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/45551Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>

external/libepubgen/libepubgen-epub3.patch.1

@@ -4467,3 +4467,33 @@ index 3f4bf3c..cbb83b7 100644
 -- 
 2.13.6
 
+From 631b21834883aa8f2ee83a20717dd37900331696 Mon Sep 17 00:00:00 2001
+From: Miklos Vajna <vmiklos@collabora.co.uk>
+Date: Tue, 21 Nov 2017 11:52:03 +0100
+Subject: [PATCH] EPUBGenerator: allow overwriting dc:identifier default
+
+All other types had a way to be overwritten.
+---
+ src/lib/EPUBGenerator.cpp          | 5 ++++-
+ src/test/EPUBTextGeneratorTest.cpp | 3 +++
+ 2 files changed, 7 insertions(+), 1 deletion(-)
+
+diff --git a/src/lib/EPUBGenerator.cpp b/src/lib/EPUBGenerator.cpp
+index 62dac6e..1cb1112 100644
+--- a/src/lib/EPUBGenerator.cpp
++++ b/src/lib/EPUBGenerator.cpp
+@@ -285,7 +285,10 @@ void EPUBGenerator::writeRoot()
+   boost::uuids::uuid uuid = boost::uuids::random_generator()();
+   identifierStream << uuid;
+   std::string identifierCharactrs = identifierStream.str();
+-  sink.insertCharacters(identifierCharactrs.c_str());
++  RVNGString identifier = identifierCharactrs.c_str();
++  if (m_metadata["dc:identifier"] && !m_metadata["dc:identifier"]->getStr().empty())
++    identifier = m_metadata["dc:identifier"]->getStr();
++  sink.insertCharacters(identifier);
+   sink.closeElement("dc:identifier");
+ 
+   RVNGString title("Unknown Title");
+-- 
+2.13.6
+

writerperfect/qa/unit/EPUBExportTest.cxx

@@ -61,6 +61,7 @@ public:
     void testSpanAutostyle();
     void testParaAutostyleCharProps();
     void testMeta();
+    void testMetaXMP();
     void testCoverImage();
     void testParaNamedstyle();
     void testCharNamedstyle();
@@ -94,6 +95,7 @@ public:
     CPPUNIT_TEST(testSpanAutostyle);
     CPPUNIT_TEST(testParaAutostyleCharProps);
     CPPUNIT_TEST(testMeta);
+    CPPUNIT_TEST(testMetaXMP);
     CPPUNIT_TEST(testCoverImage);
     CPPUNIT_TEST(testParaNamedstyle);
     CPPUNIT_TEST(testCharNamedstyle);
@@ -335,6 +337,19 @@ void EPUBExportTest::testMeta()
     CPPUNIT_ASSERT(mxZipFile->hasByName("OEBPS/images/image0001.png"));
 }
 
+void EPUBExportTest::testMetaXMP()
+{
+    createDoc("meta-xmp.fodt", {});
+    mpXmlDoc = parseExport("OEBPS/content.opf");
+
+    // These were the libepubgen default values, metadata from a matching .xmp file was not picked up.
+    assertXPathContent(mpXmlDoc, "/opf:package/opf:metadata/dc:identifier", "deadbeef-e394-4cd6-9b83-7172794612e5");
+    assertXPathContent(mpXmlDoc, "/opf:package/opf:metadata/dc:title", "unknown title from xmp");
+    assertXPathContent(mpXmlDoc, "/opf:package/opf:metadata/dc:creator", "unknown author from xmp");
+    assertXPathContent(mpXmlDoc, "/opf:package/opf:metadata/dc:language", "nl");
+    assertXPathContent(mpXmlDoc, "/opf:package/opf:metadata/opf:meta[@property='dcterms:modified']", "2016-11-20T17:16:07Z");
+}
+
 void EPUBExportTest::testCoverImage()
 {
     OUString aCoverURL = m_directories.getURLFromSrc(DATA_DIRECTORY) + "meta.cover-image.png";

writerperfect/qa/unit/data/writer/epubexport/meta-xmp.fodt

+<?xml version="1.0" encoding="UTF-8"?>
+<office:document office:mimetype="application/vnd.oasis.opendocument.text" office:version="1.2" xmlns:office="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:text="urn:oasis:names:tc:opendocument:xmlns:text:1.0">
+  <office:body>
+    <office:text>
+      <text:p><text:span>Hello world!</text:span></text:p>
+    </office:text>
+  </office:body>
+</office:document>

writerperfect/qa/unit/data/writer/epubexport/meta-xmp.xmp

+<?xml version="1.0" encoding="UTF-8"?>
+<x:xmpmeta xmlns:x="adobe:ns:meta/">
+  <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
+    <rdf:Description rdf:about="" xmlns:dc="http://purl.org/dc/elements/1.1/">
+      <dc:identifier>deadbeef-e394-4cd6-9b83-7172794612e5</dc:identifier>
+      <dc:title>
+        <rdf:Alt>
+          <rdf:li>unknown title from xmp</rdf:li>
+        </rdf:Alt>
+      </dc:title>
+      <dc:creator>
+        <rdf:Seq>
+          <rdf:li>unknown author from xmp</rdf:li>
+        </rdf:Seq>
+      </dc:creator>
+      <dc:language>
+        <rdf:Bag>
+          <rdf:li>nl</rdf:li>
+        </rdf:Bag>
+      </dc:language>
+      <dc:date>
+        <rdf:Seq>
+          <rdf:li>2016-11-20T17:16:07Z</rdf:li>
+        </rdf:Seq>
+      </dc:date>
+    </rdf:Description>
+  </rdf:RDF>
+</x:xmpmeta>

writerperfect/source/writer/EPUBExportFilter.cxx

@@ -79,7 +79,7 @@ sal_Bool EPUBExportFilter::filter(const uno::Sequence<beans::PropertyValue> &rDe
     uno::Reference<frame::XModel> xSourceModel(mxSourceDocument, uno::UNO_QUERY);
     if (xSourceModel.is())
         aSourceURL = xSourceModel->getURL();
-    uno::Reference<xml::sax::XDocumentHandler> xExportHandler(new exp::XMLImport(aGenerator, aSourceURL, rDescriptor));
+    uno::Reference<xml::sax::XDocumentHandler> xExportHandler(new exp::XMLImport(mxContext, aGenerator, aSourceURL, rDescriptor));
 
     uno::Reference<lang::XInitialization> xInitialization(mxContext->getServiceManager()->createInstanceWithContext("com.sun.star.comp.Writer.XMLOasisExporter", mxContext), uno::UNO_QUERY);
     xInitialization->initialize({uno::makeAny(xExportHandler)});

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")

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;

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
 

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