sw/qa/extras/ooxmlexport/data/custom-styles-TOC-comma.docx |binary
sw/qa/extras/ooxmlexport/ooxmlexport17.cxx | 25 +++++++++++
writerfilter/source/dmapper/DomainMapper_Impl.cxx | 4 +
writerfilter/source/dmapper/StyleSheetTable.cxx | 29 +++++++++++++
4 files changed, 57 insertions(+), 1 deletion(-)
New commits:
commit d42e7d50225d5ff7359cfc765ec9694cf83c9749
Author: Michael Stahl <michael.st...@allotropia.de>
AuthorDate: Wed Jan 18 15:41:57 2023 +0100
Commit: Miklos Vajna <vmik...@collabora.com>
CommitDate: Thu Jan 19 12:43:03 2023 +0000
tdf#153083 writerfilter: import locale-dependent TOC \t style names
The bugdoc contains this style:
<w:style w:type="paragraph" w:styleId="IntensivesZitat">
<w:name w:val="Intense Quote"/>
<w:basedOn w:val="Standard"/>
<w:next w:val="Standard"/>
<w:link w:val="IntensivesZitatZchn"/>
...
which is referred to by:
TOC \o "1-3" \h \z \t "Intensives Zitat;3;Custom1;3;_MyStyle0;3"
Word in an "en" locale is unable to match the "Intensives Zitat" in the
TOC field with the style "Intense Quote", which is a built-in style
in Word (no equivalent in Writer).
At first glance nothing in styles.xml matches the localised built-in
style name in the TOC field.
But it looks like the w:styleId value is somehow generated from the
localised style name by omitting certain characters like SPACE and
non-ASCII letters.
Change-Id: I2050f7cf7f8d80bee1f667ee53b7f9981bbf7b49
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/145745
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.st...@allotropia.de>
(cherry picked from commit ecbad22fdf81c6f072b6c9f9c16dbba47fe4748c)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/145721
Tested-by: Michael Stahl <michael.st...@allotropia.de>
Reviewed-by: Miklos Vajna <vmik...@collabora.com>
diff --git a/sw/qa/extras/ooxmlexport/data/custom-styles-TOC-comma.docx
b/sw/qa/extras/ooxmlexport/data/custom-styles-TOC-comma.docx
new file mode 100644
index 000000000000..1b013086dcc6
Binary files /dev/null and
b/sw/qa/extras/ooxmlexport/data/custom-styles-TOC-comma.docx differ
diff --git a/sw/qa/extras/ooxmlexport/ooxmlexport17.cxx
b/sw/qa/extras/ooxmlexport/ooxmlexport17.cxx
index d9d5802e0606..b216e14d22fe 100644
--- a/sw/qa/extras/ooxmlexport/ooxmlexport17.cxx
+++ b/sw/qa/extras/ooxmlexport/ooxmlexport17.cxx
@@ -15,6 +15,7 @@
#include <com/sun/star/style/ParagraphAdjust.hpp>
#include <com/sun/star/text/WritingMode2.hpp>
#include <com/sun/star/text/XBookmarksSupplier.hpp>
+#include <com/sun/star/text/XDocumentIndex.hpp>
#include <com/sun/star/text/XFootnotesSupplier.hpp>
#include <com/sun/star/text/XTextFieldsSupplier.hpp>
#include <com/sun/star/text/XTextField.hpp>
@@ -701,6 +702,30 @@ DECLARE_OOXMLEXPORT_TEST(testTdf148361, "tdf148361.docx")
CPPUNIT_ASSERT_EQUAL(OUString("[Type text]"), aActual);
}
+DECLARE_OOXMLEXPORT_TEST(testTdf153082_comma, "custom-styles-TOC-comma.docx")
+{
+ uno::Reference<text::XDocumentIndexesSupplier> xIndexSupplier(mxComponent,
uno::UNO_QUERY);
+ uno::Reference<container::XIndexAccess> xIndexes =
xIndexSupplier->getDocumentIndexes();
+ uno::Reference<text::XDocumentIndex> xTOC(xIndexes->getByIndex(0),
uno::UNO_QUERY);
+ // check styles
+ uno::Reference<container::XIndexAccess> xParaStyles =
+ getProperty<uno::Reference<container::XIndexAccess>>(xTOC,
"LevelParagraphStyles");
+ uno::Sequence<OUString> styles;
+ xParaStyles->getByIndex(0) >>= styles;
+ CPPUNIT_ASSERT_EQUAL(uno::Sequence<OUString>{"_MyStyle0"}, styles);
+ xParaStyles->getByIndex(1) >>= styles;
+ CPPUNIT_ASSERT_EQUAL(uno::Sequence<OUString>{"Custom1"}, styles);
+ xParaStyles->getByIndex(2) >>= styles;
+ // the first one is built-in Word style that was localised DE "Intensives
Zitat" in the file
+ CPPUNIT_ASSERT_EQUAL(uno::Sequence<OUString>{"Intense Quote"}, styles);
+ xTOC->update();
+ OUString const tocContent(xTOC->getAnchor()->getString());
+ CPPUNIT_ASSERT(tocContent.startsWith("Table of Contents"));
+ CPPUNIT_ASSERT(tocContent.indexOf("Lorem ipsum dolor sit amet,
consectetuer adipiscing elit.") != -1);
+ CPPUNIT_ASSERT(tocContent.indexOf("Fusce posuere, magna sed pulvinar
ultricies, purus lectus malesuada libero, sit amet commodo magna eros quis
urna.") != -1);
+ CPPUNIT_ASSERT(tocContent.indexOf("Pellentesque habitant morbi tristique
senectus et netus et malesuada fames ac turpis egestas.") != -1);
+}
+
DECLARE_OOXMLEXPORT_TEST(testTdf142407, "tdf142407.docx")
{
uno::Reference<container::XNameAccess> xPageStyles =
getStyles("PageStyles");
diff --git a/writerfilter/source/dmapper/DomainMapper_Impl.cxx
b/writerfilter/source/dmapper/DomainMapper_Impl.cxx
index bcfcbc7c72dd..3f33280f5a22 100644
--- a/writerfilter/source/dmapper/DomainMapper_Impl.cxx
+++ b/writerfilter/source/dmapper/DomainMapper_Impl.cxx
@@ -6279,7 +6279,9 @@ void DomainMapper_Impl::handleToc
uno::Sequence< OUString> aStyles( nLevelCount );
for ( auto& rStyle : asNonConstRange(aStyles) )
{
- rStyle = (aTOCStyleIter++)->second;
+ // tdf#153082 must map w:styleId to w:name
+ rStyle =
GetStyleSheetTable()->ConvertStyleName(aTOCStyleIter->second, true);
+ ++aTOCStyleIter;
}
xParaStyles->replaceByIndex(nLevel - 1, uno::Any(aStyles));
}
diff --git a/writerfilter/source/dmapper/StyleSheetTable.cxx
b/writerfilter/source/dmapper/StyleSheetTable.cxx
index 9641a2a032bd..1a767712ea94 100644
--- a/writerfilter/source/dmapper/StyleSheetTable.cxx
+++ b/writerfilter/source/dmapper/StyleSheetTable.cxx
@@ -42,6 +42,7 @@
#include <map>
#include <osl/diagnose.h>
#include <rtl/ustrbuf.hxx>
+#include <rtl/character.hxx>
#include <sal/log.hxx>
#include <comphelper/propertyvalue.hxx>
#include <comphelper/string.hxx>
@@ -1412,6 +1413,30 @@ const StyleSheetEntryPtr &
StyleSheetTable::GetCurrentEntry() const
return m_pImpl->m_pCurrentEntry;
}
+/**
+ This is a heuristic to find Word's w:styleId value from localised style name.
+ It's not clear how exactly it works, but apparently Word stores into
+ w:styleId some filtered representation of the localised style name.
+ Tragically there are references to the localised style name itself in TOC
+ fields.
+ Hopefully this works and a complete map of >100 built-in style names
+ localised to all langauges isn't needed.
+*/
+static auto FilterChars(OUString const& rStyleName) -> OUString
+{
+ OUStringBuffer ret;
+ sal_Int32 index(0);
+ while (index < rStyleName.getLength())
+ {
+ auto const c(rStyleName.iterateCodePoints(&index));
+ if (rtl::isAsciiAlphanumeric(c))
+ {
+ ret.appendUtf32(c);
+ }
+ }
+ return ret.makeStringAndClear();
+}
+
OUString StyleSheetTable::ConvertStyleName( const OUString& rWWName, bool
bExtendedSearch)
{
OUString sRet( rWWName );
@@ -1419,6 +1444,10 @@ OUString StyleSheetTable::ConvertStyleName( const
OUString& rWWName, bool bExten
{
//search for the rWWName in the IdentifierD of the existing styles and
convert the sStyleName member
auto findIt = m_pImpl->m_aStyleSheetEntriesMap.find(rWWName);
+ if (findIt == m_pImpl->m_aStyleSheetEntriesMap.end())
+ {
+ findIt =
m_pImpl->m_aStyleSheetEntriesMap.find(FilterChars(rWWName));
+ }
if (findIt != m_pImpl->m_aStyleSheetEntriesMap.end())
sRet = findIt->second->sStyleName;
}
