Merge pull request #10279 from rouault/csv_inf_nan

CSV: allow inf, -inf and nan as numeric values
OSGeo · Jul 6, 2024 · b08dfac · b08dfac
2 parents 86e203c + 3d8b1e0
commit b08dfac
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 7 deletions.
diff --git a/autotest/ogr/data/csv/inf_nan.csv b/autotest/ogr/data/csv/inf_nan.csv
@@ -0,0 +1,5 @@
+id,v
+1,10
+2,inf
+3,-inf
+4,NaN
diff --git a/autotest/ogr/ogr_csv.py b/autotest/ogr/ogr_csv.py
@@ -27,6 +27,7 @@
 # Boston, MA 02111-1307, USA.
 ###############################################################################
 
+import math
 import pathlib
 import sys
 
@@ -3133,6 +3134,26 @@ def test_ogr_csv_force_opening(tmp_vsimem):
     assert ds.GetDriver().GetDescription() == "CSV"
 
 
+###############################################################################
+# Test opening a CSV file with inf/nan numeric values
+
+
+@gdaltest.enable_exceptions()
+def test_ogr_csv_inf_nan():
+
+    ds = gdal.OpenEx("data/csv/inf_nan.csv", open_options=["AUTODETECT_TYPE=YES"])
+    lyr = ds.GetLayer(0)
+    assert lyr.GetLayerDefn().GetFieldDefn(1).GetType() == ogr.OFTReal
+    f = lyr.GetNextFeature()
+    assert f["v"] == 10.0
+    f = lyr.GetNextFeature()
+    assert f["v"] == float("inf")
+    f = lyr.GetNextFeature()
+    assert f["v"] == float("-inf")
+    f = lyr.GetNextFeature()
+    assert math.isnan(f["v"])
+
+
 ###############################################################################
 
 

diff --git a/ogr/ogrsf_frmts/csv/ogrcsvlayer.cpp b/ogr/ogrsf_frmts/csv/ogrcsvlayer.cpp
@@ -1049,7 +1049,10 @@ char **OGRCSVLayer::AutodetectFieldTypes(CSLConstList papszOpenOptions,
                 else
                     eOGRFieldType = OFTInteger;
             }
-            else if (eType == CPL_VALUE_REAL)
+            else if (eType == CPL_VALUE_REAL ||
+                     EQUAL(papszTokens[iField], "inf") ||
+                     EQUAL(papszTokens[iField], "-inf") ||
+                     EQUAL(papszTokens[iField], "nan"))
             {
                 eOGRFieldType = OFTReal;
             }
@@ -1466,14 +1469,16 @@ OGRFeature *OGRCSVLayer::GetNextUnfilteredFeature()
                     if (chComma)
                         *chComma = '.';
                 }
-                CPLValueType eType = CPLGetValueType(papszTokens[iAttr]);
-                if (eType == CPL_VALUE_INTEGER || eType == CPL_VALUE_REAL)
+                char *endptr = nullptr;
+                const double dfVal =
+                    CPLStrtodDelim(papszTokens[iAttr], &endptr, '.');
+                if (endptr == papszTokens[iAttr] + strlen(papszTokens[iAttr]))
                 {
-                    poFeature->SetField(iOGRField, papszTokens[iAttr]);
+                    poFeature->SetField(iOGRField, dfVal);
                     if (!bWarningBadTypeOrWidth &&
                         (eFieldType == OFTInteger ||
                          eFieldType == OFTInteger64) &&
-                        eType == CPL_VALUE_REAL)
+                        CPLGetValueType(papszTokens[iAttr]) == CPL_VALUE_REAL)
                     {
                         bWarningBadTypeOrWidth = true;
                         CPLError(CE_Warning, CPLE_AppDefined,
@@ -1495,8 +1500,9 @@ OGRFeature *OGRCSVLayer::GetNextUnfilteredFeature()
                                  nNextFID, poFieldDefn->GetNameRef());
                     }
                     else if (!bWarningBadTypeOrWidth &&
-                             eType == CPL_VALUE_REAL &&
-                             poFieldDefn->GetWidth() > 0)
+                             poFieldDefn->GetWidth() > 0 &&
+                             CPLGetValueType(papszTokens[iAttr]) ==
+                                 CPL_VALUE_REAL)
                     {
                         const char *pszDot = strchr(papszTokens[iAttr], '.');
                         const int nPrecision =