Resolves: tdf#102846 CSV: Detect separator, limit preview line concatenations

In CSV import preview, if a line starts with a quote character and
none of the remembered last field separators used occur in data in
conjunction with a closing quote, then reading data tried to
concatenate line by line to form a data field to be presented in
the preview, worst case the entire file..

For the preview, detect one possible not yet selected separator if
used with a quoted field (similar to commit
c807e7ea7a0725a4d8375eda07d6f70870e0d50a for tdf#56910 space
separator) and limit the number of source lines that are tried to
be concatenated if no separator was encountered after a possibly
closing field quote.

Change-Id: Iefd37a8301161e72cb607cea88d4faadad47b4ae
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/120690
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins

1 files changed, 40 insertions, 10 deletions

diff --git a/sc/source/ui/docshell/impex.cxx b/sc/source/ui/docshell/impex.cxx
index 3a798b37e50e..3c2772619dac 100644
--- a/sc/source/ui/docshell/impex.cxx
+++ b/sc/source/ui/docshell/impex.cxx
@@ -607,17 +607,40 @@ static QuoteType lcl_isFieldEndQuote( const sal_Unicode* p, const sal_Unicode* p
     // Due to broken CSV generators that don't double embedded quotes check if
     // a field separator immediately or with trailing spaces follows the quote,
     // only then end the field, or at end of string.
-    const sal_Unicode cBlank = ' ';
+    constexpr sal_Unicode cBlank = ' ';
     if (p[1] == cBlank && ScGlobal::UnicodeStrChr( pSeps, cBlank))
         return FIELDEND_QUOTE;
     // Detect a possible blank separator if it's not already in the list (which
     // was checked right above for p[1]==cBlank).
-    if (p[1] == cBlank && !rcDetectSep && p[2] && p[2] != cBlank)
-        rcDetectSep = cBlank;
+    const bool bBlankSep = (p[1] == cBlank && !rcDetectSep && p[2] && p[2] != cBlank);
     while (p[1] == cBlank)
         ++p;
     if (!p[1] || ScGlobal::UnicodeStrChr( pSeps, p[1]))
         return FIELDEND_QUOTE;
+    // Extended separator detection after a closing quote (with or without
+    // blanks). Note that nQuotes is incremented *after* the call so is not yet
+    // even here, and that with separator detection we reach here only if
+    // lcl_isEscapedOrFieldEndQuote() did not already detect FIRST_QUOTE or
+    // SECOND_QUOTE for an escaped embedded quote, thus nQuotes does not have
+    // to be checked.
+    if (!rcDetectSep)
+    {
+        constexpr sal_Unicode vSep[] = { ',', '\t', ';' };
+        for (const sal_Unicode c : vSep)
+        {
+            if (p[1] == c)
+            {
+                rcDetectSep = c;
+                return FIELDEND_QUOTE;
+            }
+        }
+    }
+    // Blank separator is least significant, after others.
+    if (bBlankSep)
+    {
+        rcDetectSep = cBlank;
+        return FIELDEND_QUOTE;
+    }
     return DONTKNOW_QUOTE;
 }
 
@@ -645,7 +668,7 @@ static QuoteType lcl_isFieldEndQuote( const sal_Unicode* p, const sal_Unicode* p
 static QuoteType lcl_isEscapedOrFieldEndQuote( sal_Int32 nQuotes, const sal_Unicode* p,
         const sal_Unicode* pSeps, sal_Unicode cStr, sal_Unicode& rcDetectSep )
 {
-    if ((nQuotes % 2) == 0)
+    if ((nQuotes & 1) == 0)
     {
         if (p[-1] == cStr)
             return SECOND_QUOTE;
@@ -2481,7 +2504,7 @@ ScImportStringStream::ScImportStringStream( const OUString& rStr )
 }
 
 OUString ReadCsvLine( SvStream &rStream, bool bEmbeddedLineBreak,
-        OUString& rFieldSeparators, sal_Unicode cFieldQuote, sal_Unicode& rcDetectSep )
+        OUString& rFieldSeparators, sal_Unicode cFieldQuote, sal_Unicode& rcDetectSep, sal_uInt32 nMaxSourceLines )
 {
     enum RetryState
     {
@@ -2506,6 +2529,8 @@ Label_RetryWithNewSep:
 
     if (bEmbeddedLineBreak)
     {
+        sal_uInt32 nLine = 0;
+
         const sal_Unicode* pSeps = rFieldSeparators.getStr();
 
         QuoteType eQuoteState = FIELDEND_QUOTE;
@@ -2544,10 +2569,11 @@ Label_RetryWithNewSep:
                         {
                             eQuoteState = lcl_isEscapedOrFieldEndQuote( nQuotes, p, pSeps, cFieldQuote, rcDetectSep);
 
-                            if (eRetryState == RetryState::ALLOW && rcDetectSep == ' ')
+                            if (eRetryState == RetryState::ALLOW && rcDetectSep)
                             {
                                 eRetryState = RetryState::RETRY;
-                                rFieldSeparators += " ";
+                                rFieldSeparators += OUStringChar(rcDetectSep);
+                                pSeps = rFieldSeparators.getStr();
                                 goto Label_RetryWithNewSep;
                             }
 
@@ -2593,10 +2619,14 @@ Label_RetryWithNewSep:
                 ++p;
             }
 
-            if (nQuotes % 2 == 0)
+            if ((nQuotes & 1) == 0)
                 // We still have a (theoretical?) problem here if due to
-                // nArbitraryLineLengthLimit we split a string right between a
-                // doubled quote pair.
+                // nArbitraryLineLengthLimit (or nMaxSourceLines below) we
+                // split a string right between a doubled quote pair.
+                break;
+            else if (++nLine >= nMaxSourceLines && nMaxSourceLines > 0)
+                // Unconditionally increment nLine even if nMaxSourceLines==0
+                // so it can be observed in debugger.
                 break;
             else
             {