nodemergejoin.c

PostgreSQL7.4.6 for Linux

/*-------------------------------------------------------------------------
 *
 * nodeMergejoin.c
 *	  routines supporting merge joins
 *
 * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
 *	  $Header: /cvsroot/pgsql/src/backend/executor/nodeMergejoin.c,v 1.62 2003/09/25 06:57:59 petere Exp $
 *
 *-------------------------------------------------------------------------
 */
/*
 * INTERFACE ROUTINES
 *		ExecMergeJoin			mergejoin outer and inner relations.
 *		ExecInitMergeJoin		creates and initializes run time states
 *		ExecEndMergeJoin		cleans up the node.
 *
 * NOTES
 *		Essential operation of the merge join algorithm is as follows:
 *
 *		Join {												   -
 *			get initial outer and inner tuples				INITIALIZE
 *			Skip Inner										SKIPINNER
 *			mark inner position								JOINMARK
 *			do forever {									   -
 *				while (outer == inner) {					JOINTEST
 *					join tuples								JOINTUPLES
 *					advance inner position					NEXTINNER
 *				}											   -
 *				advance outer position						NEXTOUTER
 *				if (outer == mark) {						TESTOUTER
 *					restore inner position to mark			TESTOUTER
 *					continue								   -
 *				} else {									   -
 *					Skip Outer								SKIPOUTER
 *					mark inner position						JOINMARK
 *				}											   -
 *			}												   -
 *		}													   -
 *
 *		Skip Outer {										SKIPOUTER_BEGIN
 *			if (inner == outer) Join Tuples					JOINTUPLES
 *			while (outer < inner)							SKIPOUTER_TEST
 *				advance outer								SKIPOUTER_ADVANCE
 *			if (outer > inner)								SKIPOUTER_TEST
 *				Skip Inner									SKIPINNER
 *		}													   -
 *
 *		Skip Inner {										SKIPINNER_BEGIN
 *			if (inner == outer) Join Tuples					JOINTUPLES
 *			while (outer > inner)							SKIPINNER_TEST
 *				advance inner								SKIPINNER_ADVANCE
 *			if (outer < inner)								SKIPINNER_TEST
 *				Skip Outer									SKIPOUTER
 *		}													   -
 *
 *		The merge join operation is coded in the fashion
 *		of a state machine.  At each state, we do something and then
 *		proceed to another state.  This state is stored in the node's
 *		execution state information and is preserved across calls to
 *		ExecMergeJoin. -cim 10/31/89
 *
 */
#include "postgres.h"

#include "access/heapam.h"
#include "access/printtup.h"
#include "catalog/pg_operator.h"
#include "executor/execdebug.h"
#include "executor/execdefs.h"
#include "executor/nodeMergejoin.h"
#include "utils/lsyscache.h"
#include "utils/syscache.h"


static bool MergeCompare(List *eqQual, List *compareQual, ExprContext *econtext);

#define MarkInnerTuple(innerTupleSlot, mergestate) \
( \
	ExecStoreTuple(heap_copytuple((innerTupleSlot)->val), \
				   (mergestate)->mj_MarkedTupleSlot, \
				   InvalidBuffer, \
				   true) \
)


/* ----------------------------------------------------------------
 *		MJFormSkipQuals
 *
 *		This takes the mergeclause which is a qualification of the
 *		form ((= expr expr) (= expr expr) ...) and forms new lists
 *		of the forms ((< expr expr) (< expr expr) ...) and
 *		((> expr expr) (> expr expr) ...).	These lists will be used
 *		by ExecMergeJoin() to determine if we should skip tuples.
 *		(We expect there to be suitable operators because the "=" operators
 *		were marked mergejoinable; however, there might be a different
 *		one needed in each qual clause.)
 * ----------------------------------------------------------------
 */
static void
MJFormSkipQuals(List *qualList, List **ltQuals, List **gtQuals,
				PlanState *parent)
{
	List	   *ltexprs,
			   *gtexprs,
			   *ltcdr,
			   *gtcdr;

	/*
	 * Make modifiable copies of the qualList.
	 */
	ltexprs = (List *) copyObject((Node *) qualList);
	gtexprs = (List *) copyObject((Node *) qualList);

	/*
	 * Scan both lists in parallel, so that we can update the operators
	 * with the minimum number of syscache searches.
	 */
	ltcdr = ltexprs;
	foreach(gtcdr, gtexprs)
	{
		OpExpr	   *ltop = (OpExpr *) lfirst(ltcdr);
		OpExpr	   *gtop = (OpExpr *) lfirst(gtcdr);

		/*
		 * The two ops should be identical, so use either one for lookup.
		 */
		if (!IsA(ltop, OpExpr))
			elog(ERROR, "mergejoin clause is not an OpExpr");

		/*
		 * Lookup the operators, and replace the data in the copied
		 * operator nodes.
		 */
		op_mergejoin_crossops(ltop->opno,
							  &ltop->opno,
							  &gtop->opno,
							  &ltop->opfuncid,
							  &gtop->opfuncid);

		ltcdr = lnext(ltcdr);
	}

	/*
	 * Prepare both lists for execution.
	 */
	*ltQuals = (List *) ExecInitExpr((Expr *) ltexprs, parent);
	*gtQuals = (List *) ExecInitExpr((Expr *) gtexprs, parent);
}

/* ----------------------------------------------------------------
 *		MergeCompare
 *
 *		Compare the keys according to 'compareQual' which is of the
 *		form: { (key1a > key2a) (key1b > key2b) ... }.
 *
 *		(actually, it could also be of the form (key1a < key2a)...)
 *
 *		This is different from calling ExecQual because ExecQual returns
 *		true only if ALL the comparison clauses are satisfied.
 *		However, there is an order of significance among the keys with
 *		the first keys being most significant. Therefore, the clauses
 *		are evaluated in order and the 'compareQual' is satisfied
 *		if (key1i > key2i) is true and (key1j = key2j) for 0 < j < i.
 *		We use the original mergeclause items to detect equality.
 * ----------------------------------------------------------------
 */
static bool
MergeCompare(List *eqQual, List *compareQual, ExprContext *econtext)
{
	bool		result;
	MemoryContext oldContext;
	List	   *clause;
	List	   *eqclause;

	/*
	 * Do expression eval in short-lived context.
	 */
	oldContext = MemoryContextSwitchTo(econtext->ecxt_per_tuple_memory);

	/*
	 * for each pair of clauses, test them until our compare conditions
	 * are satisfied. if we reach the end of the list, none of our key
	 * greater-than conditions were satisfied so we return false.
	 */
	result = false;				/* assume 'false' result */

	eqclause = eqQual;
	foreach(clause, compareQual)
	{
		Datum		const_value;
		bool		isNull;

		/*
		 * first test if our compare clause is satisfied. if so then
		 * return true.
		 *
		 * A NULL result is considered false.
		 */
		const_value = ExecEvalExpr((ExprState *) lfirst(clause),
								   econtext,
								   &isNull,
								   NULL);

		if (DatumGetBool(const_value) && !isNull)
		{
			result = true;
			break;
		}

		/*-----------
		 * ok, the compare clause failed so we test if the keys are
		 * equal... if key1 != key2, we return false. otherwise
		 * key1 = key2 so we move on to the next pair of keys.
		 *-----------
		 */
		const_value = ExecEvalExpr((ExprState *) lfirst(eqclause),
								   econtext,
								   &isNull,
								   NULL);

		if (!DatumGetBool(const_value) || isNull)
			break;				/* return false */

		eqclause = lnext(eqclause);
	}

	MemoryContextSwitchTo(oldContext);

	return result;
}

/* ----------------------------------------------------------------
 *		ExecMergeTupleDump
 *
 *		This function is called through the MJ_dump() macro
 *		when EXEC_MERGEJOINDEBUG is defined
 * ----------------------------------------------------------------
 */
#ifdef EXEC_MERGEJOINDEBUG

static void
ExecMergeTupleDumpOuter(MergeJoinState *mergestate)
{
	TupleTableSlot *outerSlot = mergestate->mj_OuterTupleSlot;

	printf("==== outer tuple ====\n");
	if (TupIsNull(outerSlot))
		printf("(nil)\n");
	else
		MJ_debugtup(outerSlot->val,
					outerSlot->ttc_tupleDescriptor);
}

static void
ExecMergeTupleDumpInner(MergeJoinState *mergestate)
{
	TupleTableSlot *innerSlot = mergestate->mj_InnerTupleSlot;

	printf("==== inner tuple ====\n");
	if (TupIsNull(innerSlot))
		printf("(nil)\n");
	else
		MJ_debugtup(innerSlot->val,
					innerSlot->ttc_tupleDescriptor);
}

static void
ExecMergeTupleDumpMarked(MergeJoinState *mergestate)
{
	TupleTableSlot *markedSlot = mergestate->mj_MarkedTupleSlot;

	printf("==== marked tuple ====\n");
	if (TupIsNull(markedSlot))
		printf("(nil)\n");
	else
		MJ_debugtup(markedSlot->val,
					markedSlot->ttc_tupleDescriptor);
}

static void
ExecMergeTupleDump(MergeJoinState *mergestate)
{
	printf("******** ExecMergeTupleDump ********\n");

	ExecMergeTupleDumpOuter(mergestate);
	ExecMergeTupleDumpInner(mergestate);
	ExecMergeTupleDumpMarked(mergestate);

	printf("******** \n");
}
#endif

/* ----------------------------------------------------------------
 *		ExecMergeJoin
 *
 * old comments
 *		Details of the merge-join routines:
 *
 *		(1) ">" and "<" operators
 *
 *		Merge-join is done by joining the inner and outer tuples satisfying
 *		the join clauses of the form ((= outerKey innerKey) ...).
 *		The join clauses is provided by the query planner and may contain
 *		more than one (= outerKey innerKey) clauses (for composite key).
 *
 *		However, the query executor needs to know whether an outer
 *		tuple is "greater/smaller" than an inner tuple so that it can
 *		"synchronize" the two relations. For e.g., consider the following
 *		relations:
 *
 *				outer: (0 ^1 1 2 5 5 5 6 6 7)	current tuple: 1
 *				inner: (1 ^3 5 5 5 5 6)			current tuple: 3
 *
 *		To continue the merge-join, the executor needs to scan both inner
 *		and outer relations till the matching tuples 5. It needs to know
 *		that currently inner tuple 3 is "greater" than outer tuple 1 and
 *		therefore it should scan the outer relation first to find a
 *		matching tuple and so on.
 *
 *		Therefore, when initializing the merge-join node, the executor
 *		creates the "greater/smaller" clause by substituting the "="
 *		operator in the join clauses with the corresponding ">" operator.
 *		The opposite "smaller/greater" clause is formed by substituting "<".
 *
 *		Note: prior to v6.5, the relational clauses were formed using the
 *		sort op used to sort the inner relation, which of course would fail
 *		if the outer and inner keys were of different data types.
 *		In the current code, we instead assume that operators named "<" and ">"
 *		will do the right thing.  This should be true since the mergejoin "="
 *		operator's pg_operator entry will have told the planner to sort by
 *		"<" for each of the left and right sides.
 *
 *		(2) repositioning inner "cursor"
 *
 *		Consider the above relations and suppose that the executor has
 *		just joined the first outer "5" with the last inner "5". The
 *		next step is of course to join the second outer "5" with all
 *		the inner "5's". This requires repositioning the inner "cursor"
 *		to point at the first inner "5". This is done by "marking" the
 *		first inner 5 and restore the "cursor" to it before joining
 *		with the second outer 5. The access method interface provides
 *		routines to mark and restore to a tuple.
 * ----------------------------------------------------------------
 */
TupleTableSlot *
ExecMergeJoin(MergeJoinState *node)
{
	EState	   *estate;
	ScanDirection direction;
	List	   *innerSkipQual;
	List	   *outerSkipQual;
	List	   *mergeclauses;
	List	   *joinqual;
	List	   *otherqual;
	bool		qualResult;
	bool		compareResult;
	PlanState  *innerPlan;
	TupleTableSlot *innerTupleSlot;
	PlanState  *outerPlan;
	TupleTableSlot *outerTupleSlot;
	ExprContext *econtext;
	bool		doFillOuter;
	bool		doFillInner;

	/*
	 * get information from node
	 */
	estate = node->js.ps.state;
	direction = estate->es_direction;
	innerPlan = innerPlanState(node);
	outerPlan = outerPlanState(node);
	econtext = node->js.ps.ps_ExprContext;
	mergeclauses = node->mergeclauses;
	joinqual = node->js.joinqual;
	otherqual = node->js.ps.qual;

	switch (node->js.jointype)
	{
		case JOIN_INNER:
		case JOIN_IN:
			doFillOuter = false;
			doFillInner = false;
			break;
		case JOIN_LEFT:
			doFillOuter = true;
			doFillInner = false;
			break;
		case JOIN_FULL:
			doFillOuter = true;
			doFillInner = true;
			break;
		case JOIN_RIGHT:
			doFillOuter = false;
			doFillInner = true;
			break;
		default:
			elog(ERROR, "unrecognized join type: %d",
				 (int) node->js.jointype);
			doFillOuter = false;	/* keep compiler quiet */
			doFillInner = false;
			break;
	}

	if (ScanDirectionIsForward(direction))
	{
		outerSkipQual = node->mj_OuterSkipQual;
		innerSkipQual = node->mj_InnerSkipQual;
	}
	else
	{
		outerSkipQual = node->mj_InnerSkipQual;
		innerSkipQual = node->mj_OuterSkipQual;
	}

	/*
	 * Check to see if we're still projecting out tuples from a previous
	 * join tuple (because there is a function-returning-set in the
	 * projection expressions).  If so, try to project another one.
	 */
	if (node->js.ps.ps_TupFromTlist)
	{
		TupleTableSlot *result;
		ExprDoneCond isDone;

		result = ExecProject(node->js.ps.ps_ProjInfo, &isDone);
		if (isDone == ExprMultipleResult)
			return result;
		/* Done with that source tuple... */
		node->js.ps.ps_TupFromTlist = false;
	}

	/*
	 * Reset per-tuple memory context to free any expression evaluation
	 * storage allocated in the previous tuple cycle.  Note this can't
	 * happen until we're done projecting out tuples from a join tuple.
	 */
	ResetExprContext(econtext);

	/*
	 * ok, everything is setup.. let's go to work
	 */
	for (;;)
	{
		/*
		 * get the current state of the join and do things accordingly.
		 * Note: The join states are highlighted with 32-* comments for
		 * improved readability.
		 */
		MJ_dump(node);

		switch (node->mj_JoinState)
		{
				/*
				 * EXEC_MJ_INITIALIZE means that this is the first time
				 * ExecMergeJoin() has been called and so we have to fetch
				 * the first tuple for both outer and inner subplans. If
				 * we fail to get a tuple here, then that subplan is
				 * empty, and we either end the join or go to one of the
				 * fill-remaining-tuples states.
				 */
			case EXEC_MJ_INITIALIZE:
				MJ_printf("ExecMergeJoin: EXEC_MJ_INITIALIZE\n");

				outerTupleSlot = ExecProcNode(outerPlan);
				node->mj_OuterTupleSlot = outerTupleSlot;
				if (TupIsNull(outerTupleSlot))
				{
					MJ_printf("ExecMergeJoin: outer subplan is empty\n");
					if (doFillInner)
					{
						/*
						 * Need to emit right-join tuples for remaining
						 * inner tuples.  We set MatchedInner = true to
						 * force the ENDOUTER state to advance inner.
						 */
						node->mj_JoinState = EXEC_MJ_ENDOUTER;
						node->mj_MatchedInner = true;
						break;
					}
					/* Otherwise we're done. */
					return NULL;
				}

				innerTupleSlot = ExecProcNode(innerPlan);
				node->mj_InnerTupleSlot = innerTupleSlot;
				if (TupIsNull(innerTupleSlot))
				{
					MJ_printf("ExecMergeJoin: inner subplan is empty\n");
					if (doFillOuter)
					{
						/*
						 * Need to emit left-join tuples for all outer
						 * tuples, including the one we just fetched.  We
						 * set MatchedOuter = false to force the ENDINNER
						 * state to emit this tuple before advancing
						 * outer.
						 */
						node->mj_JoinState = EXEC_MJ_ENDINNER;
						node->mj_MatchedOuter = false;
						break;
					}
					/* Otherwise we're done. */
					return NULL;
				}

				/*
				 * OK, we have the initial tuples.	Begin by skipping
				 * unmatched inner tuples.
				 */
				node->mj_JoinState = EXEC_MJ_SKIPINNER_BEGIN;
				break;

				/*
				 * EXEC_MJ_JOINMARK means we have just found a new outer
				 * tuple and a possible matching inner tuple. This is the
				 * case after the INITIALIZE, SKIPOUTER or SKIPINNER
				 * states.
				 */
			case EXEC_MJ_JOINMARK:
				MJ_printf("ExecMergeJoin: EXEC_MJ_JOINMARK\n");

				ExecMarkPos(innerPlan);

				MarkInnerTuple(node->mj_InnerTupleSlot, node);

				node->mj_JoinState = EXEC_MJ_JOINTEST;
				break;

				/*
				 * EXEC_MJ_JOINTEST means we have two tuples which might
				 * satisfy the merge clause, so we test them.
				 *