Use QRegularExpression::UseUnicodePropertiesOption in regular express…

…ions (#41507) Should fix: #41453 in which values holding UTF diacritics were not handled. While these expressions are used for (international) text values, I think they should be usable with Unicode (instead of ASCII only) See: https://doc.qt.io/qt-5/qregularexpression.html and: https://doc.qt.io/qt-5/qregularexpression.html#PatternOption-enum
qgis · Feb 12, 2021 · 1f752e0 · 1f752e0
1 parent b5970a3
commit 1f752e0
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 7 deletions.
diff --git a/resources/function_help/json/regexp_match b/resources/function_help/json/regexp_match
@@ -2,9 +2,9 @@
   "name": "regexp_match",
   "type": "function",
   "groups": ["Conditionals", "String"],
-  "description": "Return the first matching position matching a regular expression within a string, or 0 if the substring is not found.",
+  "description": "Return the first matching position matching a regular expression within an unicode string, or 0 if the substring is not found.",
   "arguments": [ {"arg":"input_string","description":"the string to test against the regular expression"},
-  {"arg":"regex","description":"The regular expression to test against. Backslash characters must be double escaped (e.g., \"\\\\\\\\s\" to match a white space character)."}
+  {"arg":"regex","description":"The regular expression to test against. Backslash characters must be double escaped (e.g., \"\\\\\\\\s\" to match a white space character or  \"\\\\\\\\b\" to a match word boundary)."}
   ],
-  "examples": [ { "expression":"regexp_match('QGIS ROCKS','\\\\\\\\sROCKS')", "returns":"5"}]
+  "examples": [ { "expression":"regexp_match('QGIS ROCKS','\\\\\\\\sROCKS')", "returns":"5"}, { "expression":"regexp_match('Budač','udač\\\\\\\\b')", "returns":"2"}]
 }
diff --git a/src/core/expression/qgsexpressionfunction.cpp b/src/core/expression/qgsexpressionfunction.cpp
@@ -1421,7 +1421,7 @@ static QVariant fcnRegexpReplace( const QVariantList &values, const QgsExpressio
   QString regexp = QgsExpressionUtils::getStringValue( values.at( 1 ), parent );
   QString after = QgsExpressionUtils::getStringValue( values.at( 2 ), parent );
 
-  QRegularExpression re( regexp );
+  QRegularExpression re( regexp, QRegularExpression::UseUnicodePropertiesOption );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
@@ -1435,7 +1435,7 @@ static QVariant fcnRegexpMatch( const QVariantList &values, const QgsExpressionC
   QString str = QgsExpressionUtils::getStringValue( values.at( 0 ), parent );
   QString regexp = QgsExpressionUtils::getStringValue( values.at( 1 ), parent );
 
-  QRegularExpression re( regexp );
+  QRegularExpression re( regexp, QRegularExpression::UseUnicodePropertiesOption );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
@@ -1450,7 +1450,7 @@ static QVariant fcnRegexpMatches( const QVariantList &values, const QgsExpressio
   QString regexp = QgsExpressionUtils::getStringValue( values.at( 1 ), parent );
   QString empty = QgsExpressionUtils::getStringValue( values.at( 2 ), parent );
 
-  QRegularExpression re( regexp );
+  QRegularExpression re( regexp, QRegularExpression::UseUnicodePropertiesOption );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
@@ -1482,7 +1482,7 @@ static QVariant fcnRegexpSubstr( const QVariantList &values, const QgsExpression
   QString str = QgsExpressionUtils::getStringValue( values.at( 0 ), parent );
   QString regexp = QgsExpressionUtils::getStringValue( values.at( 1 ), parent );
 
-  QRegularExpression re( regexp );
+  QRegularExpression re( regexp, QRegularExpression::UseUnicodePropertiesOption );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );

diff --git a/tests/src/core/testqgsexpression.cpp b/tests/src/core/testqgsexpression.cpp
@@ -1405,6 +1405,11 @@ class TestQgsExpression: public QObject
       QTest::newRow( "nullif substitute double" ) << "nullif(3.3, 3.3)" << false << QVariant();
       QTest::newRow( "nullif substitute int" ) << "nullif(0, 0)" << false << QVariant();
       QTest::newRow( "regexp match" ) << "regexp_match('abc','.b.')" << false << QVariant( 1 );
+      // testing unicode and \b, see #41453. \b tests for a 'word boundary'
+      QTest::newRow( "regexp match unicode" ) << "regexp_match('Budač','Buda\\\\b')" << false << QVariant( 0 );
+      QTest::newRow( "regexp match unicode 2" ) << "regexp_match('Buda','Buda\\\\b')" << false << QVariant( 1 );
+      QTest::newRow( "regexp match unicode 3" ) << "regexp_match('Budač','Budač\\\\b')" << false << QVariant( 1 );
+
       QTest::newRow( "regexp match invalid" ) << "regexp_match('abc DEF','[[[')" << true << QVariant();
       QTest::newRow( "regexp match escaped" ) << "regexp_match('abc DEF','\\\\s[A-Z]+')" << false << QVariant( 4 );
       QTest::newRow( "regexp match false" ) << "regexp_match('abc DEF','\\\\s[a-z]+')" << false << QVariant( 0 );