src/libcaptive/rtl/unicode.c

   1 /* $Id$
   2  * Unicode add-ons to reactos ntoskrnl/rtl/unicode.c for libcaptive
   3  * Copyright (C) 2002 Jan Kratochvil <project-captive@jankratochvil.net>
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; exactly version 2 of June 1991 is required
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, write to the Free Software
  16  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  17  */
  18
  19
  20 #include "config.h"
  21
  22 #include "captive/unicode.h"    /* self */
  23 #include "captive/unicode_reactos.h"    /* for captive_ucs2 */
  24 #include <glib/gtypes.h>
  25 #include <glib/gmessages.h>
  26 #include <glib/gunicode.h>
  27 #include <glib/gmem.h>
  28 #include "reactos/napi/types.h"  /* for PUNICODE_STRING etc. */
  29 #include "reactos/unicode.h"
  30 #include "captive/macros.h"
  31 #include <glib/gstrfuncs.h>
  32 #include <wchar.h>      /* for wcslen() */
  33 #include <glib/ghash.h>
  34 #include <string.h>
  35
  36
  37 /* CONFIG: */
  38 /* Use simplified g_malloc() functions as wrappers around g_alloca() ones.
  39  */
  40 #define FUNCMALLOC_FROM_ALLOCA 1
  41
  42
  43 /* compiler sanity */
  44 static gboolean captive_validate_unicode_types(void)
  45 {
  46         g_return_val_if_fail(4==sizeof(gunichar),FALSE);
  47         g_return_val_if_fail(2==sizeof(WCHAR),FALSE);
  48         g_return_val_if_fail(1==sizeof(CHAR),FALSE);
  49
  50         return TRUE;
  51 }
  52
  53
  54 /**
  55  * captive_validate_ucs4:
  56  * @string_ucs4: #const #gunichar * type string to validate.
  57  * Invalid string input is forbidden.
  58  *
  59  * Checks the validity of all 32-bit unicharacters of 0-terminated string.
  60  * It is required to have characters complying to g_unichar_validate().
  61  *
  62  * Returns: %TRUE if the string is valid.
  63  */
  64 gboolean captive_validate_ucs4(const gunichar *string_ucs4)
  65 {
  66 const gunichar *cs_ucs4;
  67
  68         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
  69         g_return_val_if_fail(string_ucs4!=NULL,FALSE);
  70
  71         for (cs_ucs4=string_ucs4;*cs_ucs4;cs_ucs4++)
  72                 g_return_val_if_fail(g_unichar_validate(*cs_ucs4),FALSE);
  73
  74         return TRUE;
  75 }
  76
  77
  78 /**
  79  * captive_validate_ucs2_fixlen:
  80  * @string_ucs2: #const #captive_ucs2 * type string to validate.
  81  * Invalid string input is forbidden.
  82  * UTF-16 encoded strings are forbidden.
  83  * @string_ucs2_fixlen: Number of characters from @string_ucs2 to check.
  84  * captive_ucs2_strlen(@string_ucs2)>=@string_ucs2_fixlen is required.
  85  * Negative value is forbidden.
  86  *
  87  * Checks the validity of first @string_ucs2_fixlen 16-bit unicharacters of @string_ucs2.
  88  * It is required to have characters complying to g_unichar_validate().
  89  * String length must be equal or larger than @string_ucs2_fixlen;
  90  *
  91  * Returns: %TRUE if the string is valid.
  92  */
  93 gboolean captive_validate_ucs2_fixlen(const captive_ucs2 *string_ucs2,glong string_ucs2_fixlen)
  94 {
  95 const captive_ucs2 *cs_ucs2;
  96
  97         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
  98         g_return_val_if_fail(string_ucs2!=NULL,FALSE);
  99         g_return_val_if_fail(string_ucs2_fixlen>=0,FALSE);
 100
 101         /* g_unichar_validate() will reject surrogates (G_UNICODE_SURROGATE) */
 102         for (cs_ucs2=string_ucs2;cs_ucs2<string_ucs2+string_ucs2_fixlen;cs_ucs2++) {
 103                 g_return_val_if_fail(*cs_ucs2!=0,FALSE);
 104                 g_return_val_if_fail(g_unichar_validate(*cs_ucs2),FALSE);
 105                 }
 106
 107         return TRUE;
 108 }
 109
 110
 111 /**
 112  * captive_validate_ucs2:
 113  * @string_ucs2: #const #captive_ucs2 * type string to validate.
 114  * Invalid string input is forbidden.
 115  * UTF-16 encoded strings are forbidden.
 116  *
 117  * Checks the validity of all 16-bit unicharacters of 0-terminated string.
 118  * It is required to have characters complying to g_unichar_validate().
 119  *
 120  * Returns: %TRUE if the string is valid.
 121  */
 122 gboolean captive_validate_ucs2(const captive_ucs2 *string_ucs2)
 123 {
 124         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 125         g_return_val_if_fail(string_ucs2!=NULL,FALSE);
 126
 127         return captive_validate_ucs2_fixlen(string_ucs2,captive_ucs2_strlen(string_ucs2));
 128 }
 129
 130
 131 /**
 132  * captive_validate_utf8:
 133  * @string_utf8: #const #gchar * utf8 type string to validate.
 134  * Invalid string input is forbidden.
 135  *
 136  * Checks the validity of all utf8 of 0-terminated string.
 137  * It is required to have characters complying to g_utf8_validate().
 138  *
 139  * Returns: %TRUE if the string is valid.
 140  */
 141 gboolean captive_validate_utf8(const gchar *string_utf8)
 142 {
 143         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 144         g_return_val_if_fail(string_utf8!=NULL,FALSE);
 145
 146         g_return_val_if_fail(g_utf8_validate(
 147                                         string_utf8,    /* str */
 148                                         -1,     /* max_len; -1 means '\0'-terminated */
 149                                         NULL),  /* end */
 150                         FALSE);
 151
 152         return TRUE;
 153 }
 154
 155
 156 /**
 157  * captive_ucs2_strlen:
 158  * @string_ucs2: String of type #const #gunichar2 * in pure UCS-2
 159  * Invalid string input is forbidden. UTF-16 encoded pairs are forbidden.
 160  *
 161  * Counts the number of characters (=2bytes) in @strings_ucs2.
 162  *
 163  * Returns: @string_ucs2 length in UCS-2 characters.
 164  */
 165 glong captive_ucs2_strlen(const captive_ucs2 *string_ucs2)
 166 {
 167 glong r;
 168
 169         /* Do not call captive_validate_ucs2(string_ucs2) as we would be looping! */
 170         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 171         g_return_val_if_fail(string_ucs2!=NULL,FALSE);
 172
 173         for (r=0;*string_ucs2;string_ucs2++)
 174                 r++;
 175
 176         return r;
 177 }
 178
 179
 180 /**
 181  * captive_validate_UnicodeString:
 182  * @string_UnicodeString: #PUNICODE_STRING type string to validate.
 183  * Invalid string input is forbidden.
 184  *
 185  * Checks the internal consistency of the given @string_UnicodeString.
 186  * It is required to have characters complying to g_unichar_validate().
 187  * @string_UnicodeString MUST be zero-terminated.
 188  *
 189  * Returns: %TRUE if the string is valid.
 190  */
 191 gboolean captive_validate_UnicodeString(const UNICODE_STRING *string_UnicodeString)
 192 {
 193         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 194         g_return_val_if_fail(sizeof(WCHAR)==sizeof(*string_UnicodeString->Buffer),FALSE);
 195         g_return_val_if_fail(string_UnicodeString!=NULL,FALSE);
 196         g_return_val_if_fail(string_UnicodeString->Length%sizeof(*string_UnicodeString->Buffer)==0,FALSE);
 197         g_return_val_if_fail(string_UnicodeString->MaximumLength
 198                         >=string_UnicodeString->Length+sizeof(*string_UnicodeString->Buffer),FALSE);
 199         g_return_val_if_fail(string_UnicodeString->Length==sizeof(*string_UnicodeString->Buffer)*
 200                         captive_ucs2_strlen(string_UnicodeString->Buffer)
 201                         ,FALSE);
 202
 203         g_return_val_if_fail(captive_validate_ucs2(string_UnicodeString->Buffer),FALSE);
 204
 205         return TRUE;
 206 }
 207
 208
 209 /**
 210  * captive_validate_UnicodeString_noterm:
 211  * @string_UnicodeString_noterm: #PUNICODE_STRING type string to validate.
 212  * Invalid string input is forbidden.
 213  *
 214  * Checks the internal consistency of the given @string_UnicodeString.
 215  * It is required to have characters complying to g_unichar_validate().
 216  * @string_UnicodeString_noterm does not neet to be zero-terminated.
 217  *
 218  * Returns: %TRUE if the string is valid.
 219  */
 220 gboolean captive_validate_UnicodeString_noterm(const UNICODE_STRING *string_UnicodeString_noterm)
 221 {
 222 const WCHAR *cwp;
 223
 224         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 225         g_return_val_if_fail(sizeof(WCHAR)==sizeof(*string_UnicodeString_noterm->Buffer),FALSE);
 226         g_return_val_if_fail(string_UnicodeString_noterm!=NULL,FALSE);
 227         g_return_val_if_fail(string_UnicodeString_noterm->Length%sizeof(*string_UnicodeString_noterm->Buffer)==0,FALSE);
 228         g_return_val_if_fail(string_UnicodeString_noterm->MaximumLength>=string_UnicodeString_noterm->Length,FALSE);
 229
 230         for (
 231                         cwp=string_UnicodeString_noterm->Buffer;
 232                         cwp<string_UnicodeString_noterm->Buffer
 233                                         +(string_UnicodeString_noterm->Length/sizeof(*string_UnicodeString_noterm->Buffer));
 234                         cwp++)
 235                 g_return_val_if_fail(*cwp!=0,FALSE);
 236
 237         g_return_val_if_fail(captive_validate_ucs2_fixlen(string_UnicodeString_noterm->Buffer,
 238                                         string_UnicodeString_noterm->Length/sizeof(*string_UnicodeString_noterm->Buffer)),
 239                         FALSE);
 240
 241         return TRUE;
 242 }
 243
 244
 245 /**
 246  * captive_validate_AnsiString:
 247  * @string_AnsiString: #PANSI_STRING type string to validate.
 248  * Invalid string input is forbidden.
 249  *
 250  * Checks the internal consistency of the given @string_AnsiString.
 251  *
 252  * Returns: %TRUE if the string is valid.
 253  */
 254 gboolean captive_validate_AnsiString(const ANSI_STRING *string_AnsiString)
 255 {
 256         g_return_val_if_fail(captive_validate_unicode_types(),FALSE);
 257         g_return_val_if_fail(sizeof(CHAR)==sizeof(*string_AnsiString->Buffer),FALSE);
 258         g_return_val_if_fail(string_AnsiString!=NULL,FALSE);
 259         g_return_val_if_fail(string_AnsiString->MaximumLength>=string_AnsiString->Length+1,FALSE);
 260         g_return_val_if_fail(string_AnsiString->Length==strlen(string_AnsiString->Buffer),FALSE);
 261
 262         return TRUE;
 263 }
 264
 265
 266 /* detect required memory size for g_alloca() */
 267 size_t _captive_UnicodeString_to_utf8_alloca_internal_sizeof(const UNICODE_STRING *string_UnicodeString)
 268 {
 269 glong length;
 270 size_t r;
 271 const WCHAR *cwcharp;
 272
 273         g_return_val_if_fail(captive_validate_UnicodeString(string_UnicodeString),1);
 274
 275         /* measure 'string_UnicodeString->Buffer' length in UTF-8 to 'r' */
 276         cwcharp=string_UnicodeString->Buffer;
 277         r=0;
 278         for (length=string_UnicodeString->Length/sizeof(*string_UnicodeString->Buffer);length;length--) {
 279 gint utf8len;
 280
 281                 utf8len=g_unichar_to_utf8(
 282                                 *cwcharp++,     /* c */
 283                                 NULL);  /* outbuf=NULL => just the length will be computed */
 284                 g_assert(utf8len>=0);
 285                 r+=utf8len;
 286                 }
 287         g_assert(*cwcharp==0);
 288         r++;    /* '\0'-termination */
 289
 290         /* utf8 byte-size */
 291         return r;
 292 }
 293
 294 /* transfer 'string_UnicodeString' to memory in 'mem' as utf8 w/o any further allocations */
 295 void _captive_UnicodeString_to_utf8_alloca_internal_fill(gchar *mem,const UNICODE_STRING *string_UnicodeString)
 296 {
 297 const WCHAR *cwcharp;
 298 #ifndef G_DISABLE_ASSERT
 299 gchar *mem_orig=mem;
 300 #endif /* G_DISABLE_ASSERT */
 301
 302         g_return_if_fail(mem!=NULL);
 303         if (!captive_validate_UnicodeString(string_UnicodeString)) {
 304                 *mem='\0';
 305                 g_return_if_reached();
 306                 }
 307
 308         /* We can't use any glib string conversions as UNICODE_STRING uses ucs2! */
 309         /* We can't use any glib string conversions as we need to write the string
 310          * to our supplied memory storage but glib always g_malloc()s it
 311          */
 312         /* copy 'string_UnicodeString->Buffer' to 'mem' */
 313         for (cwcharp=string_UnicodeString->Buffer;*cwcharp;cwcharp++) {
 314 gint utf8len;
 315
 316                 utf8len=g_unichar_to_utf8(
 317                                 (gunichar)*cwcharp,     /* c */
 318                                 mem);   /* outbuf */
 319                 g_assert(utf8len>=0);
 320                 mem+=utf8len;
 321                 }
 322         *mem='\0';
 323
 324         g_assert((size_t)((mem+1)-mem_orig) == _captive_UnicodeString_to_utf8_alloca_internal_sizeof(string_UnicodeString));
 325         g_assert(captive_validate_utf8(mem_orig));
 326 }
 327
 328
 329 /**
 330  * captive_UnicodeString_to_utf8_malloc:
 331  * @string_UnicodeString: #PUNICODE_STRING type of string to convert.
 332  *
 333  * g_malloc()-based conversion from #PUNICODE_STRING to plain #utf8 string.
 334  * You must free the result with g_free() function.
 335  *
 336  * Returns: #const #gchar * g_malloc()ed converted string @string_UnicodeString.
 337  */
 338 gchar *captive_UnicodeString_to_utf8_malloc(const UNICODE_STRING *string_UnicodeString)
 339 {
 340 gchar *r;
 341 #ifndef FUNCMALLOC_FROM_ALLOCA
 342 glong utf16_read,utf8_written;
 343 GError *err;
 344 #endif /* !FUNCMALLOC_FROM_ALLOCA */
 345
 346         g_return_val_if_fail(captive_validate_UnicodeString(string_UnicodeString),g_strdup(""));
 347
 348 #ifdef FUNCMALLOC_FROM_ALLOCA
 349
 350         r=g_malloc(_captive_UnicodeString_to_utf8_alloca_internal_sizeof(string_UnicodeString));
 351         _captive_UnicodeString_to_utf8_alloca_internal_fill(r,string_UnicodeString);
 352
 353 #else
 354
 355         err=NULL;       /* not precleared by g_utf8_to_utf16()! */
 356         r=g_utf16_to_utf8(
 357                         (const gunichar2 *)string_UnicodeString->Buffer,        /* str */
 358                         -1,     /* len=>'\0'-terminated */
 359                         &utf16_read,    /* items_read; counted in unichar2 (NOT UTF-16 characters or bytes!) */
 360                         &utf8_written,  /* items_written; counted in bytes (NOT UTF-8 characters!) */
 361                         &err);
 362         if (err) {
 363                 g_warning("%s: utf16_read=%ld,utf8_written=%ld: %s",G_STRLOC,
 364                                 (long)utf16_read,(long)utf8_written,err->message);
 365                 g_error_free(err);
 366                 g_assert(r==NULL);
 367                 g_return_val_if_reached(g_strdup(""));
 368                 }
 369         g_assert(r!=NULL);
 370
 371         g_assert(utf16_read==(glong)(string_UnicodeString->length/sizeof(*string_UnicodeString->Buffer)));
 372         g_assert(utf6_written==strlen(r));
 373
 374 #endif /* !FUNCMALLOC_FROM_ALLOCA */
 375
 376         g_assert(captive_validate_utf8(r));
 377
 378         return r;
 379 }
 380
 381
 382 /* detect required memory size for g_alloca() */
 383 size_t _captive_utf8_to_UnicodeString_alloca_internal_sizeof(const gchar *string_utf8)
 384 {
 385         g_return_val_if_fail(captive_validate_utf8(string_utf8),1);
 386
 387         /* find the value for PUNICODE_STRING->MaximumLength */
 388         return 0
 389                         +sizeof(UNICODE_STRING)
 390                         +sizeof(WCHAR)*(g_utf8_strlen(string_utf8,
 391                                         -1      /* max; -1 means '\0'-terminated */
 392                                         )+1);   /* '\0'-termination */
 393 }
 394
 395 static void terminate_static_UnicodeString(UNICODE_STRING *string_UnicodeString,glong length)
 396 {
 397         /* 'string_UnicodeString' is not yet valid in this point! */
 398         g_return_if_fail(string_UnicodeString!=NULL);
 399         g_return_if_fail(length>=0);
 400
 401         string_UnicodeString->Length=length*sizeof(WCHAR);
 402         string_UnicodeString->MaximumLength=(length+1)*sizeof(WCHAR);
 403         string_UnicodeString->Buffer[length]=0;
 404
 405         g_assert(captive_validate_UnicodeString(string_UnicodeString));
 406 }
 407
 408 /* transfer 'string_UnicodeString' to memory in 'mem' w/o any further allocations */
 409 void _captive_utf8_to_UnicodeString_alloca_internal_fill(UNICODE_STRING *mem,const gchar *string_utf8)
 410 {
 411 gunichar2 *utf16;
 412 captive_ucs2 *ucs2;
 413 glong utf8_read,utf16_written;
 414 GError *err;
 415
 416         g_return_if_fail(mem!=NULL);
 417         mem->Buffer=(PWSTR)(((char *)mem)+sizeof(*mem));        /* for terminate_static_UnicodeString() below */
 418         if (!captive_validate_utf8(string_utf8)) {
 419                 terminate_static_UnicodeString(mem,0);
 420                 g_return_if_reached();
 421                 }
 422
 423         err=NULL;       /* not precleared by g_utf8_to_utf16()! */
 424         utf16=g_utf8_to_utf16(
 425                         string_utf8,    /* str */
 426                         -1,     /* len=>'\0'-terminated */
 427                         &utf8_read,     /* items_read; counted in bytes (NOT chars!) */
 428                         &utf16_written, /* items_written; counted in UTF-16 characters (NOT unichar2 or bytes!) */
 429                         &err);
 430         if (err) {
 431                 g_warning("%s: utf8_read=%ld,utf16_written=%ld: %s",G_STRLOC,
 432                                 (long)utf8_read,(long)utf16_written,err->message);
 433                 g_error_free(err);
 434                 g_assert(utf16==NULL);
 435                 terminate_static_UnicodeString(mem,0);
 436                 g_return_if_reached();
 437                 }
 438         g_assert(utf16!=NULL);
 439
 440         /* Check for UCS-2 compliance (reject if surrogates inside) */
 441         g_assert(captive_validate_ucs2((const captive_ucs2 *)utf16));
 442         /* valid UCS-2 */
 443         ucs2=(captive_ucs2 *)utf16;
 444
 445         g_assert(utf8_read==(glong)strlen(string_utf8));
 446         g_assert(utf16_written==captive_ucs2_strlen(ucs2));
 447
 448         /* check of validity of _captive_utf8_to_UnicodeString_alloca_internal_sizeof() result */
 449         g_assert((gchar *)(mem->Buffer+(utf16_written+1))       /* +1 => '\0'-termination */
 450                         == ((gchar *)mem)+_captive_utf8_to_UnicodeString_alloca_internal_sizeof(string_utf8));
 451
 452         memcpy(mem->Buffer,ucs2,sizeof(WCHAR)*(utf16_written+1));
 453         g_free(ucs2);
 454         terminate_static_UnicodeString(mem,utf16_written);
 455
 456         g_assert(captive_validate_UnicodeString(mem));
 457 }
 458
 459
 460 /**
 461  * captive_utf8_to_UnicodeString_malloc:
 462  * @string_utf8: #const #gchar * string in #utf8 to convert.
 463  *
 464  * g_malloc()-based conversion from plain #utf8 string to #PUNICODE_STRING.
 465  * You must free the result with g_free() function.
 466  *
 467  * Returns: #PUNICODE_STRING g_malloc()ed converted string @string_utf8.
 468  */
 469 PUNICODE_STRING captive_utf8_to_UnicodeString_malloc(const gchar *string_utf8)
 470 {
 471 UNICODE_STRING *r;
 472 #ifndef FUNCMALLOC_FROM_ALLOCA
 473 gunichar *ucs4;
 474 glong utf8_read,ucs4_written;
 475 GError *err;
 476 #endif /* !FUNCMALLOC_FROM_ALLOCA */
 477
 478         g_return_val_if_fail(captive_validate_utf8(string_utf8),captive_utf8_to_UnicodeString_malloc(""));
 479
 480 #ifdef FUNCMALLOC_FROM_ALLOCA
 481
 482         r=g_malloc(_captive_utf8_to_UnicodeString_alloca_internal_sizeof(string_utf8));
 483         _captive_utf8_to_UnicodeString_alloca_internal_fill(r,string_utf8);
 484
 485 #else
 486
 487 #error "FIXME: NOT IMPLEMENTED"
 488
 489 #endif /* !FUNCMALLOC_FROM_ALLOCA */
 490
 491         g_assert(captive_validate_UnicodeString(r));
 492
 493         return r;
 494 }
 495
 496
 497 /* map: (const gunichar *) -> (const gunichar2 *); UCS-4 -> UTF-16 */
 498 static GHashTable *captive_ucs4_to_utf16_hash;
 499
 500 static void captive_ucs4_to_utf16_hash_init(void)
 501 {
 502         if (captive_ucs4_to_utf16_hash)
 503                 return;
 504         captive_ucs4_to_utf16_hash=g_hash_table_new_full(
 505                         g_direct_hash,  /* hash_func */
 506                         g_direct_equal, /* key_equal_func */
 507                         (GDestroyNotify)NULL,   /* key_destroy_func; we require persistent strings as input */
 508                         (GDestroyNotify)g_free);        /* value_destroy_func; result of g_ucs4_to_utf16() */
 509 }
 510
 511 /**
 512  * captive_ucs4_to_utf16_const:
 513  * @string_ucs4: #const #gunichar * type of persistent string to convert.
 514  * This string MUST remain readable with the same content forever.
 515  *
 516  * Constant string conversion from 32-bit #wchar_t to 16-bit (possible pairs of) UTF-16.
 517  * You may not modify the result in any way.
 518  *
 519  * It is guaranteed to get two different string addresses for two different
 520  * input addresses even if the input strings content is the same.
 521  * Otherwise we would behave as #GCC option %-fmerge-constants which
 522  * results in %C non-conforming behaviour.
 523  *
 524  * FIXME: UTF-16 encoding IS NOT IMPLEMENTED.
 525  *
 526  * See also captive_ucs4_to_ucs2_const().
 527  *
 528  * Returns: #const #gunichar2 * converted string @string_ucs4.
 529  */
 530 const gunichar2 *captive_ucs4_to_utf16_const(const gunichar *string_ucs4)
 531 {
 532 glong ucs4_read,utf16_written;
 533 GError *err;
 534 const gunichar2 *r_lookup;
 535 gunichar2 *r;
 536
 537         g_return_val_if_fail(captive_validate_ucs4(string_ucs4),captive_ucs4_to_utf16_const((const gunichar *)L""));
 538
 539         captive_ucs4_to_utf16_hash_init();
 540
 541         /* found already existing item in the table */
 542         if ((r_lookup=g_hash_table_lookup(captive_ucs4_to_utf16_hash,
 543                         string_ucs4)    /* key */
 544                         )) {
 545                 return r_lookup;
 546                 }
 547
 548         /* Prepare 'r' as UTF-16 */
 549         err=NULL;       /* not precleared by g_ucs4_to_utf16()! */
 550         r=g_ucs4_to_utf16(
 551                         (const gunichar *)string_ucs4,  /* str */
 552                         -1,     /* len; -1 means '\0'-termination */
 553                         &ucs4_read,     /* items_read; counted in chars (==unichars; NOT bytes!) */
 554                         &utf16_written, /* items_written; counted in gunichar2 (NOT chars or bytes!) */
 555                         &err);
 556         if (err) {
 557                 g_warning("%s: ucs4_read=%ld,utf16_written=%ld: %s",G_STRLOC,
 558                                 (long)ucs4_read,(long)utf16_written,err->message);
 559                 g_error_free(err);
 560                 g_assert(r==NULL);
 561                 g_return_val_if_reached(captive_ucs4_to_utf16_const((const gunichar *)L""));
 562                 }
 563         g_assert(r!=NULL);
 564         g_assert(ucs4_read==(glong)wcslen((const wchar_t *)string_ucs4));
 565         /* FIXME: We don't have captive_utf16_strlen() */
 566         g_assert(utf16_written==(glong)captive_ucs2_strlen((const gunichar2 *)r));
 567         /* (ucs4_read==utf16_written) check would discard any double-pair UTF-16 encodings
 568          * but this function is designed as UTF-16 compliant.
 569          */
 570
 571         /* store new item to the table */
 572         g_hash_table_insert(captive_ucs4_to_utf16_hash,
 573                         (gpointer)string_ucs4,  /* key; de-const */
 574                         r);     /* value */
 575
 576 #if 0   /* We don't have captive_validate_utf16() */
 577         g_assert(captive_validate_utf16(r));
 578 #endif
 579
 580         return r;
 581 }