sw/inc/swscanner.hxx | 4 ++++
sw/source/core/txtnode/txtedt.cxx | 25 +++++++++++++++++--------
2 files changed, 21 insertions(+), 8 deletions(-)
New commits:
commit 50e722923d4efaadac7a11ec523501cebd341cb1
Author: Noel Grandin <noel.gran...@collabora.co.uk>
AuthorDate: Fri Mar 14 15:49:32 2025 +0200
Commit: Caolán McNamara <caolan.mcnam...@collabora.com>
CommitDate: Fri Mar 14 18:01:51 2025 +0100
cool#1125 Reduce cost of spell checking in writer
Loading a new CharClass for each NextWord() call can add up.
Change-Id: Ie15f2adcd4c603074320b49d5c0276f558aab3e9
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/182920
Tested-by: Jenkins CollaboraOffice <jenkinscollaboraoff...@gmail.com>
Reviewed-by: Caolán McNamara <caolan.mcnam...@collabora.com>
Tested-by: Caolán McNamara <caolan.mcnam...@collabora.com>
diff --git a/sw/inc/swscanner.hxx b/sw/inc/swscanner.hxx
index 88433bfe4bb1..8a26085bc40f 100644
--- a/sw/inc/swscanner.hxx
+++ b/sw/inc/swscanner.hxx
@@ -22,8 +22,10 @@
#include <i18nlangtag/lang.h>
#include "modeltoviewhelper.hxx"
+#include <unotools/charclass.hxx>
#include <functional>
+#include <optional>
class SwTextNode;
@@ -43,6 +45,8 @@ class SwScanner
sal_Int32 m_nBegin;
sal_Int32 m_nLength;
sal_Int32 m_nOverriddenDashCount;
+ // caches the CharClass for m_aCurrentLang, which can be expensive to
repeatedly retrieve
+ std::optional<CharClass> moCharClass;
LanguageType m_aCurrentLang;
sal_uInt16 m_nWordType;
bool m_bClip;
diff --git a/sw/source/core/txtnode/txtedt.cxx
b/sw/source/core/txtnode/txtedt.cxx
index df0582e1f058..87ef9fa945e1 100644
--- a/sw/source/core/txtnode/txtedt.cxx
+++ b/sw/source/core/txtnode/txtedt.cxx
@@ -35,7 +35,6 @@
#include <osl/diagnose.h>
#include <officecfg/Office/Writer.hxx>
#include <unotools/transliterationwrapper.hxx>
-#include <unotools/charclass.hxx>
#include <sal/log.hxx>
#include <swmodule.hxx>
#include <splargs.hxx>
@@ -778,15 +777,21 @@
SwScanner::SwScanner(std::function<LanguageType(sal_Int32, sal_Int32, bool)> aGe
assert(m_aPreDashReplacementText.getLength() == m_aText.getLength());
+ LanguageType aNewLang;
if ( m_pLanguage )
{
- m_aCurrentLang = *m_pLanguage;
+ aNewLang = *m_pLanguage;
}
else
{
ModelToViewHelper::ModelPosition aModelBeginPos =
m_ModelToView.ConvertToModelPosition( m_nBegin );
- m_aCurrentLang = m_pGetLangOfChar(aModelBeginPos.mnPos, 0, true);
+ aNewLang = m_pGetLangOfChar(aModelBeginPos.mnPos, 0, true);
+ }
+ if (m_aCurrentLang != aNewLang)
+ {
+ m_aCurrentLang = aNewLang;
+ moCharClass.reset();
}
}
@@ -855,8 +860,6 @@ bool SwScanner::NextWord()
m_nBegin = m_nBegin + m_nLength;
Boundary aBound;
- std::optional<CharClass> xLocalCharClass;
-
while ( true )
{
// skip non-letter characters:
@@ -869,13 +872,19 @@ bool SwScanner::NextWord()
const sal_uInt16 nNextScriptType =
g_pBreakIt->GetBreakIter()->getScriptType( m_aText, m_nBegin );
ModelToViewHelper::ModelPosition aModelBeginPos =
m_ModelToView.ConvertToModelPosition( m_nBegin );
- m_aCurrentLang = m_pGetLangOfChar(aModelBeginPos.mnPos,
nNextScriptType, false);
+ LanguageType aNewLang =
m_pGetLangOfChar(aModelBeginPos.mnPos, nNextScriptType, false);
+ if (aNewLang != m_aCurrentLang)
+ {
+ m_aCurrentLang = aNewLang;
+ moCharClass.reset();
+ }
}
if ( m_nWordType != i18n::WordType::WORD_COUNT )
{
- xLocalCharClass.emplace(LanguageTag(
g_pBreakIt->GetLocale( m_aCurrentLang ) ));
- if (
xLocalCharClass->isLetterNumeric(OUString(m_aText[m_nBegin])) )
+ if (!moCharClass)
+ moCharClass.emplace(LanguageTag(
g_pBreakIt->GetLocale( m_aCurrentLang ) ));
+ if (
moCharClass->isLetterNumeric(OUString(m_aText[m_nBegin])) )
break;
}
else
