From 5c5d546a393e68203ab853a99575d643672655cc Mon Sep 17 00:00:00 2001
From: Dimitri van Heesch <doxygen@gmail.com>
Date: Sun, 22 Dec 2019 21:44:19 +0100
Subject: Optimize UTF-8 nbsp conversion in markdown

---
 src/markdown.cpp | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

diff --git a/src/markdown.cpp b/src/markdown.cpp
index d63492e..0a71aae 100644
--- a/src/markdown.cpp
+++ b/src/markdown.cpp
@@ -108,8 +108,8 @@ static Entry         *g_current;
 static QCString       g_fileName;
 static int            g_lineNr;
 static int            g_indentLevel=0;  // 0 is outside markdown, -1=page level
-static char           g_doxynbsputf8[3] = {'\xc2', '\xa0', '\0'}; // UTF-8 nbsp
-static char          *g_doxynbsp = "&_doxy_nbsp;";
+static const char     g_utf8_nbsp[3] = {'\xc2', '\xa0', '\0'}; // UTF-8 nbsp
+static const char    *g_doxy_nsbp = "&_doxy_nbsp;";            // doxygen escape command for UTF-8 nbsp
 //----------
 
 const int codeBlockIndent = 4;
@@ -1029,6 +1029,17 @@ static int processCodeSpan(GrowBuf &out, const char *data, int /*offset*/, int s
   return end;
 }
 
+static void addStrEscapeUtf8Nbsp(GrowBuf &out,const char *s,int len)
+{
+  if (strnstr(s,g_doxy_nsbp,len)==0) // no escape needed -> fast
+  {
+    out.addStr(s,len);
+  }
+  else // escape needed -> slow
+  {
+    out.addStr(substitute(QCString(s).left(len),g_doxy_nsbp,g_utf8_nbsp));
+  }
+}
 
 static int processSpecialCommand(GrowBuf &out, const char *data, int offset, int size)
 {
@@ -1045,7 +1056,7 @@ static int processSpecialCommand(GrowBuf &out, const char *data, int offset, int
         if (qstrncmp(&data[i+1],endBlockName,l)==0)
         {
           //printf("found end at %d\n",i);
-          out.addStr(substitute(QCString(data).left(i+1+l),g_doxynbsp,g_doxynbsputf8));
+          addStrEscapeUtf8Nbsp(out,data,i+1+l);
           return i+1+l;
         }
       }
@@ -2175,7 +2186,7 @@ static void writeFencedCodeBlock(GrowBuf &out,const char *data,const char *lng,
   {
     out.addStr("{"+lang+"}");
   }
-  out.addStr(substitute(QCString(data+blockStart).left(blockEnd-blockStart),g_doxynbsp,g_doxynbsputf8));
+  addStrEscapeUtf8Nbsp(out,data+blockStart,blockEnd-blockStart);
   out.addStr("\n");
   out.addStr("@endcode\n");
 }
@@ -2485,7 +2496,7 @@ static QCString detab(const QCString &s,int &refIndent)
           // special handling of the UTF-8 nbsp character 0xc2 0xa0
           if (c == '\xc2' && data[i] == '\xa0')
           {
-            out.addStr(g_doxynbsp);
+            out.addStr(g_doxy_nsbp);
             i++;
           }
           else
@@ -2562,7 +2573,7 @@ QCString processMarkdown(const QCString &fileName,const int lineNr,Entry *e,cons
   processInline(out,s,s.length());
   out.addChar(0);
   Debug::print(Debug::Markdown,0,"======== Markdown =========\n---- input ------- \n%s\n---- output -----\n%s\n=========\n",qPrint(input),qPrint(out.get()));
-  return substitute(out.get(),g_doxynbsp,"&nbsp;");
+  return substitute(out.get(),g_doxy_nsbp,"&nbsp;");
 }
 
 //---------------------------------------------------------------------------
-- 
cgit v0.12