Merge pull request #3723 from nirvn/regexp_matches

[FEATURE] add regexp_matches() function
qgis · Nov 7, 2016 · 3c530a6 · 3c530a6
2 parents 4e4bb2b + 1d245b2
commit 3c530a6
Show file tree

Hide file tree

Showing 5 changed files with 55 additions and 2 deletions.
diff --git a/resources/function_help/json/array_to_string b/resources/function_help/json/array_to_string
@@ -5,7 +5,7 @@
   "arguments": [
     {"arg":"array", "description":"the input array"},
     {"arg":"delimiter","optional":true,"default":"','","description":"the string delimiter used to separate concatenated array elements"},
-    {"arg":"emptyvalue","optional":true,"default":"''","description":"the optional string to use as replacement to empty values"}],
+    {"arg":"empty_value","optional":true,"default":"''","description":"the optional string to use as replacement for empty (zero length) matches"}],
   "examples": [ { "expression":"array_to_string(array('1','2','3'),',')", "returns":"'1,2,3'"},
                 { "expression":"array_to_string(array('1','','3'),',','0')", "returns":"'1,0,3'"}
   ]

diff --git a/resources/function_help/json/regexp_matches b/resources/function_help/json/regexp_matches
@@ -0,0 +1,12 @@
+{
+  "name": "regexp_matches",
+  "type": "function",
+  "description": "Returns an array of all strings captured by capturing groups, in the order the groups themselves appear in the supplied regular expression against a string.",
+  "arguments": [
+    {"arg":"string", "description":"the string to capture groups from against the regular expression"},
+    {"arg":"regex","description":"the regular expression used to capture groups"},
+    {"arg":"empty_value","optional":true,"default":"''","description":"the optional string to use as replacement for empty (zero length) matches"}],
+  "examples": [ { "expression":"regexp_matches('qgis=>rocks','(.*)=>(.*)')", "returns":"array: 'qgis', 'rocks'"},
+                { "expression":"regexp_matches('key=>','(.*)=>(.*)','empty value')", "returns":"array: 'key', 'empty value'"}
+  ]
+}
diff --git a/resources/function_help/json/string_to_array b/resources/function_help/json/string_to_array
@@ -5,7 +5,7 @@
   "arguments": [
     {"arg":"string", "description":"the input string"},
     {"arg":"delimiter","optional":true,"default":"','","description":"the string delimiter used to split the input string"},
-    {"arg":"emptyvalue","optional":true,"default":"''","description":"the optional string to use as replacement to empty values"}],
+    {"arg":"empty_value","optional":true,"default":"''","description":"the optional string to use as replacement for empty (zero length) matches"}],
   "examples": [ { "expression":"string_to_array('1,2,3',',')", "returns":"array: '1', '2', '3'"},
                 { "expression":"string_to_array('1,,3',',','0')", "returns":"array: '1', '0', '3'"}
   ]

diff --git a/src/core/qgsexpression.cpp b/src/core/qgsexpression.cpp
@@ -1335,6 +1335,39 @@ static QVariant fcnRegexpMatch( const QVariantList& values, const QgsExpressionC
   return QVariant( str.contains( re ) ? 1 : 0 );
 }
 
+static QVariant fcnRegexpMatches( const QVariantList& values, const QgsExpressionContext*, QgsExpression* parent )
+{
+  QString str = getStringValue( values.at( 0 ), parent );
+  QString regexp = getStringValue( values.at( 1 ), parent );
+  QString empty = getStringValue( values.at( 2 ), parent );
+
+  QRegularExpression re( regexp );
+  if ( !re.isValid() )
+  {
+    parent->setEvalErrorString( QObject::tr( "Invalid regular expression '%1': %2" ).arg( regexp, re.errorString() ) );
+    return QVariant();
+  }
+
+  QRegularExpressionMatch matches = re.match( str );
+  if ( matches.hasMatch() )
+  {
+    QVariantList array;
+    QStringList list = matches.capturedTexts();
+
+    // Skip the first string to only return captured groups
+    for ( QStringList::const_iterator it = ++list.constBegin(); it != list.constEnd(); ++it )
+    {
+      array += ( *it ).isEmpty() == false ? *it : empty;
+    }
+
+    return QVariant( array );
+  }
+  else
+  {
+    return QVariant();
+  }
+}
+
 static QVariant fcnRegexpSubstr( const QVariantList& values, const QgsExpressionContext*, QgsExpression* parent )
 {
   QString str = getStringValue( values.at( 0 ), parent );
@@ -3745,6 +3778,8 @@ const QList<QgsExpression::Function*>& QgsExpression::Functions()
     << new StaticFunction( QStringLiteral( "concatenate" ), aggParams << Parameter( QStringLiteral( "concatenator" ), true ), fcnAggregateStringConcat, QStringLiteral( "Aggregates" ), QString(), False, QSet<QString>(), true )
 
     << new StaticFunction( QStringLiteral( "regexp_match" ), ParameterList() << Parameter( QStringLiteral( "string" ) ) << Parameter( QStringLiteral( "regex" ) ), fcnRegexpMatch, QStringList() << QStringLiteral( "Conditionals" ) << QStringLiteral( "String" ) )
+    << new StaticFunction( QStringLiteral( "regexp_matches" ), ParameterList() << Parameter( QStringLiteral( "string" ) ) << Parameter( QStringLiteral( "regex" ) ) << Parameter( QStringLiteral( "emptyvalue" ), true, "" ), fcnRegexpMatches, QStringLiteral( "Arrays" ) )
+
     << new StaticFunction( QStringLiteral( "now" ), 0, fcnNow, QStringLiteral( "Date and Time" ), QString(), false, QSet<QString>(), false, QStringList() << QStringLiteral( "$now" ) )
     << new StaticFunction( QStringLiteral( "age" ), 2, fcnAge, QStringLiteral( "Date and Time" ) )
     << new StaticFunction( QStringLiteral( "year" ), 1, fcnYear, QStringLiteral( "Date and Time" ) )

diff --git a/tests/src/core/testqgsexpression.cpp b/tests/src/core/testqgsexpression.cpp
@@ -841,6 +841,11 @@ class TestQgsExpression: public QObject
       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( "regexp_matches" ) << "array_get(regexp_matches('qgis=>rOcks;hello=>world','qgis=>(.*)[;$]'),0)" << false << QVariant( "rOcks" );
+      QTest::newRow( "regexp_matches empty custom value" ) << "array_get(regexp_matches('qgis=>;hello=>world','qgis=>(.*)[;$]','empty'),0)" << false << QVariant( "empty" );
+      QTest::newRow( "regexp_matches no match" ) << "regexp_matches('123','no()match')" << false << QVariant();
+      QTest::newRow( "regexp_matches no capturing group" ) << "regexp_matches('some string','.*')" << false << QVariant( QVariantList() );
+      QTest::newRow( "regexp_matches invalid" ) << "regexp_matches('invalid','(')" << true << QVariant();
       QTest::newRow( "strpos" ) << "strpos('Hello World','World')" << false << QVariant( 7 );
       QTest::newRow( "strpos outside" ) << "strpos('Hello World','blah')" << false << QVariant( 0 );
       QTest::newRow( "left" ) << "left('Hello World',5)" << false << QVariant( "Hello" );
@@ -2266,6 +2271,7 @@ class TestQgsExpression: public QObject
       builderExpected << "world";
       QCOMPARE( QgsExpression( "array('hello', 'world')" ).evaluate( &context ), QVariant( builderExpected ) );
       QCOMPARE( QgsExpression( "string_to_array('hello,world',',')" ).evaluate( &context ), QVariant( builderExpected ) );
+      QCOMPARE( QgsExpression( "regexp_matches('hello=>world','([A-Za-z]*)=>([A-Za-z]*)')" ).evaluate( &context ), QVariant( builderExpected ) );
 
       QCOMPARE( QgsExpression( "array_length(\"strings\")" ).evaluate( &context ), QVariant( 2 ) );