Fix xmlns:xsi placement for ELEMENTS XSINIL with ROOT in FOR XML RAW/PATH (#4855)

When using ELEMENTS XSINIL together with ROOT in FOR XML RAW or PATH mode, Babelfish was placing the
xmlns:xsi namespace declaration on each row element instead of on the ROOT element where it is expected to be at
ROOT element only.

Before:

<data xmlns:xsi="..."><row xmlns:xsi="..."><a>1</a><b xsi:nil="true"/></row></data>

After:

<data xmlns:xsi="..."><row><a>1</a><b xsi:nil="true"/></row></data>

When ROOT already carries the xmlns:xsi declaration, the per-row declaration is now suppressed. Without ROOT,
the per-row declaration is still emitted (matching T-SQL behavior for the no-ROOT case).

The aggregate state now tracks a has_root flag, set when the ROOT open-tag is emitted, and threads it into the
RAW and PATH emitters. When ROOT is present they skip the row-level xmlns:xsi since the ROOT already carries it;
without ROOT the existing emission is unchanged. The same flag also handles PATH('') with XSINIL (where xmlns
would otherwise repeat on each column element) and RAW('') with XSINIL — there the ROOT tag itself now
correctly includes xmlns:xsi.

Task: BABEL-6429

Signed-off-by: Japleen Kaur <amjj@amazon.com>

Author: JapleenKaur3

contrib/babelfishpg_tsql/src/tsql_for/forxml.c

@@ -89,11 +89,12 @@ typedef struct forxml_state
 {
 	StringInfo			xml_output;		/* Accumulated XML output */
 	forxml_auto_state  *auto_state;		/* AUTO mode state, NULL for other modes */
+	bool				has_root;		/* True if ROOT clause was specified */
 } forxml_state;
 
 static StringInfo for_xml_ffunc(PG_FUNCTION_ARGS);
-static void tsql_row_to_xml_raw(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool elements, bool xsinil);
-static void tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool xsinil);
+static void tsql_row_to_xml_raw(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool elements, bool xsinil, bool has_root);
+static void tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool xsinil, bool has_root);
 static void tsql_row_to_xml_auto(StringInfo state, Datum record, bool elements, bool xsinil, forxml_auto_state *auto_state);
 static void update_tsql_datatype_and_val(HeapTuple tuple, TupleDesc tupdesc, Oid *datatype_oid, Datum *colval, bool binary_base64, int i);
 static char *tsql_escape_xml(const char *str);
@@ -197,7 +198,8 @@ tsql_query_to_xml_sfunc(PG_FUNCTION_ARGS)
 		fstate->auto_state = NULL;
 		state = fstate->xml_output;
 		root_name = PG_ARGISNULL(5) ? NULL : text_to_cstring(PG_GETARG_TEXT_PP(5));
-		if (root_name != NULL && strlen(root_name) > 0)
+		fstate->has_root = (root_name != NULL && strlen(root_name) > 0);
+		if (fstate->has_root)
 		{
 			/*
 			 * We need to add an extra token to the beginning so that the
@@ -248,7 +250,7 @@ tsql_query_to_xml_sfunc(PG_FUNCTION_ARGS)
 	switch (mode)
 	{
 		case TSQL_FORXML_RAW:	/* FOR XML RAW */
-			tsql_row_to_xml_raw(state, record, element_name, binary_base64, elements, xsinil);
+			tsql_row_to_xml_raw(state, record, element_name, binary_base64, elements, xsinil, fstate->has_root);
 			break;
 		case TSQL_FORXML_AUTO:	/* FOR XML AUTO */
 			{
@@ -263,7 +265,7 @@ tsql_query_to_xml_sfunc(PG_FUNCTION_ARGS)
 			}
 			break;
 		case TSQL_FORXML_PATH:	/* FOR XML PATH */
-			tsql_row_to_xml_path(state, record, element_name, binary_base64, xsinil);
+			tsql_row_to_xml_path(state, record, element_name, binary_base64, xsinil, fstate->has_root);
 			break;
 		case TSQL_FORXML_EXPLICIT:
 
@@ -380,7 +382,7 @@ for_xml_ffunc(PG_FUNCTION_ARGS)
  * Map an SQL row to an XML element in RAW mode.
  */
 static void
-tsql_row_to_xml_raw(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool elements, bool xsinil)
+tsql_row_to_xml_raw(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool elements, bool xsinil, bool has_root)
 {
 	HeapTupleHeader td;
 	Oid             tupType;
@@ -419,8 +421,13 @@ tsql_row_to_xml_raw(StringInfo state, Datum record, const char *element_name, bo
 	{
 		if (elements)
 		{
-			/* ELEMENTS mode: <row><col>value</col></row> */
-			if (xsinil)
+			/*
+			 * ELEMENTS mode: <row><col>value</col></row>.
+			 * When XSINIL is set, xmlns:xsi normally goes on the row tag,
+			 * but when ROOT is present we already emitted it there, so
+			 * suppress the per-row declaration to match T-SQL behavior.
+			 */
+			if (xsinil && !has_root)
 				appendStringInfo(state, "<%s " XML_XMLNS_XSI ">", element_name);
 			else
 				appendStringInfo(state, "<%s>", element_name);
@@ -564,7 +571,7 @@ validate_attribute_centric_col_names_xml(const char *element_name, TupleDesc tup
  * Map an SQL row to an XML element in PATH mode.
  */
 static void
-tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool xsinil)
+tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, bool binary_base64, bool xsinil, bool has_root)
 {
 	HeapTupleHeader td;
 	Oid             tupType;
@@ -593,21 +600,25 @@ tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, b
 
 	/*
 	 * each tuple is either contained in a "row" tag, or standalone if the
-	 * element_name is an empty string
+	 * element_name is an empty string.
+	 *
+	 * When XSINIL is set, xmlns:xsi normally goes on the row tag,
+	 * but when ROOT is present we already emitted it there, so suppress
+	 * the per-row declaration to match T-SQL behavior.
 	 */
 	if (element_name && strlen(element_name) > 0)
 	{
 		/* if "''" is the input path, ignore it per TSQL behavior */
 		if (has_att_centric)
 		{
-			if (xsinil)
+			if (xsinil && !has_root)
 				appendStringInfo(state, "<%s " XML_XMLNS_XSI, element_name);
 			else
 				appendStringInfo(state, "<%s", element_name);
 		}
 		else
 		{
-			if (xsinil)
+			if (xsinil && !has_root)
 				appendStringInfo(state, "<%s " XML_XMLNS_XSI ">", element_name);
 			else
 				appendStringInfo(state, "<%s>", element_name);
@@ -656,8 +667,13 @@ tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, b
 				}
 				else
 				{
-					/* When PATH('') is used with XSINIL, add xmlns to each element */
-					if ((element_name && strlen(element_name) == 0) && xsinil)
+					/*
+					 * PATH('') + XSINIL emits xmlns:xsi on each column
+					 * element since there is no row tag to carry it.
+					 * When ROOT already declared it, suppress the per-column
+					 * declaration to match T-SQL behavior.
+					 */
+					if ((element_name && strlen(element_name) == 0) && xsinil && !has_root)
 						appendStringInfo(state, "<%s " XML_XMLNS_XSI ">%s</%s>",
 										 colname,
 										 map_sql_value_to_xml_value(colval, datatype_oid, true),
@@ -689,8 +705,13 @@ tsql_row_to_xml_path(StringInfo state, Datum record, const char *element_name, b
 
 				if (strncmp(NameStr(att->attname), "?column?", 8) != 0)
 				{
-					/* When PATH('') is used with XSINIL, add xmlns to each element */
-					if (element_name && strlen(element_name) == 0)
+					/*
+					 * PATH('') + XSINIL emits xmlns:xsi on each column
+					 * element since there is no row tag to carry it.
+					 * When ROOT already declared it, suppress the per-column
+					 * declaration to match T-SQL behavior.
+					 */
+					if (element_name && strlen(element_name) == 0 && !has_root)
 						appendStringInfo(state, "<%s " XML_XMLNS_XSI " " XML_XSI_NIL "/>", colname);
 					else
 						appendStringInfo(state, "<%s " XML_XSI_NIL "/>", colname);