Metadata: fix incorrect person identifiers when transforming metadata

schema_transformations.py

@@ -111,7 +111,7 @@ def _default2_default3(ctx, m):
     """
     Add affiliation identifiers to creators and contributors.
 
-    Tags are renamed to Keywords and Related Datapackage to Related Resource.
+    Tags are renamed to Keywords, Related Datapackage renamed to Related Resource and improved Affiliation and Person Identifiers.
 
     :param ctx: Combined type of a callback and rei struct
     :param m:   Metadata to transform (default-2)
@@ -121,19 +121,67 @@ def _default2_default3(ctx, m):
     if m.get('Creator', False):
         # For this creator step through all its affiliations
         for creator in m['Creator']:
-            new_affiliations = []
+            affiliations = []
             for affiliation in creator['Affiliation']:
-                new_affiliations.append({"Affiliation_Name": affiliation, "Affiliation_Identifier": ""})
-            creator['Affiliation'] = new_affiliations
+                affiliations.append({"Affiliation_Name": affiliation, "Affiliation_Identifier": ""})
+            creator['Affiliation'] = affiliations
+
+            personal_identifiers = []
+            for personal_identifier in creator['Person_Identifier']:
+                if personal_identifier.get('Name_Identifier_Scheme', True):
+                    continue
+                elif personal_identifier.get('Name_Identifier_Scheme', None) == 'ORCID':
+                    # Check for incorrect ORCID format.
+                    if not re.search("^(https://orcid.org/)[0-9]{4}-[0-9]{4}-[0-9]{4}-[0-9]{3}[0-9X]$", personal_identifier.get('Name_Identifier', None)):
+                        corrected_orcid = correctify_orcid(personal_identifier['Name_Identifier'])
+                        # Only it an actual correction took place change the value and mark this data as 'changed'.
+                        if corrected_orcid != personal_identifier['Name_Identifier']:
+                            personal_identifier['Name_Identifier'] = corrected_orcid
+                elif personal_identifier.get('Name_Identifier_Scheme', None) == 'ResearcherID (Web of Science)':
+                    # Check for incorrect ResearcherID format.
+                    if not re.search("^(https://www.researcherid.com/rid/)[A-Z]-[0-9]{4}-[0-9]{4}$", personal_identifier.get('Name_Identifier', None)):
+                        corrected_researcher_id = correctify_researcher_id(personal_identifier['Name_Identifier'])
+                        # Only it an actual correction took place change the value and mark this data as 'changed'.
+                        if corrected_researcher_id != personal_identifier['Name_Identifier']:
+                            personal_identifier['Name_Identifier'] = corrected_researcher_id
+
+                personal_identifiers.append({"Name_Identifier_Scheme": personal_identifier['Name_Identifier_Scheme'], "Name_Identifier": personal_identifier['Name_Identifier']})
+
+            if len(personal_identifiers) > 0:
+                creator['Person_Identifier'] = personal_identifiers
 
     if m.get('Contributor', False):
         # For this contributor step through all its affiliations
-        for contrib in m['Contributor']:
-            new_affiliations = []
-            if contrib.get('Affiliation', False):
-                for affiliation in contrib['Affiliation']:
-                    new_affiliations.append({"Affiliation_Name": affiliation, "Affiliation_Identifier": ""})
-                contrib['Affiliation'] = new_affiliations
+        for contributor in m['Contributor']:
+            affiliations = []
+            if contributor.get('Affiliation', False):
+                for affiliation in contributor['Affiliation']:
+                    affiliations.append({"Affiliation_Name": affiliation, "Affiliation_Identifier": ""})
+                contributor['Affiliation'] = affiliations
+
+            personal_identifiers = []
+            for personal_identifier in contributor['Person_Identifier']:
+                if personal_identifier.get('Name_Identifier_Scheme', True):
+                    continue
+                elif personal_identifier.get('Name_Identifier_Scheme', None) == 'ORCID':
+                    # Check for incorrect ORCID format.
+                    if not re.search("^(https://orcid.org/)[0-9]{4}-[0-9]{4}-[0-9]{4}-[0-9]{3}[0-9X]$", personal_identifier.get('Name_Identifier', None)):
+                        corrected_orcid = correctify_orcid(personal_identifier['Name_Identifier'])
+                        # Only it an actual correction took place change the value and mark this data as 'changed'.
+                        if corrected_orcid != personal_identifier['Name_Identifier']:
+                            personal_identifier['Name_Identifier'] = corrected_orcid
+                elif personal_identifier.get('Name_Identifier_Scheme', None) == 'ResearcherID (Web of Science)':
+                    # Check for incorrect ResearcherID format.
+                    if not re.search("^(https://www.researcherid.com/rid/)[A-Z]-[0-9]{4}-[0-9]{4}$", personal_identifier.get('Name_Identifier', None)):
+                        corrected_researcher_id = correctify_researcher_id(personal_identifier['Name_Identifier'])
+                        # Only it an actual correction took place change the value and mark this data as 'changed'.
+                        if corrected_researcher_id != personal_identifier['Name_Identifier']:
+                            personal_identifier['Name_Identifier'] = corrected_researcher_id
+
+                personal_identifiers.append({"Name_Identifier_Scheme": personal_identifier['Name_Identifier_Scheme'], "Name_Identifier": personal_identifier['Name_Identifier']})
+
+            if len(personal_identifiers) > 0:
+                contributor['Person_Identifier'] = personal_identifiers
 
     # Rename Tags to Keywords
     if m.get('Tag', False):
@@ -643,3 +691,36 @@ def get(src_id, dst_id):
 
     x = transformations.get(src_id)
     return None if x is None else x.get(dst_id)
+
+
+def correctify_orcid(org_orcid):
+    """Correct illformatted ORCID."""
+    # Get rid of all spaces.
+    orcid = org_orcid.replace(' ', '')
+
+    # Upper-case X.
+    orcid = org_orcid.replace('x', 'X')
+
+    # The last part should hold a valid id like eg: 1234-1234-1234-123X.
+    # If not, it is impossible to correct it to the valid orcid format
+    orcs = orcid.split('/')
+    if not re.search("^[0-9]{4}-[0-9]{4}-[0-9]{4}-[0-9]{3}[0-9X]$", orcs[-1]):
+        # Return original value.
+        return org_orcid
+
+    return "https://orcid.org/{}".format(orcs[-1])
+
+
+def correctify_researcher_id(org_researcher_id):
+    """Correct illformatted ResearcherID."""
+    # Get rid of all spaces.
+    researcher_id = org_researcher_id.replace(' ', '')
+
+    # The last part should hold a valid id like eg: A-1234-1234
+    # If not, it is impossible to correct it to the valid ResearcherID format
+    orcs = researcher_id.split('/')
+    if not re.search("^[A-Z]-[0-9]{4}-[0-9]{4}$", orcs[-1]):
+        # Return original value.
+        return org_researcher_id
+
+    return "https://www.researcherid.com/rid/{}".format(orcs[-1])