Resolves: tdf#96561 Include embedded null-characters while assembling CSV line

... instead of prematurely ending analysis and skipping the
remainder that is still part of data read.

Change-Id: I8f34bc9672e37f9c1c1ae81ddeba464360add7c2
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/120555
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
(cherry picked from commit 7bd03074ae7362d8fffe5182529626a0b43a0ed1)
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/120572
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
diff --git a/sc/source/ui/docshell/impex.cxx b/sc/source/ui/docshell/impex.cxx
index 3f86cc8..7233b14 100644
--- a/sc/source/ui/docshell/impex.cxx
+++ b/sc/source/ui/docshell/impex.cxx
@@ -2476,8 +2476,17 @@ Label_RetryWithNewSep:
        while (!rStream.eof() && aStr.getLength() < nArbitraryLineLengthLimit)
        {
            const sal_Unicode * p = aStr.getStr() + nLastOffset;
            while (*p)
            const sal_Unicode * const pStop = aStr.getStr() + aStr.getLength();
            while (p < pStop)
            {
                if (!*p)
                {
                    // Skip embedded null-characters. They don't change
                    // anything and are handled at a higher level.
                    ++p;
                    continue;
                }

                if (nQuotes)
                {
                    if (*p == cFieldQuote)