Add QRegularExpression support to QTextDocument

Currently QTextDocuement only provides find using QRegExp. This patch
aims to add support for QRegularExpression.

[ChangeLog][QtGui][QTextDocument] Support for searching with a
QRegularExpression in a document has been added.

Change-Id: I6dba10545b83995d093407038a821fe54db3d261
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>

src/gui/text/qtextdocument.cpp

@@ -48,6 +48,7 @@
 #include "qtextlist.h"
 #include <qdebug.h>
 #include <qregexp.h>
+#include <qregularexpression.h>
 #include <qvarlengtharray.h>
 #include <qtextcodec.h>
 #include <qthread.h>
@@ -1432,6 +1433,131 @@ QTextCursor QTextDocument::find(const QRegExp &expr, const QTextCursor &from, Fi
     return find(expr, pos, options);
 }
 
+#ifndef QT_NO_REGULAREXPRESSION
+static bool findInBlock(const QTextBlock &block, const QRegularExpression &expression, int offset,
+                        QTextDocument::FindFlags options, QTextCursor *cursor)
+{
+    QRegularExpression expr(expression);
+    if (!(options & QTextDocument::FindCaseSensitively))
+        expr.setPatternOptions(expr.patternOptions() | QRegularExpression::CaseInsensitiveOption);
+    else
+        expr.setPatternOptions(expr.patternOptions() & ~QRegularExpression::CaseInsensitiveOption);
+
+    QString text = block.text();
+    text.replace(QChar::Nbsp, QLatin1Char(' '));
+    QRegularExpressionMatch match;
+    int idx = -1;
+
+    while (offset >= 0 && offset <= text.length()) {
+        idx = (options & QTextDocument::FindBackward) ?
+               text.lastIndexOf(expr, offset, &match) : text.indexOf(expr, offset, &match);
+        if (idx == -1)
+            return false;
+
+        if (options & QTextDocument::FindWholeWords) {
+            const int start = idx;
+            const int end = start + match.capturedLength();
+            if ((start != 0 && text.at(start - 1).isLetterOrNumber())
+                || (end != text.length() && text.at(end).isLetterOrNumber())) {
+                //if this is not a whole word, continue the search in the string
+                offset = (options & QTextDocument::FindBackward) ? idx-1 : end+1;
+                idx = -1;
+                continue;
+            }
+        }
+        //we have a hit, return the cursor for that.
+        break;
+    }
+    if (idx == -1)
+        return false;
+    *cursor = qMove(QTextCursor(block.docHandle(), block.position() + idx));
+    cursor->setPosition(cursor->position() + match.capturedLength(), QTextCursor::KeepAnchor);
+    return true;
+}
+
+/*!
+    \since 5.5
+
+    Finds the next occurrence, matching the regular expression, \a expr, in the document.
+    The search starts at the given \a from position, and proceeds forwards
+    through the document unless specified otherwise in the search options.
+    The \a options control the type of search performed.
+
+    Returns a cursor with the match selected if a match was found; otherwise
+    returns a null cursor.
+
+    If the \a from position is 0 (the default) the search begins from the beginning
+    of the document; otherwise it begins at the specified position.
+*/
+QTextCursor QTextDocument::find(const QRegularExpression &expr, int from, FindFlags options) const
+{
+    Q_D(const QTextDocument);
+
+    if (!expr.isValid())
+        return QTextCursor();
+
+    int pos = from;
+    //the cursor is positioned between characters, so for a backward search
+    //do not include the character given in the position.
+    if (options & FindBackward) {
+        --pos ;
+        if (pos < 0)
+            return QTextCursor();
+    }
+
+    QTextCursor cursor;
+    QTextBlock block = d->blocksFind(pos);
+
+    if (!(options & FindBackward)) {
+        int blockOffset = qMax(0, pos - block.position());
+        while (block.isValid()) {
+            if (findInBlock(block, expr, blockOffset, options, &cursor))
+                return cursor;
+            blockOffset = 0;
+            block = block.next();
+        }
+    } else {
+        int blockOffset = pos - block.position();
+        while (block.isValid()) {
+            if (findInBlock(block, expr, blockOffset, options, &cursor))
+                return cursor;
+            block = block.previous();
+            blockOffset = block.length() - 1;
+        }
+    }
+
+    return QTextCursor();
+}
+
+/*!
+    \since 5.5
+
+    Finds the next occurrence, matching the regular expression, \a expr, in the document.
+    The search starts at the position of the given \a from cursor, and proceeds
+    forwards through the document unless specified otherwise in the search
+    options. The \a options control the type of search performed.
+
+    Returns a cursor with the match selected if a match was found; otherwise
+    returns a null cursor.
+
+    If the given \a from cursor has a selection, the search begins after the
+    selection; otherwise it begins at the cursor's position.
+
+    By default the search is case-sensitive, and can match text anywhere in the
+    document.
+*/
+QTextCursor QTextDocument::find(const QRegularExpression &expr, const QTextCursor &from, FindFlags options) const
+{
+    int pos = 0;
+    if (!from.isNull()) {
+        if (options & QTextDocument::FindBackward)
+            pos = from.selectionStart();
+        else
+            pos = from.selectionEnd();
+    }
+    return find(expr, pos, options);
+}
+#endif // QT_NO_REGULAREXPRESSION
 
 /*!
     \fn QTextObject *QTextDocument::createObject(const QTextFormat &format)

src/gui/text/qtextdocument.h

@@ -169,10 +169,14 @@ public:
 
     QTextCursor find(const QString &subString, int from = 0, FindFlags options = 0) const;
     QTextCursor find(const QString &subString, const QTextCursor &from, FindFlags options = 0) const;
-
     QTextCursor find(const QRegExp &expr, int from = 0, FindFlags options = 0) const;
     QTextCursor find(const QRegExp &expr, const QTextCursor &from, FindFlags options = 0) const;
 
+#ifndef QT_NO_REGULAREXPRESSION
+    QTextCursor find(const QRegularExpression &expr, int from = 0, FindFlags options = 0) const;
+    QTextCursor find(const QRegularExpression &expr, const QTextCursor &from, FindFlags options = 0) const;
+#endif
+
     QTextFrame *frameAt(int pos) const;
     QTextFrame *rootFrame() const;
 

tests/auto/gui/text/qtextdocument/tst_qtextdocument.cpp

@@ -86,6 +86,8 @@ private slots:
     void find2();
     void findWithRegExp_data();
     void findWithRegExp();
+    void findWithRegularExpression_data();
+    void findWithRegularExpression();
     void findMultiple();
     void basicIsModifiedChecks();
     void moreIsModified();
@@ -196,6 +198,7 @@ private slots:
 
 private:
     void backgroundImage_checkExpectedHtml(const QTextDocument &doc);
+    void buildRegExpData();
 
     QTextDocument *doc;
     QTextCursor cursor;
@@ -352,21 +355,7 @@ void tst_QTextDocument::find()
 
 void tst_QTextDocument::findWithRegExp_data()
 {
-    QTest::addColumn<QString>("haystack");
-    QTest::addColumn<QString>("needle");
-    QTest::addColumn<int>("flags");
-    QTest::addColumn<int>("from");
-    QTest::addColumn<int>("anchor");
-    QTest::addColumn<int>("position");
-
-    // match integers 0 to 99
-    QTest::newRow("1") << "23" << "^\\d\\d?$" << int(QTextDocument::FindCaseSensitively) << 0 << 0 << 2;
-    // match ampersands but not &amp;
-    QTest::newRow("2") << "His &amp; hers & theirs" << "&(?!amp;)"<< int(QTextDocument::FindCaseSensitively) << 0 << 15 << 16;
-    //backward search
-    QTest::newRow("3") << QString::fromLatin1("HelloBlahWorld Blah Hah")
-                              << "h" << int(QTextDocument::FindBackward) << 18 << 8 << 9;
-
+    buildRegExpData();
 }
 
 void tst_QTextDocument::findWithRegExp()
@@ -393,6 +382,34 @@ void tst_QTextDocument::findWithRegExp()
     }
 }
 
+void tst_QTextDocument::findWithRegularExpression_data()
+{
+    buildRegExpData();
+}
+
+void tst_QTextDocument::findWithRegularExpression()
+{
+    QFETCH(QString, haystack);
+    QFETCH(QString, needle);
+    QFETCH(int, flags);
+    QFETCH(int, from);
+    QFETCH(int, anchor);
+    QFETCH(int, position);
+
+    cursor.insertText(haystack);
+    //search using a regular expression
+    QRegularExpression expr(needle);
+    QTextDocument::FindFlags flg(flags);
+    cursor = doc->find(expr, from, flg);
+
+    if (anchor != -1) {
+        QCOMPARE(cursor.anchor(), anchor);
+        QCOMPARE(cursor.position(), position);
+    } else {
+        QVERIFY(cursor.isNull());
+    }
+}
+
 void tst_QTextDocument::find2()
 {
     doc->setPlainText("aaa");
@@ -2602,6 +2619,24 @@ void tst_QTextDocument::backgroundImage_checkExpectedHtml(const QTextDocument &d
     QCOMPARE(doc.toHtml(), expectedHtml);
 }
 
+void tst_QTextDocument::buildRegExpData()
+{
+    QTest::addColumn<QString>("haystack");
+    QTest::addColumn<QString>("needle");
+    QTest::addColumn<int>("flags");
+    QTest::addColumn<int>("from");
+    QTest::addColumn<int>("anchor");
+    QTest::addColumn<int>("position");
+
+    // match integers 0 to 99
+    QTest::newRow("1") << "23" << "^\\d\\d?$" << int(QTextDocument::FindCaseSensitively) << 0 << 0 << 2;
+    // match ampersands but not &amp;
+    QTest::newRow("2") << "His &amp; hers & theirs" << "&(?!amp;)"<< int(QTextDocument::FindCaseSensitively) << 0 << 15 << 16;
+    //backward search
+    QTest::newRow("3") << QString::fromLatin1("HelloBlahWorld Blah Hah")
+                              << "h" << int(QTextDocument::FindBackward) << 18 << 8 << 9;
+}
+
 void tst_QTextDocument::backgroundImage_toHtml()
 {
     CREATE_DOC_AND_CURSOR();