ToUnicode map also in Type3 fonts

PdfForQtLib/sources/pdffont.cpp

@@ -514,7 +514,8 @@ void PDFRealizedFontImpl::fillTextSequence(const QByteArray& byteArray, TextSequ
                     reporter->reportRenderError(RenderErrorType::Warning, PDFTranslationContext::tr("Glyph for simple font character code '%1' not found.").arg(static_cast<uint8_t>(byteArray[i])));
                     if (glyphWidth > 0)
                     {
-                        textSequence.items.emplace_back(nullptr, QChar(), glyphWidth * m_pixelSize * FONT_WIDTH_MULTIPLIER);
+                        const QPainterPath* nullpath = nullptr;
+                        textSequence.items.emplace_back(nullpath, QChar(), glyphWidth * m_pixelSize * FONT_WIDTH_MULTIPLIER);
                     }
                 }
             }
@@ -555,7 +556,8 @@ void PDFRealizedFontImpl::fillTextSequence(const QByteArray& byteArray, TextSequ
                     {
                         // We do not multiply advance with font size and FONT_WIDTH_MULTIPLIER, because in the code,
                         // "advance" is treated as in font space.
-                        textSequence.items.emplace_back(nullptr, QChar(), -glyphWidth);
+                        const QPainterPath* nullpath = nullptr;
+                        textSequence.items.emplace_back(nullpath, QChar(), -glyphWidth);
                     }
                 }
             }
@@ -1406,8 +1408,21 @@ PDFFontPointer PDFFont::createFont(const PDFObject& object, const PDFDocument* d
                 }
             }
 
+            PDFFontCMap toUnicodeCMap;
+            const PDFObject& toUnicode = document->getObject(fontDictionary->get("ToUnicode"));
+            if (toUnicode.isName())
+            {
+                toUnicodeCMap = PDFFontCMap::createFromName(toUnicode.getString());
+            }
+            else if (toUnicode.isStream())
+            {
+                const PDFStream* stream = toUnicode.getStream();
+                QByteArray decodedStream = document->getDecodedStream(stream);
+                toUnicodeCMap = PDFFontCMap::createFromData(decodedStream);
+            }
+
             std::vector<PDFReal> widths = fontLoader.readNumberArrayFromDictionary(fontDictionary, "Widths");
-            return PDFFontPointer(new PDFType3Font(qMove(fontDescriptor), firstChar, lastChar, fontMatrix, qMove(characterContentStreams), qMove(widths), document->getObject(fontDictionary->get("Resources"))));
+            return PDFFontPointer(new PDFType3Font(qMove(fontDescriptor), firstChar, lastChar, fontMatrix, qMove(characterContentStreams), qMove(widths), document->getObject(fontDictionary->get("Resources")), qMove(toUnicodeCMap)));
         }
 
         default:
@@ -2146,14 +2161,16 @@ PDFType3Font::PDFType3Font(FontDescriptor fontDescriptor,
                            QMatrix fontMatrix,
                            std::map<int, QByteArray>&& characterContentStreams,
                            std::vector<double>&& widths,
-                           const PDFObject& resources) :
+                           const PDFObject& resources,
+                           PDFFontCMap toUnicode) :
     PDFFont(qMove(fontDescriptor)),
     m_firstCharacterIndex(firstCharacterIndex),
     m_lastCharacterIndex(lastCharacterIndex),
     m_fontMatrix(fontMatrix),
     m_characterContentStreams(qMove(characterContentStreams)),
     m_widths(qMove(widths)),
-    m_resources(resources)
+    m_resources(resources),
+    m_toUnicode(qMove(toUnicode))
 {
 
 }
@@ -2203,11 +2220,12 @@ void PDFRealizedType3FontImpl::fillTextSequence(const QByteArray& byteArray, Tex
     {
         int index = static_cast<uint8_t>(byteArray[i]);
         const QByteArray* contentStream = parentFont->getContentStream(index);
+        QChar character = parentFont->getUnicode(index);
         const double width = parentFont->getWidth(index);
 
         if (contentStream)
         {
-            textSequence.items.emplace_back(contentStream, width);
+            textSequence.items.emplace_back(contentStream, character, width);
         }
         else
         {

PdfForQtLib/sources/pdffont.h

@@ -61,7 +61,7 @@ struct TextSequenceItem
     inline explicit TextSequenceItem() = default;
     inline explicit TextSequenceItem(const QPainterPath* glyph, QChar character, PDFReal advance) : glyph(glyph), character(character), advance(advance) { }
     inline explicit TextSequenceItem(PDFReal advance) : character(), advance(advance) { }
-    inline explicit TextSequenceItem(const QByteArray* characterContentStream, double advance) : characterContentStream(characterContentStream), advance(advance) { }
+    inline explicit TextSequenceItem(const QByteArray* characterContentStream, QChar character, PDFReal advance) : characterContentStream(characterContentStream), character(character), advance(advance) { }
 
     inline bool isContentStream() const { return characterContentStream; }
     inline bool isCharacter() const { return glyph; }
@@ -354,39 +354,6 @@ public:
     virtual FontType getFontType() const override;
 };
 
-class PDFType3Font : public PDFFont
-{
-public:
-    explicit PDFType3Font(FontDescriptor fontDescriptor,
-                          int firstCharacterIndex,
-                          int lastCharacterIndex,
-                          QMatrix fontMatrix,
-                          std::map<int, QByteArray>&& characterContentStreams,
-                          std::vector<double>&& widths,
-                          const PDFObject& resources);
-
-    virtual FontType getFontType() const override;
-    virtual void dumpFontToTreeItem(QTreeWidgetItem*item) const override;
-
-    /// Returns width of the character. If character doesn't exist, then zero is returned.
-    double getWidth(int characterIndex) const;
-
-    /// Return content stream for the character. If character doesn't exist, then nullptr
-    /// is returned.
-    const QByteArray* getContentStream(int characterIndex) const;
-
-    const QMatrix& getFontMatrix() const { return m_fontMatrix; }
-    const PDFObject& getResources() const { return m_resources; }
-
-private:
-    int m_firstCharacterIndex;
-    int m_lastCharacterIndex;
-    QMatrix m_fontMatrix;
-    std::map<int, QByteArray> m_characterContentStreams;
-    std::vector<double> m_widths;
-    PDFObject m_resources;
-};
-
 /// Font cache which caches both fonts, and realized fonts. Cache has individual limit
 /// for fonts, and realized fonts.
 class PDFFontCache
@@ -542,6 +509,46 @@ private:
     bool m_vertical = false;
 };
 
+class PDFType3Font : public PDFFont
+{
+public:
+    explicit PDFType3Font(FontDescriptor fontDescriptor,
+                          int firstCharacterIndex,
+                          int lastCharacterIndex,
+                          QMatrix fontMatrix,
+                          std::map<int, QByteArray>&& characterContentStreams,
+                          std::vector<double>&& widths,
+                          const PDFObject& resources,
+                          PDFFontCMap toUnicode);
+
+    virtual FontType getFontType() const override;
+    virtual void dumpFontToTreeItem(QTreeWidgetItem*item) const override;
+
+    /// Returns width of the character. If character doesn't exist, then zero is returned.
+    double getWidth(int characterIndex) const;
+
+    /// Return content stream for the character. If character doesn't exist, then nullptr
+    /// is returned.
+    const QByteArray* getContentStream(int characterIndex) const;
+
+    const QMatrix& getFontMatrix() const { return m_fontMatrix; }
+    const PDFObject& getResources() const { return m_resources; }
+    const PDFFontCMap& getToUnicode() const { return m_toUnicode; }
+
+    /// Returns unicode character for given character index. If unicode mapping is not
+    /// present, empty (null) character is returned.
+    QChar getUnicode(int characterIndex) const { return m_toUnicode.getToUnicode(characterIndex); }
+
+private:
+    int m_firstCharacterIndex;
+    int m_lastCharacterIndex;
+    QMatrix m_fontMatrix;
+    std::map<int, QByteArray> m_characterContentStreams;
+    std::vector<double> m_widths;
+    PDFObject m_resources;
+    PDFFontCMap m_toUnicode;
+};
+
 /// Composite font (CID-keyed font)
 class PDFType0Font : public PDFFont
 {