_license_ref)
-
-                    except Exception as e:
-                        raise ValueError(
-                            "SKOSXL.prefLabel is missing in the RDF-file: %s"
-                            % e)
+                self._bind_namespaces(g)
+                self._parse_graph(g)
 
+                license_homepages_literal_mapping, \
+                license_ref_literal_mapping, \
+                license_homepage_ref_mapping = self._process_graph(g)
                 self._license_cache = (license_homepages_literal_mapping,
                                        license_ref_literal_mapping,
                                        license_homepage_ref_mapping)
@@ -229,49 +240,48 @@ def _get_license_values(self):
 
     def get_license_ref_uri_by_name(self, vocabulary_name):
         _, license_ref_literal_vocabulary, _ = self._get_license_values()
-        for key, value in license_ref_literal_vocabulary.items():
-            if unicode(vocabulary_name) == value:
-                return key
-        return None
+        return next((key for key, value in
+                     license_ref_literal_vocabulary.items()
+                     if unicode(vocabulary_name) == value),
+                    None)
 
     def get_license_ref_uri_by_homepage_uri(self, vocabulary_name):
         _, _, license_homepage_ref_vocabulary = self._get_license_values()
-        for key, value in license_homepage_ref_vocabulary.items():
-            if unicode(vocabulary_name) == key:
-                return value
-        return None
+        return next((value for key, value in
+                     license_homepage_ref_vocabulary.items()
+                     if unicode(vocabulary_name) == key),
+                    None)
 
     def get_license_name_by_ref_uri(self, vocabulary_uri):
         _, license_ref_literal_vocabulary, _ = self._get_license_values()
-        for key, value in license_ref_literal_vocabulary.items():
-            if unicode(vocabulary_uri) == key:
-                return value
-        return None
+        return next((value for key, value in
+                     license_ref_literal_vocabulary.items()
+                     if unicode(vocabulary_uri) == key),
+                    None)
 
     def get_license_name_by_homepage_uri(self, vocabulary_uri):
         license_homepages_literal_vocabulary, _, _ = self._get_license_values()
-        for key, value in license_homepages_literal_vocabulary.items():
-            if unicode(vocabulary_uri) == key:
-                return value
-        return None
+        return next((value for key, value in
+                     license_homepages_literal_vocabulary.items()
+                     if unicode(vocabulary_uri) == key),
+                    None)
 
     def get_license_homepage_uri_by_name(self, vocabulary_name):
         license_homepages_literal_vocabulary, _, _ = self._get_license_values()
-        for key, value in license_homepages_literal_vocabulary.items():
-            if unicode(vocabulary_name) == value:
-                return key
-        return None
+        return next((key for key, value in
+                     license_homepages_literal_vocabulary.items()
+                     if unicode(vocabulary_name) == value),
+                    None)
 
     def get_license_homepage_uri_by_uri(self, vocabulary_uri):
         _, _, license_homepage_ref_vocabulary = self._get_license_values()
         license_homepages = list(license_homepage_ref_vocabulary.keys())
         if unicode(vocabulary_uri) in license_homepages:
             return unicode(vocabulary_uri)
-        else:
-            for key, value in license_homepage_ref_vocabulary.items():
-                if unicode(vocabulary_uri) == value:
-                    return key
-        return None
+        return next((key for key, value in
+                     license_homepage_ref_vocabulary.items()
+                     if unicode(vocabulary_uri) == value),
+                    None)
 
 
 def get_license_values():

From d7efda16e0d0f8c37b6379870e5911249e8f78ce Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 14:41:54 +0200
Subject: [PATCH 16/19] fix: Remove blank line

---
 ckanext/dcatapchharvest/dcat_helpers.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 17c279c..fde330b 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -219,7 +219,6 @@ def _process_graph(self, graph):
         return (license_homepages_literal_mapping,
                 license_ref_literal_mapping, license_homepage_ref_mapping)
 
-
     def _get_license_values(self):
         if self._license_cache is None:
             try:
@@ -227,9 +226,10 @@ def _get_license_values(self):
                 self._bind_namespaces(g)
                 self._parse_graph(g)
 
-                license_homepages_literal_mapping, \
-                license_ref_literal_mapping, \
-                license_homepage_ref_mapping = self._process_graph(g)
+                (license_homepages_literal_mapping,
+                 license_ref_literal_mapping,
+                 license_homepage_ref_mapping) = self._process_graph(g)
+
                 self._license_cache = (license_homepages_literal_mapping,
                                        license_ref_literal_mapping,
                                        license_homepage_ref_mapping)

From f723c7633f672c42eb0bb3dd981f8763eb7c4a45 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 21:47:05 +0200
Subject: [PATCH 17/19] fix: Remove unused function

---
 ckanext/dcatapchharvest/dcat_helpers.py | 42 -------------------------
 1 file changed, 42 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index fde330b..0f8e609 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -284,48 +284,6 @@ def get_license_homepage_uri_by_uri(self, vocabulary_uri):
                     None)
 
 
-def get_license_values():
-    g = Graph()
-    license_ref_literal_mapping = {}
-    license_homepages_literal_mapping = {}
-    license_homepage_ref_mapping = {}
-
-    for prefix, namespace in license_namespaces.items():
-        g.bind(prefix, namespace)
-    file = os.path.join(__location__, 'license.ttl')
-    g.parse(file, format='turtle')
-    for ogdch_license_ref in g.subjects(predicate=RDF.type,
-                                        object=SKOS.Concept):
-        license_homepage = None
-        for homepage in g.objects(subject=ogdch_license_ref,
-                                  predicate=FOAF.homepage):
-            license_homepage = homepage
-            break  # Assume one homepage per concept
-
-        license_literal = None
-        try:
-            for license_pref_label in g.objects(subject=ogdch_license_ref,
-                                                predicate=SKOSXL.prefLabel):
-                license_literal = next(g.objects(subject=license_pref_label,
-                                                 predicate=SKOSXL.literalForm))
-                if license_literal is not None:
-                    break  # Assume one literal per concept
-
-            license_homepages_literal_mapping[unicode(license_homepage)] = \
-                unicode(license_literal)
-            license_ref_literal_mapping[unicode(ogdch_license_ref)] = \
-                unicode(license_literal)
-            license_homepage_ref_mapping[unicode(license_homepage)] = \
-                unicode(ogdch_license_ref)
-
-        except Exception as e:
-            raise ValueError("SKOSXL.prefLabel is missing in the RDF-file: %s"
-                             % e)
-
-    return (license_homepages_literal_mapping, license_ref_literal_mapping,
-            license_homepage_ref_mapping)
-
-
 def get_theme_mapping():
     g = Graph()
     theme_mapping = {}

From 3aa4ddfff974e02dc78c8fb61a2335f79530666d Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Mon, 5 Aug 2024 21:56:47 +0200
Subject: [PATCH 18/19] fix: Return a value of a dict

---
 ckanext/dcatapchharvest/dcat_helpers.py | 17 +++++------------
 1 file changed, 5 insertions(+), 12 deletions(-)

diff --git a/ckanext/dcatapchharvest/dcat_helpers.py b/ckanext/dcatapchharvest/dcat_helpers.py
index 0f8e609..68c1d6e 100644
--- a/ckanext/dcatapchharvest/dcat_helpers.py
+++ b/ckanext/dcatapchharvest/dcat_helpers.py
@@ -247,24 +247,17 @@ def get_license_ref_uri_by_name(self, vocabulary_name):
 
     def get_license_ref_uri_by_homepage_uri(self, vocabulary_name):
         _, _, license_homepage_ref_vocabulary = self._get_license_values()
-        return next((value for key, value in
-                     license_homepage_ref_vocabulary.items()
-                     if unicode(vocabulary_name) == key),
-                    None)
+        return license_homepage_ref_vocabulary.get(unicode(vocabulary_name))
 
     def get_license_name_by_ref_uri(self, vocabulary_uri):
         _, license_ref_literal_vocabulary, _ = self._get_license_values()
-        return next((value for key, value in
-                     license_ref_literal_vocabulary.items()
-                     if unicode(vocabulary_uri) == key),
-                    None)
+        return license_ref_literal_vocabulary.get(
+            unicode(vocabulary_uri))
 
     def get_license_name_by_homepage_uri(self, vocabulary_uri):
         license_homepages_literal_vocabulary, _, _ = self._get_license_values()
-        return next((value for key, value in
-                     license_homepages_literal_vocabulary.items()
-                     if unicode(vocabulary_uri) == key),
-                    None)
+        return license_homepages_literal_vocabulary.get(
+            unicode(vocabulary_uri))
 
     def get_license_homepage_uri_by_name(self, vocabulary_name):
         license_homepages_literal_vocabulary, _, _ = self._get_license_values()

From 2e2ed151badac04f11b65db7fd761d88c3ba3c49 Mon Sep 17 00:00:00 2001
From: kovalch <nataliia.kovalchuk@liip.ch>
Date: Wed, 7 Aug 2024 15:52:00 +0200
Subject: [PATCH 19/19] fix: Remove language lable from the uri

---
 ckanext/dcatapchharvest/license.ttl                   | 6 +++---
 ckanext/dcatapchharvest/tests/fixtures/dataset.json   | 2 +-
 ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py | 8 ++++----
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/ckanext/dcatapchharvest/license.ttl b/ckanext/dcatapchharvest/license.ttl
index 660043a..720da79 100644
--- a/ckanext/dcatapchharvest/license.ttl
+++ b/ckanext/dcatapchharvest/license.ttl
@@ -46,7 +46,7 @@
     skosxl:literalForm "NonCommercialAllowed-CommercialAllowed-ReferenceRequired"@en ;
     rdfs:label "NonCommercialAllowed-CommercialAllowed-ReferenceRequired"@de
   ] ;
-  foaf:homepage <https://opendata.swiss/en/terms-of-use/#terms_by> .
+  foaf:homepage <https://opendata.swiss/terms-of-use/#terms_by> .
 
 <http://dcat-ap.ch/vocabulary/licenses/terms_ask>
   a skos:Concept ;
@@ -61,7 +61,7 @@
     skosxl:literalForm "NonCommercialAllowed-CommercialWithPermission-ReferenceNotRequired"@en ;
     rdfs:label "NonCommercialAllowed-CommercialWithPermission-ReferenceNotRequired"@de
   ] ;
-  foaf:homepage <https://opendata.swiss/en/terms-of-use/#terms_ask> .
+  foaf:homepage <https://opendata.swiss/terms-of-use/#terms_ask> .
 
 <http://dcat-ap.ch/vocabulary/licenses/terms_by_ask>
   a skos:Concept ;
@@ -76,7 +76,7 @@
     skosxl:literalForm "NonCommercialAllowed-CommercialWithPermission-ReferenceRequired"@en ;
     rdfs:label "NonCommercialAllowed-CommercialWithPermission-ReferenceRequired"@de
   ] ;
-  foaf:homepage <https://opendata.swiss/en/terms-of-use/#terms_by_ask> .
+  foaf:homepage <https://opendata.swiss/terms-of-use/#terms_by_ask> .
 
 <https://creativecommons.org/publicdomain/zero/1.0/>
   a skos:Concept, cc:License ;
diff --git a/ckanext/dcatapchharvest/tests/fixtures/dataset.json b/ckanext/dcatapchharvest/tests/fixtures/dataset.json
index 7bc5cbc..ebd1883 100644
--- a/ckanext/dcatapchharvest/tests/fixtures/dataset.json
+++ b/ckanext/dcatapchharvest/tests/fixtures/dataset.json
@@ -135,7 +135,7 @@
         "https://example.com/documentation-resource-2"
       ],
       "rights": "http://dcat-ap.ch/vocabulary/licenses/terms_by",
-      "license": "https://opendata.swiss/en/terms-of-use/#terms_by",
+      "license": "https://opendata.swiss/terms-of-use/#terms_by",
       "format": "HTML"
     },
     {
diff --git a/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py b/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
index d729b38..5866fd8 100644
--- a/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
+++ b/ckanext/dcatapchharvest/tests/test_dcatap_ch_parse.py
@@ -32,8 +32,8 @@ def test_rights_license(self):
              # Resources
             eq_(len(dataset['resources']), 1)
             resource = dataset['resources'][0]
-            eq_(unicode(resource['rights']), u'https://opendata.swiss/en/terms-of-use/#terms_by')
-            eq_(unicode(resource['license']), u'https://opendata.swiss/en/terms-of-use/#terms_by_ask')
+            eq_(unicode(resource['rights']), u'https://opendata.swiss/terms-of-use/#terms_by')
+            eq_(unicode(resource['license']), u'https://opendata.swiss/terms-of-use/#terms_by_ask')
 
     def test_dataset_all_fields(self):
 
@@ -146,7 +146,7 @@ def test_dataset_all_fields(self):
         eq_(resource['format'], u'html')
         eq_(resource['media_type'], u'text/html')
         eq_(resource['identifier'], u'346265-fr@bundesamt-fur-statistik-bfs')
-        eq_(resource['license'], u'https://opendata.swiss/en/terms-of-use/#terms_by')
+        eq_(resource['license'], u'https://opendata.swiss/terms-of-use/#terms_by')
         eq_(resource['rights'], u'http://www.opendefinition.org/licenses/cc-zero')
         eq_(resource['language'], [u'fr'])
         eq_(resource['issued'], u'1900-12-31T00:00:00')
@@ -402,7 +402,7 @@ def test_multiple_rights_statements(self):
         dataset = [d for d in p.datasets()][0]
         resource = dataset["resources"][0]
 
-        eq_(unicode(resource['rights']), u"https://opendata.swiss/en/terms-of-use/#terms_by_ask")
+        eq_(unicode(resource['rights']), u"https://opendata.swiss/terms-of-use/#terms_by_ask")
 
     def test_eu_themes_mapping(self):
         contents = self._get_file_contents('catalog-themes.xml')