Tool for exporting internal format of PDF to xml file

PdfForQtLib/sources/pdfencoding.cpp

@@ -18,6 +18,7 @@
 #include "pdfencoding.h"
 
 #include <QTimeZone>
+#include <QTextCodec>
 
 #include <cctype>
 
@@ -2187,6 +2188,21 @@ bool PDFEncoding::canConvertToEncoding(const QString& string, PDFEncoding::Encod
     return true;
 }
 
+bool PDFEncoding::canConvertFromEncoding(const QByteArray& stream, PDFEncoding::Encoding encoding)
+{
+    const encoding::EncodingTable* table = getTableForEncoding(encoding);
+    for (const unsigned char index : stream)
+    {
+        QChar character = (*table)[index];
+        if (character == QChar(0xfffd))
+        {
+            return false;
+        }
+    }
+
+    return true;
+}
+
 QString PDFEncoding::convertTextString(const QByteArray& stream)
 {
     if (hasUnicodeLeadMarkings(stream))
@@ -2333,6 +2349,60 @@ const encoding::EncodingTable* PDFEncoding::getTableForEncoding(Encoding encodin
     return nullptr;
 }
 
+QString PDFEncoding::convertSmartFromByteStringToUnicode(const QByteArray& stream, bool* isBinary)
+{
+    if (isBinary)
+    {
+        *isBinary = false;
+    }
+
+    if (hasUnicodeLeadMarkings(stream))
+    {
+        QTextCodec::ConverterState state = { };
+
+        {
+            QTextCodec* codec = QTextCodec::codecForName("UTF-16BE");
+            QString text = codec->toUnicode(stream.constData(), stream.length(), &state);
+            if (state.invalidChars == 0)
+            {
+                return text;
+            }
+        }
+
+        {
+            QTextCodec* codec = QTextCodec::codecForName("UTF-16LE");
+            QString text = codec->toUnicode(stream.constData(), stream.length(), &state);
+            if (state.invalidChars == 0)
+            {
+                return text;
+            }
+        }
+    }
+
+    if (hasUTF8LeadMarkings(stream))
+    {
+        QTextCodec::ConverterState state = { };
+
+        QTextCodec* codec = QTextCodec::codecForName("UTF-8");
+        QString text = codec->toUnicode(stream.constData(), stream.length(), &state);
+        if (state.invalidChars == 0)
+        {
+            return text;
+        }
+    }
+
+    if (canConvertFromEncoding(stream, Encoding::PDFDoc))
+    {
+        return convert(stream, Encoding::PDFDoc);
+    }
+
+    if (isBinary)
+    {
+        *isBinary = true;
+    }
+    return QString::fromLatin1(stream.toHex()).toUpper();
+}
+
 bool PDFEncoding::hasUnicodeLeadMarkings(const QByteArray& stream)
 {
     if (stream.size() >= 2)

PdfForQtLib/sources/pdfencoding.h

@@ -77,6 +77,13 @@ public:
     /// \param encoding Encoding used in verification of conversion
     static bool canConvertToEncoding(const QString& string, Encoding encoding);
 
+    /// Checks, if stream can be converted to string using encoding (i.e. all
+    /// characters are defined). If all characters are valid, then true is
+    /// returned. This is only guess.
+    /// \param stream Stream
+    /// \param encoding Target encoding
+    static bool canConvertFromEncoding(const QByteArray& stream, Encoding encoding);
+
     /// Convert text string to the unicode string, using either PDFDocEncoding,
     /// or UTF-16BE encoding. Please see PDF Reference 1.7, Chapter 3.8.1. If
     /// UTF-16BE encoding is used, then leading bytes should be 0xFE and 0xFF
@@ -104,6 +111,15 @@ public:
     /// \param encoding Encoding
     static const encoding::EncodingTable* getTableForEncoding(Encoding encoding);
 
+    /// Tries to convert stream to unicode string. Stream can be binary.
+    /// If this is the case, then hexadecimal representation of stream is returned.
+    /// Function checks if stream can be converted to unicode by heuristic
+    /// way, it is not always reliable.
+    /// \param stream Stream
+    /// \param isBinary If specified, it is set to true if conversion failed
+    /// \returns Unicode string or string converted to hexadecimal representation
+    static QString convertSmartFromByteStringToUnicode(const QByteArray& stream, bool* isBinary);
+
 private:
     /// Returns true, if byte array has UTF-16BE/LE unicode marking bytes at the
     /// stream start. If they are present, then byte stream is probably encoded

PdfForQtLib/sources/pdfobject.h

@@ -111,7 +111,7 @@ struct PDFInplaceString
 };
 
 /// Reference to the string implementations
-struct PDFStringRef
+struct PDFFORQTLIBSHARED_EXPORT PDFStringRef
 {
     const PDFInplaceString* inplaceString = nullptr;
     const PDFString* memoryString = nullptr;

PdfForQtLib/sources/pdfvisitor.h

@@ -32,7 +32,7 @@ namespace pdf
 {
 
 /// Abstract visitor, can iterate trough object tree
-class PDFAbstractVisitor
+class PDFFORQTLIBSHARED_EXPORT PDFAbstractVisitor
 {
 public: