Trim strings automatically in Wikibase schema. Closes #1781.

extensions/wikidata/module/langs/translation-en.json

@@ -169,14 +169,6 @@
                         "title": "No language provided for monolingual text.",
                         "body": "Some label, description, alias or monolingual text value have been skipped because no language was provided. Example value: <span class=\"wb-issue-preformat\">{example_text}</span>."
                 },
-                "leading-whitespace": {
-                        "title": "Leading whitespace in strings.",
-                        "body": "Strings such as <span class=\"wb-issue-preformat\">{example_string}</span> have leading whitespace."
-                },
-                "trailing-whitespace": {
-                        "title": "Trailing whitespace in strings.",
-                        "body": "Strings such as <span class=\"wb-issue-preformat\">{example_string}</span> have trailing whitespace."
-                },
                 "duplicate-whitespace": {
                         "title": "Duplicate whitespace in strings.",
                         "body": "Strings such as <span class=\"wb-issue-preformat\">{example_string}</span> contain duplicate whitespace."

extensions/wikidata/module/langs/translation-fr.json

@@ -167,14 +167,6 @@
             "title": "Pas de langue fournie pour des textes monolingues.",
             "body": "Des libellés, descriptions, alias ou textes monolingues ont été ignorés car aucune langue n'a été fournie. Exemple: <span class=\"wb-issue-preformat\">{example_text}</span>."
         },
-        "leading-whitespace": {
-            "title": "Espaces au début de chaînes de caractères.",
-            "body": "Des chaînes telles que <span class=\"wb-issue-preformat\">{example_string}</span> ont des espaces au début."
-        },
-        "trailing-whitespace": {
-            "title": "Espaces à la fin de chaînes de caractères.",
-            "body": "Des chaînes telles que <span class=\"wb-issue-preformat\">{example_string}</span> ont des espaces à la fin."
-        },
         "duplicate-whitespace": {
             "title": "Espaces dédoublées dans des chaînes de caractères.",
             "body": "Des chaînes telles que <span class=\"wb-issue-preformat\">{example_string}</span> contiennent des espaces dédoublées."

extensions/wikidata/module/langs/translation-jp.json

@@ -180,14 +180,6 @@
             "title": "言語指定がありません.",
             "body": "言語指定がないので、ラベル・記述・別名・単一言語テキストが無視されました。例えば: <span class=\"wb-issue-preformat\">{example_text}</span>."
         },
-        "leading-whitespace": {
-            "title": "文頭に空白文字があります.",
-            "body": "<span class=\"wb-issue-preformat\">{example_string}</span>の文頭に空白文字があります."
-        },
-        "trailing-whitespace": {
-            "title": "文末に空白文字があります.",
-            "body": "<span class=\"wb-issue-preformat\">{example_string}</span>の文末に空白文字があります."
-        },
         "duplicate-whitespace": {
             "title": "二重の空白文字があります.",
             "body": "<span class=\"wb-issue-preformat\">{example_string}</span>には二重の空白文字があります."

extensions/wikidata/src/org/openrefine/wikidata/qa/scrutinizers/WhitespaceScrutinizer.java

@@ -43,15 +43,11 @@ public class WhitespaceScrutinizer extends ValueScrutinizer {
 
     private Map<String, Pattern> _issuesMap;
 
-    public static final String leadingWhitespaceType = "leading-whitespace";
-    public static final String trailingWhitespaceType = "trailing-whitespace";
     public static final String duplicateWhitespaceType = "duplicate-whitespace";
     public static final String nonPrintableCharsType = "non-printable-characters";
 
     public WhitespaceScrutinizer() {
         _issuesMap = new HashMap<>();
-        _issuesMap.put(leadingWhitespaceType, Pattern.compile("^\\s"));
-        _issuesMap.put(trailingWhitespaceType, Pattern.compile("\\s$"));
         _issuesMap.put(duplicateWhitespaceType, Pattern.compile("\\s\\s"));
 
         // https://stackoverflow.com/questions/14565934/regular-expression-to-remove-all-non-printable-characters

extensions/wikidata/src/org/openrefine/wikidata/schema/WbStringVariable.java

@@ -58,7 +58,7 @@ public class WbStringVariable extends WbVariableExpr<StringValue> {
     public StringValue fromCell(Cell cell, ExpressionContext ctxt)
             throws SkipSchemaExpressionException {
         if (!cell.value.toString().isEmpty()) {
-            return Datamodel.makeStringValue(cell.value.toString());
+            return Datamodel.makeStringValue(cell.value.toString().trim());
         }
         throw new SkipSchemaExpressionException();
     }

extensions/wikidata/tests/src/org/openrefine/wikidata/qa/scrutinizers/WhitespaceScrutinizerTest.java

@@ -33,18 +33,6 @@ public class WhitespaceScrutinizerTest extends ValueScrutinizerTest {
         return new WhitespaceScrutinizer();
     }
 
-    @Test
-    public void testLeadingWhitespace() {
-        scrutinize(Datamodel.makeStringValue(" a"));
-        assertWarningsRaised(WhitespaceScrutinizer.leadingWhitespaceType);
-    }
-
-    @Test
-    public void testTrailingWhitespace() {
-        scrutinize(Datamodel.makeStringValue("a\t"));
-        assertWarningsRaised(WhitespaceScrutinizer.trailingWhitespaceType);
-    }
-
     @Test
     public void testDuplicateWhitespace() {
         scrutinize(Datamodel.makeStringValue("a\t b"));
@@ -65,14 +53,13 @@ public class WhitespaceScrutinizerTest extends ValueScrutinizerTest {
 
     @Test
     public void testMultipleIssues() {
-        scrutinize(Datamodel.makeStringValue(" a\t b "));
-        assertWarningsRaised(WhitespaceScrutinizer.duplicateWhitespaceType, WhitespaceScrutinizer.leadingWhitespaceType,
-                WhitespaceScrutinizer.trailingWhitespaceType);
+        scrutinize(Datamodel.makeStringValue("a\t b\u0003"));
+        assertWarningsRaised(WhitespaceScrutinizer.duplicateWhitespaceType, WhitespaceScrutinizer.nonPrintableCharsType);
     }
 
     @Test
     public void testMonolingualTextValue() {
-        scrutinizeLabel(Datamodel.makeMonolingualTextValue(" a", "fr"));
-        assertWarningsRaised(WhitespaceScrutinizer.leadingWhitespaceType);
+        scrutinizeLabel(Datamodel.makeMonolingualTextValue("a  b", "fr"));
+        assertWarningsRaised(WhitespaceScrutinizer.duplicateWhitespaceType);
     }
 }

extensions/wikidata/tests/src/org/openrefine/wikidata/schema/WbStringConstantTest.java

@@ -20,6 +20,11 @@ public class WbStringConstantTest extends WbExpressionTest<StringValue> {
         evaluatesTo(Datamodel.makeStringValue("hello world"), constant);
     }
     
+    @Test
+    public void testTrim() {
+        evaluatesTo(Datamodel.makeStringValue("hello world"), new WbStringConstant(" hello world "));
+    }
+    
     @Test(expectedExceptions = IllegalArgumentException.class)
     public void testEmpty() {
         new WbStringConstant("");

extensions/wikidata/tests/src/org/openrefine/wikidata/schema/WbStringVariableTest.java

@@ -45,17 +45,16 @@ public class WbStringVariableTest extends WbVariableTest<StringValue> {
     }
 
     /**
-     * It is not up to the evaluator to clean up the strings it gets. This is
-     * flagged later on by scrutinizers.
+     *  The evaluator cleans up leading and trailing whitespace, but not duplicate spaces
      */
     @Test
     public void testTrailingWhitespace() {
-        evaluatesTo(Datamodel.makeStringValue("dirty \t"), "dirty \t");
+        evaluatesTo(Datamodel.makeStringValue("dirty"), "dirty \t");
     }
 
     @Test
     public void testLeadingWhitespace() {
-        evaluatesTo(Datamodel.makeStringValue(" dirty"), " dirty");
+        evaluatesTo(Datamodel.makeStringValue("dirty"), " dirty");
     }
 
     @Test