1 /* $NetBSD: ucstr.c,v 1.2 2020/08/11 13:15:38 christos Exp $ */ 2 3 /* $OpenLDAP$ */ 4 /* This work is part of OpenLDAP Software <http://www.openldap.org/>. 5 * 6 * Copyright 1998-2020 The OpenLDAP Foundation. 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted only as authorized by the OpenLDAP 11 * Public License. 12 * 13 * A copy of this license is available in file LICENSE in the 14 * top-level directory of the distribution or, alternatively, at 15 * <http://www.OpenLDAP.org/license.html>. 16 */ 17 18 #include <sys/cdefs.h> 19 __RCSID("$NetBSD: ucstr.c,v 1.2 2020/08/11 13:15:38 christos Exp $"); 20 21 #include "portable.h" 22 23 #include <ac/bytes.h> 24 #include <ac/ctype.h> 25 #include <ac/string.h> 26 #include <ac/stdlib.h> 27 28 #include <lber_pvt.h> 29 30 #include <ldap_utf8.h> 31 #include <ldap_pvt_uc.h> 32 33 #define malloc(x) ber_memalloc_x(x,ctx) 34 #define realloc(x,y) ber_memrealloc_x(x,y,ctx) 35 #define free(x) ber_memfree_x(x,ctx) 36 37 int ucstrncmp( 38 const ldap_unicode_t *u1, 39 const ldap_unicode_t *u2, 40 ber_len_t n ) 41 { 42 for(; 0 < n; ++u1, ++u2, --n ) { 43 if( *u1 != *u2 ) { 44 return *u1 < *u2 ? -1 : +1; 45 } 46 if ( *u1 == 0 ) { 47 return 0; 48 } 49 } 50 return 0; 51 } 52 53 int ucstrncasecmp( 54 const ldap_unicode_t *u1, 55 const ldap_unicode_t *u2, 56 ber_len_t n ) 57 { 58 for(; 0 < n; ++u1, ++u2, --n ) { 59 ldap_unicode_t uu1 = uctolower( *u1 ); 60 ldap_unicode_t uu2 = uctolower( *u2 ); 61 62 if( uu1 != uu2 ) { 63 return uu1 < uu2 ? -1 : +1; 64 } 65 if ( uu1 == 0 ) { 66 return 0; 67 } 68 } 69 return 0; 70 } 71 72 ldap_unicode_t * ucstrnchr( 73 const ldap_unicode_t *u, 74 ber_len_t n, 75 ldap_unicode_t c ) 76 { 77 for(; 0 < n; ++u, --n ) { 78 if( *u == c ) { 79 return (ldap_unicode_t *) u; 80 } 81 } 82 83 return NULL; 84 } 85 86 ldap_unicode_t * ucstrncasechr( 87 const ldap_unicode_t *u, 88 ber_len_t n, 89 ldap_unicode_t c ) 90 { 91 c = uctolower( c ); 92 for(; 0 < n; ++u, --n ) { 93 if( uctolower( *u ) == c ) { 94 return (ldap_unicode_t *) u; 95 } 96 } 97 98 return NULL; 99 } 100 101 void ucstr2upper( 102 ldap_unicode_t *u, 103 ber_len_t n ) 104 { 105 for(; 0 < n; ++u, --n ) { 106 *u = uctoupper( *u ); 107 } 108 } 109 110 struct berval * UTF8bvnormalize( 111 struct berval *bv, 112 struct berval *newbv, 113 unsigned flags, 114 void *ctx ) 115 { 116 int i, j, len, clen, outpos, ucsoutlen, outsize, last; 117 int didnewbv = 0; 118 char *out, *outtmp, *s; 119 ac_uint4 *ucs, *p, *ucsout; 120 121 static unsigned char mask[] = { 122 0, 0x7f, 0x1f, 0x0f, 0x07, 0x03, 0x01 }; 123 124 unsigned casefold = flags & LDAP_UTF8_CASEFOLD; 125 unsigned approx = flags & LDAP_UTF8_APPROX; 126 127 if ( bv == NULL ) { 128 return NULL; 129 } 130 131 s = bv->bv_val; 132 len = bv->bv_len; 133 134 if ( len == 0 ) { 135 return ber_dupbv_x( newbv, bv, ctx ); 136 } 137 138 if ( !newbv ) { 139 newbv = ber_memalloc_x( sizeof(struct berval), ctx ); 140 if ( !newbv ) return NULL; 141 didnewbv = 1; 142 } 143 144 /* Should first check to see if string is already in proper 145 * normalized form. This is almost as time consuming as 146 * the normalization though. 147 */ 148 149 /* finish off everything up to character before first non-ascii */ 150 if ( LDAP_UTF8_ISASCII( s ) ) { 151 if ( casefold ) { 152 outsize = len + 7; 153 out = (char *) ber_memalloc_x( outsize, ctx ); 154 if ( out == NULL ) { 155 fail: 156 if ( didnewbv ) 157 ber_memfree_x( newbv, ctx ); 158 return NULL; 159 } 160 outpos = 0; 161 162 for ( i = 1; (i < len) && LDAP_UTF8_ISASCII(s + i); i++ ) { 163 out[outpos++] = TOLOWER( s[i-1] ); 164 } 165 if ( i == len ) { 166 out[outpos++] = TOLOWER( s[len-1] ); 167 out[outpos] = '\0'; 168 newbv->bv_val = out; 169 newbv->bv_len = outpos; 170 return newbv; 171 } 172 } else { 173 for ( i = 1; (i < len) && LDAP_UTF8_ISASCII(s + i); i++ ) { 174 /* empty */ 175 } 176 177 if ( i == len ) { 178 return ber_str2bv_x( s, len, 1, newbv, ctx ); 179 } 180 181 outsize = len + 7; 182 out = (char *) ber_memalloc_x( outsize, ctx ); 183 if ( out == NULL ) { 184 goto fail; 185 } 186 outpos = i - 1; 187 memcpy(out, s, outpos); 188 } 189 } else { 190 outsize = len + 7; 191 out = (char *) ber_memalloc_x( outsize, ctx ); 192 if ( out == NULL ) { 193 goto fail; 194 } 195 outpos = 0; 196 i = 0; 197 } 198 199 p = ucs = ber_memalloc_x( len * sizeof(*ucs), ctx ); 200 if ( ucs == NULL ) { 201 ber_memfree_x(out, ctx); 202 goto fail; 203 } 204 205 /* convert character before first non-ascii to ucs-4 */ 206 if ( i > 0 ) { 207 *p = casefold ? TOLOWER( s[i-1] ) : s[i-1]; 208 p++; 209 } 210 211 /* s[i] is now first non-ascii character */ 212 for (;;) { 213 /* s[i] is non-ascii */ 214 /* convert everything up to next ascii to ucs-4 */ 215 while ( i < len ) { 216 clen = LDAP_UTF8_CHARLEN2( s + i, clen ); 217 if ( clen == 0 ) { 218 ber_memfree_x( ucs, ctx ); 219 ber_memfree_x( out, ctx ); 220 goto fail; 221 } 222 if ( clen == 1 ) { 223 /* ascii */ 224 break; 225 } 226 *p = s[i] & mask[clen]; 227 i++; 228 for( j = 1; j < clen; j++ ) { 229 if ( (s[i] & 0xc0) != 0x80 ) { 230 ber_memfree_x( ucs, ctx ); 231 ber_memfree_x( out, ctx ); 232 goto fail; 233 } 234 *p <<= 6; 235 *p |= s[i] & 0x3f; 236 i++; 237 } 238 if ( casefold ) { 239 *p = uctolower( *p ); 240 } 241 p++; 242 } 243 /* normalize ucs of length p - ucs */ 244 uccompatdecomp( ucs, p - ucs, &ucsout, &ucsoutlen, ctx ); 245 if ( approx ) { 246 for ( j = 0; j < ucsoutlen; j++ ) { 247 if ( ucsout[j] < 0x80 ) { 248 out[outpos++] = ucsout[j]; 249 } 250 } 251 } else { 252 ucsoutlen = uccanoncomp( ucsout, ucsoutlen ); 253 /* convert ucs to utf-8 and store in out */ 254 for ( j = 0; j < ucsoutlen; j++ ) { 255 /* allocate more space if not enough room for 256 6 bytes and terminator */ 257 if ( outsize - outpos < 7 ) { 258 outsize = ucsoutlen - j + outpos + 6; 259 outtmp = (char *) ber_memrealloc_x( out, outsize, ctx ); 260 if ( outtmp == NULL ) { 261 ber_memfree_x( ucsout, ctx ); 262 ber_memfree_x( ucs, ctx ); 263 ber_memfree_x( out, ctx ); 264 goto fail; 265 } 266 out = outtmp; 267 } 268 outpos += ldap_x_ucs4_to_utf8( ucsout[j], &out[outpos] ); 269 } 270 } 271 272 ber_memfree_x( ucsout, ctx ); 273 ucsout = NULL; 274 275 if ( i == len ) { 276 break; 277 } 278 279 last = i; 280 281 /* Allocate more space in out if necessary */ 282 if (len - i >= outsize - outpos) { 283 outsize += 1 + ((len - i) - (outsize - outpos)); 284 outtmp = (char *) ber_memrealloc_x(out, outsize, ctx); 285 if (outtmp == NULL) { 286 ber_memfree_x( ucs, ctx ); 287 ber_memfree_x( out, ctx ); 288 goto fail; 289 } 290 out = outtmp; 291 } 292 293 /* s[i] is ascii */ 294 /* finish off everything up to char before next non-ascii */ 295 for ( i++; (i < len) && LDAP_UTF8_ISASCII(s + i); i++ ) { 296 out[outpos++] = casefold ? TOLOWER( s[i-1] ) : s[i-1]; 297 } 298 if ( i == len ) { 299 out[outpos++] = casefold ? TOLOWER( s[len-1] ) : s[len-1]; 300 break; 301 } 302 303 /* convert character before next non-ascii to ucs-4 */ 304 *ucs = casefold ? TOLOWER( s[i-1] ) : s[i-1]; 305 p = ucs + 1; 306 } 307 308 ber_memfree_x( ucs, ctx ); 309 out[outpos] = '\0'; 310 newbv->bv_val = out; 311 newbv->bv_len = outpos; 312 return newbv; 313 } 314 315 /* compare UTF8-strings, optionally ignore casing */ 316 /* slow, should be optimized */ 317 int UTF8bvnormcmp( 318 struct berval *bv1, 319 struct berval *bv2, 320 unsigned flags, 321 void *ctx ) 322 { 323 int i, l1, l2, len, ulen, res = 0; 324 char *s1, *s2, *done; 325 ac_uint4 *ucs, *ucsout1, *ucsout2; 326 327 unsigned casefold = flags & LDAP_UTF8_CASEFOLD; 328 unsigned norm1 = flags & LDAP_UTF8_ARG1NFC; 329 unsigned norm2 = flags & LDAP_UTF8_ARG2NFC; 330 331 if (bv1 == NULL) { 332 return bv2 == NULL ? 0 : -1; 333 334 } else if (bv2 == NULL) { 335 return 1; 336 } 337 338 l1 = bv1->bv_len; 339 l2 = bv2->bv_len; 340 341 len = (l1 < l2) ? l1 : l2; 342 if (len == 0) { 343 return l1 == 0 ? (l2 == 0 ? 0 : -1) : 1; 344 } 345 346 s1 = bv1->bv_val; 347 s2 = bv2->bv_val; 348 done = s1 + len; 349 350 while ( (s1 < done) && LDAP_UTF8_ISASCII(s1) && LDAP_UTF8_ISASCII(s2) ) { 351 if (casefold) { 352 char c1 = TOLOWER(*s1); 353 char c2 = TOLOWER(*s2); 354 res = c1 - c2; 355 } else { 356 res = *s1 - *s2; 357 } 358 s1++; 359 s2++; 360 if (res) { 361 /* done unless next character in s1 or s2 is non-ascii */ 362 if (s1 < done) { 363 if (!LDAP_UTF8_ISASCII(s1) || !LDAP_UTF8_ISASCII(s2)) { 364 break; 365 } 366 } else if (((len < l1) && !LDAP_UTF8_ISASCII(s1)) || 367 ((len < l2) && !LDAP_UTF8_ISASCII(s2))) 368 { 369 break; 370 } 371 return res; 372 } 373 } 374 375 /* We have encountered non-ascii or strings equal up to len */ 376 377 /* set i to number of iterations */ 378 i = s1 - done + len; 379 /* passed through loop at least once? */ 380 if (i > 0) { 381 if (!res && (s1 == done) && 382 ((len == l1) || LDAP_UTF8_ISASCII(s1)) && 383 ((len == l2) || LDAP_UTF8_ISASCII(s2))) { 384 /* all ascii and equal up to len */ 385 return l1 - l2; 386 } 387 388 /* rewind one char, and do normalized compare from there */ 389 s1--; 390 s2--; 391 l1 -= i - 1; 392 l2 -= i - 1; 393 } 394 395 /* Should first check to see if strings are already in 396 * proper normalized form. 397 */ 398 ucs = malloc( ( ( norm1 || l1 > l2 ) ? l1 : l2 ) * sizeof(*ucs) ); 399 if ( ucs == NULL ) { 400 return l1 > l2 ? 1 : -1; /* what to do??? */ 401 } 402 403 /* 404 * XXYYZ: we convert to ucs4 even though -llunicode 405 * expects ucs2 in an ac_uint4 406 */ 407 408 /* convert and normalize 1st string */ 409 for ( i = 0, ulen = 0; i < l1; i += len, ulen++ ) { 410 ucs[ulen] = ldap_x_utf8_to_ucs4( s1 + i ); 411 if ( ucs[ulen] == LDAP_UCS4_INVALID ) { 412 free( ucs ); 413 return -1; /* what to do??? */ 414 } 415 len = LDAP_UTF8_CHARLEN( s1 + i ); 416 } 417 418 if ( norm1 ) { 419 ucsout1 = ucs; 420 l1 = ulen; 421 ucs = malloc( l2 * sizeof(*ucs) ); 422 if ( ucs == NULL ) { 423 free( ucsout1 ); 424 return l1 > l2 ? 1 : -1; /* what to do??? */ 425 } 426 } else { 427 uccompatdecomp( ucs, ulen, &ucsout1, &l1, ctx ); 428 l1 = uccanoncomp( ucsout1, l1 ); 429 } 430 431 /* convert and normalize 2nd string */ 432 for ( i = 0, ulen = 0; i < l2; i += len, ulen++ ) { 433 ucs[ulen] = ldap_x_utf8_to_ucs4( s2 + i ); 434 if ( ucs[ulen] == LDAP_UCS4_INVALID ) { 435 free( ucsout1 ); 436 free( ucs ); 437 return 1; /* what to do??? */ 438 } 439 len = LDAP_UTF8_CHARLEN( s2 + i ); 440 } 441 442 if ( norm2 ) { 443 ucsout2 = ucs; 444 l2 = ulen; 445 } else { 446 uccompatdecomp( ucs, ulen, &ucsout2, &l2, ctx ); 447 l2 = uccanoncomp( ucsout2, l2 ); 448 free( ucs ); 449 } 450 451 res = casefold 452 ? ucstrncasecmp( ucsout1, ucsout2, l1 < l2 ? l1 : l2 ) 453 : ucstrncmp( ucsout1, ucsout2, l1 < l2 ? l1 : l2 ); 454 free( ucsout1 ); 455 free( ucsout2 ); 456 457 if ( res != 0 ) { 458 return res; 459 } 460 if ( l1 == l2 ) { 461 return 0; 462 } 463 return l1 > l2 ? 1 : -1; 464 } 465