Closes #2596 In legacy datasets, our brand new ORCID validation may c… (

#2605) Closes #2596 In legacy datasets, our brand new ORCID validation may create confusing errors when trying to save a draft after file edit (upload or changes in metadata or licenses) --------- Co-authored-by: lbownik <[email protected]>
CeON · Dec 10, 2024 · 363170c · 363170c
1 parent 3268227
commit 363170c
Show file tree

Hide file tree

Showing 2 changed files with 93 additions and 0 deletions.
diff --git a/scripts/database/delete_unfixable_orcids.sql b/scripts/database/delete_unfixable_orcids.sql
@@ -0,0 +1,52 @@
+-- !!!!!!!  run fix_fixable_orcids.sql before this script,
+-- !!!!!!!  otherwse fixable ORCIDs will be lost
+DO $$
+DECLARE
+    ids numeric[];
+BEGIN
+    -- find indetifiers of all dataset fields with supposed ORCID values 
+    -- that don't match ORCIS format
+    SELECT ARRAY(select id
+        from datasetfield 
+         where 
+         datasetfieldtype_id in (select id 
+                        from datasetfieldtype 
+                        where 
+                        name = 'authorIdentifier')  
+        and
+        datasetfieldparent_id in 
+            (select datasetfieldparent_id 
+             from datasetfield  
+             where 
+                id in (select datasetfield_id 
+                    from datasetfield_controlledvocabularyvalue 
+                    where  
+                    controlledvocabularyvalues_id in (select controlledvocabularyvalue.id   
+                                                      from controlledvocabularyvalue join datasetfieldtype ON controlledvocabularyvalue.datasetfieldtype_id = datasetfieldtype.id
+                                                      where controlledvocabularyvalue.strvalue = 'ORCID' and datasetfieldtype.name = 'authorIdentifierScheme'))) 
+        and fieldvalue !~ '^(\d{4}-){3}\d{3}[\dX]$'
+        and fieldvalue is not null)
+    INTO ids;
+
+    -- clear all unfixable orcid values
+    update datasetfield 
+    set 
+    fieldvalue = null
+    where 
+    id = any(ids);
+
+    -- clear indetified schema selections for deleted ORCIDs
+    delete 
+    from datasetfield_controlledvocabularyvalue
+    where datasetfield_id in (select id from datasetfield 
+                              where
+                              datasetfieldparent_id in (select datasetfieldparent_id 
+                                                        from datasetfield 
+                                                        where
+                                                        id = any(ids))
+                              and
+                              datasetfieldtype_id in (select id 
+                                                      from datasetfieldtype 
+                                                      where
+                                                      name = 'authorIdentifierScheme'));
+END $$;;
diff --git a/scripts/database/fix_fixable_orcids.sql b/scripts/database/fix_fixable_orcids.sql
@@ -0,0 +1,41 @@
+-- fixes ORCID values
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'ORCID: ') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'ORCID: %';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'ORCID ') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'ORCID 0%';
+
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'http://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'http://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'https://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'https://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'https://orcid.org/ ') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'https://orcid.org/ %';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'hhttps://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'hhttps://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'Warsaw, Poland https://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'Warsaw, Poland https://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'ORCID logo https://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'ORCID logo https://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'orcid logo https://orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'orcid logo https://orcid.org/%';
+update datasetfield set fieldvalue = LTRIM(fieldvalue, 'orcid.org/') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like 'orcid.org/%';
+
+update datasetfield set fieldvalue = RTRIM(fieldvalue, 'Zbiór') 
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like '%Zbiór';
+
+update datasetfield set fieldvalue = ('0' || fieldvalue)  
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like '000-%';
+
+update datasetfield set fieldvalue = LTRIM(fieldvalue, '/')
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like '/0000-%';
+
+update datasetfield set fieldvalue = RTRIM(fieldvalue, '.')
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like '%.';
+
+update datasetfield set fieldvalue = RTRIM(fieldvalue, ',')
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier') and fieldvalue like '%,';
+
+
+update datasetfield set fieldvalue = TRIM(fieldvalue)
+    where datasetfieldtype_id  in (select id from datasetfieldtype where name = 'authorIdentifier');