9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * CDDL HEADER START
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * The contents of this file are subject to the terms of the
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Common Development and Distribution License (the "License").
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * You may not use this file except in compliance with the License.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * or http://www.opensolaris.org/os/licensing.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * See the License for the specific language governing permissions
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * and limitations under the License.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * When distributing Covered Code, include this CDDL HEADER in each
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * If applicable, add the following below this CDDL HEADER, with the
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * fields enclosed by brackets "[]" replaced with your own identifying
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * information: Portions Copyright [yyyy] [name of copyright owner]
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * CDDL HEADER END
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Use is subject to license terms.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Unicode conversions (yet more)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <stdio.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <stdlib.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <string.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <errno.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <iconv.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <libintl.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <sys/u8_textprep.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include <netsmb/smb_lib.h>
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross#include "charsets.h"
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Number of unicode symbols in the string,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * not including the 2-byte null terminator.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * (multiply by two for storage size)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rosssize_t
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossunicode_strlen(const uint16_t *us)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross size_t len = 0;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross while (*us++)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross len++;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (len);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossstatic char *convert_ucs2xx_to_utf8(iconv_t, const uint16_t *);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Convert (native) Unicode string to UTF-8.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Returns allocated memory.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rosschar *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_unicode_to_utf8(uint16_t *us)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross static iconv_t cd1 = (iconv_t)-1;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Get conversion descriptor (to, from) */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd1 == (iconv_t)-1)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross cd1 = iconv_open("UTF-8", "UCS-2");
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (convert_ucs2xx_to_utf8(cd1, us));
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Convert little-endian Unicode string to UTF-8.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Returns allocated memory.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rosschar *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_leunicode_to_utf8(unsigned short *us)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross static iconv_t cd2 = (iconv_t)-1;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Get conversion descriptor (to, from) */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd2 == (iconv_t)-1)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross cd2 = iconv_open("UTF-8", "UCS-2LE");
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (convert_ucs2xx_to_utf8(cd2, us));
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossstatic char *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_ucs2xx_to_utf8(iconv_t cd, const uint16_t *us)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross char *obuf, *optr;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross const char *iptr;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross size_t ileft, obsize, oleft, ret;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd == (iconv_t)-1) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv_open(UTF-8/UCS-2)"), -1);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (NULL);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross iptr = (const char *)us;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross ileft = unicode_strlen(us);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross ileft *= 2; /* now bytes */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Worst-case output size is 2x input size. */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross oleft = ileft * 2;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross obsize = oleft + 2; /* room for null */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross obuf = malloc(obsize);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (!obuf)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (NULL);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross optr = obuf;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross ret = iconv(cd, &iptr, &ileft, &optr, &oleft);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *optr = '\0';
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (ret == (size_t)-1) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv(%s) failed"), errno, obuf);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (ileft) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv(%s) failed"), -1, obuf);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * XXX: What's better? return NULL?
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * The truncated string? << for now
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (obuf);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossstatic uint16_t *convert_utf8_to_ucs2xx(iconv_t, const char *);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Convert UTF-8 string to Unicode.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Returns allocated memory.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossuint16_t *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_utf8_to_unicode(const char *utf8_string)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross static iconv_t cd3 = (iconv_t)-1;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Get conversion descriptor (to, from) */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd3 == (iconv_t)-1)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross cd3 = iconv_open("UCS-2", "UTF-8");
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (convert_utf8_to_ucs2xx(cd3, utf8_string));
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross/*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Convert UTF-8 string to little-endian Unicode.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * Returns allocated memory.
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossuint16_t *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_utf8_to_leunicode(const char *utf8_string)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross static iconv_t cd4 = (iconv_t)-1;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Get conversion descriptor (to, from) */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd4 == (iconv_t)-1)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross cd4 = iconv_open("UCS-2LE", "UTF-8");
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (convert_utf8_to_ucs2xx(cd4, utf8_string));
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossstatic uint16_t *
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Rossconvert_utf8_to_ucs2xx(iconv_t cd, const char *utf8_string)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross{
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross uint16_t *obuf, *optr;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross const char *iptr;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross size_t ileft, obsize, oleft, ret;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (cd == (iconv_t)-1) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv_open(UCS-2/UTF-8)"), -1);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (NULL);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross iptr = utf8_string;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross ileft = strlen(iptr);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /* Worst-case output size is 2x input size. */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross oleft = ileft * 2;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross obsize = oleft + 2; /* room for null */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross obuf = malloc(obsize);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (!obuf)
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (NULL);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross optr = obuf;
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross ret = iconv(cd, &iptr, &ileft, (char **)&optr, &oleft);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross *optr = '\0';
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (ret == (size_t)-1) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv(%s) failed"), errno, utf8_string);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross if (ileft) {
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross smb_error(dgettext(TEXT_DOMAIN,
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross "iconv(%s) failed"), -1, utf8_string);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross /*
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * XXX: What's better? return NULL?
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross * The truncated string? << for now
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross */
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross }
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross return (obuf);
9c9af2590af49bb395bc8d2eace0f2d4ea16d165Gordon Ross}
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross/*
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * A simple wrapper around u8_textprep_str() that returns the Unicode
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * upper-case version of some string. Returns memory from malloc.
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * Borrowed from idmapd.
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross */
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rossstatic char *
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rossutf8_str_to_upper_or_lower(const char *s, int upper_lower)
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross{
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross char *res = NULL;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross char *outs;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross size_t inlen, outlen, inbleft, outbleft;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross int rc, err;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross /*
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * u8_textprep_str() does not allocate memory. The input and
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * output buffers may differ in size (though that would be more
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * likely when normalization is done). We have to loop over it...
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross *
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * To improve the chances that we can avoid looping we add 10
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross * bytes of output buffer room the first go around.
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross */
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross inlen = inbleft = strlen(s);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross outlen = outbleft = inlen + 10;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross if ((res = malloc(outlen)) == NULL)
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (NULL);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross outs = res;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross while ((rc = u8_textprep_str((char *)s, &inbleft, outs,
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross &outbleft, upper_lower, U8_UNICODE_LATEST, &err)) < 0 &&
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross err == E2BIG) {
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross if ((res = realloc(res, outlen + inbleft)) == NULL)
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (NULL);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross /* adjust input/output buffer pointers */
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross s += (inlen - inbleft);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross outs = res + outlen - outbleft;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross /* adjust outbleft and outlen */
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross outlen += inbleft;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross outbleft += inbleft;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross }
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross if (rc < 0) {
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross free(res);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross res = NULL;
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (NULL);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross }
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross res[outlen - outbleft] = '\0';
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (res);
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross}
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rosschar *
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rossutf8_str_toupper(const char *s)
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross{
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (utf8_str_to_upper_or_lower(s, U8_TEXTPREP_TOUPPER));
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross}
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rosschar *
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Rossutf8_str_tolower(const char *s)
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross{
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross return (utf8_str_to_upper_or_lower(s, U8_TEXTPREP_TOLOWER));
613a2f6ba31e891e3d947a356daf5e563d43c1ceGordon Ross}