granicus.if.org Git - postgresql/blob - src/backend/parser/parse_node.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * parse_node.c
   4  *        various routines that make nodes for querytrees
   5  *
   6  * Portions Copyright (c) 1996-2016, PostgreSQL Global Development Group
   7  * Portions Copyright (c) 1994, Regents of the University of California
   8  *
   9  *
  10  * IDENTIFICATION
  11  *        src/backend/parser/parse_node.c
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15 #include "postgres.h"
  16
  17 #include "access/heapam.h"
  18 #include "access/htup_details.h"
  19 #include "catalog/pg_type.h"
  20 #include "mb/pg_wchar.h"
  21 #include "nodes/makefuncs.h"
  22 #include "nodes/nodeFuncs.h"
  23 #include "parser/parsetree.h"
  24 #include "parser/parse_coerce.h"
  25 #include "parser/parse_expr.h"
  26 #include "parser/parse_relation.h"
  27 #include "utils/builtins.h"
  28 #include "utils/int8.h"
  29 #include "utils/lsyscache.h"
  30 #include "utils/syscache.h"
  31 #include "utils/varbit.h"
  32
  33
  34 static void pcb_error_callback(void *arg);
  35
  36
  37 /*
  38  * make_parsestate
  39  *              Allocate and initialize a new ParseState.
  40  *
  41  * Caller should eventually release the ParseState via free_parsestate().
  42  */
  43 ParseState *
  44 make_parsestate(ParseState *parentParseState)
  45 {
  46         ParseState *pstate;
  47
  48         pstate = palloc0(sizeof(ParseState));
  49
  50         pstate->parentParseState = parentParseState;
  51
  52         /* Fill in fields that don't start at null/false/zero */
  53         pstate->p_next_resno = 1;
  54
  55         if (parentParseState)
  56         {
  57                 pstate->p_sourcetext = parentParseState->p_sourcetext;
  58                 /* all hooks are copied from parent */
  59                 pstate->p_pre_columnref_hook = parentParseState->p_pre_columnref_hook;
  60                 pstate->p_post_columnref_hook = parentParseState->p_post_columnref_hook;
  61                 pstate->p_paramref_hook = parentParseState->p_paramref_hook;
  62                 pstate->p_coerce_param_hook = parentParseState->p_coerce_param_hook;
  63                 pstate->p_ref_hook_state = parentParseState->p_ref_hook_state;
  64         }
  65
  66         return pstate;
  67 }
  68
  69 /*
  70  * free_parsestate
  71  *              Release a ParseState and any subsidiary resources.
  72  */
  73 void
  74 free_parsestate(ParseState *pstate)
  75 {
  76         /*
  77          * Check that we did not produce too many resnos; at the very least we
  78          * cannot allow more than 2^16, since that would exceed the range of a
  79          * AttrNumber. It seems safest to use MaxTupleAttributeNumber.
  80          */
  81         if (pstate->p_next_resno - 1 > MaxTupleAttributeNumber)
  82                 ereport(ERROR,
  83                                 (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
  84                                  errmsg("target lists can have at most %d entries",
  85                                                 MaxTupleAttributeNumber)));
  86
  87         if (pstate->p_target_relation != NULL)
  88                 heap_close(pstate->p_target_relation, NoLock);
  89
  90         pfree(pstate);
  91 }
  92
  93
  94 /*
  95  * parser_errposition
  96  *              Report a parse-analysis-time cursor position, if possible.
  97  *
  98  * This is expected to be used within an ereport() call.  The return value
  99  * is a dummy (always 0, in fact).
 100  *
 101  * The locations stored in raw parsetrees are byte offsets into the source
 102  * string.  We have to convert them to 1-based character indexes for reporting
 103  * to clients.  (We do things this way to avoid unnecessary overhead in the
 104  * normal non-error case: computing character indexes would be much more
 105  * expensive than storing token offsets.)
 106  */
 107 int
 108 parser_errposition(ParseState *pstate, int location)
 109 {
 110         int                     pos;
 111
 112         /* No-op if location was not provided */
 113         if (location < 0)
 114                 return 0;
 115         /* Can't do anything if source text is not available */
 116         if (pstate == NULL || pstate->p_sourcetext == NULL)
 117                 return 0;
 118         /* Convert offset to character number */
 119         pos = pg_mbstrlen_with_len(pstate->p_sourcetext, location) + 1;
 120         /* And pass it to the ereport mechanism */
 121         return errposition(pos);
 122 }
 123
 124
 125 /*
 126  * setup_parser_errposition_callback
 127  *              Arrange for non-parser errors to report an error position
 128  *
 129  * Sometimes the parser calls functions that aren't part of the parser
 130  * subsystem and can't reasonably be passed a ParseState; yet we would
 131  * like any errors thrown in those functions to be tagged with a parse
 132  * error location.  Use this function to set up an error context stack
 133  * entry that will accomplish that.  Usage pattern:
 134  *
 135  *              declare a local variable "ParseCallbackState pcbstate"
 136  *              ...
 137  *              setup_parser_errposition_callback(&pcbstate, pstate, location);
 138  *              call function that might throw error;
 139  *              cancel_parser_errposition_callback(&pcbstate);
 140  */
 141 void
 142 setup_parser_errposition_callback(ParseCallbackState *pcbstate,
 143                                                                   ParseState *pstate, int location)
 144 {
 145         /* Setup error traceback support for ereport() */
 146         pcbstate->pstate = pstate;
 147         pcbstate->location = location;
 148         pcbstate->errcallback.callback = pcb_error_callback;
 149         pcbstate->errcallback.arg = (void *) pcbstate;
 150         pcbstate->errcallback.previous = error_context_stack;
 151         error_context_stack = &pcbstate->errcallback;
 152 }
 153
 154 /*
 155  * Cancel a previously-set-up errposition callback.
 156  */
 157 void
 158 cancel_parser_errposition_callback(ParseCallbackState *pcbstate)
 159 {
 160         /* Pop the error context stack */
 161         error_context_stack = pcbstate->errcallback.previous;
 162 }
 163
 164 /*
 165  * Error context callback for inserting parser error location.
 166  *
 167  * Note that this will be called for *any* error occurring while the
 168  * callback is installed.  We avoid inserting an irrelevant error location
 169  * if the error is a query cancel --- are there any other important cases?
 170  */
 171 static void
 172 pcb_error_callback(void *arg)
 173 {
 174         ParseCallbackState *pcbstate = (ParseCallbackState *) arg;
 175
 176         if (geterrcode() != ERRCODE_QUERY_CANCELED)
 177                 (void) parser_errposition(pcbstate->pstate, pcbstate->location);
 178 }
 179
 180
 181 /*
 182  * make_var
 183  *              Build a Var node for an attribute identified by RTE and attrno
 184  */
 185 Var *
 186 make_var(ParseState *pstate, RangeTblEntry *rte, int attrno, int location)
 187 {
 188         Var                *result;
 189         int                     vnum,
 190                                 sublevels_up;
 191         Oid                     vartypeid;
 192         int32           type_mod;
 193         Oid                     varcollid;
 194
 195         vnum = RTERangeTablePosn(pstate, rte, &sublevels_up);
 196         get_rte_attribute_type(rte, attrno, &vartypeid, &type_mod, &varcollid);
 197         result = makeVar(vnum, attrno, vartypeid, type_mod, varcollid, sublevels_up);
 198         result->location = location;
 199         return result;
 200 }
 201
 202 /*
 203  * transformArrayType()
 204  *              Identify the types involved in a subscripting operation
 205  *
 206  * On entry, arrayType/arrayTypmod identify the type of the input value
 207  * to be subscripted (which could be a domain type).  These are modified
 208  * if necessary to identify the actual array type and typmod, and the
 209  * array's element type is returned.  An error is thrown if the input isn't
 210  * an array type.
 211  */
 212 Oid
 213 transformArrayType(Oid *arrayType, int32 *arrayTypmod)
 214 {
 215         Oid                     origArrayType = *arrayType;
 216         Oid                     elementType;
 217         HeapTuple       type_tuple_array;
 218         Form_pg_type type_struct_array;
 219
 220         /*
 221          * If the input is a domain, smash to base type, and extract the actual
 222          * typmod to be applied to the base type.  Subscripting a domain is an
 223          * operation that necessarily works on the base array type, not the domain
 224          * itself.  (Note that we provide no method whereby the creator of a
 225          * domain over an array type could hide its ability to be subscripted.)
 226          */
 227         *arrayType = getBaseTypeAndTypmod(*arrayType, arrayTypmod);
 228
 229         /*
 230          * We treat int2vector and oidvector as though they were domains over
 231          * int2[] and oid[].  This is needed because array slicing could create an
 232          * array that doesn't satisfy the dimensionality constraints of the
 233          * xxxvector type; so we want the result of a slice operation to be
 234          * considered to be of the more general type.
 235          */
 236         if (*arrayType == INT2VECTOROID)
 237                 *arrayType = INT2ARRAYOID;
 238         else if (*arrayType == OIDVECTOROID)
 239                 *arrayType = OIDARRAYOID;
 240
 241         /* Get the type tuple for the array */
 242         type_tuple_array = SearchSysCache1(TYPEOID, ObjectIdGetDatum(*arrayType));
 243         if (!HeapTupleIsValid(type_tuple_array))
 244                 elog(ERROR, "cache lookup failed for type %u", *arrayType);
 245         type_struct_array = (Form_pg_type) GETSTRUCT(type_tuple_array);
 246
 247         /* needn't check typisdefined since this will fail anyway */
 248
 249         elementType = type_struct_array->typelem;
 250         if (elementType == InvalidOid)
 251                 ereport(ERROR,
 252                                 (errcode(ERRCODE_DATATYPE_MISMATCH),
 253                                  errmsg("cannot subscript type %s because it is not an array",
 254                                                 format_type_be(origArrayType))));
 255
 256         ReleaseSysCache(type_tuple_array);
 257
 258         return elementType;
 259 }
 260
 261 /*
 262  * transformArraySubscripts()
 263  *              Transform array subscripting.  This is used for both
 264  *              array fetch and array assignment.
 265  *
 266  * In an array fetch, we are given a source array value and we produce an
 267  * expression that represents the result of extracting a single array element
 268  * or an array slice.
 269  *
 270  * In an array assignment, we are given a destination array value plus a
 271  * source value that is to be assigned to a single element or a slice of
 272  * that array.  We produce an expression that represents the new array value
 273  * with the source data inserted into the right part of the array.
 274  *
 275  * For both cases, if the source array is of a domain-over-array type,
 276  * the result is of the base array type or its element type; essentially,
 277  * we must fold a domain to its base type before applying subscripting.
 278  * (Note that int2vector and oidvector are treated as domains here.)
 279  *
 280  * pstate               Parse state
 281  * arrayBase    Already-transformed expression for the array as a whole
 282  * arrayType    OID of array's datatype (should match type of arrayBase,
 283  *                              or be the base type of arrayBase's domain type)
 284  * elementType  OID of array's element type (fetch with transformArrayType,
 285  *                              or pass InvalidOid to do it here)
 286  * arrayTypMod  typmod for the array (which is also typmod for the elements)
 287  * indirection  Untransformed list of subscripts (must not be NIL)
 288  * assignFrom   NULL for array fetch, else transformed expression for source.
 289  */
 290 ArrayRef *
 291 transformArraySubscripts(ParseState *pstate,
 292                                                  Node *arrayBase,
 293                                                  Oid arrayType,
 294                                                  Oid elementType,
 295                                                  int32 arrayTypMod,
 296                                                  List *indirection,
 297                                                  Node *assignFrom)
 298 {
 299         bool            isSlice = false;
 300         List       *upperIndexpr = NIL;
 301         List       *lowerIndexpr = NIL;
 302         ListCell   *idx;
 303         ArrayRef   *aref;
 304
 305         /*
 306          * Caller may or may not have bothered to determine elementType.  Note
 307          * that if the caller did do so, arrayType/arrayTypMod must be as modified
 308          * by transformArrayType, ie, smash domain to base type.
 309          */
 310         if (!OidIsValid(elementType))
 311                 elementType = transformArrayType(&arrayType, &arrayTypMod);
 312
 313         /*
 314          * A list containing only simple subscripts refers to a single array
 315          * element.  If any of the items are slice specifiers (lower:upper), then
 316          * the subscript expression means an array slice operation.  In this case,
 317          * we convert any non-slice items to slices by treating the single
 318          * subscript as the upper bound and supplying an assumed lower bound of 1.
 319          * We have to prescan the list to see if there are any slice items.
 320          */
 321         foreach(idx, indirection)
 322         {
 323                 A_Indices  *ai = (A_Indices *) lfirst(idx);
 324
 325                 if (ai->is_slice)
 326                 {
 327                         isSlice = true;
 328                         break;
 329                 }
 330         }
 331
 332         /*
 333          * Transform the subscript expressions.
 334          */
 335         foreach(idx, indirection)
 336         {
 337                 A_Indices  *ai = (A_Indices *) lfirst(idx);
 338                 Node       *subexpr;
 339
 340                 Assert(IsA(ai, A_Indices));
 341                 if (isSlice)
 342                 {
 343                         if (ai->lidx)
 344                         {
 345                                 subexpr = transformExpr(pstate, ai->lidx, pstate->p_expr_kind);
 346                                 /* If it's not int4 already, try to coerce */
 347                                 subexpr = coerce_to_target_type(pstate,
 348                                                                                                 subexpr, exprType(subexpr),
 349                                                                                                 INT4OID, -1,
 350                                                                                                 COERCION_ASSIGNMENT,
 351                                                                                                 COERCE_IMPLICIT_CAST,
 352                                                                                                 -1);
 353                                 if (subexpr == NULL)
 354                                         ereport(ERROR,
 355                                                         (errcode(ERRCODE_DATATYPE_MISMATCH),
 356                                                          errmsg("array subscript must have type integer"),
 357                                                 parser_errposition(pstate, exprLocation(ai->lidx))));
 358                         }
 359                         else if (!ai->is_slice)
 360                         {
 361                                 /* Make a constant 1 */
 362                                 subexpr = (Node *) makeConst(INT4OID,
 363                                                                                          -1,
 364                                                                                          InvalidOid,
 365                                                                                          sizeof(int32),
 366                                                                                          Int32GetDatum(1),
 367                                                                                          false,
 368                                                                                          true);         /* pass by value */
 369                         }
 370                         else
 371                         {
 372                                 /* Slice with omitted lower bound, put NULL into the list */
 373                                 subexpr = NULL;
 374                         }
 375                         lowerIndexpr = lappend(lowerIndexpr, subexpr);
 376                 }
 377                 else
 378                         Assert(ai->lidx == NULL && !ai->is_slice);
 379
 380                 if (ai->uidx)
 381                 {
 382                         subexpr = transformExpr(pstate, ai->uidx, pstate->p_expr_kind);
 383                         /* If it's not int4 already, try to coerce */
 384                         subexpr = coerce_to_target_type(pstate,
 385                                                                                         subexpr, exprType(subexpr),
 386                                                                                         INT4OID, -1,
 387                                                                                         COERCION_ASSIGNMENT,
 388                                                                                         COERCE_IMPLICIT_CAST,
 389                                                                                         -1);
 390                         if (subexpr == NULL)
 391                                 ereport(ERROR,
 392                                                 (errcode(ERRCODE_DATATYPE_MISMATCH),
 393                                                  errmsg("array subscript must have type integer"),
 394                                                  parser_errposition(pstate, exprLocation(ai->uidx))));
 395                 }
 396                 else
 397                 {
 398                         /* Slice with omitted upper bound, put NULL into the list */
 399                         Assert(isSlice && ai->is_slice);
 400                         subexpr = NULL;
 401                 }
 402                 upperIndexpr = lappend(upperIndexpr, subexpr);
 403         }
 404
 405         /*
 406          * If doing an array store, coerce the source value to the right type.
 407          * (This should agree with the coercion done by transformAssignedExpr.)
 408          */
 409         if (assignFrom != NULL)
 410         {
 411                 Oid                     typesource = exprType(assignFrom);
 412                 Oid                     typeneeded = isSlice ? arrayType : elementType;
 413                 Node       *newFrom;
 414
 415                 newFrom = coerce_to_target_type(pstate,
 416                                                                                 assignFrom, typesource,
 417                                                                                 typeneeded, arrayTypMod,
 418                                                                                 COERCION_ASSIGNMENT,
 419                                                                                 COERCE_IMPLICIT_CAST,
 420                                                                                 -1);
 421                 if (newFrom == NULL)
 422                         ereport(ERROR,
 423                                         (errcode(ERRCODE_DATATYPE_MISMATCH),
 424                                          errmsg("array assignment requires type %s"
 425                                                         " but expression is of type %s",
 426                                                         format_type_be(typeneeded),
 427                                                         format_type_be(typesource)),
 428                                  errhint("You will need to rewrite or cast the expression."),
 429                                          parser_errposition(pstate, exprLocation(assignFrom))));
 430                 assignFrom = newFrom;
 431         }
 432
 433         /*
 434          * Ready to build the ArrayRef node.
 435          */
 436         aref = makeNode(ArrayRef);
 437         aref->refarraytype = arrayType;
 438         aref->refelemtype = elementType;
 439         aref->reftypmod = arrayTypMod;
 440         /* refcollid will be set by parse_collate.c */
 441         aref->refupperindexpr = upperIndexpr;
 442         aref->reflowerindexpr = lowerIndexpr;
 443         aref->refexpr = (Expr *) arrayBase;
 444         aref->refassgnexpr = (Expr *) assignFrom;
 445
 446         return aref;
 447 }
 448
 449 /*
 450  * make_const
 451  *
 452  *      Convert a Value node (as returned by the grammar) to a Const node
 453  *      of the "natural" type for the constant.  Note that this routine is
 454  *      only used when there is no explicit cast for the constant, so we
 455  *      have to guess what type is wanted.
 456  *
 457  *      For string literals we produce a constant of type UNKNOWN ---- whose
 458  *      representation is the same as cstring, but it indicates to later type
 459  *      resolution that we're not sure yet what type it should be considered.
 460  *      Explicit "NULL" constants are also typed as UNKNOWN.
 461  *
 462  *      For integers and floats we produce int4, int8, or numeric depending
 463  *      on the value of the number.  XXX We should produce int2 as well,
 464  *      but additional cleanup is needed before we can do that; there are
 465  *      too many examples that fail if we try.
 466  */
 467 Const *
 468 make_const(ParseState *pstate, Value *value, int location)
 469 {
 470         Const      *con;
 471         Datum           val;
 472         int64           val64;
 473         Oid                     typeid;
 474         int                     typelen;
 475         bool            typebyval;
 476         ParseCallbackState pcbstate;
 477
 478         switch (nodeTag(value))
 479         {
 480                 case T_Integer:
 481                         val = Int32GetDatum(intVal(value));
 482
 483                         typeid = INT4OID;
 484                         typelen = sizeof(int32);
 485                         typebyval = true;
 486                         break;
 487
 488                 case T_Float:
 489                         /* could be an oversize integer as well as a float ... */
 490                         if (scanint8(strVal(value), true, &val64))
 491                         {
 492                                 /*
 493                                  * It might actually fit in int32. Probably only INT_MIN can
 494                                  * occur, but we'll code the test generally just to be sure.
 495                                  */
 496                                 int32           val32 = (int32) val64;
 497
 498                                 if (val64 == (int64) val32)
 499                                 {
 500                                         val = Int32GetDatum(val32);
 501
 502                                         typeid = INT4OID;
 503                                         typelen = sizeof(int32);
 504                                         typebyval = true;
 505                                 }
 506                                 else
 507                                 {
 508                                         val = Int64GetDatum(val64);
 509
 510                                         typeid = INT8OID;
 511                                         typelen = sizeof(int64);
 512                                         typebyval = FLOAT8PASSBYVAL;            /* int8 and float8 alike */
 513                                 }
 514                         }
 515                         else
 516                         {
 517                                 /* arrange to report location if numeric_in() fails */
 518                                 setup_parser_errposition_callback(&pcbstate, pstate, location);
 519                                 val = DirectFunctionCall3(numeric_in,
 520                                                                                   CStringGetDatum(strVal(value)),
 521                                                                                   ObjectIdGetDatum(InvalidOid),
 522                                                                                   Int32GetDatum(-1));
 523                                 cancel_parser_errposition_callback(&pcbstate);
 524
 525                                 typeid = NUMERICOID;
 526                                 typelen = -1;   /* variable len */
 527                                 typebyval = false;
 528                         }
 529                         break;
 530
 531                 case T_String:
 532
 533                         /*
 534                          * We assume here that UNKNOWN's internal representation is the
 535                          * same as CSTRING
 536                          */
 537                         val = CStringGetDatum(strVal(value));
 538
 539                         typeid = UNKNOWNOID;    /* will be coerced later */
 540                         typelen = -2;           /* cstring-style varwidth type */
 541                         typebyval = false;
 542                         break;
 543
 544                 case T_BitString:
 545                         /* arrange to report location if bit_in() fails */
 546                         setup_parser_errposition_callback(&pcbstate, pstate, location);
 547                         val = DirectFunctionCall3(bit_in,
 548                                                                           CStringGetDatum(strVal(value)),
 549                                                                           ObjectIdGetDatum(InvalidOid),
 550                                                                           Int32GetDatum(-1));
 551                         cancel_parser_errposition_callback(&pcbstate);
 552                         typeid = BITOID;
 553                         typelen = -1;
 554                         typebyval = false;
 555                         break;
 556
 557                 case T_Null:
 558                         /* return a null const */
 559                         con = makeConst(UNKNOWNOID,
 560                                                         -1,
 561                                                         InvalidOid,
 562                                                         -2,
 563                                                         (Datum) 0,
 564                                                         true,
 565                                                         false);
 566                         con->location = location;
 567                         return con;
 568
 569                 default:
 570                         elog(ERROR, "unrecognized node type: %d", (int) nodeTag(value));
 571                         return NULL;            /* keep compiler quiet */
 572         }
 573
 574         con = makeConst(typeid,
 575                                         -1,                     /* typmod -1 is OK for all cases */
 576                                         InvalidOid, /* all cases are uncollatable types */
 577                                         typelen,
 578                                         val,
 579                                         false,
 580                                         typebyval);
 581         con->location = location;
 582
 583         return con;
 584 }