1
0
mirror of https://github.com/postgres/postgres.git synced 2025-05-17 06:41:24 +03:00
Peter Eisentraut 5e1963fb76 Collations with nondeterministic comparison
This adds a flag "deterministic" to collations.  If that is false,
such a collation disables various optimizations that assume that
strings are equal only if they are byte-wise equal.  That then allows
use cases such as case-insensitive or accent-insensitive comparisons
or handling of strings with different Unicode normal forms.

This functionality is only supported with the ICU provider.  At least
glibc doesn't appear to have any locales that work in a
nondeterministic way, so it's not worth supporting this for the libc
provider.

The term "deterministic comparison" in this context is from Unicode
Technical Standard #10
(https://unicode.org/reports/tr10/#Deterministic_Comparison).

This patch makes changes in three areas:

- CREATE COLLATION DDL changes and system catalog changes to support
  this new flag.

- Many executor nodes and auxiliary code are extended to track
  collations.  Previously, this code would just throw away collation
  information, because the eventually-called user-defined functions
  didn't use it since they only cared about equality, which didn't
  need collation information.

- String data type functions that do equality comparisons and hashing
  are changed to take the (non-)deterministic flag into account.  For
  comparison, this just means skipping various shortcuts and tie
  breakers that use byte-wise comparison.  For hashing, we first need
  to convert the input string to a canonical "sort key" using the ICU
  analogue of strxfrm().

Reviewed-by: Daniel Verite <daniel@manitou-mail.org>
Reviewed-by: Peter Geoghegan <pg@bowt.ie>
Discussion: https://www.postgresql.org/message-id/flat/1ccc668f-4cbc-0bef-af67-450b47cdfee7@2ndquadrant.com
2019-03-22 12:12:43 +01:00

256 lines
6.1 KiB
C

/*-------------------------------------------------------------------------
*
* nodeGroup.c
* Routines to handle group nodes (used for queries with GROUP BY clause).
*
* Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* DESCRIPTION
* The Group node is designed for handling queries with a GROUP BY clause.
* Its outer plan must deliver tuples that are sorted in the order
* specified by the grouping columns (ie. tuples from the same group are
* consecutive). That way, we just have to compare adjacent tuples to
* locate group boundaries.
*
* IDENTIFICATION
* src/backend/executor/nodeGroup.c
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#include "executor/executor.h"
#include "executor/nodeGroup.h"
#include "miscadmin.h"
#include "utils/memutils.h"
/*
* ExecGroup -
*
* Return one tuple for each group of matching input tuples.
*/
static TupleTableSlot *
ExecGroup(PlanState *pstate)
{
GroupState *node = castNode(GroupState, pstate);
ExprContext *econtext;
TupleTableSlot *firsttupleslot;
TupleTableSlot *outerslot;
CHECK_FOR_INTERRUPTS();
/*
* get state info from node
*/
if (node->grp_done)
return NULL;
econtext = node->ss.ps.ps_ExprContext;
/*
* The ScanTupleSlot holds the (copied) first tuple of each group.
*/
firsttupleslot = node->ss.ss_ScanTupleSlot;
/*
* We need not call ResetExprContext here because ExecQualAndReset() will
* reset the per-tuple memory context once per input tuple.
*/
/*
* If first time through, acquire first input tuple and determine whether
* to return it or not.
*/
if (TupIsNull(firsttupleslot))
{
outerslot = ExecProcNode(outerPlanState(node));
if (TupIsNull(outerslot))
{
/* empty input, so return nothing */
node->grp_done = true;
return NULL;
}
/* Copy tuple into firsttupleslot */
ExecCopySlot(firsttupleslot, outerslot);
/*
* Set it up as input for qual test and projection. The expressions
* will access the input tuple as varno OUTER.
*/
econtext->ecxt_outertuple = firsttupleslot;
/*
* Check the qual (HAVING clause); if the group does not match, ignore
* it and fall into scan loop.
*/
if (ExecQual(node->ss.ps.qual, econtext))
{
/*
* Form and return a projection tuple using the first input tuple.
*/
return ExecProject(node->ss.ps.ps_ProjInfo);
}
else
InstrCountFiltered1(node, 1);
}
/*
* This loop iterates once per input tuple group. At the head of the
* loop, we have finished processing the first tuple of the group and now
* need to scan over all the other group members.
*/
for (;;)
{
/*
* Scan over all remaining tuples that belong to this group
*/
for (;;)
{
outerslot = ExecProcNode(outerPlanState(node));
if (TupIsNull(outerslot))
{
/* no more groups, so we're done */
node->grp_done = true;
return NULL;
}
/*
* Compare with first tuple and see if this tuple is of the same
* group. If so, ignore it and keep scanning.
*/
econtext->ecxt_innertuple = firsttupleslot;
econtext->ecxt_outertuple = outerslot;
if (!ExecQualAndReset(node->eqfunction, econtext))
break;
}
/*
* We have the first tuple of the next input group. See if we want to
* return it.
*/
/* Copy tuple, set up as input for qual test and projection */
ExecCopySlot(firsttupleslot, outerslot);
econtext->ecxt_outertuple = firsttupleslot;
/*
* Check the qual (HAVING clause); if the group does not match, ignore
* it and loop back to scan the rest of the group.
*/
if (ExecQual(node->ss.ps.qual, econtext))
{
/*
* Form and return a projection tuple using the first input tuple.
*/
return ExecProject(node->ss.ps.ps_ProjInfo);
}
else
InstrCountFiltered1(node, 1);
}
}
/* -----------------
* ExecInitGroup
*
* Creates the run-time information for the group node produced by the
* planner and initializes its outer subtree
* -----------------
*/
GroupState *
ExecInitGroup(Group *node, EState *estate, int eflags)
{
GroupState *grpstate;
const TupleTableSlotOps *tts_ops;
/* check for unsupported flags */
Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
/*
* create state structure
*/
grpstate = makeNode(GroupState);
grpstate->ss.ps.plan = (Plan *) node;
grpstate->ss.ps.state = estate;
grpstate->ss.ps.ExecProcNode = ExecGroup;
grpstate->grp_done = false;
/*
* create expression context
*/
ExecAssignExprContext(estate, &grpstate->ss.ps);
/*
* initialize child nodes
*/
outerPlanState(grpstate) = ExecInitNode(outerPlan(node), estate, eflags);
/*
* Initialize scan slot and type.
*/
tts_ops = ExecGetResultSlotOps(outerPlanState(&grpstate->ss), NULL);
ExecCreateScanSlotFromOuterPlan(estate, &grpstate->ss, tts_ops);
/*
* Initialize result slot, type and projection.
*/
ExecInitResultTupleSlotTL(&grpstate->ss.ps, &TTSOpsVirtual);
ExecAssignProjectionInfo(&grpstate->ss.ps, NULL);
/*
* initialize child expressions
*/
grpstate->ss.ps.qual =
ExecInitQual(node->plan.qual, (PlanState *) grpstate);
/*
* Precompute fmgr lookup data for inner loop
*/
grpstate->eqfunction =
execTuplesMatchPrepare(ExecGetResultType(outerPlanState(grpstate)),
node->numCols,
node->grpColIdx,
node->grpOperators,
node->grpCollations,
&grpstate->ss.ps);
return grpstate;
}
/* ------------------------
* ExecEndGroup(node)
*
* -----------------------
*/
void
ExecEndGroup(GroupState *node)
{
PlanState *outerPlan;
ExecFreeExprContext(&node->ss.ps);
/* clean up tuple table */
ExecClearTuple(node->ss.ss_ScanTupleSlot);
outerPlan = outerPlanState(node);
ExecEndNode(outerPlan);
}
void
ExecReScanGroup(GroupState *node)
{
PlanState *outerPlan = outerPlanState(node);
node->grp_done = false;
/* must clear first tuple */
ExecClearTuple(node->ss.ss_ScanTupleSlot);
/*
* if chgParam of subnode is not null then plan will be re-scanned by
* first ExecProcNode.
*/
if (outerPlan->chgParam == NULL)
ExecReScan(outerPlan);
}