[Libreoffice-commits] core.git: sax/source

[dante (via logerrit)]

 sax/source/fastparser/fastparser.cxx |   52 +++++++++++++++++++++++------------
 1 file changed, 35 insertions(+), 17 deletions(-)

New commits:
commit 9edd554eb634bcb88add30e3290cb9a8650ce38f
Author:     dante <dante19031999 at gmail.com>
AuthorDate: Tue Dec 15 15:39:21 2020 +0100
Commit:     Noel Grandin <noel.grandin at collabora.co.uk>
CommitDate: Thu Dec 17 07:59:07 2020 +0100

    Sort custom entity names on fast parser
    
    When there are lots of entities and a lot of cases to replace, it becomes a ballast to search the whole array.
    So in order to avoid it, now uses ordered search and stops when OUString order implies that it can't be further.
    The entity list is sorted before the parse by quick sort.
    
    Change-Id: I9c91338ad67ddea1c273e329542549a904a0e563
    Reviewed-on: https://gerrit.libreoffice.org/c/core/+/107774
    Tested-by: Jenkins
    Reviewed-by: Noel Grandin <noel.grandin at collabora.co.uk>

diff --git a/sax/source/fastparser/fastparser.cxx b/sax/source/fastparser/fastparser.cxx
index f0fe5aea2f67..94d2651d4dff 100644
--- a/sax/source/fastparser/fastparser.cxx
+++ b/sax/source/fastparser/fastparser.cxx
@@ -200,6 +200,21 @@ struct Entity : public ParserData
     Event& getEvent( CallbackType aType );
 };
 
+// Stuff for custom entity names
+struct ReplacementPair
+{
+    OUString name;
+    OUString replacement;
+};
+inline bool operator<(const ReplacementPair& lhs, const ReplacementPair& rhs)
+{
+    return lhs.name < rhs.name;
+}
+inline bool operator<(const ReplacementPair& lhs, const char* rhs)
+{
+    return lhs.name.compareToAscii(rhs) < 0;
+}
+
 } // namespace
 
 namespace sax_fastparser {
@@ -211,8 +226,7 @@ public:
     ~FastSaxParserImpl();
 
 private:
-    ::css::uno::Sequence< ::rtl::OUString > mEntityNames;
-    ::css::uno::Sequence< ::rtl::OUString > mEntityReplacements;
+    std::vector<ReplacementPair> m_Replacements;
 
 public:
     // XFastParser
@@ -934,10 +948,18 @@ void FastSaxParserImpl::setNamespaceHandler( const Reference< XFastNamespaceHand
     maData.mxNamespaceHandler = Handler;
 }
 
-void FastSaxParserImpl::setCustomEntityNames( const ::css::uno::Sequence< ::rtl::OUString >& names, const ::css::uno::Sequence< ::rtl::OUString >& replacements )
+void FastSaxParserImpl::setCustomEntityNames(
+    const ::css::uno::Sequence<::rtl::OUString>& names,
+    const ::css::uno::Sequence<::rtl::OUString>& replacements)
 {
-    mEntityNames = names;
-    mEntityReplacements = replacements;
+    m_Replacements.resize(names.size());
+    for (size_t i = 0; i < names.size(); ++i)
+    {
+        m_Replacements[i].name = names[i];
+        m_Replacements[i].replacement = replacements[i];
+    }
+    if (names.size() > 1)
+        std::sort(m_Replacements.begin(), m_Replacements.end());
 }
 
 void FastSaxParserImpl::deleteUsedEvents()
@@ -1372,18 +1394,13 @@ xmlEntityPtr FastSaxParserImpl::callbackGetEntity( const xmlChar *name )
     int lname = strlen(dname);
     if( lname == 0 )
         return xmlGetPredefinedEntity(name);
-    if (mEntityNames.hasElements())
+    if (m_Replacements.size() > 0)
     {
-        for (size_t i = 0; i < mEntityNames.size(); ++i)
-        {
-            if (mEntityNames[i].compareToAscii(dname) == 0)
-            {
-                return xmlNewEntity(
-                    nullptr, name, XML_INTERNAL_GENERAL_ENTITY, nullptr, nullptr,
-                    BAD_CAST(
-                        OUStringToOString(mEntityReplacements[i], RTL_TEXTENCODING_UTF8).getStr()));
-            }
-        }
+        auto it = std::lower_bound(m_Replacements.begin(), m_Replacements.end(), dname);
+        if (it != m_Replacements.end() && it->name.compareToAscii(dname) == 0)
+            return xmlNewEntity(
+                nullptr, name, XML_INTERNAL_GENERAL_ENTITY, nullptr, nullptr,
+                BAD_CAST(OUStringToOString(it->replacement, RTL_TEXTENCODING_UTF8).getStr()));
     }
     if( lname < 2 )
         return xmlGetPredefinedEntity(name);
@@ -1495,7 +1512,8 @@ OUString FastSaxParser::getImplementationName()
     return "com.sun.star.comp.extensions.xml.sax.FastParser";
 }
 
-void FastSaxParser::setCustomEntityNames( const ::css::uno::Sequence< ::rtl::OUString >& names, const ::css::uno::Sequence< ::rtl::OUString >& replacements )
+void FastSaxParser::setCustomEntityNames(const ::css::uno::Sequence<::rtl::OUString>& names,
+                                         const ::css::uno::Sequence<::rtl::OUString>& replacements)
 {
     assert(names.size() == replacements.size());
     mpImpl->setCustomEntityNames(names, replacements);