mirror of
				https://github.com/postgres/postgres.git
				synced 2025-11-03 09:13:20 +03:00 
			
		
		
		
	As benchmarks show, using libc's string-to-integer conversion is pretty slow. At least part of the reason for that is that strtol[l] have to be more generic than what largely is required inside pg. This patch considerably speeds up int2/int4 input (int8 already was already using hand-rolled code). Most of the existing pg_atoi callers have been converted. But as one requires pg_atoi's custom delimiter functionality, and as it seems likely that there's external pg_atoi users, it seems sensible to just keep pg_atoi around. Author: Andres Freund Reviewed-By: Robert Haas Discussion: https://postgr.es/m/20171208214437.qgn6zdltyq5hmjpk@alap3.anarazel.de
		
			
				
	
	
		
			235 lines
		
	
	
		
			5.6 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
			
		
		
	
	
			235 lines
		
	
	
		
			5.6 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
/*-------------------------------------------------------------------------
 | 
						|
 *
 | 
						|
 * arrayutils.c
 | 
						|
 *	  This file contains some support routines required for array functions.
 | 
						|
 *
 | 
						|
 * Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
 | 
						|
 * Portions Copyright (c) 1994, Regents of the University of California
 | 
						|
 *
 | 
						|
 *
 | 
						|
 * IDENTIFICATION
 | 
						|
 *	  src/backend/utils/adt/arrayutils.c
 | 
						|
 *
 | 
						|
 *-------------------------------------------------------------------------
 | 
						|
 */
 | 
						|
 | 
						|
#include "postgres.h"
 | 
						|
 | 
						|
#include "catalog/pg_type.h"
 | 
						|
#include "utils/array.h"
 | 
						|
#include "utils/builtins.h"
 | 
						|
#include "utils/memutils.h"
 | 
						|
 | 
						|
 | 
						|
/*
 | 
						|
 * Convert subscript list into linear element number (from 0)
 | 
						|
 *
 | 
						|
 * We assume caller has already range-checked the dimensions and subscripts,
 | 
						|
 * so no overflow is possible.
 | 
						|
 */
 | 
						|
int
 | 
						|
ArrayGetOffset(int n, const int *dim, const int *lb, const int *indx)
 | 
						|
{
 | 
						|
	int			i,
 | 
						|
				scale = 1,
 | 
						|
				offset = 0;
 | 
						|
 | 
						|
	for (i = n - 1; i >= 0; i--)
 | 
						|
	{
 | 
						|
		offset += (indx[i] - lb[i]) * scale;
 | 
						|
		scale *= dim[i];
 | 
						|
	}
 | 
						|
	return offset;
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Same, but subscripts are assumed 0-based, and use a scale array
 | 
						|
 * instead of raw dimension data (see mda_get_prod to create scale array)
 | 
						|
 */
 | 
						|
int
 | 
						|
ArrayGetOffset0(int n, const int *tup, const int *scale)
 | 
						|
{
 | 
						|
	int			i,
 | 
						|
				lin = 0;
 | 
						|
 | 
						|
	for (i = 0; i < n; i++)
 | 
						|
		lin += tup[i] * scale[i];
 | 
						|
	return lin;
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Convert array dimensions into number of elements
 | 
						|
 *
 | 
						|
 * This must do overflow checking, since it is used to validate that a user
 | 
						|
 * dimensionality request doesn't overflow what we can handle.
 | 
						|
 *
 | 
						|
 * We limit array sizes to at most about a quarter billion elements,
 | 
						|
 * so that it's not necessary to check for overflow in quite so many
 | 
						|
 * places --- for instance when palloc'ing Datum arrays.
 | 
						|
 *
 | 
						|
 * The multiplication overflow check only works on machines that have int64
 | 
						|
 * arithmetic, but that is nearly all platforms these days, and doing check
 | 
						|
 * divides for those that don't seems way too expensive.
 | 
						|
 */
 | 
						|
int
 | 
						|
ArrayGetNItems(int ndim, const int *dims)
 | 
						|
{
 | 
						|
	int32		ret;
 | 
						|
	int			i;
 | 
						|
 | 
						|
#define MaxArraySize ((Size) (MaxAllocSize / sizeof(Datum)))
 | 
						|
 | 
						|
	if (ndim <= 0)
 | 
						|
		return 0;
 | 
						|
	ret = 1;
 | 
						|
	for (i = 0; i < ndim; i++)
 | 
						|
	{
 | 
						|
		int64		prod;
 | 
						|
 | 
						|
		/* A negative dimension implies that UB-LB overflowed ... */
 | 
						|
		if (dims[i] < 0)
 | 
						|
			ereport(ERROR,
 | 
						|
					(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 | 
						|
					 errmsg("array size exceeds the maximum allowed (%d)",
 | 
						|
							(int) MaxArraySize)));
 | 
						|
 | 
						|
		prod = (int64) ret * (int64) dims[i];
 | 
						|
 | 
						|
		ret = (int32) prod;
 | 
						|
		if ((int64) ret != prod)
 | 
						|
			ereport(ERROR,
 | 
						|
					(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 | 
						|
					 errmsg("array size exceeds the maximum allowed (%d)",
 | 
						|
							(int) MaxArraySize)));
 | 
						|
	}
 | 
						|
	Assert(ret >= 0);
 | 
						|
	if ((Size) ret > MaxArraySize)
 | 
						|
		ereport(ERROR,
 | 
						|
				(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 | 
						|
				 errmsg("array size exceeds the maximum allowed (%d)",
 | 
						|
						(int) MaxArraySize)));
 | 
						|
	return (int) ret;
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Compute ranges (sub-array dimensions) for an array slice
 | 
						|
 *
 | 
						|
 * We assume caller has validated slice endpoints, so overflow is impossible
 | 
						|
 */
 | 
						|
void
 | 
						|
mda_get_range(int n, int *span, const int *st, const int *endp)
 | 
						|
{
 | 
						|
	int			i;
 | 
						|
 | 
						|
	for (i = 0; i < n; i++)
 | 
						|
		span[i] = endp[i] - st[i] + 1;
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Compute products of array dimensions, ie, scale factors for subscripts
 | 
						|
 *
 | 
						|
 * We assume caller has validated dimensions, so overflow is impossible
 | 
						|
 */
 | 
						|
void
 | 
						|
mda_get_prod(int n, const int *range, int *prod)
 | 
						|
{
 | 
						|
	int			i;
 | 
						|
 | 
						|
	prod[n - 1] = 1;
 | 
						|
	for (i = n - 2; i >= 0; i--)
 | 
						|
		prod[i] = prod[i + 1] * range[i + 1];
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * From products of whole-array dimensions and spans of a sub-array,
 | 
						|
 * compute offset distances needed to step through subarray within array
 | 
						|
 *
 | 
						|
 * We assume caller has validated dimensions, so overflow is impossible
 | 
						|
 */
 | 
						|
void
 | 
						|
mda_get_offset_values(int n, int *dist, const int *prod, const int *span)
 | 
						|
{
 | 
						|
	int			i,
 | 
						|
				j;
 | 
						|
 | 
						|
	dist[n - 1] = 0;
 | 
						|
	for (j = n - 2; j >= 0; j--)
 | 
						|
	{
 | 
						|
		dist[j] = prod[j] - 1;
 | 
						|
		for (i = j + 1; i < n; i++)
 | 
						|
			dist[j] -= (span[i] - 1) * prod[i];
 | 
						|
	}
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * Generates the tuple that is lexicographically one greater than the current
 | 
						|
 * n-tuple in "curr", with the restriction that the i-th element of "curr" is
 | 
						|
 * less than the i-th element of "span".
 | 
						|
 *
 | 
						|
 * Returns -1 if no next tuple exists, else the subscript position (0..n-1)
 | 
						|
 * corresponding to the dimension to advance along.
 | 
						|
 *
 | 
						|
 * We assume caller has validated dimensions, so overflow is impossible
 | 
						|
 */
 | 
						|
int
 | 
						|
mda_next_tuple(int n, int *curr, const int *span)
 | 
						|
{
 | 
						|
	int			i;
 | 
						|
 | 
						|
	if (n <= 0)
 | 
						|
		return -1;
 | 
						|
 | 
						|
	curr[n - 1] = (curr[n - 1] + 1) % span[n - 1];
 | 
						|
	for (i = n - 1; i && curr[i] == 0; i--)
 | 
						|
		curr[i - 1] = (curr[i - 1] + 1) % span[i - 1];
 | 
						|
 | 
						|
	if (i)
 | 
						|
		return i;
 | 
						|
	if (curr[0])
 | 
						|
		return 0;
 | 
						|
 | 
						|
	return -1;
 | 
						|
}
 | 
						|
 | 
						|
/*
 | 
						|
 * ArrayGetIntegerTypmods: verify that argument is a 1-D cstring array,
 | 
						|
 * and get the contents converted to integers.  Returns a palloc'd array
 | 
						|
 * and places the length at *n.
 | 
						|
 */
 | 
						|
int32 *
 | 
						|
ArrayGetIntegerTypmods(ArrayType *arr, int *n)
 | 
						|
{
 | 
						|
	int32	   *result;
 | 
						|
	Datum	   *elem_values;
 | 
						|
	int			i;
 | 
						|
 | 
						|
	if (ARR_ELEMTYPE(arr) != CSTRINGOID)
 | 
						|
		ereport(ERROR,
 | 
						|
				(errcode(ERRCODE_ARRAY_ELEMENT_ERROR),
 | 
						|
				 errmsg("typmod array must be type cstring[]")));
 | 
						|
 | 
						|
	if (ARR_NDIM(arr) != 1)
 | 
						|
		ereport(ERROR,
 | 
						|
				(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
 | 
						|
				 errmsg("typmod array must be one-dimensional")));
 | 
						|
 | 
						|
	if (array_contains_nulls(arr))
 | 
						|
		ereport(ERROR,
 | 
						|
				(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
 | 
						|
				 errmsg("typmod array must not contain nulls")));
 | 
						|
 | 
						|
	/* hardwired knowledge about cstring's representation details here */
 | 
						|
	deconstruct_array(arr, CSTRINGOID,
 | 
						|
					  -2, false, 'c',
 | 
						|
					  &elem_values, NULL, n);
 | 
						|
 | 
						|
	result = (int32 *) palloc(*n * sizeof(int32));
 | 
						|
 | 
						|
	for (i = 0; i < *n; i++)
 | 
						|
		result[i] = pg_strtoint32(DatumGetCString(elem_values[i]));
 | 
						|
 | 
						|
	pfree(elem_values);
 | 
						|
 | 
						|
	return result;
 | 
						|
}
 |