tdf133647 tdf123386 tdf123389 Improved .docx table formula import

Converts table formula syntax from MS Word to LibreOffice.
This version uses the list separator of the document for the
formula regexen; however, it does not convert the decimal or
list separators in the case where the person opening the document
is using a different locale from the author.

Change-Id: I9600a0bea060a76705a7ad6b051ed4fdd50b9d40
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/98614
Tested-by: Jenkins
Tested-by: László Németh <nemeth@numbertext.org>
Reviewed-by: László Németh <nemeth@numbertext.org>

5 files changed, 95 insertions, 3 deletions

diff --git a/writerfilter/Library_writerfilter.mk b/writerfilter/Library_writerfilter.mk
index 61e62d66e10d..cbda03e9dcdd 100644
--- a/writerfilter/Library_writerfilter.mk
+++ b/writerfilter/Library_writerfilter.mk
@@ -51,7 +51,9 @@ $(eval $(call gb_Library_use_libraries,writerfilter,\
 
 $(eval $(call gb_Library_use_externals,writerfilter,\
 	boost_headers \
-	icu_headers \
+    icui18n \
+    icuuc \
+    icu_headers \
 	libxml2 \
 ))
 
diff --git a/writerfilter/source/dmapper/DomainMapper_Impl.cxx b/writerfilter/source/dmapper/DomainMapper_Impl.cxx
index e33a6d753364..4e8afddc8d50 100644
--- a/writerfilter/source/dmapper/DomainMapper_Impl.cxx
+++ b/writerfilter/source/dmapper/DomainMapper_Impl.cxx
@@ -88,6 +88,8 @@
 #include <map>
 #include <tuple>
 #include <unordered_map>
+#include <regex>
+#include <algorithm>
 
 #include <officecfg/Office/Common.hxx>
 #include <filter/msfilter/util.hxx>
@@ -100,6 +102,9 @@
 #include <tools/diagnose_ex.h>
 #include <sal/log.hxx>
 
+#include <unicode/errorcode.h>
+#include <unicode/regex.h>
+
 using namespace ::com::sun::star;
 using namespace oox;
 namespace writerfilter::dmapper{
@@ -4196,6 +4201,70 @@ void DomainMapper_Impl::handleFieldAsk
     }
 }
 
+/**
+ * Converts a Microsoft Word field formula into LibreOffice syntax
+ * @param input The Microsoft Word field formula, with no leading '=' sign
+ * @return An equivalent LibreOffice field formula
+ */
+OUString DomainMapper_Impl::convertFieldFormula(const OUString& input) {
+
+    OUString listSeparator = m_pSettingsTable->GetListSeparator();
+
+    /* Replace logical condition functions with LO equivalent operators */
+    OUString changed = input.replaceAll(" <> ", " NEQ ");
+    changed = changed.replaceAll(" <= ", " LEQ ");
+    changed = changed.replaceAll(" >= ", " GEQ ");
+    changed = changed.replaceAll(" = " , " EQ ");
+    changed = changed.replaceAll(" < " , " L ");
+    changed = changed.replaceAll(" > " , " G ");
+
+    changed = changed.replaceAll("<>", " NEQ ");
+    changed = changed.replaceAll("<=", " LEQ ");
+    changed = changed.replaceAll(">=", " GEQ ");
+    changed = changed.replaceAll("=" , " EQ ");
+    changed = changed.replaceAll("<" , " L ");
+    changed = changed.replaceAll(">" , " G ");
+
+    /* Replace function calls with infix keywords for AND(), OR(), and ROUND(). Nothing needs to be
+     * done for NOT(). This simple regex will work properly with most common cases. However, it may
+     * not work correctly when the arguments are nested subcalls to other functions, like
+     * ROUND(MIN(1,2),MAX(3,4)). See TDF#134765.  */
+    icu::ErrorCode status;
+    icu::UnicodeString usInput(changed.getStr());
+    const uint32_t rMatcherFlags = UREGEX_CASE_INSENSITIVE;
+    OUString regex = "\\b(AND|OR|ROUND)\\s*\\(\\s*([^" + listSeparator + "]+)\\s*" + listSeparator + "\\s*([^)]+)\\s*\\)";
+    icu::UnicodeString usRegex(regex.getStr());
+    icu::RegexMatcher rmatch1(usRegex, usInput, rMatcherFlags, status);
+    usInput = rmatch1.replaceAll(icu::UnicodeString("(($2) $1 ($3))"), status);
+
+    /* Assumes any remaining list separators separate arguments to functions that accept lists
+     * (SUM, MIN, MAX, MEAN, etc.) */
+    usInput.findAndReplace(icu::UnicodeString(listSeparator.getStr()), "|");
+
+    /* Surround single cell references with angle brackets.
+     * If there is ever added a function name that ends with a digit, this regex will need to be revisited. */
+    icu::RegexMatcher rmatch2("\\b([A-Z]{1,3}[0-9]+)\\b(?![(])", usInput, rMatcherFlags, status);
+    usInput = rmatch2.replaceAll(icu::UnicodeString("<$1>"), status);
+
+    /* Cell references must be upper case */
+    icu::RegexMatcher rmatch3("<[a-z]{1,3}[0-9]+>", usInput, rMatcherFlags, status);
+    icu::UnicodeString replacedCellRefs;
+    while (rmatch3.find(status) && status.isSuccess()) {
+        rmatch3.appendReplacement(replacedCellRefs, rmatch3.group(status).toUpper(), status);
+    }
+    rmatch3.appendTail(replacedCellRefs);
+
+    /* Fix up cell ranges */
+    icu::RegexMatcher rmatch4("<([A-Z]{1,3}[0-9]+)>:<([A-Z]{1,3}[0-9]+)>", replacedCellRefs, rMatcherFlags, status);
+    usInput = rmatch4.replaceAll(icu::UnicodeString("<$1:$2>"), status);
+
+    /* Fix up user defined names */
+    icu::RegexMatcher rmatch5("DEFINED\\s*\\(<([A-Z]+[0-9]+)>\\)", usInput, rMatcherFlags, status);
+    usInput = rmatch5.replaceAll(icu::UnicodeString("DEFINED($1)"), status);
+
+    return OUString(usInput.getTerminatedBuffer());
+}
+
 void DomainMapper_Impl::handleFieldFormula
     (const FieldContextPtr& pContext,
      uno::Reference< beans::XPropertySet > const& xFieldProperties)
@@ -4215,7 +4284,7 @@ void DomainMapper_Impl::handleFieldFormula
         return;
 
     // we don't copy the = symbol from the command
-    OUString formula = command.copy(1);
+    OUString formula = convertFieldFormula(command.copy(1));
 
     xFieldProperties->setPropertyValue(getPropertyName(PROP_CONTENT), uno::makeAny(formula));
     xFieldProperties->setPropertyValue(getPropertyName(PROP_NUMBER_FORMAT), uno::makeAny(sal_Int32(0)));
diff --git a/writerfilter/source/dmapper/DomainMapper_Impl.hxx b/writerfilter/source/dmapper/DomainMapper_Impl.hxx
index 2597f6877f34..54a0f6c7ac25 100644
--- a/writerfilter/source/dmapper/DomainMapper_Impl.hxx
+++ b/writerfilter/source/dmapper/DomainMapper_Impl.hxx
@@ -818,7 +818,8 @@ public:
         (const FieldContextPtr& pContext,
         css::uno::Reference< css::uno::XInterface > & xFieldInterface,
         css::uno::Reference< css::beans::XPropertySet > const& xFieldProperties);
-    static void handleFieldFormula
+    OUString convertFieldFormula(const OUString& input);
+    void handleFieldFormula
         (const FieldContextPtr& pContext,
         css::uno::Reference< css::beans::XPropertySet > const& xFieldProperties);
     void handleAutoNum
diff --git a/writerfilter/source/dmapper/SettingsTable.cxx b/writerfilter/source/dmapper/SettingsTable.cxx
index 5ac2400cd493..2042ade7dc66 100644
--- a/writerfilter/source/dmapper/SettingsTable.cxx
+++ b/writerfilter/source/dmapper/SettingsTable.cxx
@@ -265,6 +265,8 @@ struct SettingsTable_Impl
     bool                m_bReadOnly;
     bool                m_bDisplayBackgroundShape;
     bool                m_bNoLeading = false;
+    OUString            m_sDecimalSymbol;
+    OUString            m_sListSeparator;
 
     uno::Sequence<beans::PropertyValue> m_pThemeFontLangProps;
 
@@ -303,6 +305,8 @@ struct SettingsTable_Impl
     , m_sRedlineProtectionKey()
     , m_bReadOnly(false)
     , m_bDisplayBackgroundShape(false)
+    , m_sDecimalSymbol(".")
+    , m_sListSeparator(",")
     , m_pThemeFontLangProps(3)
     , m_pCurrentCompatSetting(3)
     {}
@@ -479,8 +483,10 @@ void SettingsTable::lcl_sprm(Sprm& rSprm)
     case NS_ooxml::LN_CT_Settings_doNotIncludeSubdocsInStats: //  92554; // Do Not Include Content in Text Boxes, Footnotes, and Endnotes in Document Statistics)
     break;
     case NS_ooxml::LN_CT_Settings_decimalSymbol: //  92562;
+        m_pImpl->m_sDecimalSymbol = pValue->getString();
     break;
     case NS_ooxml::LN_CT_Settings_listSeparator: //  92563;
+        m_pImpl->m_sListSeparator = pValue->getString();
     break;
     case NS_ooxml::LN_CT_Settings_rsids: //  92549; revision save Ids - probably not necessary
     break;
@@ -687,6 +693,17 @@ sal_Int16 SettingsTable::GetHypenationZone() const
     return m_pImpl->m_nHyphenationZone;
 }
 
+OUString SettingsTable::GetDecimalSymbol() const
+{
+    return m_pImpl->m_sDecimalSymbol;
+}
+
+OUString SettingsTable::GetListSeparator() const
+{
+    return m_pImpl->m_sListSeparator;
+}
+
+
 uno::Sequence<beans::PropertyValue> const & SettingsTable::GetThemeFontLangProperties() const
 {
     return m_pImpl->m_pThemeFontLangProps;
diff --git a/writerfilter/source/dmapper/SettingsTable.hxx b/writerfilter/source/dmapper/SettingsTable.hxx
index d91db71ab1f2..2ff62576a980 100644
--- a/writerfilter/source/dmapper/SettingsTable.hxx
+++ b/writerfilter/source/dmapper/SettingsTable.hxx
@@ -78,6 +78,9 @@ class SettingsTable : public LoggedProperties, public LoggedTable
     bool GetNoHyphenateCaps() const;
     sal_Int16 GetHypenationZone() const;
 
+    OUString GetDecimalSymbol() const;
+    OUString GetListSeparator() const;
+
     css::uno::Sequence<css::beans::PropertyValue> const & GetThemeFontLangProperties() const;
 
     css::uno::Sequence<css::beans::PropertyValue> GetCompatSettings() const;