From 0d73e6266a8841202938ba4e3b201c1e8c91c601 Mon Sep 17 00:00:00 2001
From: Even Rouault <even.rouault@spatialys.com>
Date: Wed, 15 Jan 2025 23:21:07 +0100
Subject: [PATCH] CSV: fix parsing files with double-quote inside a field value

Fixes #11660
---
 .../double_quotes_in_middle_of_field_bis.csv  |  3 +
 autotest/ogr/ogr_csv.py                       | 20 ++++++
 port/cpl_csv.cpp                              | 63 ++++++++++++-------
 3 files changed, 65 insertions(+), 21 deletions(-)
 create mode 100644 autotest/ogr/data/csv/double_quotes_in_middle_of_field_bis.csv

diff --git a/autotest/ogr/data/csv/double_quotes_in_middle_of_field_bis.csv b/autotest/ogr/data/csv/double_quotes_in_middle_of_field_bis.csv
new file mode 100644
index 000000000000..050d95d81045
--- /dev/null
+++ b/autotest/ogr/data/csv/double_quotes_in_middle_of_field_bis.csv
@@ -0,0 +1,3 @@
+first,second,third
+1,two"with quote,3
+10,twenty"with quote,30
diff --git a/autotest/ogr/ogr_csv.py b/autotest/ogr/ogr_csv.py
index 0ab07472aed8..977c87952769 100755
--- a/autotest/ogr/ogr_csv.py
+++ b/autotest/ogr/ogr_csv.py
@@ -2800,6 +2800,26 @@ def test_ogr_csv_double_quotes_in_middle_of_field():
     assert f["str"] == "foo"
 
 
+###############################################################################
+# Test bugfix for https://github.com/OSGeo/gdal/issues/11660
+
+
+def test_ogr_csv_double_quotes_in_middle_of_field_bis():
+
+    ds = ogr.Open("data/csv/double_quotes_in_middle_of_field_bis.csv")
+    lyr = ds.GetLayer(0)
+
+    f = lyr.GetNextFeature()
+    assert f["first"] == "1"
+    assert f["second"] == """two"with quote"""
+    assert f["third"] == "3"
+
+    f = lyr.GetNextFeature()
+    assert f["first"] == "10"
+    assert f["second"] == """twenty"with quote"""
+    assert f["third"] == "30"
+
+
 ###############################################################################
 
 
diff --git a/port/cpl_csv.cpp b/port/cpl_csv.cpp
index 6acf55fe234c..36e0216f39ff 100644
--- a/port/cpl_csv.cpp
+++ b/port/cpl_csv.cpp
@@ -647,45 +647,66 @@ CSVReadParseLineGeneric(void *fp, const char *(*pfnReadLine)(void *, size_t),
         return CSVSplitLine(pszLine, pszDelimiter, bKeepLeadingAndClosingQuotes,
                             bMergeDelimiter);
 
+    const size_t nDelimiterLength = strlen(pszDelimiter);
+    bool bInString = false;           // keep in that scope !
+    std::string osWorkLine(pszLine);  // keep in that scope !
+    size_t i = 0;                     // keep in that scope !
+
     try
     {
-        // We must now count the quotes in our working string, and as
-        // long as it is odd, keep adding new lines.
-        std::string osWorkLine(pszLine);
-
-        size_t i = 0;
-        int nCount = 0;
-
         while (true)
         {
-            for (; i < osWorkLine.size(); i++)
+            for (; i < osWorkLine.size(); ++i)
             {
                 if (osWorkLine[i] == '\"')
-                    nCount++;
+                {
+                    if (!bInString)
+                    {
+                        // Only consider " as the start of a quoted string
+                        // if it is the first character of the line, or
+                        // if it is immediately after the field delimiter.
+                        if (i == 0 ||
+                            (i >= nDelimiterLength &&
+                             osWorkLine.compare(i - nDelimiterLength,
+                                                nDelimiterLength, pszDelimiter,
+                                                nDelimiterLength) == 0))
+                        {
+                            bInString = true;
+                        }
+                    }
+                    else if (i + 1 < osWorkLine.size() &&
+                             osWorkLine[i + 1] == '"')
+                    {
+                        // Escaped double quote in a quoted string
+                        ++i;
+                    }
+                    else
+                    {
+                        bInString = false;
+                    }
+                }
             }
 
-            if (nCount % 2 == 0)
-                break;
+            if (!bInString)
+            {
+                return CSVSplitLine(osWorkLine.c_str(), pszDelimiter,
+                                    bKeepLeadingAndClosingQuotes,
+                                    bMergeDelimiter);
+            }
 
-            pszLine = pfnReadLine(fp, nMaxLineSize);
-            if (pszLine == nullptr)
+            const char *pszNewLine = pfnReadLine(fp, nMaxLineSize);
+            if (pszNewLine == nullptr)
                 break;
 
             osWorkLine.append("\n");
-            osWorkLine.append(pszLine);
+            osWorkLine.append(pszNewLine);
         }
-
-        char **papszReturn =
-            CSVSplitLine(osWorkLine.c_str(), pszDelimiter,
-                         bKeepLeadingAndClosingQuotes, bMergeDelimiter);
-
-        return papszReturn;
     }
     catch (const std::exception &e)
     {
         CPLError(CE_Failure, CPLE_OutOfMemory, "%s", e.what());
-        return nullptr;
     }
+    return nullptr;
 }
 
 /************************************************************************/