From a8a0334dfc94432f22f0e1452dfaf2dabf2fe780 Mon Sep 17 00:00:00 2001
From: Stephan Bergmann <sbergman@redhat.com>
Date: Fri, 7 Oct 2016 15:02:47 +0200
Subject: [PATCH] rhbz#1382401: Support surrogate pairs in HTMLOutFuncs

(cherry picked from commit 375b99cad4a79d26a6cbcd0f71bc12b312d95818)
Conflicts:
	svtools/source/svhtml/htmlout.cxx
plus cherry-pick of 6131bf9c96fb2ae37decf13e453f27304707271b "Minor performance
improvement of previous patch"

Change-Id: Ib578f758e4f5f355a79a014c2ad4660924dd34a4
---
 include/svtools/htmlout.hxx       |  2 +-
 svtools/source/svhtml/htmlout.cxx | 40 +++++++++++++++++++++++++++------------
 2 files changed, 29 insertions(+), 13 deletions(-)

diff --git a/include/svtools/htmlout.hxx b/include/svtools/htmlout.hxx
index 715f0ba..e8a5d50 100644
--- a/include/svtools/htmlout.hxx
+++ b/include/svtools/htmlout.hxx
@@ -60,7 +60,7 @@ struct HTMLOutFuncs
     SVT_DLLPUBLIC static SvStream& Out_AsciiTag( SvStream&, const sal_Char* pStr,
                                    bool bOn = true,
                         rtl_TextEncoding eDestEnc = RTL_TEXTENCODING_MS_1252);
-    SVT_DLLPUBLIC static SvStream& Out_Char( SvStream&, sal_Unicode cChar,
+    SVT_DLLPUBLIC static SvStream& Out_Char( SvStream&, sal_uInt32 cChar,
                         HTMLOutContext& rContext,
                         OUString *pNonConvertableChars = nullptr );
     SVT_DLLPUBLIC static SvStream& Out_String( SvStream&, const OUString&,
diff --git a/svtools/source/svhtml/htmlout.cxx b/svtools/source/svhtml/htmlout.cxx
index 99d9e38..4305338 100644
--- a/svtools/source/svhtml/htmlout.cxx
+++ b/svtools/source/svhtml/htmlout.cxx
@@ -55,7 +55,7 @@ HTMLOutContext::~HTMLOutContext()
     rtl_destroyUnicodeToTextConverter( m_hConv );
 }
 
-static const sal_Char *lcl_svhtml_GetEntityForChar( sal_Unicode c,
+static const sal_Char *lcl_svhtml_GetEntityForChar( sal_uInt32 c,
                                              rtl_TextEncoding eDestEnc )
 {
     const sal_Char* pStr = nullptr;
@@ -388,10 +388,12 @@ static const sal_Char *lcl_svhtml_GetEntityForChar( sal_Unicode c,
     return pStr;
 }
 
-static OString lcl_ConvertCharToHTML( sal_Unicode c,
+static OString lcl_ConvertCharToHTML( sal_uInt32 c,
                             HTMLOutContext& rContext,
                             OUString *pNonConvertableChars )
 {
+    assert(rtl::isUnicodeCodePoint(c));
+
     OStringBuffer aDest;
     DBG_ASSERT( RTL_TEXTENCODING_DONTKNOW != rContext.m_eDestEnc,
                     "wrong destination encoding" );
@@ -439,8 +441,18 @@ static OString lcl_ConvertCharToHTML( sal_Unicode c,
     }
     else
     {
+        sal_Unicode utf16[2];
+        sal_Size n;
+        if (c < 0x10000) {
+            utf16[0] = c;
+            n = 1;
+        } else {
+            utf16[0] = rtl::getHighSurrogate(c);
+            utf16[1] = rtl::getLowSurrogate(c);
+            n = 2;
+        }
         sal_Size nLen = rtl_convertUnicodeToText( rContext.m_hConv,
-                                                  rContext.m_hContext, &c, 1,
+                                                  rContext.m_hContext, utf16, n,
                                                      cBuffer, TXTCONV_BUFFER_SIZE,
                                                   nFlags,
                                                   &nInfo, &nSrcChars );
@@ -466,11 +478,15 @@ static OString lcl_ConvertCharToHTML( sal_Unicode c,
             while( nLen-- )
                 aDest.append(*pBuffer++);
 
-            aDest.append('&').append('#').append(static_cast<sal_Int64>(c))
+            aDest.append('&').append('#').append(static_cast<sal_Int32>(c))
+                    // Unicode code points guaranteed to fit into sal_Int32
                  .append(';');
-            if( pNonConvertableChars &&
-                -1 == pNonConvertableChars->indexOf( c ) )
-                (*pNonConvertableChars) += OUString(c);
+            if( pNonConvertableChars )
+            {
+                OUString cs(&c, 1);
+                if( -1 == pNonConvertableChars->indexOf( cs ) )
+                    (*pNonConvertableChars) += cs;
+            }
         }
     }
     return aDest.makeStringAndClear();
@@ -505,9 +521,9 @@ OString HTMLOutFuncs::ConvertStringToHTML( const OUString& rSrc,
 {
     HTMLOutContext aContext( eDestEnc );
     OStringBuffer aDest;
-    for( sal_Int32 i=0, nLen = rSrc.getLength(); i < nLen; i++ )
+    for( sal_Int32 i=0, nLen = rSrc.getLength(); i < nLen; )
         aDest.append(lcl_ConvertCharToHTML(
-            rSrc[i], aContext, pNonConvertableChars));
+            rSrc.iterateCodePoints(&i), aContext, pNonConvertableChars));
     aDest.append(lcl_FlushToAscii(aContext));
     return aDest.makeStringAndClear();
 }
@@ -525,7 +541,7 @@ SvStream& HTMLOutFuncs::Out_AsciiTag( SvStream& rStream, const sal_Char *pStr,
     return rStream;
 }
 
-SvStream& HTMLOutFuncs::Out_Char( SvStream& rStream, sal_Unicode c,
+SvStream& HTMLOutFuncs::Out_Char( SvStream& rStream, sal_uInt32 c,
                                   HTMLOutContext& rContext,
                                   OUString *pNonConvertableChars )
 {
@@ -540,8 +556,8 @@ SvStream& HTMLOutFuncs::Out_String( SvStream& rStream, const OUString& rOUStr,
 {
     HTMLOutContext aContext( eDestEnc );
     sal_Int32 nLen = rOUStr.getLength();
-    for( sal_Int32 n = 0; n < nLen; n++ )
-        HTMLOutFuncs::Out_Char( rStream, rOUStr[n],
+    for( sal_Int32 n = 0; n < nLen; )
+        HTMLOutFuncs::Out_Char( rStream, rOUStr.iterateCodePoints(&n),
                                 aContext, pNonConvertableChars );
     HTMLOutFuncs::FlushToAscii( rStream, aContext );
     return rStream;
-- 
2.7.4