Advanced search (second part)

PdfForQtLib/sources/pdfcompiler.h

@@ -133,6 +133,9 @@ public:
     /// Returns true, if text layout is ready
     bool isTextLayoutReady() const { return m_textLayouts.has_value(); }
 
+    /// Returns text layout storage (if it is ready), or nullptr
+    const PDFTextLayoutStorage* getTextLayoutStorage() const { return isTextLayoutReady() ? &m_textLayouts.value() : nullptr; }
+
 signals:
     void textLayoutChanged();
 

PdfForQtLib/sources/pdftextlayout.cpp

@@ -571,6 +571,64 @@ void PDFTextLayoutStorage::setTextLayout(PDFInteger pageIndex, const PDFTextLayo
     layoutStream << result;
 }
 
+PDFFindResults PDFTextLayoutStorage::find(const QString& text, Qt::CaseSensitivity caseSensitivity, PDFTextFlow::FlowFlags flowFlags) const
+{
+    PDFFindResults results;
+
+    QMutex resultsMutex;
+    auto findImpl = [this, flowFlags, caseSensitivity, &results, &resultsMutex, &text](size_t pageIndex)
+    {
+        PDFTextLayout textLayout = getTextLayout(pageIndex);
+        PDFTextFlows textFlows = PDFTextFlow::createTextFlows(textLayout, flowFlags, pageIndex);
+        for (const PDFTextFlow& textFlow : textFlows)
+        {
+            PDFFindResults flowResults = textFlow.find(text, caseSensitivity);
+
+            // Jakub Melka: Do not lock mutex, if we didn't find anything. In that case, just skip to next flow.
+            if (!flowResults.empty())
+            {
+                QMutexLocker lock(&resultsMutex);
+                results.insert(results.end(), flowResults.begin(), flowResults.end());
+            }
+        }
+    };
+
+    auto range = PDFIntegerRange<size_t>(0, m_offsets.size());
+    std::for_each(std::execution::parallel_policy(), range.begin(), range.end(), findImpl);
+
+    std::sort(results.begin(), results.end());
+    return results;
+}
+
+PDFFindResults PDFTextLayoutStorage::find(const QRegularExpression& expression, PDFTextFlow::FlowFlags flowFlags) const
+{
+    PDFFindResults results;
+
+    QMutex resultsMutex;
+    auto findImpl = [this, flowFlags, &results, &resultsMutex, &expression](size_t pageIndex)
+    {
+        PDFTextLayout textLayout = getTextLayout(pageIndex);
+        PDFTextFlows textFlows = PDFTextFlow::createTextFlows(textLayout, flowFlags, pageIndex);
+        for (const PDFTextFlow& textFlow : textFlows)
+        {
+            PDFFindResults flowResults = textFlow.find(expression);
+
+            // Jakub Melka: Do not lock mutex, if we didn't find anything. In that case, just skip to next flow.
+            if (!flowResults.empty())
+            {
+                QMutexLocker lock(&resultsMutex);
+                results.insert(results.end(), flowResults.begin(), flowResults.end());
+            }
+        }
+    };
+
+    auto range = PDFIntegerRange<size_t>(0, m_offsets.size());
+    std::for_each(std::execution::parallel_policy(), range.begin(), range.end(), findImpl);
+
+    std::sort(results.begin(), results.end());
+    return results;
+}
+
 QDataStream& operator<<(QDataStream& stream, const PDFTextLayoutSettings& settings)
 {
     stream << settings.samples;
@@ -622,6 +680,33 @@ PDFFindResults PDFTextFlow::find(const QString& text, Qt::CaseSensitivity caseSe
     return results;
 }
 
+PDFFindResults PDFTextFlow::find(const QRegularExpression& expression) const
+{
+    PDFFindResults results;
+
+    QRegularExpressionMatchIterator iterator = expression.globalMatch(m_text, 0, QRegularExpression::NormalMatch, QRegularExpression::NoMatchOption);
+    while (iterator.hasNext())
+    {
+        QRegularExpressionMatch match = iterator.next();
+
+        Q_ASSERT(match.hasMatch());
+        const int index = match.capturedStart();
+        const int length = match.capturedLength();
+
+        PDFFindResult result;
+        result.matched = match.captured();
+        result.textSelectionItems = getTextSelectionItems(index, length);
+        result.context = getContext(index, length);
+
+        if (!result.textSelectionItems.empty())
+        {
+            results.emplace_back(qMove(result));
+        }
+    }
+
+    return results;
+}
+
 void PDFTextFlow::merge(const PDFTextFlow& next)
 {
     m_text += next.m_text;
@@ -728,8 +813,25 @@ PDFTextSelectionItems PDFTextFlow::getTextSelectionItems(size_t index, size_t le
 
     auto it = std::next(m_characterPointers.cbegin(), index);
     auto itEnd = std::next(m_characterPointers.cbegin(), index + length);
-    s
+    while (it != itEnd)
+    {
+        // Skip invalid items, find first valid
+        if (!it->isValid())
+        {
+            ++it;
+            continue;
+        }
 
+        auto itSelectionStart = it;
+        while (it != itEnd && it->isValid() && it->hasSameBlock(*itSelectionStart))
+        {
+            ++it;
+        }
+        auto itSelectionEnd = std::prev(it);
+        items.emplace_back(*itSelectionStart, *itSelectionEnd);
+    }
+
+    std::sort(items.begin(), items.end());
     return items;
 }
 
@@ -754,9 +856,22 @@ QString PDFTextFlow::getContext(size_t index, size_t length) const
     return m_text.mid(int(index), int(length));
 }
 
+bool PDFCharacterPointer::hasSameBlock(const PDFCharacterPointer& other) const
+{
+    return pageIndex == other.pageIndex && blockIndex == other.blockIndex;
+}
+
 bool PDFCharacterPointer::hasSameLine(const PDFCharacterPointer& other) const
 {
-    return pageIndex == other.pageIndex && blockIndex == other.blockIndex && lineIndex == other.lineIndex;
+    return hasSameBlock(other) && lineIndex == other.lineIndex;
+}
+
+bool PDFFindResult::operator<(const PDFFindResult& other) const
+{
+    Q_ASSERT(!textSelectionItems.empty());
+    Q_ASSERT(!other.textSelectionItems.empty());
+
+    return textSelectionItems.front() < other.textSelectionItems.front();
 }
 
 }   // namespace pdf

PdfForQtLib/sources/pdftextlayout.h

@@ -163,6 +163,9 @@ struct PDFCharacterPointer
     /// Returns true, if character pointer is valid and points to the correct location
     bool isValid() const { return pageIndex > -1; }
 
+    /// Returns true, if character belongs to same block
+    bool hasSameBlock(const PDFCharacterPointer& other) const;
+
     /// Returns true, if character belongs to same line
     bool hasSameLine(const PDFCharacterPointer& other) const;
 
@@ -187,6 +190,8 @@ private:
 
 struct PDFFindResult
 {
+    bool operator<(const PDFFindResult& other) const;
+
     /// Matched string during search
     QString matched;
 
@@ -222,6 +227,10 @@ public:
     /// \param caseSensitivity Case sensitivity
     PDFFindResults find(const QString& text, Qt::CaseSensitivity caseSensitivity) const;
 
+    /// Finds regular expression matches in current text flow. All text occurences are returned.
+    /// \param expression Regular expression to be matched
+    PDFFindResults find(const QRegularExpression& expression) const;
+
     /// Merge data from \p next flow (i.e. connect two consecutive flows)
     void merge(const PDFTextFlow& next);
 
@@ -297,7 +306,7 @@ private:
 /// For writing, mutex is used to synchronize asynchronous writes, for reading
 /// no mutex is used at all. For this reason, both reading/writing at the same time
 /// is prohibited, it is not thread safe.
-class PDFTextLayoutStorage
+class PDFFORQTLIBSHARED_EXPORT PDFTextLayoutStorage
 {
 public:
     explicit inline PDFTextLayoutStorage() = default;
@@ -320,6 +329,17 @@ public:
     /// \param mutex Mutex for locking (calls of setTextLayout from multiple threads)
     void setTextLayout(PDFInteger pageIndex, const PDFTextLayout& layout, QMutex* mutex);
 
+    /// Finds simple text in all pages. All text occurences are returned.
+    /// \param text Text to be found
+    /// \param caseSensitivity Case sensitivity
+    /// \param flowFlags Text flow flags
+    PDFFindResults find(const QString& text, Qt::CaseSensitivity caseSensitivity, PDFTextFlow::FlowFlags flowFlags) const;
+
+    /// Finds regular expression matches in current text flow. All text occurences are returned.
+    /// \param expression Regular expression to be matched
+    /// \param flowFlags Text flow flags
+    PDFFindResults find(const QRegularExpression& expression, PDFTextFlow::FlowFlags flowFlags) const;
+
 private:
     std::vector<int> m_offsets;
     QByteArray m_textLayouts;

PdfForQtViewer/pdfadvancedfindwidget.cpp

@@ -35,6 +35,8 @@ PDFAdvancedFindWidget::PDFAdvancedFindWidget(pdf::PDFDrawWidgetProxy* proxy, QWi
 {
     ui->setupUi(this);
 
+    ui->resultsTableWidget->setHorizontalHeaderLabels({ tr("Page No."), tr("Phrase"), tr("Context") });
+
     connect(ui->regularExpressionsCheckbox, &QCheckBox::clicked, this, &PDFAdvancedFindWidget::updateUI);
     connect(m_proxy, &pdf::PDFDrawWidgetProxy::textLayoutChanged, this, &PDFAdvancedFindWidget::performSearch);
     updateUI();
@@ -62,6 +64,7 @@ void PDFAdvancedFindWidget::on_searchButton_clicked()
     m_parameters.isRegularExpression = ui->regularExpressionsCheckbox->isChecked();
     m_parameters.isDotMatchingEverything = ui->dotMatchesEverythingCheckBox->isChecked();
     m_parameters.isMultiline = ui->multilineMatchingCheckBox->isChecked();
+    m_parameters.isSoftHyphenRemoved = ui->removeSoftHyphenCheckBox->isChecked();
     m_parameters.isSearchFinished = m_parameters.phrase.isEmpty();
 
     if (m_parameters.isSearchFinished)
@@ -85,6 +88,9 @@ void PDFAdvancedFindWidget::on_searchButton_clicked()
         }
     }
 
+    m_findResults.clear();
+    updateResultsUI();
+
     pdf::PDFAsynchronousTextLayoutCompiler* compiler = m_proxy->getTextLayoutCompiler();
     if (compiler->isTextLayoutReady())
     {
@@ -104,6 +110,25 @@ void PDFAdvancedFindWidget::updateUI()
     ui->regularExpressionSettingsGroupBox->setEnabled(enableRegularExpressionUI);
 }
 
+void PDFAdvancedFindWidget::updateResultsUI()
+{
+    ui->tabWidget->setTabText(ui->tabWidget->indexOf(ui->resultsTab), !m_findResults.empty() ? tr("Results (%1)").arg(m_findResults.size()) : tr("Results"));
+    ui->resultsTableWidget->setRowCount(static_cast<int>(m_findResults.size()));
+
+    for (int i = 0, rowCount = int(m_findResults.size()); i < rowCount; ++i)
+    {
+        const pdf::PDFFindResult& findResult = m_findResults[i];
+        ui->resultsTableWidget->setItem(i, 0, new QTableWidgetItem(QString::number(findResult.textSelectionItems.front().first.pageIndex + 1)));
+        ui->resultsTableWidget->setItem(i, 1, new QTableWidgetItem(findResult.matched));
+        ui->resultsTableWidget->setItem(i, 2, new QTableWidgetItem(findResult.context));
+    }
+
+    if (!m_findResults.empty())
+    {
+        ui->tabWidget->setCurrentWidget(ui->resultsTab);
+    }
+}
+
 void PDFAdvancedFindWidget::performSearch()
 {
     if (m_parameters.isSearchFinished)
@@ -120,7 +145,62 @@ void PDFAdvancedFindWidget::performSearch()
         return;
     }
 
+    // Prepare string to search
+    bool useRegularExpression = m_parameters.isRegularExpression;
+    QString expression = m_parameters.phrase;
 
+    if (m_parameters.isWholeWordsOnly)
+    {
+        if (useRegularExpression)
+        {
+            expression = QString("\\b%1\\b").arg(expression);
+        }
+        else
+        {
+            expression = QString("\\b%1\\b").arg(QRegularExpression::escape(expression));
+        }
+        useRegularExpression = true;
+    }
+
+    pdf::PDFTextFlow::FlowFlags flowFlags = pdf::PDFTextFlow::SeparateBlocks;
+    if (m_parameters.isSoftHyphenRemoved)
+    {
+        flowFlags |= pdf::PDFTextFlow::RemoveSoftHyphen;
+    }
+    if (m_parameters.isRegularExpression)
+    {
+        flowFlags |= pdf::PDFTextFlow::AddLineBreaks;
+    }
+
+    const pdf::PDFTextLayoutStorage* textLayoutStorage = compiler->getTextLayoutStorage();
+    if (!useRegularExpression)
+    {
+        // Use simple text search
+        Qt::CaseSensitivity caseSensitivity = m_parameters.isCaseSensitive ? Qt::CaseSensitive : Qt::CaseInsensitive;
+        m_findResults = textLayoutStorage->find(expression, caseSensitivity, flowFlags);
+    }
+    else
+    {
+        // Use regular expression search
+        QRegularExpression::PatternOptions patternOptions = QRegularExpression::UseUnicodePropertiesOption | QRegularExpression::OptimizeOnFirstUsageOption;
+        if (!m_parameters.isCaseSensitive)
+        {
+            patternOptions |= QRegularExpression::CaseInsensitiveOption;
+        }
+        if (m_parameters.isDotMatchingEverything)
+        {
+            patternOptions |= QRegularExpression::DotMatchesEverythingOption;
+        }
+        if (m_parameters.isMultiline)
+        {
+            patternOptions |= QRegularExpression::MultilineOption;
+        }
+
+        QRegularExpression regularExpression(expression, patternOptions);
+        m_findResults = textLayoutStorage->find(regularExpression, flowFlags);
+    }
+
+    updateResultsUI();
 }
 
 }   // namespace pdfviewer

PdfForQtViewer/pdfadvancedfindwidget.h

@@ -19,6 +19,7 @@
 #define PDFADVANCEDFINDWIDGET_H
 
 #include "pdfglobal.h"
+#include "pdftextlayout.h"
 
 #include <QWidget>
 
@@ -51,6 +52,7 @@ private slots:
 
 private:
     void updateUI();
+    void updateResultsUI();
     void performSearch();
 
     struct SearchParameters
@@ -62,6 +64,7 @@ private:
         bool isDotMatchingEverything = false;
         bool isMultiline = false;
         bool isSearchFinished = false;
+        bool isSoftHyphenRemoved = false;
     };
 
     Ui::PDFAdvancedFindWidget* ui;
@@ -69,6 +72,7 @@ private:
     pdf::PDFDrawWidgetProxy* m_proxy;
     const pdf::PDFDocument* m_document;
     SearchParameters m_parameters;
+    pdf::PDFFindResults m_findResults;
 };
 
 }   // namespace pdfviewer

PdfForQtViewer/pdfadvancedfindwidget.ui

@@ -30,27 +30,13 @@
           <string>Search Settings</string>
          </property>
          <layout class="QGridLayout" name="searchSettingsGroupBoxLayout" columnstretch="0,1,0">
-          <item row="0" column="0">
-           <widget class="QLabel" name="searchLabel">
-            <property name="text">
-             <string>Search for:</string>
-            </property>
-           </widget>
-          </item>
           <item row="0" column="1" colspan="2">
            <widget class="QLineEdit" name="searchPhraseEdit"/>
           </item>
-          <item row="4" column="2">
-           <widget class="QPushButton" name="searchButton">
+          <item row="3" column="1" colspan="2">
+           <widget class="QCheckBox" name="regularExpressionsCheckbox">
             <property name="text">
-             <string>Search</string>
-            </property>
-           </widget>
-          </item>
-          <item row="1" column="1" colspan="2">
-           <widget class="QCheckBox" name="caseSensitiveCheckBox">
-            <property name="text">
-             <string>Case sensitive</string>
+             <string>Use regular expressions</string>
             </property>
            </widget>
           </item>
@@ -61,10 +47,31 @@
             </property>
            </widget>
           </item>
-          <item row="3" column="1" colspan="2">
-           <widget class="QCheckBox" name="regularExpressionsCheckbox">
+          <item row="0" column="0">
+           <widget class="QLabel" name="searchLabel">
             <property name="text">
-             <string>Use regular expressions</string>
+             <string>Search for:</string>
+            </property>
+           </widget>
+          </item>
+          <item row="1" column="1" colspan="2">
+           <widget class="QCheckBox" name="caseSensitiveCheckBox">
+            <property name="text">
+             <string>Case sensitive</string>
+            </property>
+           </widget>
+          </item>
+          <item row="4" column="1">
+           <widget class="QCheckBox" name="removeSoftHyphenCheckBox">
+            <property name="text">
+             <string>Remove soft hyphen at end of line</string>
+            </property>
+           </widget>
+          </item>
+          <item row="5" column="2">
+           <widget class="QPushButton" name="searchButton">
+            <property name="text">
+             <string>Search</string>
             </property>
            </widget>
           </item>
@@ -118,12 +125,28 @@
       </attribute>
       <layout class="QVBoxLayout" name="verticalLayout_3">
        <item>
-        <widget class="QTreeWidget" name="treeWidget">
-         <column>
-          <property name="text">
-           <string notr="true">1</string>
-          </property>
-         </column>
+        <widget class="QTableWidget" name="resultsTableWidget">
+         <property name="editTriggers">
+          <set>QAbstractItemView::NoEditTriggers</set>
+         </property>
+         <property name="alternatingRowColors">
+          <bool>true</bool>
+         </property>
+         <property name="selectionBehavior">
+          <enum>QAbstractItemView::SelectRows</enum>
+         </property>
+         <property name="sortingEnabled">
+          <bool>true</bool>
+         </property>
+         <property name="columnCount">
+          <number>3</number>
+         </property>
+         <attribute name="horizontalHeaderStretchLastSection">
+          <bool>true</bool>
+         </attribute>
+         <column/>
+         <column/>
+         <column/>
         </widget>
        </item>
       </layout>