[expressions] Allow non-greedy regex by switching to QRegularExpression

qgis · Oct 26, 2016 · 6f82740 · 6f82740 · nirvn · Oct 26, 2016
1 parent 323c658
commit 6f82740
Show file tree

Hide file tree

Showing 5 changed files with 13 additions and 9 deletions.
diff --git a/resources/function_help/json/regexp_match b/resources/function_help/json/regexp_match
@@ -3,7 +3,7 @@
   "type": "function",
   "description": "Returns true if any part of a string matches the supplied regular expression.",
   "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 (eg \"\\\\\\\\s\" to match a white space character). Non-greedy regular expressions are not supported."}
+  {"arg":"regex","description":"The regular expression to test against. Backslash characters must be double escaped (eg \"\\\\\\\\s\" to match a white space character)."}
   ],
   "examples": [ { "expression":"regexp_match('QGIS ROCKS','\\\\\\\\sROCKS')", "returns":"true"}]
 }
diff --git a/resources/function_help/json/regexp_replace b/resources/function_help/json/regexp_replace
@@ -3,7 +3,7 @@
   "type": "function",
   "description": "Returns a string with the supplied regular expression replaced.",
   "arguments": [ {"arg":"input_string","description":"the string to replace matches in"},
-  {"arg":"regex","description":"The regular expression to replace. Backslash characters must be double escaped (eg \"\\\\\\\\s\" to match a white space character). Non-greedy regular expressions are not supported."},
+  {"arg":"regex","description":"The regular expression to replace. Backslash characters must be double escaped (eg \"\\\\\\\\s\" to match a white space character)."},
   {"arg":"replacement","description":"The string that will replace any matching occurrences of the supplied regular expression. Captured groups can be inserted into the replacement string using \\\\\\\\1, \\\\\\\\2, etc."}
   ],
   "examples": [ { "expression":"regexp_replace('QGIS SHOULD ROCK','\\\\\\\\sSHOULD\\\\\\\\s',' DOES ')", "returns":"'QGIS DOES ROCK'"}]

diff --git a/resources/function_help/json/regexp_substr b/resources/function_help/json/regexp_substr
@@ -3,7 +3,7 @@
   "type": "function",
   "description": "Returns the portion of a string which matches a supplied regular expression.",
   "arguments": [ {"arg":"input_string","description":"the string to find matches in"},
-  {"arg":"regex","description":"The regular expression to match against. Backslash characters must be double escaped (eg \"\\\\\\\\s\" to match a white space character). Non-greedy regular expressions are not supported."}
+  {"arg":"regex","description":"The regular expression to match against. Backslash characters must be double escaped (eg \"\\\\\\\\s\" to match a white space character)."}
   ],
   "examples": [ { "expression":"regexp_substr('abc123','(\\\\\\\\d+)')", "returns":"'123'"}]
 }
diff --git a/src/core/qgsexpression.cpp b/src/core/qgsexpression.cpp
@@ -1225,7 +1225,7 @@ static QVariant fcnRegexpReplace( const QVariantList& values, const QgsExpressio
   QString regexp = getStringValue( values.at( 1 ), parent );
   QString after = getStringValue( values.at( 2 ), parent );
 
-  QRegExp re( regexp );
+  QRegularExpression re( regexp );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
@@ -1239,7 +1239,7 @@ static QVariant fcnRegexpMatch( const QVariantList& values, const QgsExpressionC
   QString str = getStringValue( values.at( 0 ), parent );
   QString regexp = getStringValue( values.at( 1 ), parent );
 
-  QRegExp re( regexp );
+  QRegularExpression re( regexp );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
@@ -1253,19 +1253,19 @@ static QVariant fcnRegexpSubstr( const QVariantList& values, const QgsExpression
   QString str = getStringValue( values.at( 0 ), parent );
   QString regexp = getStringValue( values.at( 1 ), parent );
 
-  QRegExp re( regexp );
+  QRegularExpression re( regexp );
   if ( !re.isValid() )
   {
     parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
     return QVariant();
   }
 
   // extract substring
-  ( void )re.indexIn( str );
-  if ( re.captureCount() > 0 )
+  QRegularExpressionMatch match = re.match( str );
+  if ( match.hasMatch() )
   {
     // return first capture
-    return QVariant( re.capturedTexts().at( 1 ) );
+    return QVariant( match.captured( 0 ) );
   }
   else
   {

diff --git a/tests/src/core/testqgsexpression.cpp b/tests/src/core/testqgsexpression.cpp
@@ -822,10 +822,14 @@ class TestQgsExpression: public QObject
       QTest::newRow( "length" ) << "length('HeLLo')" << false << QVariant( 5 );
       QTest::newRow( "replace" ) << "replace('HeLLo', 'LL', 'xx')" << false << QVariant( "Hexxo" );
       QTest::newRow( "regexp_replace" ) << "regexp_replace('HeLLo','[eL]+', '-')" << false << QVariant( "H-o" );
+      QTest::newRow( "regexp_replace greedy" ) << "regexp_replace('HeLLo','(?<=H).*L', '-')" << false << QVariant( "H-o" );
+      QTest::newRow( "regexp_replace non greedy" ) << "regexp_replace('HeLLo','(?<=H).*?L', '-')" << false << QVariant( "H-Lo" );
+      QTest::newRow( "regexp_replace cap group" ) << "regexp_replace('HeLLo','(eL)', 'x\\\\1x')" << false << QVariant( "HxeLxLo" );
       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 non-greedy" ) << "regexp_substr('abc123','(\\\\d+?)')" << false << QVariant( "1" );
       QTest::newRow( "regexp_substr no hit" ) << "regexp_substr('abcdef','(\\\\d+)')" << false << QVariant( "" );
       QTest::newRow( "regexp_substr invalid" ) << "regexp_substr('abc123','([[[')" << true << QVariant();
       QTest::newRow( "strpos" ) << "strpos('Hello World','World')" << false << QVariant( 7 );