1/*-------------------------------------------------------------------------
4 * Declarations for Postgres arrays.
6 * A standard varlena array has the following internal structure:
7 * <vl_len_> - standard varlena header word
8 * <ndim> - number of dimensions of the array
9 * <dataoffset> - offset to stored data, or 0 if no nulls bitmap
10 * <elemtype> - element type OID
11 * <dimensions> - length of each array axis (C array of int)
12 * <lower bnds> - lower boundary of each dimension (C array of int)
13 * <null bitmap> - bitmap showing locations of nulls (OPTIONAL)
14 * <actual data> - whatever is the stored data
16 * The <dimensions> and <lower bnds> arrays each have ndim elements.
18 * The <null bitmap> may be omitted if the array contains no NULL elements.
19 * If it is absent, the <dataoffset> field is zero and the offset to the
20 * stored data must be computed on-the-fly. If the bitmap is present,
21 * <dataoffset> is nonzero and is equal to the offset from the array start
22 * to the first data element (including any alignment padding). The bitmap
23 * follows the same conventions as tuple null bitmaps, ie, a 1 indicates
24 * a non-null entry and the LSB of each bitmap byte is used first.
26 * The actual data starts on a MAXALIGN boundary. Individual items in the
27 * array are aligned as specified by the array element type. They are
28 * stored in row-major order (last subscript varies most rapidly).
30 * NOTE: it is important that array elements of toastable datatypes NOT be
31 * toasted, since the tupletoaster won't know they are there. (We could
32 * support compressed toasted items; only out-of-line items are dangerous.
33 * However, it seems preferable to store such items uncompressed and allow
34 * the toaster to compress the whole array as one input.)
37 * The OIDVECTOR and INT2VECTOR datatypes are storage-compatible with
38 * generic arrays, but they support only one-dimensional arrays with no
39 * nulls (and no null bitmap). They don't support being toasted, either.
41 * There are also some "fixed-length array" datatypes, such as NAME and
42 * POINT. These are simply a sequence of a fixed number of items each
43 * of a fixed-length datatype, with no overhead; the item size must be
44 * a multiple of its alignment requirement, because we do no padding.
45 * We support subscripting on these types, but array_in() and array_out()
46 * only work with varlena arrays.
48 * In addition, arrays are a major user of the "expanded object" TOAST
49 * infrastructure. This allows a varlena array to be converted to a
50 * separate representation that may include "deconstructed" Datum/isnull
51 * arrays holding the elements.
54 * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group
55 * Portions Copyright (c) 1994, Regents of the University of California
57 * src/include/utils/array.h
59 *-------------------------------------------------------------------------
67/* avoid including execnodes.h here */
73 * Maximum number of array subscripts (arbitrary limit)
78 * Maximum number of elements in an array. We limit this to at most about a
79 * quarter billion elements, so that it's not necessary to check for overflow
80 * in quite so many places --- for instance when palloc'ing Datum arrays.
82 #define MaxArraySize ((Size) (MaxAllocSize / sizeof(Datum)))
85 * Arrays are varlena objects, so must meet the varlena convention that
86 * the first int32 of the object contains the total object size in bytes.
87 * Be sure to use VARSIZE() and SET_VARSIZE() to access it, though!
89 * CAUTION: if you change the header for ordinary arrays you will also
90 * need to change the headers for oidvector and int2vector!
95 int ndim;
/* # of dimensions */
101 * An expanded array is contained within a private memory context (as
102 * all expanded objects must be) and has a control structure as below.
104 * The expanded array might contain a regular "flat" array if that was the
105 * original input and we've not modified it significantly. Otherwise, the
106 * contents are represented by Datum/isnull arrays plus dimensionality and
107 * type information. We could also have both forms, if we've deconstructed
108 * the original array for access purposes but not yet changed it. For pass-
109 * by-reference element types, the Datums would point into the flat array in
110 * this situation. Once we start modifying array elements, new pass-by-ref
111 * elements are separately palloc'd within the memory context.
113 #define EA_MAGIC 689375833 /* ID for debugging crosschecks */
117 /* Standard header for expanded objects */
120 /* Magic value identifying an expanded array (for debugging only) */
123 /* Dimensionality info (always valid) */
125 int *
dims;
/* array dimensions */
126 int *
lbound;
/* index lower bounds for each dimension */
128 /* Element type info (always valid) */
135 * If we have a Datum-array representation of the array, it's kept here;
136 * else dvalues/dnulls are NULL. The dvalues and dnulls arrays are always
137 * palloc'd within the object private context, but may change size from
138 * time to time. For pass-by-ref element types, dvalues entries might
139 * point either into the fstartptr..fendptr area, or to separately
140 * palloc'd chunks. Elements should always be fully detoasted, as they
141 * are in the standard flat representation.
143 * Even when dvalues is valid, dnulls can be NULL if there are no null
147 bool *
dnulls;
/* array of is-null flags for Datums */
149 int nelems;
/* number of valid entries in above arrays */
152 * flat_size is the current space requirement for the flat equivalent of
153 * the expanded array, if known; otherwise it's 0. We store this to make
154 * consecutive calls of get_flat_size cheap.
159 * fvalue points to the flat representation if it is valid, else it is
160 * NULL. If we have or ever had a flat representation then
161 * fstartptr/fendptr point to the start and end+1 of its data area; this
162 * is so that we can tell which Datum pointers point into the flat
163 * representation rather than being pointers to separately palloc'd data.
171 * Functions that can handle either a "flat" varlena array or an expanded
172 * array use this union to work with their input. Don't refer to "flt";
173 * instead, cast to ArrayType. This struct nominally requires 8-byte
174 * alignment on 64-bit, but it's often used for an ArrayType having 4-byte
175 * alignment. UBSan complains about referencing "flt" in such cases.
184 * working state for accumArrayResult() and friends
185 * note that the input must be scalars (legal array elements)
191 bool *
dnulls;
/* array of is-null flags for Datums */
192 int alen;
/* allocated length of above arrays */
193 int nelems;
/* number of valid entries in above arrays */
202 * working state for accumArrayResultArr() and friends
203 * note that the input must be arrays, and the same array type is returned
208 char *
data;
/* accumulated data */
210 int abytes;
/* allocated length of "data" */
211 int nbytes;
/* number of bytes used so far */
212 int aitems;
/* allocated length of bitmap (in elements) */
213 int nitems;
/* total number of elements in result */
214 int ndims;
/* current dimensions of result */
223 * working state for accumArrayResultAny() and friends
224 * these functions handle both cases
228 /* Exactly one of these is not NULL: */
234 * structure to cache type metadata needed for array manipulation
249 * private state needed by array_map (here because caller must provide it)
257/* ArrayIteratorData is private in arrayfuncs.c */
260/* fmgr macros for regular varlena array objects */
261 #define DatumGetArrayTypeP(X) ((ArrayType *) PG_DETOAST_DATUM(X))
262 #define DatumGetArrayTypePCopy(X) ((ArrayType *) PG_DETOAST_DATUM_COPY(X))
263 #define PG_GETARG_ARRAYTYPE_P(n) DatumGetArrayTypeP(PG_GETARG_DATUM(n))
264 #define PG_GETARG_ARRAYTYPE_P_COPY(n) DatumGetArrayTypePCopy(PG_GETARG_DATUM(n))
265 #define PG_RETURN_ARRAYTYPE_P(x) PG_RETURN_POINTER(x)
267/* fmgr macros for expanded array objects */
268 #define PG_GETARG_EXPANDED_ARRAY(n) DatumGetExpandedArray(PG_GETARG_DATUM(n))
269 #define PG_GETARG_EXPANDED_ARRAYX(n, metacache) \
270 DatumGetExpandedArrayX(PG_GETARG_DATUM(n), metacache)
271 #define PG_RETURN_EXPANDED_ARRAY(x) PG_RETURN_DATUM(EOHPGetRWDatum(&(x)->hdr))
273/* fmgr macros for AnyArrayType (ie, get either varlena or expanded form) */
274 #define PG_GETARG_ANY_ARRAY_P(n) DatumGetAnyArrayP(PG_GETARG_DATUM(n))
277 * Access macros for varlena array header fields.
279 * ARR_DIMS returns a pointer to an array of array dimensions (number of
280 * elements along the various array axes).
282 * ARR_LBOUND returns a pointer to an array of array lower bounds.
284 * That is: if the third axis of an array has elements 5 through 8, then
285 * ARR_DIMS(a)[2] == 4 and ARR_LBOUND(a)[2] == 5.
287 * Unlike C, the default lower bound is 1.
289 #define ARR_SIZE(a) VARSIZE(a)
290 #define ARR_NDIM(a) ((a)->ndim)
291 #define ARR_HASNULL(a) ((a)->dataoffset != 0)
292 #define ARR_ELEMTYPE(a) ((a)->elemtype)
294 #define ARR_DIMS(a) \
295 ((int *) (((char *) (a)) + sizeof(ArrayType)))
296 #define ARR_LBOUND(a) \
297 ((int *) (((char *) (a)) + sizeof(ArrayType) + \
298 sizeof(int) * ARR_NDIM(a)))
300 #define ARR_NULLBITMAP(a) \
302 (bits8 *) (((char *) (a)) + sizeof(ArrayType) + \
303 2 * sizeof(int) * ARR_NDIM(a)) \
307 * The total array header size (in bytes) for an array with the specified
308 * number of dimensions and total number of items.
310 #define ARR_OVERHEAD_NONULLS(ndims) \
311 MAXALIGN(sizeof(ArrayType) + 2 * sizeof(int) * (ndims))
312 #define ARR_OVERHEAD_WITHNULLS(ndims, nitems) \
313 MAXALIGN(sizeof(ArrayType) + 2 * sizeof(int) * (ndims) + \
316 #define ARR_DATA_OFFSET(a) \
317 (ARR_HASNULL(a) ? (a)->dataoffset : ARR_OVERHEAD_NONULLS(ARR_NDIM(a)))
320 * Returns a pointer to the actual array data.
322 #define ARR_DATA_PTR(a) \
323 (((char *) (a)) + ARR_DATA_OFFSET(a))
326 * Macros for working with AnyArrayType inputs. Beware multiple references!
328 #define AARR_NDIM(a) \
329 (VARATT_IS_EXPANDED_HEADER(a) ? \
330 (a)->xpn.ndims : ARR_NDIM((ArrayType *) (a)))
331 #define AARR_HASNULL(a) \
332 (VARATT_IS_EXPANDED_HEADER(a) ? \
333 ((a)->xpn.dvalues != NULL ? (a)->xpn.dnulls != NULL : ARR_HASNULL((a)->xpn.fvalue)) : \
334 ARR_HASNULL((ArrayType *) (a)))
335 #define AARR_ELEMTYPE(a) \
336 (VARATT_IS_EXPANDED_HEADER(a) ? \
337 (a)->xpn.element_type : ARR_ELEMTYPE((ArrayType *) (a)))
338 #define AARR_DIMS(a) \
339 (VARATT_IS_EXPANDED_HEADER(a) ? \
340 (a)->xpn.dims : ARR_DIMS((ArrayType *) (a)))
341 #define AARR_LBOUND(a) \
342 (VARATT_IS_EXPANDED_HEADER(a) ? \
343 (a)->xpn.lbound : ARR_LBOUND((ArrayType *) (a)))
352 * prototypes for functions defined in arrayfuncs.c
364 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign,
367 Datum dataValue,
bool isNull,
368 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign);
370 int *upperIndx,
int *lowerIndx,
371 bool *upperProvided,
bool *lowerProvided,
372 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign);
374 int *upperIndx,
int *lowerIndx,
375 bool *upperProvided,
bool *lowerProvided,
376 Datum srcArrayDatum,
bool isNull,
377 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign);
380 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign,
383 Datum dataValue,
bool isNull,
384 int arraytyplen,
int elmlen,
bool elmbyval,
char elmalign);
391 const bits8 *srcbitmap,
int srcoffset,
396 int elmlen,
bool elmbyval,
char elmalign);
403 Oid elmtype,
int elmlen,
bool elmbyval,
char elmalign);
410 int elmlen,
bool elmbyval,
char elmalign,
411 Datum **elemsp,
bool **nullsp,
int *nelemsp);
414 Datum **elemsp,
bool **nullsp,
int *nelemsp);
421 bool subcontext,
int initsize);
423 Datum dvalue,
bool disnull,
434 Datum dvalue,
bool disnull,
443 Datum dvalue,
bool disnull,
454 * prototypes for functions defined in arrayutils.c
457extern int ArrayGetOffset(
int n,
const int *dim,
const int *lb,
const int *indx);
460 struct Node *escontext);
463 struct Node *escontext);
464extern void mda_get_range(
int n,
int *span,
const int *st,
const int *endp);
471 * prototypes for functions defined in array_expanded.c
struct ArrayMapState ArrayMapState
void mda_get_offset_values(int n, int *dist, const int *prod, const int *span)
int ArrayGetNItemsSafe(int ndim, const int *dims, struct Node *escontext)
ArrayType * array_set(ArrayType *array, int nSubscripts, int *indx, Datum dataValue, bool isNull, int arraytyplen, int elmlen, bool elmbyval, char elmalign)
bool array_contains_nulls(ArrayType *array)
ArrayBuildState * accumArrayResult(ArrayBuildState *astate, Datum dvalue, bool disnull, Oid element_type, MemoryContext rcontext)
struct ExpandedArrayHeader ExpandedArrayHeader
ExpandedArrayHeader * construct_empty_expanded_array(Oid element_type, MemoryContext parentcontext, ArrayMetaState *metacache)
bool array_iterate(ArrayIterator iterator, Datum *value, bool *isnull)
bool ArrayCheckBoundsSafe(int ndim, const int *dims, const int *lb, struct Node *escontext)
void array_free_iterator(ArrayIterator iterator)
int ArrayGetOffset(int n, const int *dim, const int *lb, const int *indx)
ArrayBuildStateAny * initArrayResultAny(Oid input_type, MemoryContext rcontext, bool subcontext)
ArrayBuildStateAny * accumArrayResultAny(ArrayBuildStateAny *astate, Datum dvalue, bool disnull, Oid input_type, MemoryContext rcontext)
ArrayType * construct_empty_array(Oid elmtype)
void CopyArrayEls(ArrayType *array, Datum *values, bool *nulls, int nitems, int typlen, bool typbyval, char typalign, bool freedata)
struct ArrayBuildStateAny ArrayBuildStateAny
ExpandedArrayHeader * DatumGetExpandedArrayX(Datum d, ArrayMetaState *metacache)
Datum expand_array(Datum arraydatum, MemoryContext parentcontext, ArrayMetaState *metacache)
Datum makeArrayResultArr(ArrayBuildStateArr *astate, MemoryContext rcontext, bool release)
void mda_get_range(int n, int *span, const int *st, const int *endp)
int ArrayGetNItems(int ndim, const int *dims)
Datum makeArrayResultAny(ArrayBuildStateAny *astate, MemoryContext rcontext, bool release)
ArrayType * construct_array(Datum *elems, int nelems, Oid elmtype, int elmlen, bool elmbyval, char elmalign)
Datum array_map(Datum arrayd, ExprState *exprstate, ExprContext *econtext, Oid retType, ArrayMapState *amstate)
Datum array_set_element(Datum arraydatum, int nSubscripts, int *indx, Datum dataValue, bool isNull, int arraytyplen, int elmlen, bool elmbyval, char elmalign)
Datum makeMdArrayResult(ArrayBuildState *astate, int ndims, int *dims, int *lbs, MemoryContext rcontext, bool release)
void mda_get_prod(int n, const int *range, int *prod)
PGDLLIMPORT bool Array_nulls
ArrayBuildStateArr * initArrayResultArr(Oid array_type, Oid element_type, MemoryContext rcontext, bool subcontext)
void ArrayCheckBounds(int ndim, const int *dims, const int *lb)
ArrayBuildStateArr * accumArrayResultArr(ArrayBuildStateArr *astate, Datum dvalue, bool disnull, Oid array_type, MemoryContext rcontext)
struct ArrayMetaState ArrayMetaState
ArrayIterator array_create_iterator(ArrayType *arr, int slice_ndim, ArrayMetaState *mstate)
int mda_next_tuple(int n, int *curr, const int *span)
struct ArrayIteratorData * ArrayIterator
Datum array_ref(ArrayType *array, int nSubscripts, int *indx, int arraytyplen, int elmlen, bool elmbyval, char elmalign, bool *isNull)
ArrayBuildState * initArrayResultWithSize(Oid element_type, MemoryContext rcontext, bool subcontext, int initsize)
struct ArrayBuildStateArr ArrayBuildStateArr
void deconstruct_array(ArrayType *array, Oid elmtype, int elmlen, bool elmbyval, char elmalign, Datum **elemsp, bool **nullsp, int *nelemsp)
void deconstruct_array_builtin(ArrayType *array, Oid elmtype, Datum **elemsp, bool **nullsp, int *nelemsp)
void deconstruct_expanded_array(ExpandedArrayHeader *eah)
ArrayType * construct_md_array(Datum *elems, bool *nulls, int ndims, int *dims, int *lbs, Oid elmtype, int elmlen, bool elmbyval, char elmalign)
ArrayBuildState * initArrayResult(Oid element_type, MemoryContext rcontext, bool subcontext)
ExpandedArrayHeader * DatumGetExpandedArray(Datum d)
ArrayType * construct_array_builtin(Datum *elems, int nelems, Oid elmtype)
Datum makeArrayResult(ArrayBuildState *astate, MemoryContext rcontext)
struct ArrayType ArrayType
AnyArrayType * DatumGetAnyArrayP(Datum d)
void array_bitmap_copy(bits8 *destbitmap, int destoffset, const bits8 *srcbitmap, int srcoffset, int nitems)
struct ArrayBuildState ArrayBuildState
Datum array_get_element(Datum arraydatum, int nSubscripts, int *indx, int arraytyplen, int elmlen, bool elmbyval, char elmalign, bool *isNull)
Datum array_get_slice(Datum arraydatum, int nSubscripts, int *upperIndx, int *lowerIndx, bool *upperProvided, bool *lowerProvided, int arraytyplen, int elmlen, bool elmbyval, char elmalign)
Datum array_set_slice(Datum arraydatum, int nSubscripts, int *upperIndx, int *lowerIndx, bool *upperProvided, bool *lowerProvided, Datum srcArrayDatum, bool isNull, int arraytyplen, int elmlen, bool elmbyval, char elmalign)
union AnyArrayType AnyArrayType
int32 * ArrayGetIntegerTypmods(ArrayType *arr, int *n)
static Datum values[MAXATTR]
static struct cvec * range(struct vars *v, chr a, chr b, int cases)
ArrayBuildStateArr * arraystate
ArrayBuildState * scalarstate