Merge pull request #325 from NASA-IMPACT/entrytitle-la

Entrytitle - Revised

Author: lavanya3k

pyQuARC/code/schema_validator.py

@@ -172,7 +172,8 @@ def run_json_validator(self, content_to_validate):
         errors = {}
         resolver = RefResolver.from_schema(schema, store=schema_store)
         validator = Draft7Validator(
-            schema, format_checker=Draft7Validator.FORMAT_CHECKER, resolver=resolver
+            schema, 
+            format_checker=Draft7Validator.FORMAT_CHECKER, resolver=resolver
         )
 
         for error in sorted(
@@ -220,7 +221,7 @@ def _build_errors(error_log, paths):
             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/string_validator.py

@@ -1,7 +1,7 @@
 from .base_validator import BaseValidator
 from .gcmd_validator import GcmdValidator
 from .utils import cmr_request, collection_in_cmr, if_arg, set_cmr_prms
-
+import re
 
 class StringValidator(BaseValidator):
     """
@@ -38,15 +38,21 @@ def length_check(string, extent, relation):
     def compare(first, second, relation):
         """
         Compares two strings based on the relationship
-
         Returns:
-            (dict) An object with the validity of the check and the instance
+           (dict) An object with the validity of the check and the instance
         """
+  
+        # Check if 'first' and 'second' contain any special characters
+        first_clean = re.sub(r'[^a-zA-Z0-9]', '', first).upper()
+        second_clean = re.sub(r'[^a-zA-Z0-9]', '', second).upper()
+  
+        # If either string contains special characters, return a warning or handle as needed
         return {
-            "valid": BaseValidator.compare(first.upper(), second.upper(), relation),
+            "valid": BaseValidator.compare(first_clean, second_clean, relation),
             "value": (first, second),
         }
 
+
     @staticmethod
     @if_arg
     def controlled_keywords_check(value, keywords_list):

pyQuARC/schemas/check_messages.json

@@ -61,7 +61,7 @@
             "message": "",
             "url": "https://wiki.earthdata.nasa.gov/display/CMR/Entry+Title"
         },
-        "remediation": "The EntryTitle/DataSetId should not be identical to the ShortName. Recommend providing a descriptive, formal title for the dataset. "
+        "remediation": "Recommend providing a more descriptive title for the dataset. "
     },
     "abstract_length_check": {
         "failure": "The abstract provided may be inadequate based on length.",

pyQuARC/schemas/check_messages_override.json

@@ -1 +1,2 @@
 {}
+

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",