move and rename sanitizeString for reuse

Change-Id: I8378f3df79e511cf2d385ace1cd7964ab1c76e7b
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/144388
Tested-by: Caolán McNamara <caolanm@redhat.com>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>

3 files changed, 37 insertions, 27 deletions

diff --git a/comphelper/source/misc/string.cxx b/comphelper/source/misc/string.cxx
index d0f7cb9ca423..da5c8b92c05c 100644
--- a/comphelper/source/misc/string.cxx
+++ b/comphelper/source/misc/string.cxx
@@ -32,6 +32,7 @@
 #include <rtl/ustrbuf.hxx>
 #include <rtl/string.hxx>
 #include <rtl/strbuf.hxx>
+#include <sal/log.hxx>
 #include <sal/types.h>
 
 #include <comphelper/string.hxx>
@@ -652,6 +653,32 @@ void replaceAt(OUStringBuffer& rIn, sal_Int32 nIndex, sal_Int32 nCount, std::u16
     rIn.setLength(nNewLength);
 }
 
+OUString sanitizeStringSurrogates(const OUString& rString)
+{
+    sal_Int32 i=0;
+    while (i < rString.getLength())
+    {
+        sal_Unicode c = rString[i];
+        if (rtl::isHighSurrogate(c))
+        {
+            if (i+1 == rString.getLength()
+                || !rtl::isLowSurrogate(rString[i+1]))
+            {
+                SAL_WARN("comphelper", "Surrogate error: high without low");
+                return rString.copy(0, i);
+            }
+            ++i;    //skip correct low
+        }
+        if (rtl::isLowSurrogate(c)) //bare low without preceding high
+        {
+            SAL_WARN("comphelper", "Surrogate error: low without high");
+            return rString.copy(0, i);
+        }
+        ++i;
+    }
+    return rString;
+}
+
 }
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/include/comphelper/string.hxx b/include/comphelper/string.hxx
index b43086a77d30..06c825b36832 100644
--- a/include/comphelper/string.hxx
+++ b/include/comphelper/string.hxx
@@ -375,6 +375,15 @@ COMPHELPER_DLLPUBLIC bool isdigitAsciiString(std::string_view rString);
  */
 COMPHELPER_DLLPUBLIC bool isdigitAsciiString(std::u16string_view rString);
 
+/** Santitize an OUString to not have invalid surrogates
+
+    @param rString  An OUString
+
+    @return         same string if no surrogates or surrogates are valid.
+                    Otherwise the string truncated to the valid sequence.
+ */
+COMPHELPER_DLLPUBLIC OUString sanitizeStringSurrogates(const OUString& rString);
+
 } // namespace comphelper::string
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sw/source/filter/ww8/ww8par3.cxx b/sw/source/filter/ww8/ww8par3.cxx
index f02a2972e058..445e5442a0fe 100644
--- a/sw/source/filter/ww8/ww8par3.cxx
+++ b/sw/source/filter/ww8/ww8par3.cxx
@@ -490,32 +490,6 @@ WW8LSTInfo* WW8ListManager::GetLSTByListId( sal_uInt32 nIdLst ) const
     return aResult->get();
 }
 
-static OUString sanitizeString(const OUString& rString)
-{
-    sal_Int32 i=0;
-    while (i < rString.getLength())
-    {
-        sal_Unicode c = rString[i];
-        if (rtl::isHighSurrogate(c))
-        {
-            if (i+1 == rString.getLength()
-                || !rtl::isLowSurrogate(rString[i+1]))
-            {
-                SAL_WARN("sw.ww8", "Surrogate error: high without low");
-                return rString.copy(0, i);
-            }
-            ++i;    //skip correct low
-        }
-        if (rtl::isLowSurrogate(c)) //bare low without preceding high
-        {
-            SAL_WARN("sw.ww8", "Surrogate error: low without high");
-            return rString.copy(0, i);
-        }
-        ++i;
-    }
-    return rString;
-}
-
 SvxNumType WW8ListManager::GetSvxNumTypeFromMSONFC(sal_uInt16 nNFC)
 {
     SvxNumType nType(SVX_NUM_ARABIC);
@@ -874,7 +848,7 @@ bool WW8ListManager::ReadLVL(SwNumFormat& rNumFormat, std::unique_ptr<SfxItemSet
 
     // 4. Read numbering String. Results in prefix and postfix
 
-    OUString sNumString(sanitizeString(read_uInt16_PascalString(m_rSt)));
+    OUString sNumString(comphelper::string::sanitizeStringSurrogates(read_uInt16_PascalString(m_rSt)));
 
     // 5. convert read values into Writer syntax