Add regexp_substr function for returning matched part of string against regex

Author: nyalldawson

resources/function_help/regexp_substr-en_US

@@ -0,0 +1,14 @@
+<h3>regexp_substr() function</h3>
+Returns the portion of a string which matches a supplied regular expression.
+
+<p><h4>Syntax</h4>
+     regexp_substr(<i>string,regex</i>)</p>
+
+<p><h4>Arguments</h4>
+<!-- List args for functions here-->
+<i>  string</i> &rarr; is string.  The input string.<br>
+<i>  regex</i> &rarr; is string.  The regular expression to match against. Backslash characters must be double escaped (eg "\\s" to match a white space character).<br>
+
+<p><h4>Example</h4>
+<!-- Show example of function.-->
+     regexp_substr('abc123','(\\d+)') &rarr; '123'</p>

src/core/qgsexpression.cpp

@@ -521,6 +521,31 @@ static QVariant fcnRegexpMatch( const QVariantList& values, QgsFeature* , QgsExp
   return QVariant( str.contains( re ) ? 1 : 0 );
 }
 
+static QVariant fcnRegexpSubstr( const QVariantList& values, QgsFeature* , QgsExpression* parent )
+{
+  QString str = getStringValue( values.at( 0 ), parent );
+  QString regexp = getStringValue( values.at( 1 ), parent );
+
+  QRegExp re( regexp );
+  if ( !re.isValid() )
+  {
+    parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp ).arg( re.errorString() ) );
+    return QVariant();
+  }
+
+  // extract substring
+  re.indexIn( str );
+  if ( re.captureCount() > 0 )
+  {
+    // return first capture
+    return QVariant( re.capturedTexts()[0] );
+  }
+  else
+  {
+    return QVariant( "" );
+  }
+}
+
 static QVariant fcnSubstr( const QVariantList& values, QgsFeature* , QgsExpression* parent )
 {
   QString str = getStringValue( values.at( 0 ), parent );
@@ -1233,7 +1258,7 @@ const QStringList &QgsExpression::BuiltinFunctions()
     << "coalesce" << "regexp_match" << "$now" << "age" << "year"
     << "month" << "week" << "day" << "hour"
     << "minute" << "second" << "lower" << "upper"
-    << "title" << "length" << "replace" << "regexp_replace"
+    << "title" << "length" << "replace" << "regexp_replace" << "regexp_substr"
     << "substr" << "concat" << "strpos" << "left"
     << "right" << "rpad" << "lpad"
     << "format_number" << "format_date"
@@ -1294,6 +1319,7 @@ const QList<QgsExpression::Function*> &QgsExpression::Functions()
     << new StaticFunction( "length", 1, fcnLength, QObject::tr( "String" ) )
     << new StaticFunction( "replace", 3, fcnReplace, QObject::tr( "String" ) )
     << new StaticFunction( "regexp_replace", 3, fcnRegexpReplace, QObject::tr( "String" ) )
+    << new StaticFunction( "regexp_substr", 2, fcnRegexpSubstr, QObject::tr( "String" ) )
     << new StaticFunction( "substr", 3, fcnSubstr, QObject::tr( "String" ) )
     << new StaticFunction( "concat", -1, fcnConcat, QObject::tr( "String" ) )
     << new StaticFunction( "strpos", 2, fcnStrpos, QObject::tr( "String" ) )

tests/src/core/testqgsexpression.cpp

@@ -274,6 +274,8 @@ class TestQgsExpression: public QObject
       QTest::newRow( "regexp_replace invalid" ) << "regexp_replace('HeLLo','[[[', '-')" << true << QVariant();
       QTest::newRow( "substr" ) << "substr('HeLLo', 3,2)" << false << QVariant( "LL" );
       QTest::newRow( "substr outside" ) << "substr('HeLLo', -5,2)" << false << QVariant( "" );
+      QTest::newRow( "regexp_substr" ) << "regexp_substr('abc123','(\\\\d+)')" << false << QVariant( "123" );
+      QTest::newRow( "regexp_substr invalid" ) << "regexp_substr('abc123','([[[')" << true << QVariant();
       QTest::newRow( "strpos" ) << "strpos('Hello World','World')" << false << QVariant( 6 );
       QTest::newRow( "strpos outside" ) << "strpos('Hello World','blah')" << false << QVariant( -1 );
       QTest::newRow( "left" ) << "left('Hello World',5)" << false << QVariant( "Hello" );