Datacenter email - revised

pyQuARC/code/schema_validator.py

@@ -136,11 +136,10 @@ 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(r"(\{http[^}]*\})", line)
             for namespace in namespaces:
                 line = line.replace(namespace, "")
-            field_name = re.search(r"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(r"Element\s'.+':\s(\[.*\])?(.*)", line)[2].strip()

pyQuARC/code/url_validator.py

@@ -117,3 +117,14 @@ def doi_link_update(value, bad_urls):
             validity = False
 
         return {"valid": validity, "value": value}
+
+    @staticmethod
+    @if_arg
+    def url_update_email_check(url, bad_urls):
+        validity = True
+        # Check if the URL matches '[email protected]'
+        if url in bad_urls or url == "[email protected]":
+            # Update the URL
+            url = "[email protected]"
+            validity = False  # Mark as invalid if the URL was updated
+        return {"valid": validity, "value": url}

pyQuARC/schemas/check_messages.json

@@ -47,6 +47,14 @@
         },
         "remediation": "This often indicates a broken link. If the URL is broken, recommend revising."
     },
+    "url_update_email_check": {
+        "failure": "The listed email contact information must be updated.",
+        "help": {
+            "message": "Recommend providing the updated contact information as per the data product.",
+            "url": "https://wiki.earthdata.nasa.gov/display/CMR/Data+Center"
+        },
+        "remediation": "Recommend changing the contact information to '[email protected]'. "
+    },
     "shortname_uniqueness": {
         "failure": "The EntryTitle/DataSetId `{}` is identical to the ShortName `{}`.",
         "help": {

pyQuARC/schemas/checks.json

@@ -24,6 +24,11 @@
         "check_function": "health_and_status_check",
         "available": true
     },
+    "url_update_email_check": {
+        "data_type": "url",
+        "check_function": "url_update_email_check",
+        "available": true
+    },
     "string_compare": {
         "data_type": "string",
         "check_function": "compare",

pyQuARC/schemas/rule_mapping.json

@@ -5564,5 +5564,20 @@
         },
         "severity": "warning",
         "check_id": "one_item_presence_check"
-    }
+    },
+    "url_update_email_check": {
+        "rule_name": "URL Email address check",
+        "fields_to_apply": {
+            "umm-c": [
+                {
+                    "fields": [
+                        "DataCenters/ContactGroups/ContactInformation/ContactMechanisms/Value",
+                        "DataCenters/ContactGroups/ContactInformation/ContactInstruction"
+                    ]
+                }
+            ]
+        },
+        "severity": "info",
+        "check_id": "url_update_email_check"
+}
 }

tests/test_downloader.py

@@ -9,11 +9,11 @@ class TestDownloader:
     def setup_method(self):
         self.concept_ids = {
             "collection": {
-                "real": "C1000000010-CDDIS",
+                "real": "C1000000042-CDDIS",
                 "dummy": "C123456-LPDAAC_ECS",
             },
             "granule": {
-                "real": "G1001434969-CDDIS",
+                "real": "G1018577631-CDDIS",
                 "dummy": "G1000000002-CMR_PROV",
             },
             "invalid": "asdfasdf",