XLSX: support documents whose XML elements have a prefix

Fixes duckdb/duckdb-spatial#362
OSGeo · Aug 10, 2024 · 5b61064 · 5b61064
1 parent 1e95085
commit 5b61064
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 14 deletions.
diff --git a/autotest/ogr/data/xlsx/with_xml_prefix.xlsx b/autotest/ogr/data/xlsx/with_xml_prefix.xlsx
diff --git a/autotest/ogr/ogr_xlsx.py b/autotest/ogr/ogr_xlsx.py
@@ -650,3 +650,18 @@ def test_ogr_xlsx_write_sheet_without_row():
     assert ds.GetLayer(2).GetFeatureCount() == 1
     ds = None
     gdal.Unlink(tmpfilename)
+
+
+###############################################################################
+# Test reading a XLSX file with XML element prefixes
+
+
+def test_ogr_xlsx_read_xml_prefix():
+
+    ds = ogr.Open("data/xlsx/with_xml_prefix.xlsx")
+    lyr = ds.GetLayer(0)
+    assert lyr.GetLayerDefn().GetFieldDefn(0).GetName() == "Col1"
+    assert lyr.GetLayerDefn().GetFieldDefn(1).GetName() == "Col2"
+    f = lyr.GetNextFeature()
+    assert f["Col1"] == "foo"
+    assert f["Col2"] == "bar"
diff --git a/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp b/ogr/ogrsf_frmts/xlsx/ogrxlsxdatasource.cpp
@@ -418,6 +418,18 @@ int OGRXLSXDataSource::Create(const char *pszFilename,
     return TRUE;
 }
 
+/************************************************************************/
+/*                           GetUnprefixed()                            */
+/************************************************************************/
+
+static const char *GetUnprefixed(const char *pszStr)
+{
+    const char *pszColumn = strchr(pszStr, ':');
+    if (pszColumn)
+        return pszColumn + 1;
+    return pszStr;
+}
+
 /************************************************************************/
 /*                           startElementCbk()                          */
 /************************************************************************/
@@ -434,6 +446,8 @@ void OGRXLSXDataSource::startElementCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     switch (stateStack[nStackDepth].eVal)
     {
@@ -474,6 +488,8 @@ void OGRXLSXDataSource::endElementCbk(const char *pszNameIn)
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
 
     nDepth--;
@@ -1346,6 +1362,8 @@ void OGRXLSXDataSource::startElementSSCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     switch (stateStack[nStackDepth].eVal)
     {
@@ -1381,11 +1399,14 @@ static void XMLCALL endElementSSCbk(void *pUserData, const char *pszNameIn)
     ((OGRXLSXDataSource *)pUserData)->endElementSSCbk(pszNameIn);
 }
 
-void OGRXLSXDataSource::endElementSSCbk(CPL_UNUSED const char *pszNameIn)
+void OGRXLSXDataSource::endElementSSCbk(const char * /*pszNameIn*/)
 {
     if (bStopParsing)
         return;
 
+    // If we were to use pszNameIn, then we need:
+    // pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
 
     nDepth--;
@@ -1529,6 +1550,8 @@ void OGRXLSXDataSource::startElementWBRelsCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "Relationship") == 0)
     {
@@ -1593,18 +1616,6 @@ void OGRXLSXDataSource::AnalyseWorkbookRels(VSILFILE *fpWorkbookRels)
     VSIFCloseL(fpWorkbookRels);
 }
 
-/************************************************************************/
-/*                           GetUnprefixed()                            */
-/************************************************************************/
-
-static const char *GetUnprefixed(const char *pszStr)
-{
-    const char *pszColumn = strchr(pszStr, ':');
-    if (pszColumn)
-        return pszColumn + 1;
-    return pszStr;
-}
-
 /************************************************************************/
 /*                          startElementWBCbk()                         */
 /************************************************************************/
@@ -1621,8 +1632,10 @@ void OGRXLSXDataSource::startElementWBCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
-    if (strcmp(GetUnprefixed(pszNameIn), "sheet") == 0)
+    if (strcmp(pszNameIn, "sheet") == 0)
     {
         const char *pszSheetName = GetAttributeValue(ppszAttr, "name", nullptr);
         const char *pszId = GetAttributeValue(ppszAttr, "r:id", nullptr);
@@ -1725,6 +1738,8 @@ void OGRXLSXDataSource::startElementStylesCbk(const char *pszNameIn,
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "numFmt") == 0)
     {
@@ -1810,6 +1825,8 @@ void OGRXLSXDataSource::endElementStylesCbk(const char *pszNameIn)
     if (bStopParsing)
         return;
 
+    pszNameIn = GetUnprefixed(pszNameIn);
+
     nWithoutEventCounter = 0;
     if (strcmp(pszNameIn, "cellXfs") == 0)
     {