Support XMLTABLE query expression

XMLTABLE is defined by the SQL/XML standard as a feature that allows
turning XML-formatted data into relational form, so that it can be used
as a <table primary> in the FROM clause of a query.

This new construct provides significant simplicity and performance
benefit for XML data processing; what in a client-side custom
implementation was reported to take 20 minutes can be executed in 400ms
using XMLTABLE.  (The same functionality was said to take 10 seconds
using nested PostgreSQL XPath function calls, and 5 seconds using
XMLReader under PL/Python).

The implemented syntax deviates slightly from what the standard
requires.  First, the standard indicates that the PASSING clause is
optional and that multiple XML input documents may be given to it; we
make it mandatory and accept a single document only.  Second, we don't
currently support a default namespace to be specified.

This implementation relies on a new executor node based on a hardcoded
method table.  (Because the grammar is fixed, there is no extensibility
in the current approach; further constructs can be implemented on top of
this such as JSON_TABLE, but they require changes to core code.)

Author: Pavel Stehule, Álvaro Herrera
Extensively reviewed by: Craig Ringer
Discussion: https://postgr.es/m/CAFj8pRAgfzMD-LoSmnMGybD0WsEznLHWap8DO79+-GTRAPR4qA@mail.gmail.com

2 files changed, 91 insertions, 0 deletions

diff --git a/src/include/executor/nodeTableFuncscan.h b/src/include/executor/nodeTableFuncscan.h
new file mode 100644
index 00000000000..529c9299931
--- /dev/null
+++ b/src/include/executor/nodeTableFuncscan.h
@@ -0,0 +1,24 @@
+/*-------------------------------------------------------------------------
+ *
+ * nodeTableFuncscan.h
+ *
+ *
+ *
+ * Portions Copyright (c) 1996-2017, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/executor/nodeTableFuncscan.h
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef NODETABLEFUNCSCAN_H
+#define NODETABLEFUNCSCAN_H
+
+#include "nodes/execnodes.h"
+
+extern TableFuncScanState *ExecInitTableFuncScan(TableFuncScan *node, EState *estate, int eflags);
+extern TupleTableSlot *ExecTableFuncScan(TableFuncScanState *node);
+extern void ExecEndTableFuncScan(TableFuncScanState *node);
+extern void ExecReScanTableFuncScan(TableFuncScanState *node);
+
+#endif   /* NODETABLEFUNCSCAN_H */
diff --git a/src/include/executor/tablefunc.h b/src/include/executor/tablefunc.h
new file mode 100644
index 00000000000..89d6381220e
--- /dev/null
+++ b/src/include/executor/tablefunc.h
@@ -0,0 +1,67 @@
+/*-------------------------------------------------------------------------
+ *
+ * tablefunc.h
+ *				interface for TableFunc executor node
+ *
+ * Portions Copyright (c) 1996-2017, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/executor/tablefunc.h
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef _TABLEFUNC_H
+#define _TABLEFUNC_H
+
+/* Forward-declare this to avoid including execnodes.h here */
+struct TableFuncScanState;
+
+/*
+ * TableFuncRoutine holds function pointers used for generating content of
+ * table-producer functions, such as XMLTABLE.
+ *
+ * InitBuilder initialize table builder private objects.  The output tuple
+ * descriptor, input functions for the columns, and typioparams are passed
+ * from executor state.
+ *
+ * SetDoc is called to define the input document.  The table builder may
+ * apply additional transformations not exposed outside the table builder
+ * context.
+ *
+ * SetNamespace is called to pass namespace declarations from the table
+ * expression.  This function may be NULL if namespaces are not supported by
+ * the table builder.  Namespaces must be given before setting the row and
+ * column filters.  If the name is given as NULL, the entry shall be for the
+ * default namespace.
+ *
+ * SetRowFilter is called do define the row-generating filter, which shall be
+ * used to extract each row from the input document.
+ *
+ * SetColumnFilter is called once for each column, to define the column-
+ * generating filter for the given column.
+ *
+ * FetchRow shall be called repeatedly until it returns that no more rows are
+ * found in the document.  On each invocation it shall set state in the table
+ * builder context such that each subsequent GetValue call returns the values
+ * for the indicated column for the row being processed.
+ *
+ * DestroyBuilder shall release all resources associated with a table builder
+ * context.  It may be called either because all rows have been consumed, or
+ * because an error ocurred while processing the table expression.
+ */
+typedef struct TableFuncRoutine
+{
+	void		(*InitOpaque) (struct TableFuncScanState *state, int natts);
+	void		(*SetDocument) (struct TableFuncScanState *state, Datum value);
+	void		(*SetNamespace) (struct TableFuncScanState *state, char *name,
+											 char *uri);
+	void		(*SetRowFilter) (struct TableFuncScanState *state, char *path);
+	void		(*SetColumnFilter) (struct TableFuncScanState *state,
+									char *path, int colnum);
+	bool		(*FetchRow) (struct TableFuncScanState *state);
+	Datum		(*GetValue) (struct TableFuncScanState *state, int colnum,
+							 Oid typid, int32 typmod, bool *isnull);
+	void		(*DestroyOpaque) (struct TableFuncScanState *state);
+} TableFuncRoutine;
+
+#endif   /* _TABLEFUNC_H */