Merge branch 'master' into python-311

.github/workflows/ci.yml

@@ -91,5 +91,10 @@ jobs:
       run: |
         python -m pip install --upgrade pip setuptools wheel
         python -m pip install hatch
+
+    # TODO: remove, once dependencies are updated
+    - run: hatch run python -m pip list
+    - run: hatch run python -m pip list --outdated
+
     - name: Run test suite with coverage
       run: hatch run cov

fuji_server/harvester/metadata_harvester.py

@@ -476,13 +476,13 @@ def parse_signposting_http_link_format(self, signposting_link_format_text):
             for link_prop in parsed_link[1:]:
                 link_prop = str(link_prop).strip()
                 if link_prop.startswith('anchor'):
-                    anchor_match = re.search('anchor\s*=\s*\"?([^,;"]+)\"?', link_prop)
+                    anchor_match = re.search(r'anchor\s*=\s*\"?([^,;"]+)\"?', link_prop)
                 if link_prop.startswith('rel'):
-                    rel_match = re.search('rel\s*=\s*\"?([^,;"]+)\"?', link_prop)
+                    rel_match = re.search(r'rel\s*=\s*\"?([^,;"]+)\"?', link_prop)
                 elif link_prop.startswith('type'):
-                    type_match = re.search('type\s*=\s*\"?([^,;"]+)\"?', link_prop)
+                    type_match = re.search(r'type\s*=\s*\"?([^,;"]+)\"?', link_prop)
                 elif link_prop.startswith('formats'):
-                    formats_match = re.search('formats\s*=\s*\"?([^,;"]+)\"?', link_prop)
+                    formats_match = re.search(r'formats\s*=\s*\"?([^,;"]+)\"?', link_prop)
             if type_match:
                 found_type = type_match[1]
             if rel_match:

fuji_server/helper/identifier_helper.py

@@ -121,7 +121,7 @@ def __init__(self, idstring, logger = None):
                 #workaround to identify arks properly:
                 self.identifier = self.identifier.replace('/ark:' , '/ark:/' )
                 self.identifier = self.identifier.replace('/ark://', '/ark:/')
-                generic_identifiers_org_pattern = '^([a-z0-9\._]+):(.+)'
+                generic_identifiers_org_pattern = r'^([a-z0-9\._]+):(.+)'
 
                 if self.is_uuid():
                     self.identifier_schemes = ['uuid']

fuji_server/helper/metadata_collector_xml.py

@@ -314,7 +314,7 @@ def get_mapped_xml_metadata(self, tree, mapping):
                         res[prop] = propcontent[0].get('tree').text
                     else:
                         res[prop] = lxml.etree.tostring(propcontent[0].get('tree'), method='text', encoding='unicode')
-                        res[prop] = re.sub('\s+', ' ', res[prop])
+                        res[prop] = re.sub(r'\s+', ' ', res[prop])
                         res[prop] = res[prop].strip()
                 else:
                     for propelem in propcontent:
@@ -324,7 +324,7 @@ def get_mapped_xml_metadata(self, tree, mapping):
                             res[prop].append(propelem.get('tree').text)
                         else:
                             resprop = lxml.etree.tostring(propelem.get('tree'), method='text', encoding='unicode')
-                            resprop = re.sub('\s+', ' ', resprop)
+                            resprop = re.sub(r'\s+', ' ', resprop)
                             resprop = resprop.strip()
                             res[prop].append(resprop)
 

pyproject.toml

@@ -16,33 +16,31 @@ classifiers = [
   "Topic :: Scientific/Engineering :: Information Analysis"
 ]
 dependencies = [
-  "beautifulsoup4~=4.8.2",
-  "configparser~=5.0.2",
-  "connexion[swagger-ui]~=2.9.0",
+  "beautifulsoup4~=4.12",
+  "configparser~=6.0",
+  "connexion[swagger-ui]~=2.14",
   "extruct~=0.13.0",
-  "feedparser~=6.0.8",
-  "flask~=1.1.4",
-  "flask-cors~=3.0.10",
+  "feedparser~=6.0",
+  "flask~=2.2.5",  # pin due to flask 2.3.0 json_encoder issue
+  "flask-cors~=4.0",
   "flask-limiter<=2.0.0",
   "hashid~=3.1.4",
-  "idutils~=1.1.5",
-  "jmespath~=0.10.0",
-  "levenshtein~=0.12.0",
-  "lxml~=4.7",
-  "markupsafe~=2.0.1",
+  "idutils~=1.2",
+  "jmespath~=1.0",
+  "levenshtein~=0.21.1",
+  "lxml~=4.9",
   "pandas~=1.3",
-  "pyRdfa3~=3.5.3",
-  "pyld~=2.0.3",
-  "pyyaml!=6.0.0,!=5.4.0,!=5.4.1",
-  "rapidfuzz~=3.1.2",
-  "rdflib~=6.1.1",
-  "requests~=2.24.0",
-  "sparqlwrapper~=1.8.5",
+  "pyRdfa3~=3.5",
+  "pyld~=2.0",
+  "pyyaml!=6.0.0,!=5.4.0,!=5.4.1",  # 6.0.1
+  "rapidfuzz~=3.3",
+  "rdflib~=6.1",
+  "requests~=2.31",
+  "sparqlwrapper~=1.8",
   "tika~=1.24",
-  "tldextract~=3.1.2",
-  "urlextract~=1.2.0",
-  "waitress~=2.1.1",
-  "werkzeug~=1.0"
+  "tldextract~=3.1",
+  "urlextract~=1.2",
+  "waitress~=2.1"
 ]
 description = "FUJI (FAIRsFAIR Data Objects Assessment Service), A service to evaluate FAIR data objects based on FAIRsFAIR Metrics"
 keywords = [

tests/functional/test_evaluation.py

@@ -13,7 +13,8 @@
 HTTP_200_OK = 200
 
 
-@pytest.mark.vcr
+# TODO: Re-enable the vcr canning
+# @pytest.mark.vcr
 def test_evaluation(client: FlaskClient) -> None:
     """Functional test of the /evaluate endpoint.
 

tests/helper/cassettes/test_preprocessor/test_set_mime_types.yaml

@@ -9,7 +9,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.24.0
+      - python-requests/2.31.0
     method: GET
     uri: https://raw.githubusercontent.com/jshttp/mime-db/master/db.json
   response:
@@ -430,13 +430,13 @@ interactions:
       Cross-Origin-Resource-Policy:
       - cross-origin
       Date:
-      - Fri, 22 Sep 2023 10:46:04 GMT
+      - Mon, 25 Sep 2023 18:07:29 GMT
       ETag:
       - W/"96f0b8a97b963debd2b8196710f6fb55a22c1b719160c77161bc768dee120c17"
       Expires:
-      - Fri, 22 Sep 2023 10:51:04 GMT
+      - Mon, 25 Sep 2023 18:12:29 GMT
       Source-Age:
-      - '118'
+      - '41'
       Strict-Transport-Security:
       - max-age=31536000
       Vary:
@@ -450,15 +450,15 @@ interactions:
       X-Content-Type-Options:
       - nosniff
       X-Fastly-Request-ID:
-      - e606b501f70c3be36f71c3aede5d9c042e8f9e8e
+      - b8cdb072c5d83461fe5450e6872cd4f9e3022cf1
       X-Frame-Options:
       - deny
       X-GitHub-Request-Id:
-      - 808A:881C:10E2BFD:1194D78:650D1E28
+      - DEAC:5405:253AA4D:26E6CD0:651112C5
       X-Served-By:
-      - cache-fra-eddf8230101-FRA
+      - cache-fra-eddf8230078-FRA
       X-Timer:
-      - S1695379565.574322,VS0,VE1
+      - S1695665249.046498,VS0,VE1
       X-XSS-Protection:
       - 1; mode=block
     status: