postgresql/src/backend/parser/parse_expr.c

1520 lines
40 KiB
C
Raw Normal View History

/*-------------------------------------------------------------------------
*
* parse_expr.c
* handle expressions in parser
*
2002-06-20 22:29:54 +02:00
* Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/backend/parser/parse_expr.c,v 1.156 2003/07/18 23:20:32 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#include "catalog/pg_operator.h"
#include "catalog/pg_proc.h"
2003-06-27 19:07:03 +02:00
#include "commands/dbcommands.h"
#include "miscadmin.h"
#include "nodes/makefuncs.h"
#include "nodes/params.h"
#include "nodes/plannodes.h"
#include "parser/analyze.h"
#include "parser/gramparse.h"
1999-07-16 07:00:38 +02:00
#include "parser/parse_coerce.h"
#include "parser/parse_expr.h"
#include "parser/parse_func.h"
#include "parser/parse_oper.h"
#include "parser/parse_relation.h"
2000-06-15 05:33:12 +02:00
#include "parser/parse_type.h"
#include "utils/builtins.h"
#include "utils/lsyscache.h"
#include "utils/syscache.h"
int max_expr_depth = DEFAULT_MAX_EXPR_DEPTH;
static int expr_depth_counter = 0;
bool Transform_null_equals = false;
static Node *typecast_expression(ParseState *pstate, Node *expr,
TypeName *typename);
static Node *transformColumnRef(ParseState *pstate, ColumnRef *cref);
static Node *transformIndirection(ParseState *pstate, Node *basenode,
List *indirection);
/*
* Initialize for parsing a new query.
*
* We reset the expression depth counter here, in case it was left nonzero
* due to elog()'ing out of the last parsing operation.
*/
void
parse_expr_init(void)
{
expr_depth_counter = 0;
}
/*
* transformExpr -
* Analyze and transform expressions. Type checking and type casting is
* done here. The optimizer and the executor cannot handle the original
* (raw) expressions collected by the parse tree. Hence the transformation
* here.
*
* NOTE: there are various cases in which this routine will get applied to
* an already-transformed expression. Some examples:
* 1. At least one construct (BETWEEN/AND) puts the same nodes
* into two branches of the parse tree; hence, some nodes
* are transformed twice.
* 2. Another way it can happen is that coercion of an operator or
* function argument to the required type (via coerce_type())
* can apply transformExpr to an already-transformed subexpression.
* An example here is "SELECT count(*) + 1.0 FROM table".
* While it might be possible to eliminate these cases, the path of
* least resistance so far has been to ensure that transformExpr() does
* no damage if applied to an already-transformed tree. This is pretty
* easy for cases where the transformation replaces one node type with
* another, such as A_Const => Const; we just do nothing when handed
* a Const. More care is needed for node types that are used as both
* input and output of transformExpr; see SubLink for example.
*/
Node *
transformExpr(ParseState *pstate, Node *expr)
{
Node *result = NULL;
if (expr == NULL)
return NULL;
/*
* Guard against an overly complex expression leading to coredump due
* to stack overflow here, or in later recursive routines that
* traverse expression trees. Note that this is very unlikely to
* happen except with pathological queries; but we don't want someone
* to be able to crash the backend quite that easily...
*/
if (++expr_depth_counter > max_expr_depth)
elog(ERROR, "Expression too complex: nesting depth exceeds max_expr_depth = %d",
max_expr_depth);
switch (nodeTag(expr))
{
case T_ColumnRef:
{
result = transformColumnRef(pstate, (ColumnRef *) expr);
break;
}
case T_ParamRef:
{
ParamRef *pref = (ParamRef *) expr;
int paramno = pref->number;
ParseState *toppstate;
Param *param;
List *fields;
/*
* Find topmost ParseState, which is where paramtype info
* lives.
*/
toppstate = pstate;
while (toppstate->parentParseState != NULL)
toppstate = toppstate->parentParseState;
/* Check parameter number is in range */
if (paramno <= 0) /* probably can't happen? */
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_PARAMETER),
errmsg("there is no parameter $%d", paramno)));
if (paramno > toppstate->p_numparams)
{
if (!toppstate->p_variableparams)
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_PARAMETER),
errmsg("there is no parameter $%d",
paramno)));
/* Okay to enlarge param array */
if (toppstate->p_paramtypes)
toppstate->p_paramtypes =
(Oid *) repalloc(toppstate->p_paramtypes,
paramno * sizeof(Oid));
else
toppstate->p_paramtypes =
(Oid *) palloc(paramno * sizeof(Oid));
/* Zero out the previously-unreferenced slots */
MemSet(toppstate->p_paramtypes + toppstate->p_numparams,
0,
(paramno - toppstate->p_numparams) * sizeof(Oid));
toppstate->p_numparams = paramno;
}
if (toppstate->p_variableparams)
{
/* If not seen before, initialize to UNKNOWN type */
if (toppstate->p_paramtypes[paramno-1] == InvalidOid)
toppstate->p_paramtypes[paramno-1] = UNKNOWNOID;
}
param = makeNode(Param);
param->paramkind = PARAM_NUM;
param->paramid = (AttrNumber) paramno;
param->paramtype = toppstate->p_paramtypes[paramno-1];
result = (Node *) param;
/* handle qualification, if any */
foreach(fields, pref->fields)
{
result = ParseFuncOrColumn(pstate,
makeList1(lfirst(fields)),
makeList1(result),
false, false, true);
}
/* handle subscripts, if any */
result = transformIndirection(pstate, result,
pref->indirection);
break;
}
case T_A_Const:
{
A_Const *con = (A_Const *) expr;
Value *val = &con->val;
result = (Node *) make_const(val);
if (con->typename != NULL)
result = typecast_expression(pstate, result,
con->typename);
break;
}
case T_ExprFieldSelect:
{
ExprFieldSelect *efs = (ExprFieldSelect *) expr;
List *fields;
result = transformExpr(pstate, efs->arg);
/* handle qualification, if any */
foreach(fields, efs->fields)
{
result = ParseFuncOrColumn(pstate,
makeList1(lfirst(fields)),
makeList1(result),
false, false, true);
}
/* handle subscripts, if any */
result = transformIndirection(pstate, result,
efs->indirection);
break;
}
case T_TypeCast:
{
TypeCast *tc = (TypeCast *) expr;
Node *arg = transformExpr(pstate, tc->arg);
result = typecast_expression(pstate, arg, tc->typename);
break;
}
case T_A_Expr:
{
A_Expr *a = (A_Expr *) expr;
switch (a->kind)
{
case AEXPR_OP:
{
/*
* Special-case "foo = NULL" and "NULL = foo"
* for compatibility with standards-broken
* products (like Microsoft's). Turn these
* into IS NULL exprs.
*/
2001-09-21 01:31:08 +02:00
if (Transform_null_equals &&
length(a->name) == 1 &&
2002-09-04 22:31:48 +02:00
strcmp(strVal(lfirst(a->name)), "=") == 0 &&
2001-09-21 01:31:08 +02:00
(exprIsNullConstant(a->lexpr) ||
exprIsNullConstant(a->rexpr)))
{
NullTest *n = makeNode(NullTest);
n->nulltesttype = IS_NULL;
if (exprIsNullConstant(a->lexpr))
n->arg = (Expr *) a->rexpr;
else
n->arg = (Expr *) a->lexpr;
2001-09-21 01:31:08 +02:00
result = transformExpr(pstate,
(Node *) n);
}
else
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
result = (Node *) make_op(pstate,
a->name,
2001-09-21 01:31:08 +02:00
lexpr,
rexpr);
}
}
break;
case AEXPR_AND:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
lexpr = coerce_to_boolean(pstate, lexpr, "AND");
rexpr = coerce_to_boolean(pstate, rexpr, "AND");
1997-12-23 20:36:20 +01:00
result = (Node *) makeBoolExpr(AND_EXPR,
makeList2(lexpr,
rexpr));
}
break;
case AEXPR_OR:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
lexpr = coerce_to_boolean(pstate, lexpr, "OR");
rexpr = coerce_to_boolean(pstate, rexpr, "OR");
result = (Node *) makeBoolExpr(OR_EXPR,
makeList2(lexpr,
rexpr));
}
break;
case AEXPR_NOT:
{
Node *rexpr = transformExpr(pstate,
a->rexpr);
rexpr = coerce_to_boolean(pstate, rexpr, "NOT");
result = (Node *) makeBoolExpr(NOT_EXPR,
makeList1(rexpr));
}
break;
case AEXPR_OP_ANY:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
result = (Node *) make_scalar_array_op(pstate,
a->name,
true,
lexpr,
rexpr);
}
break;
case AEXPR_OP_ALL:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
result = (Node *) make_scalar_array_op(pstate,
a->name,
false,
lexpr,
rexpr);
}
break;
case AEXPR_DISTINCT:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
2002-09-04 22:31:48 +02:00
result = (Node *) make_op(pstate,
a->name,
lexpr,
rexpr);
if (((OpExpr *) result)->opresulttype != BOOLOID)
elog(ERROR, "IS DISTINCT FROM requires = operator to yield boolean");
/*
* We rely on DistinctExpr and OpExpr being same struct
*/
NodeSetTag(result, T_DistinctExpr);
}
break;
case AEXPR_NULLIF:
{
Node *lexpr = transformExpr(pstate,
a->lexpr);
Node *rexpr = transformExpr(pstate,
a->rexpr);
result = (Node *) make_op(pstate,
a->name,
lexpr,
rexpr);
if (((OpExpr *) result)->opresulttype != BOOLOID)
elog(ERROR, "NULLIF requires = operator to yield boolean");
/*
* We rely on NullIfExpr and OpExpr being same struct
*/
NodeSetTag(result, T_NullIfExpr);
}
break;
case AEXPR_OF:
{
/*
* Checking an expression for match to type.
* Will result in a boolean constant node.
*/
2002-09-04 22:31:48 +02:00
List *telem;
A_Const *n;
Oid ltype,
rtype;
bool matched = FALSE;
Node *lexpr = transformExpr(pstate,
a->lexpr);
2002-09-04 22:31:48 +02:00
ltype = exprType(lexpr);
foreach(telem, (List *) a->rexpr)
{
rtype = LookupTypeName(lfirst(telem));
matched = (rtype == ltype);
2002-09-04 22:31:48 +02:00
if (matched)
break;
}
2002-09-04 22:31:48 +02:00
/*
* Expect two forms: equals or not equals.
* Flip the sense of the result for not
* equals.
*/
if (strcmp(strVal(lfirst(a->name)), "!=") == 0)
2002-09-04 22:31:48 +02:00
matched = (!matched);
n = makeNode(A_Const);
n->val.type = T_String;
2002-09-04 22:31:48 +02:00
n->val.val.str = (matched ? "t" : "f");
n->typename = SystemTypeName("bool");
result = transformExpr(pstate, (Node *) n);
}
break;
}
break;
}
case T_FuncCall:
{
FuncCall *fn = (FuncCall *) expr;
List *targs;
List *args;
/*
* Transform the list of arguments. We use a shallow
* list copy and then transform-in-place to avoid O(N^2)
* behavior from repeated lappend's.
*/
targs = listCopy(fn->args);
foreach(args, targs)
{
lfirst(args) = transformExpr(pstate,
(Node *) lfirst(args));
}
result = ParseFuncOrColumn(pstate,
1998-12-04 16:34:49 +01:00
fn->funcname,
targs,
fn->agg_star,
fn->agg_distinct,
false);
break;
}
case T_SubLink:
{
SubLink *sublink = (SubLink *) expr;
List *qtrees;
Query *qtree;
1998-02-06 17:46:29 +01:00
/* If we already transformed this node, do nothing */
if (IsA(sublink->subselect, Query))
{
result = expr;
break;
}
1998-02-06 17:46:29 +01:00
pstate->p_hasSubLinks = true;
qtrees = parse_sub_analyze(sublink->subselect, pstate);
if (length(qtrees) != 1)
elog(ERROR, "Bad query in subselect");
qtree = (Query *) lfirst(qtrees);
if (qtree->commandType != CMD_SELECT ||
qtree->resultRelation != 0)
elog(ERROR, "Bad query in subselect");
sublink->subselect = (Node *) qtree;
if (sublink->subLinkType == EXISTS_SUBLINK)
{
/*
* EXISTS needs no lefthand or combining operator.
* These fields should be NIL already, but make sure.
*/
sublink->lefthand = NIL;
sublink->operName = NIL;
sublink->operOids = NIL;
sublink->useOr = FALSE;
}
else if (sublink->subLinkType == EXPR_SUBLINK ||
sublink->subLinkType == ARRAY_SUBLINK)
{
List *tlist = qtree->targetList;
/*
* Make sure the subselect delivers a single column
* (ignoring resjunk targets).
*/
if (tlist == NIL ||
((TargetEntry *) lfirst(tlist))->resdom->resjunk)
elog(ERROR, "Subselect must have a field");
while ((tlist = lnext(tlist)) != NIL)
{
if (!((TargetEntry *) lfirst(tlist))->resdom->resjunk)
elog(ERROR, "Subselect must have only one field");
}
/*
* EXPR and ARRAY need no lefthand or combining operator.
* These fields should be NIL already, but make sure.
*/
sublink->lefthand = NIL;
sublink->operName = NIL;
sublink->operOids = NIL;
sublink->useOr = FALSE;
}
else
{
/* ALL, ANY, or MULTIEXPR: generate operator list */
List *left_list = sublink->lefthand;
List *right_list = qtree->targetList;
int row_length = length(left_list);
bool needNot = false;
List *op = sublink->operName;
char *opname = strVal(llast(op));
List *elist;
1998-01-20 06:05:08 +01:00
/* transform lefthand expressions */
foreach(elist, left_list)
lfirst(elist) = transformExpr(pstate, lfirst(elist));
/*
* If the expression is "<> ALL" (with unqualified opname)
* then convert it to "NOT IN". This is a hack to improve
* efficiency of expressions output by pre-7.4 Postgres.
*/
if (sublink->subLinkType == ALL_SUBLINK &&
length(op) == 1 && strcmp(opname, "<>") == 0)
{
sublink->subLinkType = ANY_SUBLINK;
opname = pstrdup("=");
op = makeList1(makeString(opname));
sublink->operName = op;
needNot = true;
}
/* Set useOr if op is "<>" (possibly qualified) */
if (strcmp(opname, "<>") == 0)
sublink->useOr = TRUE;
else
sublink->useOr = FALSE;
/* Combining operators other than =/<> is dubious... */
if (row_length != 1 &&
strcmp(opname, "=") != 0 &&
strcmp(opname, "<>") != 0)
elog(ERROR, "Row comparison cannot use operator %s",
opname);
/*
* To build the list of combining operator OIDs, we must
* scan subquery's targetlist to find values that will
* be matched against lefthand values. We need to
* ignore resjunk targets, so doing the outer
* iteration over right_list is easier than doing it
* over left_list.
*/
sublink->operOids = NIL;
while (right_list != NIL)
1998-01-20 06:05:08 +01:00
{
TargetEntry *tent = (TargetEntry *) lfirst(right_list);
Node *lexpr;
Operator optup;
Form_pg_operator opform;
1998-01-20 06:05:08 +01:00
right_list = lnext(right_list);
if (tent->resdom->resjunk)
continue;
if (left_list == NIL)
elog(ERROR, "Subselect has too many fields");
lexpr = lfirst(left_list);
left_list = lnext(left_list);
Clean up two rather nasty bugs in operator selection code. 1. If there is exactly one pg_operator entry of the right name and oprkind, oper() and related routines would return that entry whether its input type had anything to do with the request or not. This is just premature optimization: we shouldn't return the single candidate until after we verify that it really is a valid candidate, ie, is at least coercion-compatible with the given types. 2. oper() and related routines only promise a coercion-compatible result. Unfortunately, there were quite a few callers that assumed the returned operator is binary-compatible with the given datatype; they would proceed to call it without making any datatype coercions. These callers include sorting, grouping, aggregation, and VACUUM ANALYZE. In general I think it is appropriate for these callers to require an exact or binary-compatible match, so I've added a new routine compatible_oper() that only succeeds if it can find an operator that doesn't require any run-time conversions. Callers now call oper() or compatible_oper() depending on whether they are prepared to deal with type conversion or not. The upshot of these bugs is revealed by the following silliness in PL/Tcl's selftest: it creates an operator @< on int4, and then tries to use it to sort a char(N) column. The system would let it do that :-( (and evidently has done so since 6.3 :-( :-(). The result in this case was just a silly sort order, but the reverse combination would've provoked coredump from trying to dereference integers. With this fix you get more reasonable behavior: pltcl_test=# select * from T_pkey1 order by key1, key2 using @<; ERROR: Unable to identify an operator '@<' for types 'bpchar' and 'bpchar' You will have to retype this query using an explicit cast
2001-02-16 04:16:58 +01:00
/*
2001-03-22 05:01:46 +01:00
* It's OK to use oper() not compatible_oper()
* here, because make_subplan() will insert type
* coercion calls if needed.
Clean up two rather nasty bugs in operator selection code. 1. If there is exactly one pg_operator entry of the right name and oprkind, oper() and related routines would return that entry whether its input type had anything to do with the request or not. This is just premature optimization: we shouldn't return the single candidate until after we verify that it really is a valid candidate, ie, is at least coercion-compatible with the given types. 2. oper() and related routines only promise a coercion-compatible result. Unfortunately, there were quite a few callers that assumed the returned operator is binary-compatible with the given datatype; they would proceed to call it without making any datatype coercions. These callers include sorting, grouping, aggregation, and VACUUM ANALYZE. In general I think it is appropriate for these callers to require an exact or binary-compatible match, so I've added a new routine compatible_oper() that only succeeds if it can find an operator that doesn't require any run-time conversions. Callers now call oper() or compatible_oper() depending on whether they are prepared to deal with type conversion or not. The upshot of these bugs is revealed by the following silliness in PL/Tcl's selftest: it creates an operator @< on int4, and then tries to use it to sort a char(N) column. The system would let it do that :-( (and evidently has done so since 6.3 :-( :-(). The result in this case was just a silly sort order, but the reverse combination would've provoked coredump from trying to dereference integers. With this fix you get more reasonable behavior: pltcl_test=# select * from T_pkey1 order by key1, key2 using @<; ERROR: Unable to identify an operator '@<' for types 'bpchar' and 'bpchar' You will have to retype this query using an explicit cast
2001-02-16 04:16:58 +01:00
*/
2003-06-25 23:30:34 +02:00
optup = oper(op,
exprType(lexpr),
exprType((Node *) tent->expr),
false);
opform = (Form_pg_operator) GETSTRUCT(optup);
if (opform->oprresult != BOOLOID)
elog(ERROR, "%s has result type of %s, but must return %s"
" to be used with quantified predicate subquery",
2002-09-04 22:31:48 +02:00
opname, format_type_be(opform->oprresult),
format_type_be(BOOLOID));
if (get_func_retset(opform->oprcode))
elog(ERROR, "%s must not return a set"
" to be used with quantified predicate subquery",
opname);
sublink->operOids = lappendo(sublink->operOids,
oprid(optup));
ReleaseSysCache(optup);
1998-01-20 06:05:08 +01:00
}
if (left_list != NIL)
elog(ERROR, "Subselect has too few fields");
if (needNot)
{
expr = coerce_to_boolean(pstate, expr, "NOT");
expr = (Node *) makeBoolExpr(NOT_EXPR,
makeList1(expr));
}
}
1998-02-13 04:41:23 +01:00
result = (Node *) expr;
break;
}
1998-12-04 16:34:49 +01:00
case T_CaseExpr:
{
CaseExpr *c = (CaseExpr *) expr;
CaseExpr *newc = makeNode(CaseExpr);
List *newargs = NIL;
List *typeids = NIL;
1998-12-04 16:34:49 +01:00
List *args;
Node *defresult;
1998-12-04 16:34:49 +01:00
Oid ptype;
/* transform the list of arguments */
foreach(args, c->args)
{
CaseWhen *w = (CaseWhen *) lfirst(args);
CaseWhen *neww = makeNode(CaseWhen);
Node *warg;
Assert(IsA(w, CaseWhen));
warg = (Node *) w->expr;
1998-12-04 16:34:49 +01:00
if (c->arg != NULL)
{
/* shorthand form was specified, so expand... */
warg = (Node *) makeSimpleA_Expr(AEXPR_OP, "=",
(Node *) c->arg,
warg);
1998-12-04 16:34:49 +01:00
}
neww->expr = (Expr *) transformExpr(pstate, warg);
neww->expr = (Expr *) coerce_to_boolean(pstate,
(Node *) neww->expr,
"CASE/WHEN");
/*
* result is NULL for NULLIF() construct - thomas
* 1998-11-11
*/
warg = (Node *) w->result;
if (warg == NULL)
{
A_Const *n = makeNode(A_Const);
n->val.type = T_Null;
warg = (Node *) n;
}
neww->result = (Expr *) transformExpr(pstate, warg);
newargs = lappend(newargs, neww);
typeids = lappendo(typeids, exprType((Node *) neww->result));
1998-12-04 16:34:49 +01:00
}
newc->args = newargs;
1999-05-25 18:15:34 +02:00
/*
* It's not shorthand anymore, so drop the implicit
* argument. This is necessary to keep any re-application
* of transformExpr from doing the wrong thing.
*/
newc->arg = NULL;
/* transform the default clause */
defresult = (Node *) c->defresult;
if (defresult == NULL)
1998-12-04 16:34:49 +01:00
{
1999-05-25 18:15:34 +02:00
A_Const *n = makeNode(A_Const);
1998-12-04 16:34:49 +01:00
n->val.type = T_Null;
defresult = (Node *) n;
1998-12-04 16:34:49 +01:00
}
newc->defresult = (Expr *) transformExpr(pstate, defresult);
2001-03-22 05:01:46 +01:00
/*
* Note: default result is considered the most significant
2001-03-22 05:01:46 +01:00
* type in determining preferred type. This is how the
* code worked before, but it seems a little bogus to me
* --- tgl
*/
typeids = lconso(exprType((Node *) newc->defresult), typeids);
1998-12-04 16:34:49 +01:00
ptype = select_common_type(typeids, "CASE");
newc->casetype = ptype;
1998-12-04 16:34:49 +01:00
/* Convert default result clause, if necessary */
newc->defresult = (Expr *)
coerce_to_common_type(pstate,
(Node *) newc->defresult,
ptype,
"CASE/ELSE");
/* Convert when-clause results, if necessary */
foreach(args, newc->args)
1998-12-04 16:34:49 +01:00
{
CaseWhen *w = (CaseWhen *) lfirst(args);
1998-12-04 16:34:49 +01:00
w->result = (Expr *)
coerce_to_common_type(pstate,
(Node *) w->result,
ptype,
"CASE/WHEN");
}
result = (Node *) newc;
1998-12-04 16:34:49 +01:00
break;
}
case T_ArrayExpr:
{
ArrayExpr *a = (ArrayExpr *) expr;
ArrayExpr *newa = makeNode(ArrayExpr);
List *newelems = NIL;
List *newcoercedelems = NIL;
List *typeids = NIL;
List *element;
Oid array_type;
Oid element_type;
int ndims;
/* Transform the element expressions */
foreach(element, a->elements)
{
Node *e = (Node *) lfirst(element);
Node *newe;
newe = transformExpr(pstate, e);
newelems = lappend(newelems, newe);
typeids = lappendo(typeids, exprType(newe));
}
/* Select a common type for the elements */
element_type = select_common_type(typeids, "ARRAY");
/* Coerce arguments to common type if necessary */
foreach(element, newelems)
{
Node *e = (Node *) lfirst(element);
Node *newe;
newe = coerce_to_common_type(pstate, e,
element_type,
"ARRAY");
newcoercedelems = lappend(newcoercedelems, newe);
}
/* Do we have an array type to use? */
array_type = get_array_type(element_type);
if (array_type != InvalidOid)
{
/* Elements are presumably of scalar type */
ndims = 1;
}
else
{
/* Must be nested array expressions */
array_type = element_type;
element_type = get_element_type(array_type);
if (!OidIsValid(element_type))
elog(ERROR, "Cannot find array type for datatype %s",
format_type_be(array_type));
/*
* make sure the element expressions all have the same
* number of dimensions
*/
ndims = 0;
foreach(element, newcoercedelems)
{
ArrayExpr *e = (ArrayExpr *) lfirst(element);
if (!IsA(e, ArrayExpr))
elog(ERROR, "Multidimensional ARRAY[] must be built from nested array expressions");
if (ndims == 0)
ndims = e->ndims;
else if (e->ndims != ndims)
elog(ERROR, "Nested array expressions must have "
"common number of dimensions");
if (e->element_typeid != element_type)
elog(ERROR, "Nested array expressions must have "
"common element type");
}
/* increment the number of dimensions */
ndims++;
/* make sure we don't have too many dimensions now */
if (ndims > MAXDIM)
elog(ERROR, "Number of array dimensions, %d, "
"exceeds the maximum allowed %d",
ndims, MAXDIM);
}
newa->array_typeid = array_type;
newa->element_typeid = element_type;
newa->elements = newcoercedelems;
newa->ndims = ndims;
result = (Node *) newa;
break;
}
case T_CoalesceExpr:
{
CoalesceExpr *c = (CoalesceExpr *) expr;
CoalesceExpr *newc = makeNode(CoalesceExpr);
List *newargs = NIL;
List *newcoercedargs = NIL;
List *typeids = NIL;
List *args;
foreach(args, c->args)
{
Node *e = (Node *) lfirst(args);
Node *newe;
newe = transformExpr(pstate, e);
newargs = lappend(newargs, newe);
typeids = lappendo(typeids, exprType(newe));
}
newc->coalescetype = select_common_type(typeids, "COALESCE");
/* Convert arguments if necessary */
foreach(args, newargs)
{
Node *e = (Node *) lfirst(args);
Node *newe;
newe = coerce_to_common_type(pstate, e,
newc->coalescetype,
"COALESCE");
newcoercedargs = lappend(newcoercedargs, newe);
}
newc->args = newcoercedargs;
result = (Node *) newc;
break;
}
case T_NullTest:
{
NullTest *n = (NullTest *) expr;
n->arg = (Expr *) transformExpr(pstate, (Node *) n->arg);
/* the argument can be any type, so don't coerce it */
result = expr;
break;
}
case T_BooleanTest:
{
BooleanTest *b = (BooleanTest *) expr;
const char *clausename;
switch (b->booltesttype)
{
case IS_TRUE:
clausename = "IS TRUE";
break;
case IS_NOT_TRUE:
clausename = "IS NOT TRUE";
break;
case IS_FALSE:
clausename = "IS FALSE";
break;
case IS_NOT_FALSE:
clausename = "IS NOT FALSE";
break;
case IS_UNKNOWN:
clausename = "IS UNKNOWN";
break;
case IS_NOT_UNKNOWN:
clausename = "IS NOT UNKNOWN";
break;
default:
elog(ERROR, "transformExpr: unexpected booltesttype %d",
(int) b->booltesttype);
2002-09-04 22:31:48 +02:00
clausename = NULL; /* keep compiler quiet */
}
b->arg = (Expr *) transformExpr(pstate, (Node *) b->arg);
b->arg = (Expr *) coerce_to_boolean(pstate,
(Node *) b->arg,
clausename);
result = expr;
break;
}
2002-09-04 22:31:48 +02:00
/*********************************************
* Quietly accept node types that may be presented when we are
* called on an already-transformed tree.
*
* Do any other node types need to be accepted? For now we are
* taking a conservative approach, and only accepting node
* types that are demonstrably necessary to accept.
*********************************************/
1998-03-26 22:08:10 +01:00
case T_Var:
case T_Const:
case T_Param:
case T_Aggref:
case T_ArrayRef:
case T_FuncExpr:
case T_OpExpr:
case T_DistinctExpr:
case T_ScalarArrayOpExpr:
case T_NullIfExpr:
case T_BoolExpr:
case T_FieldSelect:
case T_RelabelType:
case T_CoerceToDomain:
case T_CoerceToDomainValue:
case T_SetToDefault:
1998-03-26 22:08:10 +01:00
{
result = (Node *) expr;
break;
}
default:
/* should not reach here */
elog(ERROR, "transformExpr: does not know how to transform node %d"
" (internal error)", nodeTag(expr));
break;
}
expr_depth_counter--;
return result;
}
static Node *
transformIndirection(ParseState *pstate, Node *basenode, List *indirection)
{
if (indirection == NIL)
return basenode;
return (Node *) transformArraySubscripts(pstate,
basenode,
exprType(basenode),
exprTypmod(basenode),
indirection,
false,
NULL);
}
static Node *
transformColumnRef(ParseState *pstate, ColumnRef *cref)
{
int numnames = length(cref->fields);
Node *node;
RangeVar *rv;
int levels_up;
/*----------
* The allowed syntaxes are:
*
* A First try to resolve as unqualified column name;
* if no luck, try to resolve as unqual. table name (A.*).
* A.B A is an unqual. table name; B is either a
* column or function name (trying column name first).
* A.B.C schema A, table B, col or func name C.
* A.B.C.D catalog A, schema B, table C, col or func D.
* A.* A is an unqual. table name; means whole-row value.
* A.B.* whole-row value of table B in schema A.
* A.B.C.* whole-row value of table C in schema B in catalog A.
*
* We do not need to cope with bare "*"; that will only be accepted by
* the grammar at the top level of a SELECT list, and transformTargetList
* will take care of it before it ever gets here.
*
* Currently, if a catalog name is given then it must equal the current
* database name; we check it here and then discard it.
*
* For whole-row references, the result is an untransformed RangeVar,
* which will work as the argument to a function call, but not in any
* other context at present. (We could instead coerce to a whole-row Var,
* but that will fail for subselect and join RTEs, because there is no
* pg_type entry for their rowtypes.)
*----------
*/
switch (numnames)
{
case 1:
2002-09-04 22:31:48 +02:00
{
char *name = strVal(lfirst(cref->fields));
2002-09-04 22:31:48 +02:00
/* Try to identify as an unqualified column */
node = colnameToVar(pstate, name);
if (node == NULL)
{
/*
* Not known as a column of any range-table entry.
*
* Consider the possibility that it's VALUE in a domain
* check expression. (We handle VALUE as a name, not a
* keyword, to avoid breaking a lot of applications that
* have used VALUE as a column name in the past.)
*/
if (pstate->p_value_substitute != NULL &&
strcmp(name, "value") == 0)
{
node = (Node *) copyObject(pstate->p_value_substitute);
break;
}
/*
* Try to find the name as a relation ... but not if
2002-09-04 22:31:48 +02:00
* subscripts appear. Note also that only relations
* already entered into the rangetable will be
* recognized.
*
* This is a hack for backwards compatibility with
* PostQUEL-inspired syntax. The preferred form now
2002-09-04 22:31:48 +02:00
* is "rel.*".
*/
if (cref->indirection == NIL &&
refnameRangeTblEntry(pstate, NULL, name,
&levels_up) != NULL)
{
rv = makeNode(RangeVar);
rv->relname = name;
rv->inhOpt = INH_DEFAULT;
node = (Node *) rv;
}
else
elog(ERROR, "Attribute \"%s\" not found", name);
}
break;
}
case 2:
{
2002-09-04 22:31:48 +02:00
char *name1 = strVal(lfirst(cref->fields));
char *name2 = strVal(lsecond(cref->fields));
2002-09-04 22:31:48 +02:00
/* Whole-row reference? */
if (strcmp(name2, "*") == 0)
{
rv = makeNode(RangeVar);
2002-09-04 22:31:48 +02:00
rv->relname = name1;
rv->inhOpt = INH_DEFAULT;
node = (Node *) rv;
2002-09-04 22:31:48 +02:00
break;
}
2002-09-04 22:31:48 +02:00
/* Try to identify as a once-qualified column */
node = qualifiedNameToVar(pstate, NULL, name1, name2, true);
if (node == NULL)
{
/*
* Not known as a column of any range-table entry, so
* try it as a function call. Here, we will create an
* implicit RTE for tables not already entered.
*/
rv = makeNode(RangeVar);
rv->relname = name1;
rv->inhOpt = INH_DEFAULT;
node = ParseFuncOrColumn(pstate,
makeList1(makeString(name2)),
makeList1(rv),
false, false, true);
}
break;
}
case 3:
{
2002-09-04 22:31:48 +02:00
char *name1 = strVal(lfirst(cref->fields));
char *name2 = strVal(lsecond(cref->fields));
char *name3 = strVal(lthird(cref->fields));
2002-09-04 22:31:48 +02:00
/* Whole-row reference? */
if (strcmp(name3, "*") == 0)
{
rv = makeNode(RangeVar);
rv->schemaname = name1;
rv->relname = name2;
rv->inhOpt = INH_DEFAULT;
node = (Node *) rv;
break;
}
/* Try to identify as a twice-qualified column */
node = qualifiedNameToVar(pstate, name1, name2, name3, true);
if (node == NULL)
{
/* Try it as a function call */
rv = makeNode(RangeVar);
rv->schemaname = name1;
rv->relname = name2;
rv->inhOpt = INH_DEFAULT;
node = ParseFuncOrColumn(pstate,
makeList1(makeString(name3)),
makeList1(rv),
false, false, true);
}
break;
}
case 4:
{
2002-09-04 22:31:48 +02:00
char *name1 = strVal(lfirst(cref->fields));
char *name2 = strVal(lsecond(cref->fields));
char *name3 = strVal(lthird(cref->fields));
char *name4 = strVal(lfourth(cref->fields));
2002-09-04 22:31:48 +02:00
/*
* We check the catalog name and then ignore it.
*/
2003-06-27 16:45:32 +02:00
if (strcmp(name1, get_database_name(MyDatabaseId)) != 0)
2002-09-04 22:31:48 +02:00
elog(ERROR, "Cross-database references are not implemented");
/* Whole-row reference? */
if (strcmp(name4, "*") == 0)
{
rv = makeNode(RangeVar);
rv->schemaname = name2;
rv->relname = name3;
rv->inhOpt = INH_DEFAULT;
node = (Node *) rv;
break;
}
/* Try to identify as a twice-qualified column */
node = qualifiedNameToVar(pstate, name2, name3, name4, true);
if (node == NULL)
{
/* Try it as a function call */
rv = makeNode(RangeVar);
rv->schemaname = name2;
rv->relname = name3;
rv->inhOpt = INH_DEFAULT;
node = ParseFuncOrColumn(pstate,
makeList1(makeString(name4)),
makeList1(rv),
false, false, true);
}
break;
}
default:
elog(ERROR, "Invalid qualified name syntax (too many names)");
node = NULL; /* keep compiler quiet */
break;
}
return transformIndirection(pstate, node, cref->indirection);
}
/*
* exprType -
* returns the Oid of the type of the expression. (Used for typechecking.)
*/
Oid
exprType(Node *expr)
{
Oid type;
1998-12-04 16:34:49 +01:00
if (!expr)
return InvalidOid;
1998-12-04 16:34:49 +01:00
switch (nodeTag(expr))
{
case T_Var:
type = ((Var *) expr)->vartype;
break;
case T_Const:
type = ((Const *) expr)->consttype;
break;
case T_Param:
type = ((Param *) expr)->paramtype;
break;
1999-01-24 01:28:37 +01:00
case T_Aggref:
type = ((Aggref *) expr)->aggtype;
break;
case T_ArrayRef:
type = ((ArrayRef *) expr)->refrestype;
break;
case T_FuncExpr:
type = ((FuncExpr *) expr)->funcresulttype;
break;
case T_OpExpr:
type = ((OpExpr *) expr)->opresulttype;
break;
case T_DistinctExpr:
type = ((DistinctExpr *) expr)->opresulttype;
break;
case T_ScalarArrayOpExpr:
type = BOOLOID;
break;
case T_BoolExpr:
type = BOOLOID;
break;
case T_SubLink:
{
1999-05-25 18:15:34 +02:00
SubLink *sublink = (SubLink *) expr;
if (sublink->subLinkType == EXPR_SUBLINK ||
sublink->subLinkType == ARRAY_SUBLINK)
{
/* get the type of the subselect's first target column */
Query *qtree = (Query *) sublink->subselect;
TargetEntry *tent;
if (!qtree || !IsA(qtree, Query))
elog(ERROR, "exprType: Cannot get type for untransformed sublink");
tent = (TargetEntry *) lfirst(qtree->targetList);
Assert(IsA(tent, TargetEntry));
Assert(!tent->resdom->resjunk);
if (sublink->subLinkType == EXPR_SUBLINK)
type = tent->resdom->restype;
else /* ARRAY_SUBLINK */
{
type = get_array_type(tent->resdom->restype);
if (!OidIsValid(type))
elog(ERROR, "Cannot find array type for datatype %s",
format_type_be(tent->resdom->restype));
}
}
else
{
/* for all other sublink types, result is boolean */
type = BOOLOID;
}
}
break;
case T_SubPlan:
{
/*
* Although the parser does not ever deal with already-planned
* expression trees, we support SubPlan nodes in this routine
* for the convenience of ruleutils.c.
*/
SubPlan *subplan = (SubPlan *) expr;
if (subplan->subLinkType == EXPR_SUBLINK ||
subplan->subLinkType == ARRAY_SUBLINK)
{
/* get the type of the subselect's first target column */
TargetEntry *tent;
tent = (TargetEntry *) lfirst(subplan->plan->targetlist);
Assert(IsA(tent, TargetEntry));
Assert(!tent->resdom->resjunk);
if (subplan->subLinkType == EXPR_SUBLINK)
type = tent->resdom->restype;
else /* ARRAY_SUBLINK */
{
type = get_array_type(tent->resdom->restype);
if (!OidIsValid(type))
elog(ERROR, "Cannot find array type for datatype %s",
format_type_be(tent->resdom->restype));
}
}
else
{
/* for all other subplan types, result is boolean */
type = BOOLOID;
}
}
break;
case T_FieldSelect:
type = ((FieldSelect *) expr)->resulttype;
break;
case T_RelabelType:
type = ((RelabelType *) expr)->resulttype;
break;
1998-12-04 16:34:49 +01:00
case T_CaseExpr:
type = ((CaseExpr *) expr)->casetype;
break;
case T_CaseWhen:
type = exprType((Node *) ((CaseWhen *) expr)->result);
1998-12-04 16:34:49 +01:00
break;
case T_ArrayExpr:
type = ((ArrayExpr *) expr)->array_typeid;
break;
case T_CoalesceExpr:
type = ((CoalesceExpr *) expr)->coalescetype;
break;
case T_NullIfExpr:
type = exprType((Node *) lfirst(((NullIfExpr *) expr)->args));
break;
case T_NullTest:
type = BOOLOID;
break;
case T_BooleanTest:
type = BOOLOID;
break;
case T_CoerceToDomain:
type = ((CoerceToDomain *) expr)->resulttype;
break;
case T_CoerceToDomainValue:
type = ((CoerceToDomainValue *) expr)->typeId;
break;
case T_SetToDefault:
type = ((SetToDefault *) expr)->typeId;
break;
case T_RangeVar:
/*
* If someone uses a bare relation name in an expression,
* we will likely first notice a problem here (see comments in
* transformColumnRef()). Issue an appropriate error message.
*/
elog(ERROR, "Relation reference \"%s\" cannot be used in an expression",
((RangeVar *) expr)->relname);
type = InvalidOid; /* keep compiler quiet */
break;
default:
elog(ERROR, "exprType: Do not know how to get type for %d node",
nodeTag(expr));
type = InvalidOid; /* keep compiler quiet */
break;
}
return type;
}
/*
* exprTypmod -
* returns the type-specific attrmod of the expression, if it can be
* determined. In most cases, it can't and we return -1.
*/
int32
exprTypmod(Node *expr)
{
if (!expr)
return -1;
switch (nodeTag(expr))
{
case T_Var:
return ((Var *) expr)->vartypmod;
case T_Const:
{
/* Be smart about string constants... */
Const *con = (Const *) expr;
switch (con->consttype)
{
case BPCHAROID:
if (!con->constisnull)
return VARSIZE(DatumGetPointer(con->constvalue));
break;
default:
break;
}
}
break;
case T_FuncExpr:
{
int32 coercedTypmod;
/* Be smart about length-coercion functions... */
if (exprIsLengthCoercion(expr, &coercedTypmod))
return coercedTypmod;
}
break;
case T_FieldSelect:
return ((FieldSelect *) expr)->resulttypmod;
case T_RelabelType:
return ((RelabelType *) expr)->resulttypmod;
case T_CaseExpr:
{
/*
* If all the alternatives agree on type/typmod, return
* that typmod, else use -1
*/
CaseExpr *cexpr = (CaseExpr *) expr;
Oid casetype = cexpr->casetype;
int32 typmod;
List *arg;
if (!cexpr->defresult)
return -1;
if (exprType((Node *) cexpr->defresult) != casetype)
return -1;
typmod = exprTypmod((Node *) cexpr->defresult);
if (typmod < 0)
return -1; /* no point in trying harder */
foreach(arg, cexpr->args)
{
CaseWhen *w = (CaseWhen *) lfirst(arg);
Assert(IsA(w, CaseWhen));
if (exprType((Node *) w->result) != casetype)
return -1;
if (exprTypmod((Node *) w->result) != typmod)
return -1;
}
return typmod;
}
break;
case T_CoalesceExpr:
{
/*
* If all the alternatives agree on type/typmod, return
* that typmod, else use -1
*/
CoalesceExpr *cexpr = (CoalesceExpr *) expr;
Oid coalescetype = cexpr->coalescetype;
int32 typmod;
List *arg;
typmod = exprTypmod((Node *) lfirst(cexpr->args));
foreach(arg, cexpr->args)
{
Node *e = (Node *) lfirst(arg);
if (exprType(e) != coalescetype)
return -1;
if (exprTypmod(e) != typmod)
return -1;
}
return typmod;
}
break;
case T_NullIfExpr:
{
NullIfExpr *nexpr = (NullIfExpr *) expr;
return exprTypmod((Node *) lfirst(nexpr->args));
}
break;
case T_CoerceToDomain:
return ((CoerceToDomain *) expr)->resulttypmod;
case T_CoerceToDomainValue:
return ((CoerceToDomainValue *) expr)->typeMod;
case T_SetToDefault:
return ((SetToDefault *) expr)->typeMod;
default:
break;
}
return -1;
}
/*
* exprIsLengthCoercion
* Detect whether an expression tree is an application of a datatype's
* typmod-coercion function. Optionally extract the result's typmod.
*
* If coercedTypmod is not NULL, the typmod is stored there if the expression
* is a length-coercion function, else -1 is stored there.
*/
bool
exprIsLengthCoercion(Node *expr, int32 *coercedTypmod)
{
FuncExpr *func;
int nargs;
Const *second_arg;
if (coercedTypmod != NULL)
*coercedTypmod = -1; /* default result on failure */
/* Is it a function-call at all? */
if (expr == NULL || !IsA(expr, FuncExpr))
return false;
func = (FuncExpr *) expr;
/*
* If it didn't come from a coercion context, reject.
*/
if (func->funcformat != COERCE_EXPLICIT_CAST &&
func->funcformat != COERCE_IMPLICIT_CAST)
return false;
/*
* If it's not a two-argument or three-argument function with the second
* argument being an int4 constant, it can't have been created from a
* length coercion (it must be a type coercion, instead).
*/
nargs = length(func->args);
if (nargs < 2 || nargs > 3)
return false;
second_arg = (Const *) lsecond(func->args);
if (!IsA(second_arg, Const) ||
second_arg->consttype != INT4OID ||
second_arg->constisnull)
return false;
/*
* OK, it is indeed a length-coercion function.
*/
if (coercedTypmod != NULL)
*coercedTypmod = DatumGetInt32(second_arg->constvalue);
return true;
}
/*
* Handle an explicit CAST construct.
*
* The given expr has already been transformed, but we need to lookup
* the type name and then apply any necessary coercion function(s).
*/
static Node *
typecast_expression(ParseState *pstate, Node *expr, TypeName *typename)
{
Oid inputType = exprType(expr);
Oid targetType;
targetType = typenameTypeId(typename);
if (inputType == InvalidOid)
return expr; /* do nothing if NULL input */
expr = coerce_to_target_type(pstate, expr, inputType,
targetType, typename->typmod,
COERCION_EXPLICIT,
COERCE_EXPLICIT_CAST);
if (expr == NULL)
elog(ERROR, "Cannot cast type %s to %s",
format_type_be(inputType),
format_type_be(targetType));
return expr;
}