Text codec used when redirecting output

PdfTool/pdfoutputformatter.cpp

@@ -17,6 +17,7 @@
 
 #include "pdfoutputformatter.h"
 
+#include <QTextCodec>
 #include <QTextStream>
 #include <QXmlStreamWriter>
 #include <QCoreApplication>
@@ -91,7 +92,7 @@ private:
 class PDFXmlOutputFormatterImpl : public PDFOutputFormatterImpl
 {
 public:
-    PDFXmlOutputFormatterImpl();
+    PDFXmlOutputFormatterImpl(QString codec);
 
     virtual void beginElement(PDFOutputFormatter::Element type, QString name, QString description, Qt::Alignment alignment, int reference) override;
     virtual void endElement() override;
@@ -109,7 +110,7 @@ private:
 class PDFHtmlOutputFormatterImpl : public PDFOutputFormatterImpl
 {
 public:
-    PDFHtmlOutputFormatterImpl();
+    PDFHtmlOutputFormatterImpl(QString codecName);
 
     virtual void beginElement(PDFOutputFormatter::Element type, QString name, QString description, Qt::Alignment alignment, int reference) override;
     virtual void endElement() override;
@@ -308,14 +309,17 @@ const PDFTextOutputFormatterImpl::TableCell& PDFTextOutputFormatterImpl::getTabl
     return dummy;
 }
 
-PDFHtmlOutputFormatterImpl::PDFHtmlOutputFormatterImpl() :
+PDFHtmlOutputFormatterImpl::PDFHtmlOutputFormatterImpl(QString codecName) :
     m_string(),
     m_streamWriter(&m_string),
     m_depth(0),
     m_headerDepth(1),
     m_elementStack()
 {
-
+    if (QTextCodec* codec = QTextCodec::codecForName(codecName.toLatin1()))
+    {
+        m_streamWriter.setCodec(codec);
+    }
 }
 
 void PDFHtmlOutputFormatterImpl::beginElement(PDFOutputFormatter::Element type, QString name, QString description, Qt::Alignment alignment, int reference)
@@ -483,7 +487,7 @@ void PDFHtmlOutputFormatterImpl::endl()
     m_streamWriter.writeEndElement();
 }
 
-PDFXmlOutputFormatterImpl::PDFXmlOutputFormatterImpl() :
+PDFXmlOutputFormatterImpl::PDFXmlOutputFormatterImpl(QString codecName) :
     m_string(),
     m_streamWriter(&m_string),
     m_depth(0)
@@ -491,6 +495,11 @@ PDFXmlOutputFormatterImpl::PDFXmlOutputFormatterImpl() :
     m_streamWriter.setAutoFormatting(true);
     m_streamWriter.setAutoFormattingIndent(2);
 
+    if (QTextCodec* codec = QTextCodec::codecForName(codecName.toLatin1()))
+    {
+        m_streamWriter.setCodec(codec);
+    }
+
     m_namespace = "https://github.com/JakubMelka/PdfForQt";
     m_prefix = "pdftool";
 }
@@ -575,7 +584,7 @@ QString PDFXmlOutputFormatterImpl::getString() const
     return m_string;
 }
 
-PDFOutputFormatter::PDFOutputFormatter(Style style) :
+PDFOutputFormatter::PDFOutputFormatter(Style style, QString codecName) :
     m_impl(nullptr)
 {
     switch (style)
@@ -585,11 +594,11 @@ PDFOutputFormatter::PDFOutputFormatter(Style style) :
             break;
 
         case Style::Xml:
-            m_impl = new PDFXmlOutputFormatterImpl();
+            m_impl = new PDFXmlOutputFormatterImpl(codecName);
             break;
 
         case Style::Html:
-            m_impl = new PDFHtmlOutputFormatterImpl();
+            m_impl = new PDFHtmlOutputFormatterImpl(codecName);
             break;
     }
 
@@ -621,19 +630,39 @@ QString PDFOutputFormatter::getString() const
     return m_impl->getString();
 }
 
-void PDFConsole::writeText(QString text)
+void PDFConsole::writeText(QString text, QString codecName)
 {
 #ifdef Q_OS_WIN
-    WriteConsoleW(GetStdHandle(STD_OUTPUT_HANDLE), text.utf16(), text.size(), nullptr, nullptr);
+    HANDLE outputHandle = GetStdHandle(STD_OUTPUT_HANDLE);
+    if (!WriteConsoleW(outputHandle, text.utf16(), text.size(), nullptr, nullptr))
+    {
+        // Write console failed. This can happen only, if outputHandle is not handle
+        // to console screen buffer, but, for example a file or a pipe.
+        if (QTextCodec* codec = QTextCodec::codecForName(codecName.toLatin1()))
+        {
+            QByteArray encodedData = codec->fromUnicode(text);
+            WriteFile(outputHandle, encodedData.constData(), encodedData.size(), nullptr, nullptr);
+        }
+    }
 #else
     QTextStream(stdout) << text;
 #endif
 }
 
-void PDFConsole::writeError(QString text)
+void PDFConsole::writeError(QString text, QString codecName)
 {
 #ifdef Q_OS_WIN
-    WriteConsoleW(GetStdHandle(STD_ERROR_HANDLE), text.utf16(), text.size(), nullptr, nullptr);
+    HANDLE outputHandle = GetStdHandle(STD_ERROR_HANDLE);
+    if (!WriteConsoleW(outputHandle, text.utf16(), text.size(), nullptr, nullptr))
+    {
+        // Write console failed. This can happen only, if outputHandle is not handle
+        // to console screen buffer, but, for example a file or a pipe.
+        if (QTextCodec* codec = QTextCodec::codecForName(codecName.toLatin1()))
+        {
+            QByteArray encodedData = codec->fromUnicode(text);
+            WriteFile(outputHandle, encodedData.constData(), encodedData.size(), nullptr, nullptr);
+        }
+    }
 #else
     QTextStream(stdout) << text;
 #endif

PdfTool/pdfoutputformatter.h

@@ -37,7 +37,7 @@ public:
         Html
     };
 
-    explicit PDFOutputFormatter(Style style);
+    explicit PDFOutputFormatter(Style style, QString codecName);
     ~PDFOutputFormatter();
 
     enum class Element
@@ -97,10 +97,10 @@ class PDFConsole
 public:
 
     /// Writes text to the console
-    static void writeText(QString text);
+    static void writeText(QString text, QString codecName);
 
     /// Writes error to the console
-    static void writeError(QString text);
+    static void writeError(QString text, QString codecName);
 
 private:
     explicit PDFConsole() = delete;

PdfTool/pdftoolabstractapplication.cpp

@@ -58,7 +58,7 @@ QString PDFToolHelpApplication::getStandardString(StandardString standardString)
 
 int PDFToolHelpApplication::execute(const PDFToolOptions& options)
 {
-    PDFOutputFormatter formatter(options.outputStyle);
+    PDFOutputFormatter formatter(options.outputStyle, options.outputCodec);
     formatter.beginDocument("help", PDFToolTranslationContext::tr("PDFTool help"));
     formatter.endl();
 
@@ -107,9 +107,27 @@ int PDFToolHelpApplication::execute(const PDFToolOptions& options)
 
     formatter.endTable();
 
+    formatter.endl();
+    formatter.beginHeader("text-output", PDFToolTranslationContext::tr("Text Encoding"));
+
+    formatter.writeText("header", PDFToolTranslationContext::tr("When you redirect console to a file, then specific codec is used to transform output text to target encoding. UTF-8 encoding is used by default. For XML output, you should use only UTF-8 codec. Available codecs:"));
+    formatter.endl();
+
+    QList<QByteArray> codecs = QTextCodec::availableCodecs();
+    QStringList codecNames;
+    for (const QByteArray& codecName : codecs)
+    {
+        codecNames << QString::fromLatin1(codecName);
+    }
+    formatter.writeText("codecs", codecNames.join(", "));
+    formatter.endl();
+    formatter.writeText("default-codec", PDFToolTranslationContext::tr("Suggested codec: UTF-8 or %1").arg(QString::fromLatin1(QTextCodec::codecForLocale()->name())));
+
+    formatter.endHeader();
+
     formatter.endDocument();
 
-    PDFConsole::writeText(formatter.getString());
+    PDFConsole::writeText(formatter.getString(), options.outputCodec);
     return ExitSuccess;
 }
 
@@ -130,6 +148,7 @@ void PDFToolAbstractApplication::initializeCommandLineParser(QCommandLineParser*
     if (optionFlags.testFlag(ConsoleFormat))
     {
         parser->addOption(QCommandLineOption("console-format", "Console output text format (valid values: text|xml|html).", "console-format", "text"));
+        parser->addOption(QCommandLineOption("text-codec", QString("Text codec used when writing text output to redirected standard output. UTF-8 is default."), "text-codec", "UTF-8"));
     }
 
     if (optionFlags.testFlag(OpenDocument))
@@ -184,11 +203,13 @@ PDFToolOptions PDFToolAbstractApplication::getOptions(QCommandLineParser* parser
         {
             if (!consoleFormat.isEmpty())
             {
-                PDFConsole::writeError(PDFToolTranslationContext::tr("Unknown console format '%1'. Defaulting to text console format.").arg(consoleFormat));
+                PDFConsole::writeError(PDFToolTranslationContext::tr("Unknown console format '%1'. Defaulting to text console format.").arg(consoleFormat), options.outputCodec);
             }
 
             options.outputStyle = PDFOutputFormatter::Style::Text;
         }
+
+        options.outputCodec = parser->value("text-codec");
     }
 
     if (optionFlags.testFlag(OpenDocument))
@@ -225,7 +246,7 @@ PDFToolOptions PDFToolAbstractApplication::getOptions(QCommandLineParser* parser
         }
         else if (!dateFormat.isEmpty())
         {
-            PDFConsole::writeError(PDFToolTranslationContext::tr("Unknown console date/time format '%1'. Defaulting to short date/time format.").arg(dateFormat));
+            PDFConsole::writeError(PDFToolTranslationContext::tr("Unknown console date/time format '%1'. Defaulting to short date/time format.").arg(dateFormat), options.outputCodec);
         }
     }
 
@@ -259,13 +280,13 @@ bool PDFToolAbstractApplication::readDocument(const PDFToolOptions& options, pdf
 
         case pdf::PDFDocumentReader::Result::Cancelled:
         {
-            PDFConsole::writeError(PDFToolTranslationContext::tr("Invalid password provided."));
+            PDFConsole::writeError(PDFToolTranslationContext::tr("Invalid password provided."), options.outputCodec);
             return false;
         }
 
         case pdf::PDFDocumentReader::Result::Failed:
         {
-            PDFConsole::writeError(PDFToolTranslationContext::tr("Error occured during document reading. %1").arg(reader.getErrorMessage()));
+            PDFConsole::writeError(PDFToolTranslationContext::tr("Error occured during document reading. %1").arg(reader.getErrorMessage()), options.outputCodec);
             return false;
         }
 
@@ -278,7 +299,7 @@ bool PDFToolAbstractApplication::readDocument(const PDFToolOptions& options, pdf
 
     for (const QString& warning : reader.getWarnings())
     {
-        PDFConsole::writeError(PDFToolTranslationContext::tr("Warning: %1").arg(warning));
+        PDFConsole::writeError(PDFToolTranslationContext::tr("Warning: %1").arg(warning), options.outputCodec);
     }
 
     return true;

PdfTool/pdftoolabstractapplication.h

@@ -42,6 +42,7 @@ struct PDFToolOptions
 {
     // For option 'ConsoleFormat'
     PDFOutputFormatter::Style outputStyle = PDFOutputFormatter::Style::Text;
+    QString outputCodec;
 
     // For option 'OpenDocument'
     QString document;

PdfTool/pdftoolverifysignatures.cpp

@@ -52,7 +52,7 @@ int PDFToolVerifySignaturesApplication::execute(const PDFToolOptions& options)
     // No document specified?
     if (options.document.isEmpty())
     {
-        PDFConsole::writeError(PDFToolTranslationContext::tr("No document specified."));
+        PDFConsole::writeError(PDFToolTranslationContext::tr("No document specified."), options.outputCodec);
         return ErrorNoDocumentSpecified;
     }
 
@@ -76,7 +76,7 @@ int PDFToolVerifySignaturesApplication::execute(const PDFToolOptions& options)
 
         case pdf::PDFDocumentReader::Result::Failed:
         {
-            PDFConsole::writeError(PDFToolTranslationContext::tr("Error occured during document reading. %1").arg(reader.getErrorMessage()));
+            PDFConsole::writeError(PDFToolTranslationContext::tr("Error occured during document reading. %1").arg(reader.getErrorMessage()), options.outputCodec);
             return ErrorDocumentReading;
         }
 
@@ -87,7 +87,7 @@ int PDFToolVerifySignaturesApplication::execute(const PDFToolOptions& options)
 
     for (const QString& warning : reader.getWarnings())
     {
-        PDFConsole::writeError(PDFToolTranslationContext::tr("Warning: %1").arg(warning));
+        PDFConsole::writeError(PDFToolTranslationContext::tr("Warning: %1").arg(warning), options.outputCodec);
     }
 
     // Verify signatures
@@ -107,7 +107,7 @@ int PDFToolVerifySignaturesApplication::execute(const PDFToolOptions& options)
     pdf::PDFForm form = pdf::PDFForm::parse(&document, document.getCatalog()->getFormObject());
     std::vector<pdf::PDFSignatureVerificationResult> signatures = pdf::PDFSignatureHandler::verifySignatures(form, reader.getSource(), parameters);
 
-    PDFOutputFormatter formatter(options.outputStyle);
+    PDFOutputFormatter formatter(options.outputStyle, options.outputCodec);
     formatter.beginDocument("signatures", PDFToolTranslationContext::tr("Digital signatures/timestamps verification of %1").arg(options.document));
     formatter.endl();
 
@@ -386,7 +386,7 @@ int PDFToolVerifySignaturesApplication::execute(const PDFToolOptions& options)
 
     formatter.endDocument();
 
-    PDFConsole::writeText(formatter.getString());
+    PDFConsole::writeText(formatter.getString(), options.outputCodec);
     return ExitSuccess;
 }
 

PdfTool/pdftoolxml.cpp

@@ -210,6 +210,11 @@ int PDFToolXmlApplication::execute(const PDFToolOptions& options)
         writer.setAutoFormattingIndent(2);
     }
 
+    if (QTextCodec* codec = QTextCodec::codecForName(options.outputCodec.toLatin1()))
+    {
+        writer.setCodec(codec);
+    }
+
     QString comment = QString("Processed by %1 %2").arg(QCoreApplication::applicationName(), QCoreApplication::applicationVersion());
     writer.writeStartDocument();
     writer.writeComment(comment);
@@ -240,7 +245,7 @@ int PDFToolXmlApplication::execute(const PDFToolOptions& options)
     writer.writeEndElement();
     writer.writeEndDocument();
 
-    PDFConsole::writeText(xmlString);
+    PDFConsole::writeText(xmlString, options.outputCodec);
     return ExitSuccess;
 }