EPUB export: allow overwriting of document metadata

Pick up overrides from <base directory>/<base name>.xmp as a start.

Change-Id: Ib64a6bbdadc53633fb1f0d4a7efdde2e3c96b5ef
Reviewed-on: https://gerrit.libreoffice.org/45551
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>

4 files changed, 222 insertions, 19 deletions

diff --git a/writerperfect/source/writer/exp/xmlimp.cxx b/writerperfect/source/writer/exp/xmlimp.cxx
index bc3b9c7693e4..57f89c870b9a 100644
--- a/writerperfect/source/writer/exp/xmlimp.cxx
+++ b/writerperfect/source/writer/exp/xmlimp.cxx
@@ -12,9 +12,12 @@
 #include <initializer_list>
 #include <unordered_map>
 
+#include <com/sun/star/xml/sax/InputSource.hpp>
+#include <com/sun/star/xml/sax/Parser.hpp>
 #include <rtl/uri.hxx>
 #include <tools/stream.hxx>
 #include <tools/urlobj.hxx>
+#include <unotools/streamwrap.hxx>
 
 #include "xmlfmt.hxx"
 #include "xmlictxt.hxx"
@@ -46,26 +49,16 @@ OUString GetMimeType(const OUString &rExtension)
 }
 
 /// Picks up a cover image from the base directory.
-OUString FindCoverImage(const OUString &rDocumentBaseURL, OUString &rMimeType, const uno::Sequence<beans::PropertyValue> &rDescriptor)
+OUString FindCoverImage(const OUString &rDocumentBaseURL, OUString &rMimeType, const uno::Sequence<beans::PropertyValue> &rFilterData)
 {
     OUString aRet;
 
     // See if filter data contains a cover image explicitly.
-    uno::Sequence<beans::PropertyValue> aFilterData;
-    for (sal_Int32 i = 0; i < rDescriptor.getLength(); ++i)
-    {
-        if (rDescriptor[i].Name == "FilterData")
-        {
-            rDescriptor[i].Value >>= aFilterData;
-            break;
-        }
-    }
-
-    for (sal_Int32 i = 0; i < aFilterData.getLength(); ++i)
+    for (sal_Int32 i = 0; i < rFilterData.getLength(); ++i)
     {
-        if (aFilterData[i].Name == "EPUBCoverImage")
+        if (rFilterData[i].Name == "EPUBCoverImage")
         {
-            aFilterData[i].Value >>= aRet;
+            rFilterData[i].Value >>= aRet;
             break;
         }
     }
@@ -99,7 +92,7 @@ OUString FindCoverImage(const OUString &rDocumentBaseURL, OUString &rMimeType, c
         }
         catch (const rtl::MalformedUriException &rException)
         {
-            SAL_WARN("writerfilter", "FindCoverImage: convertRelToAbs() failed:" << rException.getMessage());
+            SAL_WARN("writerperfect", "FindCoverImage: convertRelToAbs() failed:" << rException.getMessage());
         }
 
         if (!aRet.isEmpty())
@@ -118,6 +111,57 @@ OUString FindCoverImage(const OUString &rDocumentBaseURL, OUString &rMimeType, c
 
     return aRet;
 }
+
+/// Picks up XMP metadata from the base directory.
+void FindXMPMetadata(const uno::Reference<uno::XComponentContext> &xContext, const OUString &rDocumentBaseURL, const uno::Sequence<beans::PropertyValue> &/*rFilterData*/, librevenge::RVNGPropertyList &rMetaData)
+{
+    if (rDocumentBaseURL.isEmpty())
+        return;
+
+    INetURLObject aDocumentBaseURL(rDocumentBaseURL);
+    OUString aURL;
+    try
+    {
+        aURL = rtl::Uri::convertRelToAbs(rDocumentBaseURL, aDocumentBaseURL.GetBase() + ".xmp");
+    }
+    catch (const rtl::MalformedUriException &rException)
+    {
+        SAL_WARN("writerperfect", "FindXMPMetadata: convertRelToAbs() failed:" << rException.getMessage());
+        return;
+    }
+
+    SvFileStream aStream(aURL, StreamMode::READ);
+    if (!aStream.IsOpen())
+        return;
+
+    xml::sax::InputSource aInputSource;
+    uno::Reference<io::XInputStream> xStream(new utl::OStreamWrapper(aStream));
+    aInputSource.aInputStream = xStream;
+    uno::Reference<xml::sax::XParser> xParser = xml::sax::Parser::create(xContext);
+    rtl::Reference<XMPParser> xXMP(new XMPParser());
+    uno::Reference<xml::sax::XDocumentHandler> xDocumentHandler(xXMP.get());
+    xParser->setDocumentHandler(xDocumentHandler);
+    try
+    {
+        xParser->parseStream(aInputSource);
+    }
+    catch (const uno::Exception &rException)
+    {
+        SAL_WARN("writerperfect", "FindXMPMetadata: parseStream() failed:" << rException.Message);
+        return;
+    }
+
+    if (!xXMP->m_aIdentifier.isEmpty())
+        rMetaData.insert("dc:identifier", xXMP->m_aIdentifier.toUtf8().getStr());
+    if (!xXMP->m_aTitle.isEmpty())
+        rMetaData.insert("dc:title", xXMP->m_aTitle.toUtf8().getStr());
+    if (!xXMP->m_aCreator.isEmpty())
+        rMetaData.insert("meta:initial-creator", xXMP->m_aCreator.toUtf8().getStr());
+    if (!xXMP->m_aLanguage.isEmpty())
+        rMetaData.insert("dc:language", xXMP->m_aLanguage.toUtf8().getStr());
+    if (!xXMP->m_aDate.isEmpty())
+        rMetaData.insert("dc:date", xXMP->m_aDate.toUtf8().getStr());
+}
 }
 
 /// Handler for <office:body>.
@@ -170,11 +214,22 @@ rtl::Reference<XMLImportContext> XMLOfficeDocContext::CreateChildContext(const O
     return nullptr;
 }
 
-XMLImport::XMLImport(librevenge::RVNGTextInterface &rGenerator, const OUString &rURL, const uno::Sequence<beans::PropertyValue> &rDescriptor)
-    : mrGenerator(rGenerator)
+XMLImport::XMLImport(const uno::Reference<uno::XComponentContext> &xContext, librevenge::RVNGTextInterface &rGenerator, const OUString &rURL, const uno::Sequence<beans::PropertyValue> &rDescriptor)
+    : mrGenerator(rGenerator),
+      mxContext(xContext)
 {
+    uno::Sequence<beans::PropertyValue> aFilterData;
+    for (sal_Int32 i = 0; i < rDescriptor.getLength(); ++i)
+    {
+        if (rDescriptor[i].Name == "FilterData")
+        {
+            rDescriptor[i].Value >>= aFilterData;
+            break;
+        }
+    }
+
     OUString aMimeType;
-    OUString aCoverImage = FindCoverImage(rURL, aMimeType, rDescriptor);
+    OUString aCoverImage = FindCoverImage(rURL, aMimeType, aFilterData);
     if (!aCoverImage.isEmpty())
     {
         librevenge::RVNGBinaryData aBinaryData;
@@ -187,6 +242,8 @@ XMLImport::XMLImport(librevenge::RVNGTextInterface &rGenerator, const OUString &
         aCoverImageProperties.insert("librevenge:mime-type", aMimeType.toUtf8().getStr());
         maCoverImages.append(aCoverImageProperties);
     }
+
+    FindXMPMetadata(mxContext, rURL, aFilterData, maMetaData);
 }
 
 const librevenge::RVNGPropertyListVector &XMLImport::GetCoverImages()
@@ -194,6 +251,11 @@ const librevenge::RVNGPropertyListVector &XMLImport::GetCoverImages()
     return maCoverImages;
 }
 
+const librevenge::RVNGPropertyList &XMLImport::GetMetaData()
+{
+    return maMetaData;
+}
+
 rtl::Reference<XMLImportContext> XMLImport::CreateContext(const OUString &rName, const css::uno::Reference<css::xml::sax::XAttributeList> &/*xAttribs*/)
 {
     if (rName == "office:document")
diff --git a/writerperfect/source/writer/exp/xmlimp.hxx b/writerperfect/source/writer/exp/xmlimp.hxx
index b1008f00dbf0..90a9762bb9b6 100644
--- a/writerperfect/source/writer/exp/xmlimp.hxx
+++ b/writerperfect/source/writer/exp/xmlimp.hxx
@@ -16,6 +16,7 @@
 #include <librevenge/librevenge.h>
 
 #include <com/sun/star/beans/PropertyValue.hpp>
+#include <com/sun/star/uno/XComponentContext.hpp>
 #include <com/sun/star/xml/sax/XDocumentHandler.hpp>
 
 #include <cppuhelper/implbase.hxx>
@@ -51,9 +52,12 @@ class XMLImport : public cppu::WeakImplHelper
     std::map<OUString, librevenge::RVNGPropertyList> maAutomaticGraphicStyles;
     std::map<OUString, librevenge::RVNGPropertyList> maGraphicStyles;
     librevenge::RVNGPropertyListVector maCoverImages;
+    /// Author, date, etc -- overwrites what would be from the document out of the box.
+    librevenge::RVNGPropertyList maMetaData;
+    const css::uno::Reference<css::uno::XComponentContext> &mxContext;
 
 public:
-    XMLImport(librevenge::RVNGTextInterface &rGenerator, const OUString &rURL, const css::uno::Sequence<css::beans::PropertyValue> &rDescriptor);
+    XMLImport(const css::uno::Reference<css::uno::XComponentContext> &xContext, librevenge::RVNGTextInterface &rGenerator, const OUString &rURL, const css::uno::Sequence<css::beans::PropertyValue> &rDescriptor);
 
     rtl::Reference<XMLImportContext> CreateContext(const OUString &rName, const css::uno::Reference<css::xml::sax::XAttributeList> &xAttribs);
 
@@ -73,6 +77,7 @@ public:
     std::map<OUString, librevenge::RVNGPropertyList> &GetTableStyles();
     std::map<OUString, librevenge::RVNGPropertyList> &GetGraphicStyles();
     const librevenge::RVNGPropertyListVector &GetCoverImages();
+    const librevenge::RVNGPropertyList &GetMetaData();
 
     // XDocumentHandler
     void SAL_CALL startDocument() override;
diff --git a/writerperfect/source/writer/exp/xmlmetai.cxx b/writerperfect/source/writer/exp/xmlmetai.cxx
index ad39aa8c39f1..09e56c8c2388 100644
--- a/writerperfect/source/writer/exp/xmlmetai.cxx
+++ b/writerperfect/source/writer/exp/xmlmetai.cxx
@@ -131,6 +131,9 @@ void XMLMetaInitialCreatorContext::characters(const OUString &rChars)
 XMLMetaDocumentContext::XMLMetaDocumentContext(XMLImport &rImport)
     : XMLImportContext(rImport)
 {
+    librevenge::RVNGPropertyList::Iter it(mrImport.GetMetaData());
+    for (it.rewind(); it.next();)
+        m_aPropertyList.insert(it.key(), it()->clone());
     m_aPropertyList.insert("librevenge:cover-images", mrImport.GetCoverImages());
 }
 
@@ -154,6 +157,94 @@ void XMLMetaDocumentContext::endElement(const OUString &/*rName*/)
     mrImport.GetGenerator().setDocumentMetaData(m_aPropertyList);
 }
 
+XMPParser::XMPParser() = default;
+
+XMPParser::~XMPParser() = default;
+
+void XMPParser::startDocument()
+{
+}
+
+void XMPParser::endDocument()
+{
+}
+
+void XMPParser::startElement(const OUString &rName, const uno::Reference<xml::sax::XAttributeList> &/*xAttribs*/)
+{
+    if (rName == "dc:identifier")
+        m_bInIdentifier = true;
+    else if (rName == "dc:title")
+        m_bInTitle = true;
+    else if (rName == "dc:creator")
+        m_bInCreator = true;
+    else if (rName == "dc:language")
+        m_bInLanguage = true;
+    else if (rName == "dc:date")
+        m_bInDate = true;
+    else if (rName == "rdf:li")
+    {
+        if (m_bInTitle)
+            m_bInTitleItem = true;
+        else if (m_bInCreator)
+            m_bInCreatorItem = true;
+        else if (m_bInLanguage)
+            m_bInLanguageItem = true;
+        else if (m_bInDate)
+            m_bInDateItem = true;
+    }
+}
+
+void XMPParser::endElement(const OUString &rName)
+{
+    if (rName == "dc:identifier")
+        m_bInIdentifier = false;
+    else if (rName == "dc:title")
+        m_bInTitle = false;
+    else if (rName == "dc:creator")
+        m_bInCreator = false;
+    else if (rName == "dc:language")
+        m_bInLanguage = false;
+    else if (rName == "dc:date")
+        m_bInDate = false;
+    else if (rName == "rdf:li")
+    {
+        if (m_bInTitle)
+            m_bInTitleItem = false;
+        else if (m_bInCreator)
+            m_bInCreatorItem = false;
+        else if (m_bInLanguage)
+            m_bInLanguageItem = false;
+        else if (m_bInDate)
+            m_bInDateItem = false;
+    }
+}
+
+void XMPParser::characters(const OUString &rChars)
+{
+    if (m_bInIdentifier)
+        m_aIdentifier += rChars;
+    else if (m_bInTitleItem)
+        m_aTitle += rChars;
+    else if (m_bInCreatorItem)
+        m_aCreator += rChars;
+    else if (m_bInLanguageItem)
+        m_aLanguage += rChars;
+    else if (m_bInDateItem)
+        m_aDate += rChars;
+}
+
+void XMPParser::ignorableWhitespace(const OUString &/*rWhitespace*/)
+{
+}
+
+void XMPParser::processingInstruction(const OUString &/*rTarget*/, const OUString &/*rData*/)
+{
+}
+
+void XMPParser::setDocumentLocator(const uno::Reference<xml::sax::XLocator> &/*xLocator*/)
+{
+}
+
 } // namespace exp
 } // namespace writerperfect
 
diff --git a/writerperfect/source/writer/exp/xmlmetai.hxx b/writerperfect/source/writer/exp/xmlmetai.hxx
index 12285937a705..a011e6b83184 100644
--- a/writerperfect/source/writer/exp/xmlmetai.hxx
+++ b/writerperfect/source/writer/exp/xmlmetai.hxx
@@ -32,6 +32,51 @@ public:
     librevenge::RVNGPropertyList m_aPropertyList;
 };
 
+/// Parses an XMP file.
+class XMPParser: public cppu::WeakImplHelper
+    <
+    css::xml::sax::XDocumentHandler
+    >
+{
+public:
+    explicit XMPParser();
+    virtual ~XMPParser() override;
+
+    // XDocumentHandler
+    virtual void SAL_CALL startDocument() override;
+
+    virtual void SAL_CALL endDocument() override;
+
+    virtual void SAL_CALL startElement(const OUString &aName, const css::uno::Reference<css::xml::sax::XAttributeList> &xAttribs) override;
+
+    virtual void SAL_CALL endElement(const OUString &aName) override;
+
+    virtual void SAL_CALL characters(const OUString &aChars) override;
+
+    virtual void SAL_CALL ignorableWhitespace(const OUString &aWhitespaces) override;
+
+    virtual void SAL_CALL processingInstruction(const OUString &aTarget, const OUString &aData) override;
+
+    virtual void SAL_CALL setDocumentLocator(const css::uno::Reference<css::xml::sax::XLocator> &xLocator) override;
+
+    OUString m_aIdentifier;
+    OUString m_aTitle;
+    OUString m_aCreator;
+    OUString m_aLanguage;
+    OUString m_aDate;
+
+private:
+    bool m_bInIdentifier = false;
+    bool m_bInTitle = false;
+    bool m_bInTitleItem = false;
+    bool m_bInCreator = false;
+    bool m_bInCreatorItem = false;
+    bool m_bInLanguage = false;
+    bool m_bInLanguageItem = false;
+    bool m_bInDate = false;
+    bool m_bInDateItem = false;
+};
+
 } // namespace exp
 } // namespace writerperfect