[PATCH-3-5] speedup autocorrect database import ...

[Michael Meeks]

        Large autocorrect databases cause quite some slowness on first
keystroke for people; turns out we had quite a gratuitous N^2 on load
there involving some ICU collation goodness ;-)

        It's still not ideal to have a multi-second freeze, but at least it's
half the length it was ;-)

        Review appreciated,

        Thanks,

                Michael.

On Sat, 2012-04-28 at 06:32 +0000, wrote:
> https://bugs.freedesktop.org/show_bug.cgi?id=46805
>
> --- Comment #22 from tommy27 <ba...@quipo.it> 2012-04-27 23:32:34 PDT ---
> ok, I downloaded the:
> master~2012-04-27_21.25.23_LibO-Dev_3.6.0alpha0_Win_x86_install_en-US.msi
> 
> and made comparative testings.
...
> 1 large autocorrect database (acor_.dat)
> time to see the white space
> 
> LibO    3.5.2 --> 9 seconds
> LibOdev 3.6.0 --> 5 seconds
> 
> 2 large autocorrect databases (acor_.dat ; acor_it-IT.dat)
> time to see the white space
> 
> LibO    3.5.2 --> 14 seconds
> LibOdev 3.6.0 -->  9 seconds

-- 
michael.me...@suse.com  <><, Pseudo Engineer, itinerant idiot

>From b269dced445494891a8e6e8d1d62b931a31dddbd Mon Sep 17 00:00:00 2001
From: Michael Meeks <michael.me...@suse.com>
Date: Thu, 26 Apr 2012 11:59:02 +0100
Subject: [PATCH] fdo#46805 - special-case appending items to autocorrect
 lists

---
 editeng/source/misc/svxacorr.cxx |   14 ++++++++++++++
 1 files changed, 14 insertions(+), 0 deletions(-)

diff --git a/editeng/source/misc/svxacorr.cxx b/editeng/source/misc/svxacorr.cxx
index f459af0..ce5c71c 100644
--- a/editeng/source/misc/svxacorr.cxx
+++ b/editeng/source/misc/svxacorr.cxx
@@ -68,6 +68,7 @@
 #include <com/sun/star/ucb/NameClash.hpp>
 #include <xmloff/xmltoken.hxx>
 #include <vcl/help.hxx>
+#include <rtl/logfile.hxx>
 
 #define CHAR_HARDBLANK      ((sal_Unicode)0x00A0)
 
@@ -233,15 +234,27 @@ void SvxAutocorrWordList::DeleteAndDestroy( sal_uInt16 nP, sal_uInt16 nL )
 }
 
 
+// Keep the list sorted ...
 sal_Bool SvxAutocorrWordList::Seek_Entry( const SvxAutocorrWordPtr aE, sal_uInt16* pP ) const
 {
     register sal_uInt16 nO  = SvxAutocorrWordList_SAR::Count(),
             nM,
             nU = 0;
+
     if( nO > 0 )
     {
         CollatorWrapper& rCmp = ::GetCollatorWrapper();
         nO--;
+
+        // quick check of the end of the list
+        if (rCmp.compareString( aE->GetShort(),
+                                (*((SvxAutocorrWordPtr*)pData + nO))->GetShort() ) > 0)
+        {
+            if( pP ) *pP = nO + 1;
+            return sal_False;
+        }
+
+        // Incredibly crude sort algorithm, should use some partitioning search.
         while( nU <= nO )
         {
             nM = nU + ( nO - nU ) / 2;
@@ -2169,6 +2182,7 @@ SvxAutocorrWordList* SvxAutoCorrectLanguageLists::LoadAutocorrWordList()
         OSL_ENSURE( xXMLParser.is(), "XMLReader::Read: com.sun.star.xml.sax.Parser service missing" );
         if( xXMLParser.is() )
         {
+            RTL_LOGFILE_PRODUCT_CONTEXT( aLog, "AutoCorrect Import" );
             uno::Reference< xml::sax::XDocumentHandler > xFilter = new SvXMLAutoCorrectImport( xServiceFactory, pAutocorr_List, rAutoCorrect, xStg );
 
             // connect parser and filter
-- 
1.7.9


_______________________________________________
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice