oracle_compat.c

PostgreSQL7.4.6 for Linux

/*-------------------------------------------------------------------------
 * oracle_compat.c
 *	Oracle compatible functions.
 *
 * Copyright (c) 1996-2003, PostgreSQL Global Development Group
 *
 *	Author: Edmund Mergl <[e-mail]>
 *	Multibyte enhancement: Tatsuo Ishii <[e-mail]>
 *
 *
 * IDENTIFICATION
 *	$Header: /cvsroot/pgsql/src/backend/utils/adt/oracle_compat.c,v 1.48 2003/08/08 21:42:06 momjian Exp $
 *
 *-------------------------------------------------------------------------
 */
#include "postgres.h"

#include <ctype.h>

#include "utils/builtins.h"
#include "mb/pg_wchar.h"


static text *dotrim(const char *string, int stringlen,
	   const char *set, int setlen,
	   bool doltrim, bool dortrim);


/********************************************************************
 *
 * lower
 *
 * Syntax:
 *
 *	 text lower(text string)
 *
 * Purpose:
 *
 *	 Returns string, with all letters forced to lowercase.
 *
 ********************************************************************/

Datum
lower(PG_FUNCTION_ARGS)
{
	text	   *string = PG_GETARG_TEXT_P_COPY(0);
	char	   *ptr;
	int			m;

	/* Since we copied the string, we can scribble directly on the value */
	ptr = VARDATA(string);
	m = VARSIZE(string) - VARHDRSZ;

	while (m-- > 0)
	{
		*ptr = tolower((unsigned char) *ptr);
		ptr++;
	}

	PG_RETURN_TEXT_P(string);
}


/********************************************************************
 *
 * upper
 *
 * Syntax:
 *
 *	 text upper(text string)
 *
 * Purpose:
 *
 *	 Returns string, with all letters forced to uppercase.
 *
 ********************************************************************/

Datum
upper(PG_FUNCTION_ARGS)
{
	text	   *string = PG_GETARG_TEXT_P_COPY(0);
	char	   *ptr;
	int			m;

	/* Since we copied the string, we can scribble directly on the value */
	ptr = VARDATA(string);
	m = VARSIZE(string) - VARHDRSZ;

	while (m-- > 0)
	{
		*ptr = toupper((unsigned char) *ptr);
		ptr++;
	}

	PG_RETURN_TEXT_P(string);
}


/********************************************************************
 *
 * initcap
 *
 * Syntax:
 *
 *	 text initcap(text string)
 *
 * Purpose:
 *
 *	 Returns string, with first letter of each word in uppercase,
 *	 all other letters in lowercase. A word is delimited by white
 *	 space.
 *
 ********************************************************************/

Datum
initcap(PG_FUNCTION_ARGS)
{
	text	   *string = PG_GETARG_TEXT_P_COPY(0);
	char	   *ptr;
	int			m;

	/* Since we copied the string, we can scribble directly on the value */
	ptr = VARDATA(string);
	m = VARSIZE(string) - VARHDRSZ;

	if (m > 0)
	{
		*ptr = toupper((unsigned char) *ptr);
		ptr++;
		m--;
	}

	while (m-- > 0)
	{
		/* Oracle capitalizes after all non-alphanumeric */
		if (!isalnum((unsigned char) ptr[-1]))
			*ptr = toupper((unsigned char) *ptr);
		else
			*ptr = tolower((unsigned char) *ptr);
		ptr++;
	}

	PG_RETURN_TEXT_P(string);
}


/********************************************************************
 *
 * lpad
 *
 * Syntax:
 *
 *	 text lpad(text string1, int4 len, text string2)
 *
 * Purpose:
 *
 *	 Returns string1, left-padded to length len with the sequence of
 *	 characters in string2.  If len is less than the length of string1,
 *	 instead truncate (on the right) to len.
 *
 ********************************************************************/

Datum
lpad(PG_FUNCTION_ARGS)
{
	text	   *string1 = PG_GETARG_TEXT_P(0);
	int32		len = PG_GETARG_INT32(1);
	text	   *string2 = PG_GETARG_TEXT_P(2);
	text	   *ret;
	char	   *ptr1,
			   *ptr2,
			   *ptr2end,
			   *ptr_ret;
	int			m,
				s1len,
				s2len;

	int			bytelen;

	/* Negative len is silently taken as zero */
	if (len < 0)
		len = 0;

	s1len = VARSIZE(string1) - VARHDRSZ;
	if (s1len < 0)
		s1len = 0;				/* shouldn't happen */

	s2len = VARSIZE(string2) - VARHDRSZ;
	if (s2len < 0)
		s2len = 0;				/* shouldn't happen */

	s1len = pg_mbstrlen_with_len(VARDATA(string1), s1len);

	if (s1len > len)
		s1len = len;			/* truncate string1 to len chars */

	if (s2len <= 0)
		len = s1len;			/* nothing to pad with, so don't pad */

	bytelen = pg_database_encoding_max_length() * len;

	/* check for integer overflow */
	if (len != 0 && bytelen / pg_database_encoding_max_length() != len)
		ereport(ERROR,
				(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
				 errmsg("requested length too large")));

	ret = (text *) palloc(VARHDRSZ + bytelen);

	m = len - s1len;

	ptr2 = VARDATA(string2);
	ptr2end = ptr2 + s2len;
	ptr_ret = VARDATA(ret);

	while (m--)
	{
		int			mlen = pg_mblen(ptr2);

		memcpy(ptr_ret, ptr2, mlen);
		ptr_ret += mlen;
		ptr2 += mlen;
		if (ptr2 == ptr2end)	/* wrap around at end of s2 */
			ptr2 = VARDATA(string2);
	}

	ptr1 = VARDATA(string1);

	while (s1len--)
	{
		int			mlen = pg_mblen(ptr1);

		memcpy(ptr_ret, ptr1, mlen);
		ptr_ret += mlen;
		ptr1 += mlen;
	}

	VARATT_SIZEP(ret) = ptr_ret - (char *) ret;

	PG_RETURN_TEXT_P(ret);
}


/********************************************************************
 *
 * rpad
 *
 * Syntax:
 *
 *	 text rpad(text string1, int4 len, text string2)
 *
 * Purpose:
 *
 *	 Returns string1, right-padded to length len with the sequence of
 *	 characters in string2.  If len is less than the length of string1,
 *	 instead truncate (on the right) to len.
 *
 ********************************************************************/

Datum
rpad(PG_FUNCTION_ARGS)
{
	text	   *string1 = PG_GETARG_TEXT_P(0);
	int32		len = PG_GETARG_INT32(1);
	text	   *string2 = PG_GETARG_TEXT_P(2);
	text	   *ret;
	char	   *ptr1,
			   *ptr2,
			   *ptr2end,
			   *ptr_ret;
	int			m,
				s1len,
				s2len;

	int			bytelen;

	/* Negative len is silently taken as zero */
	if (len < 0)
		len = 0;

	s1len = VARSIZE(string1) - VARHDRSZ;
	if (s1len < 0)
		s1len = 0;				/* shouldn't happen */

	s2len = VARSIZE(string2) - VARHDRSZ;
	if (s2len < 0)
		s2len = 0;				/* shouldn't happen */

	s1len = pg_mbstrlen_with_len(VARDATA(string1), s1len);

	if (s1len > len)
		s1len = len;			/* truncate string1 to len chars */

	if (s2len <= 0)
		len = s1len;			/* nothing to pad with, so don't pad */

	bytelen = pg_database_encoding_max_length() * len;

	/* Check for integer overflow */
	if (len != 0 && bytelen / pg_database_encoding_max_length() != len)
		ereport(ERROR,
				(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
				 errmsg("requested length too large")));

	ret = (text *) palloc(VARHDRSZ + bytelen);
	m = len - s1len;

	ptr1 = VARDATA(string1);
	ptr_ret = VARDATA(ret);

	while (s1len--)
	{
		int			mlen = pg_mblen(ptr1);

		memcpy(ptr_ret, ptr1, mlen);
		ptr_ret += mlen;
		ptr1 += mlen;
	}

	ptr2 = VARDATA(string2);
	ptr2end = ptr2 + s2len;

	while (m--)
	{
		int			mlen = pg_mblen(ptr2);

		memcpy(ptr_ret, ptr2, mlen);
		ptr_ret += mlen;
		ptr2 += mlen;
		if (ptr2 == ptr2end)	/* wrap around at end of s2 */
			ptr2 = VARDATA(string2);
	}

	VARATT_SIZEP(ret) = ptr_ret - (char *) ret;

	PG_RETURN_TEXT_P(ret);
}


/********************************************************************
 *
 * btrim
 *
 * Syntax:
 *
 *	 text btrim(text string, text set)
 *
 * Purpose:
 *
 *	 Returns string with characters removed from the front and back
 *	 up to the first character not in set.
 *
 ********************************************************************/

Datum
btrim(PG_FUNCTION_ARGS)
{
	text	   *string = PG_GETARG_TEXT_P(0);
	text	   *set = PG_GETARG_TEXT_P(1);
	text	   *ret;

	ret = dotrim(VARDATA(string), VARSIZE(string) - VARHDRSZ,
				 VARDATA(set), VARSIZE(set) - VARHDRSZ,
				 true, true);

	PG_RETURN_TEXT_P(ret);
}

/********************************************************************
 *
 * btrim1 --- btrim with set fixed as ' '
 *
 ********************************************************************/

Datum
btrim1(PG_FUNCTION_ARGS)
{
	text	   *string = PG_GETARG_TEXT_P(0);
	text	   *ret;

	ret = dotrim(VARDATA(string), VARSIZE(string) - VARHDRSZ,
				 " ", 1,
				 true, true);

	PG_RETURN_TEXT_P(ret);
}

/*
 * Common implementation for btrim, ltrim, rtrim
 */
static text *
dotrim(const char *string, int stringlen,
	   const char *set, int setlen,
	   bool doltrim, bool dortrim)
{
	text	   *result;
	int			i;

	/* Nothing to do if either string or set is empty */
	if (stringlen > 0 && setlen > 0)
	{
		if (pg_database_encoding_max_length() > 1)
		{
			/*
			 * In the multibyte-encoding case, build arrays of pointers to
			 * character starts, so that we can avoid inefficient checks
			 * in the inner loops.
			 */
			const char **stringchars;
			const char **setchars;
			int		   *stringmblen;
			int		   *setmblen;
			int			stringnchars;
			int			setnchars;
			int			resultndx;
			int			resultnchars;
			const char *p;
			int			len;
			int			mblen;
			const char *str_pos;
			int			str_len;

			stringchars = (const char **) palloc(stringlen * sizeof(char *));
			stringmblen = (int *) palloc(stringlen * sizeof(int));
			stringnchars = 0;
			p = string;
			len = stringlen;
			while (len > 0)
			{
				stringchars[stringnchars] = p;
				stringmblen[stringnchars] = mblen = pg_mblen(p);
				stringnchars++;
				p += mblen;
				len -= mblen;
			}

			setchars = (const char **) palloc(setlen * sizeof(char *));
			setmblen = (int *) palloc(setlen * sizeof(int));
			setnchars = 0;
			p = set;
			len = setlen;
			while (len > 0)
			{
				setchars[setnchars] = p;
				setmblen[setnchars] = mblen = pg_mblen(p);
				setnchars++;
				p += mblen;
				len -= mblen;
			}

			resultndx = 0;		/* index in stringchars[] */
			resultnchars = stringnchars;

			if (doltrim)
			{
				while (resultnchars > 0)
				{
					str_pos = stringchars[resultndx];
					str_len = stringmblen[resultndx];
					for (i = 0; i < setnchars; i++)
					{
						if (str_len == setmblen[i] &&
							memcmp(str_pos, setchars[i], str_len) == 0)
							break;
					}
					if (i >= setnchars)
						break;	/* no match here */
					string += str_len;
					stringlen -= str_len;
					resultndx++;
					resultnchars--;