RelatedUrls description

pyQuARC/code/custom_validator.py

@@ -56,11 +56,43 @@ def mime_type_check(mime_type, url_type, controlled_list):
     @staticmethod
     def availability_check(field_value, parent_value):
         # If the parent is available, the child should be available too, else it is invalid
+
         return {
             "valid": bool(field_value) if parent_value else True,
             "value": parent_value,
         }
 
+    @staticmethod
+    def url_description_presence_check(field_value):
+        """
+        Ensures that URL descriptions are present and not empty if a URL is provided.
+        """
+        if isinstance(field_value, list):
+            # List of dictionaries (URL objects)
+            errors = []
+            for url_obj in field_value:
+                description = url_obj.get("Description", "")
+                if not description or not str(description).strip():
+                    errors.append({
+                        "valid": False,
+                        "value": url_obj,
+                    })
+
+            if errors:
+                return errors[0]  
+            else:
+                return {"valid": True, "value": field_value}
+
+
+        if not field_value or not str(field_value).strip():
+            return {
+                "valid": False,
+                "value": field_value,
+            }
+
+        return {"valid": True, "value": field_value}
+
+
     @staticmethod
     @if_arg
     def bounding_coordinate_logic_check(west, north, east, south):

pyQuARC/code/schema_validator.py

@@ -3,7 +3,7 @@
 import re
 
 from io import BytesIO
-from jsonschema import Draft7Validator, draft7_format_checker, RefResolver
+from jsonschema import Draft7Validator, RefResolver
 from lxml import etree
 from urllib.request import pathname2url
 
@@ -75,25 +75,19 @@ def run_json_validator(self, content_to_validate):
         """
         schema = self.read_json_schema()
         schema_store = {}
-
         if self.metadata_format == UMM_C:
             with open(SCHEMA_PATHS["umm-cmn-json-schema"]) as schema_file:
                 schema_base = json.load(schema_file)
-
             # workaround to read local referenced schema file (only supports uri)
             schema_store = {
                 schema_base.get("$id", "/umm-cmn-json-schema.json"): schema_base,
                 schema_base.get("$id", "umm-cmn-json-schema.json"): schema_base,
             }
-
         errors = {}
-
         resolver = RefResolver.from_schema(schema, store=schema_store)
-
         validator = Draft7Validator(
-            schema, format_checker=draft7_format_checker, resolver=resolver
+            schema, format_checker=Draft7Validator.FORMAT_CHECKER, resolver=resolver
         )
-
         for error in sorted(
             validator.iter_errors(json.loads(content_to_validate)), key=str
         ):
@@ -136,13 +130,13 @@ def _build_errors(error_log, paths):
             # For DIF, because the namespace is specified in the metadata file, lxml library
             # provides field name concatenated with the namespace,
             # the following 3 lines of code removes the namespace
-            namespaces = re.findall("(\{http[^}]*\})", line)
+            namespaces = re.findall(r"(\{http[^}]*\})", line)
             for namespace in namespaces:
                 line = line.replace(namespace, "")
-            field_name = re.search("Element\s'(.*)':", line)[1]
+            field_name = re.search(r"Element\s'(.*)':", line)[1] 
             field_paths = [abs_path for abs_path in paths if field_name in abs_path]
             field_name = field_paths[0] if len(field_paths) == 1 else field_name
-            message = re.search("Element\s'.+':\s(\[.*\])?(.*)", line)[2].strip()
+            message = re.search(r"Element\s'.+':\s(\[.*\])?(.*)", line)[2].strip()
             errors.setdefault(field_name, {})["schema"] = {
                 "message": [f"Error: {message}"],
                 "valid": False,

pyQuARC/schemas/MimeType.csv

@@ -1,23 +1,25 @@
-"Hits: 37","page_num: 1","page_size: 2000","Keyword Version: 14.3","Revision: 2022-08-26 10:35:56","Timestamp: 2022-09-15 12:28:30","Terms Of Use: https://cdn.earthdata.nasa.gov/conduit/upload/5182/KeywordsCommunityGuide_Baseline_v1_SIGNED_FINAL.pdf","The most up to date XML representations can be found here: https://gcmd.earthdata.nasa.gov/kms/concepts/concept_scheme/MimeType/?format=xml","Case native"
+"Hits: 39","page_num: 1","page_size: 2000","Keyword Version: 21.6","Revision: 2025-06-20 10:09:41","Timestamp: 2025-07-15 08:58:02","Terms Of Use: https://cdn.earthdata.nasa.gov/conduit/upload/5182/KeywordsCommunityGuide_Baseline_v1_SIGNED_FINAL.pdf","The most up to date XML representations can be found here: https://gcmd.earthdata.nasa.gov/kms/concepts/concept_scheme/MimeType/?format=xml","Case native"
 MimeType,UUID
+"application/geo+json","aa595aba-86bf-4cc6-8ab5-26ee4de68eeb"
 "application/gml+xml","40bdf6e5-780c-43e2-ab8e-e5dfae4bd779"
 "application/gzip","a8ee535a-8bc8-46fd-8b97-917bd7ea7666"
 "application/json","8542dd4a-a11b-475d-8d46-cad785a7f510"
 "application/msword","c79a0e11-2774-4cf3-a194-45b9e58a93fd"
 "application/octet-stream","b77e64ef-ce80-4dab-b552-c6062990a6e0"
 "application/opensearchdescription+xml","07bcc60e-1551-44d9-b87e-7c260d230ecb"
 "application/pdf","627269ae-ba93-492e-8c31-cc4de1d69810"
+"application/tar","84ef762f-e348-42a6-981c-563822a47806"
 "application/tar+gzip","43ca8ee0-04a5-4020-b0ec-998ec0e0f30e"
 "application/tar+zip","17e82b7c-498d-4d69-993c-fd691aa25ce8"
-"application/tar","84ef762f-e348-42a6-981c-563822a47806"
 "application/vnd.google-earth.kml+xml","80045dcb-18ee-463a-8baf-ffcabed510ea"
 "application/vnd.google-earth.kmz","f7328bf5-8ef2-4f95-a4e0-6fb16d122237"
 "application/vnd.ms-excel","7c99ff72-5239-424d-a0bf-9712c33ea76d"
 "application/vnd.opendap.dap4.dmrpp+xml","b26761fa-8d8e-4bd8-a8ba-db6575554ad7"
 "application/x-bufr","e384b8a8-8cec-4230-9ebe-4db76bbef706"
+"application/x-geotiff","53054b2b-034f-44d8-ab56-fd365947d386"
+"application/x-hdf","b0a3e733-4d1b-486f-b56c-c405a5e4367b"
 "application/x-hdf5","4e80047b-c50b-4805-ac68-789dbc38803f"
 "application/x-hdfeos","b1eac265-2b00-4c39-a429-797c13a2c640"
-"application/x-hdf","b0a3e733-4d1b-486f-b56c-c405a5e4367b"
 "application/x-netcdf","2b192915-32a8-4b68-a720-8ca8a84f04ca"
 "application/x-tar-gz","5e70beda-396e-4cc8-bdd5-70dfc8a1142e"
 "application/x-vnd.iso.19139-2+xml","c1a8dbb7-312d-4481-998e-58d126b32080"

pyQuARC/schemas/check_messages.json

@@ -973,8 +973,18 @@
             "message": "",
             "url": "https://wiki.earthdata.nasa.gov/display/CMR/Related+URLs"
         },
-        "remediation": "Descriptions should be unique to each URL. At least one of the descriptions are repeated in this record. Recommend changing the descriptions to more accurately and uniquely describe each link."
+        "remediation": "Descriptions should be unique to each URL. Several of the descriptions are repeated in this record. Recommend changing the descriptions to more accurately and uniquely describe each link"
+    },
+
+    "url_description_presence_check": {
+        "failure": "A URL description is missing.",
+        "help": {
+            "message": "",
+            "url": "https://wiki.earthdata.nasa.gov/display/CMR/Related+URLs"
+        },
+        "remediation": "Recommend providing a description for each URL"
     },
+
     "online_resource_description_uniqueness_check": {
         "failure": "A URL description is duplicated: `{}`.",
         "help": {

pyQuARC/schemas/checks.json

@@ -64,6 +64,11 @@
         "check_function": "availability_check",
         "available": true
     },
+    "url_description_presence_check": {
+        "data_type": "custom",
+        "check_function": "url_description_presence_check",
+        "available": true
+    },
     "mime_type_check": {
         "data_type": "custom",
         "check_function": "mime_type_check",