mpdm_s.c

   1 /*
   2
   3     MPDM - Minimum Profit Data Manager
   4     Copyright (C) 2003/2010 Angel Ortega <angel@triptico.com>
   5
   6     mpdm_s.c - String management
   7
   8     This program is free software; you can redistribute it and/or
   9     modify it under the terms of the GNU General Public License
  10     as published by the Free Software Foundation; either version 2
  11     of the License, or (at your option) any later version.
  12
  13     This program is distributed in the hope that it will be useful,
  14     but WITHOUT ANY WARRANTY; without even the implied warranty of
  15     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16     GNU General Public License for more details.
  17
  18     You should have received a copy of the GNU General Public License
  19     along with this program; if not, write to the Free Software
  20     Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  21
  22     http://www.triptico.com
  23
  24 */
  25
  26 #include "config.h"
  27
  28 #include <stdio.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31 #include <wchar.h>
  32 #include <locale.h>
  33 #include <wctype.h>
  34
  35 #ifdef CONFOPT_GETTEXT
  36 #include <libintl.h>
  37 #endif
  38
  39 #ifdef CONFOPT_WIN32
  40 #include <windows.h>
  41 #endif
  42
  43 #include "mpdm.h"
  44
  45
  46 /** code **/
  47
  48 void *mpdm_poke_o(void *dst, int *dsize, int *offset, const void *org, int osize, int esize)
  49 {
  50         if (org != NULL && osize) {
  51                 /* enough room? */
  52                 if (*offset + osize > *dsize) {
  53                         /* no; enlarge */
  54                         *dsize += osize;
  55
  56                         dst = realloc(dst, *dsize * esize);
  57                 }
  58
  59                 memcpy((char *)dst + (*offset * esize), org, osize * esize);
  60                 *offset += osize;
  61         }
  62
  63         return dst;
  64 }
  65
  66
  67 void *mpdm_poke(void *dst, int *dsize, const void *org, int osize, int esize)
  68 /* pokes (adds) org into dst, which is a dynamic string, making it grow */
  69 {
  70         int offset = *dsize;
  71
  72         return mpdm_poke_o(dst, dsize, &offset, org, osize, esize);
  73 }
  74
  75
  76 wchar_t *mpdm_pokewsn(wchar_t *dst, int *dsize, const wchar_t *str, int slen)
  77 /* adds a wide string to dst using mpdm_poke() with size */
  78 {
  79         if (str)
  80                 dst = mpdm_poke(dst, dsize, str, slen, sizeof(wchar_t));
  81
  82         return dst;
  83 }
  84
  85
  86 wchar_t *mpdm_pokews(wchar_t *dst, int *dsize, const wchar_t *str)
  87 /* adds a wide string to dst using mpdm_poke() */
  88 {
  89         if (str)
  90                 dst = mpdm_pokewsn(dst, dsize, str, wcslen(str));
  91
  92         return dst;
  93 }
  94
  95
  96 wchar_t *mpdm_pokev(wchar_t * dst, int *dsize, const mpdm_t v)
  97 /* adds the string in v to dst using mpdm_poke() */
  98 {
  99         if (v != NULL) {
 100                 const wchar_t *ptr = mpdm_string(v);
 101
 102                 mpdm_ref(v);
 103                 dst = mpdm_pokews(dst, dsize, ptr);
 104                 mpdm_unref(v);
 105         }
 106
 107         return dst;
 108 }
 109
 110
 111 wchar_t *mpdm_mbstowcs(const char *str, int *s, int l)
 112 /* converts an mbs to a wcs, but filling invalid chars
 113    with question marks instead of just failing */
 114 {
 115         wchar_t *ptr = NULL;
 116         char tmp[64];           /* really MB_CUR_MAX + 1 */
 117         wchar_t wc;
 118         int n, i, c, t = 0;
 119         char *cstr;
 120
 121         /* allow NULL values for s */
 122         if (s == NULL)
 123                 s = &t;
 124
 125         /* if there is a limit, duplicate and break the string */
 126         if (l >= 0) {
 127                 cstr = strdup(str);
 128                 cstr[l] = '\0';
 129         }
 130         else
 131                 cstr = (char *) str;
 132
 133         /* try first a direct conversion with mbstowcs */
 134         if ((*s = mbstowcs(NULL, cstr, 0)) != -1) {
 135                 /* direct conversion is possible; do it */
 136                 if ((ptr = malloc((*s + 1) * sizeof(wchar_t))) != NULL) {
 137                         mbstowcs(ptr, cstr, *s);
 138                         ptr[*s] = L'\0';
 139                 }
 140         }
 141         else {
 142                 /* zero everything */
 143                 *s = n = i = 0;
 144
 145                 for (;;) {
 146                         /* no more characters to process? */
 147                         if ((c = cstr[n + i]) == '\0' && i == 0)
 148                                 break;
 149
 150                         tmp[i++] = c;
 151                         tmp[i] = '\0';
 152
 153                         /* try to convert */
 154                         if (mbstowcs(&wc, tmp, 1) == (size_t) -1) {
 155                                 /* can still be an incomplete multibyte char? */
 156                                 if (c != '\0' && i <= (int) MB_CUR_MAX)
 157                                         continue;
 158                                 else {
 159                                         /* too many failing bytes; skip 1 byte */
 160                                         wc = L'?';
 161                                         i = 1;
 162                                 }
 163                         }
 164
 165                         /* skip used bytes and back again */
 166                         n += i;
 167                         i = 0;
 168
 169                         /* store new char */
 170                         if ((ptr = mpdm_poke(ptr, s, &wc, 1, sizeof(wchar_t))) == NULL)
 171                                 break;
 172                 }
 173
 174                 /* null terminate and count one less */
 175                 if (ptr != NULL) {
 176                         ptr = mpdm_poke(ptr, s, L"", 1, sizeof(wchar_t));
 177                         (*s)--;
 178                 }
 179         }
 180
 181         /* free the duplicate */
 182         if (cstr != str)
 183                 free(cstr);
 184
 185         return ptr;
 186 }
 187
 188
 189 char *mpdm_wcstombs(const wchar_t * str, int *s)
 190 /* converts a wcs to an mbs, but filling invalid chars
 191    with question marks instead of just failing */
 192 {
 193         char *ptr = NULL;
 194         char tmp[64];           /* really MB_CUR_MAX + 1 */
 195         int l, t = 0;
 196
 197         /* allow NULL values for s */
 198         if (s == NULL)
 199                 s = &t;
 200
 201         /* try first a direct conversion with wcstombs */
 202         if ((*s = wcstombs(NULL, str, 0)) != -1) {
 203                 /* direct conversion is possible; do it and return */
 204                 if ((ptr = malloc(*s + 1)) != NULL) {
 205                         wcstombs(ptr, str, *s);
 206                         ptr[*s] = '\0';
 207                 }
 208
 209                 return ptr;
 210         }
 211
 212         /* invalid encoding? convert characters one by one */
 213         *s = 0;
 214
 215         while (*str) {
 216                 if ((l = wctomb(tmp, *str)) <= 0) {
 217                         /* if char couldn't be converted,
 218                            write a question mark instead */
 219                         l = wctomb(tmp, L'?');
 220                 }
 221
 222                 tmp[l] = '\0';
 223                 if ((ptr = mpdm_poke(ptr, s, tmp, l, 1)) == NULL)
 224                         break;
 225
 226                 str++;
 227         }
 228
 229         /* null terminate and count one less */
 230         if (ptr != NULL) {
 231                 ptr = mpdm_poke(ptr, s, "", 1, 1);
 232                 (*s)--;
 233         }
 234
 235         return ptr;
 236 }
 237
 238
 239 mpdm_t mpdm_new_wcs(int flags, const wchar_t * str, int size, int cpy)
 240 /* creates a new string value from a wcs */
 241 {
 242         wchar_t *ptr;
 243
 244         /* a size of -1 means 'calculate it' */
 245         if (size == -1 && str != NULL)
 246                 size = wcslen(str);
 247
 248         /* create a copy? */
 249         if (cpy) {
 250                 /* free() on destruction */
 251                 flags |= MPDM_FREE;
 252
 253                 /* allocs */
 254                 if ((ptr = malloc((size + 1) * sizeof(wchar_t))) == NULL)
 255                         return NULL;
 256
 257                 /* if no source, reset to zeroes; otherwise, copy */
 258                 if (str == NULL)
 259                         memset(ptr, '\0', size * sizeof(wchar_t));
 260                 else {
 261                         wcsncpy(ptr, str, size);
 262                         ptr[size] = L'\0';
 263                 }
 264         }
 265         else
 266                 ptr = (wchar_t *)str;
 267
 268         /* it's a string */
 269         flags |= MPDM_STRING;
 270
 271         return mpdm_new(flags, ptr, size);
 272 }
 273
 274
 275 mpdm_t mpdm_new_mbstowcs(int flags, const char *str, int l)
 276 /* creates a new string value from an mbs */
 277 {
 278         wchar_t *ptr;
 279         int size;
 280
 281         if ((ptr = mpdm_mbstowcs(str, &size, l)) == NULL)
 282                 return NULL;
 283
 284         /* it's a string */
 285         flags |= (MPDM_STRING | MPDM_FREE);
 286
 287         return mpdm_new(flags, ptr, size);
 288 }
 289
 290
 291 mpdm_t mpdm_new_wcstombs(int flags, const wchar_t * str)
 292 /* creates a new mbs value from a wbs */
 293 {
 294         char *ptr;
 295         int size;
 296
 297         ptr = mpdm_wcstombs(str, &size);
 298
 299         flags |= MPDM_FREE;
 300
 301         /* unset the string flag; mbs,s are not 'strings' */
 302         flags &= ~MPDM_STRING;
 303
 304         return mpdm_new(flags, ptr, size);
 305 }
 306
 307
 308 mpdm_t mpdm_new_i(int ival)
 309 /* creates a new string value from an integer */
 310 {
 311         mpdm_t v;
 312         char tmp[32];
 313
 314         /* creates the visual representation */
 315         snprintf(tmp, sizeof(tmp), "%d", ival);
 316
 317         v = MPDM_MBS(tmp);
 318
 319         return mpdm_set_ival(v, ival);
 320 }
 321
 322
 323 mpdm_t mpdm_new_r(double rval)
 324 /* creates a new string value from a real number */
 325 {
 326         mpdm_t v;
 327         char tmp[128];
 328
 329         /* creates the visual representation */
 330         snprintf(tmp, sizeof(tmp), "%lf", rval);
 331
 332         /* manually strip useless zeroes */
 333         if (strchr(tmp, '.') != NULL) {
 334                 char *ptr;
 335
 336                 for (ptr = tmp + strlen(tmp) - 1; *ptr == '0'; ptr--);
 337
 338                 /* if it's over the ., strip it also */
 339                 if (*ptr != '.')
 340                         ptr++;
 341
 342                 *ptr = '\0';
 343         }
 344
 345         v = MPDM_MBS(tmp);
 346
 347         return mpdm_set_rval(v, rval);
 348 }
 349
 350
 351 /* interface */
 352
 353 /**
 354  * mpdm_string - Returns a printable representation of a value.
 355  * @v: the value
 356  *
 357  * Returns a printable representation of a value. For strings, it's
 358  * the value data itself; for any other type, a conversion to string
 359  * is returned instead. This value should be used immediately, as it
 360  * can be a pointer to a static buffer.
 361  * [Strings]
 362  */
 363 wchar_t *mpdm_string(const mpdm_t v)
 364 {
 365         static wchar_t wtmp[32];
 366         char tmp[32];
 367         static wchar_t *ret;
 368
 369         /* if it's NULL, return a constant */
 370         if (v == NULL)
 371                 ret = L"[NULL]";
 372         else
 373         /* if it's a string, return it */
 374         if (v->flags & MPDM_STRING)
 375                 ret = (wchar_t *) v->data;
 376         else {
 377                 /* otherwise, return a visual representation */
 378                 snprintf(tmp, sizeof(tmp), "%p", v);
 379                 mbstowcs(wtmp, tmp, sizeof(wtmp));
 380                 wtmp[(sizeof(wtmp) / sizeof(wchar_t)) - 1] = L'\0';
 381
 382                 ret = wtmp;
 383         }
 384
 385         return ret;
 386 }
 387
 388
 389 /**
 390  * mpdm_cmp - Compares two values.
 391  * @v1: the first value
 392  * @v2: the second value
 393  *
 394  * Compares two values. If both has the MPDM_STRING flag set,
 395  * a comparison using wcscoll() is returned; if both are arrays,
 396  * the size is compared first and, if they have the same number
 397  * elements, each one is compared; otherwise, a simple pointer
 398  * comparison is done.
 399  * [Strings]
 400  */
 401 int mpdm_cmp(const mpdm_t v1, const mpdm_t v2)
 402 {
 403         int r;
 404
 405         /* same values? */
 406         if (v1 == v2)
 407                 r = 0;
 408         else
 409         /* is any value NULL? */
 410         if (v1 == NULL)
 411                 r = -1;
 412         else
 413         if (v2 == NULL)
 414                 r = 1;
 415         else
 416         /* different values, but same content? (unlikely) */
 417         if (v1->data == v2->data)
 418                 r = 0;
 419         else
 420         if (MPDM_IS_STRING(v1) && MPDM_IS_STRING(v2))
 421                 r = wcscoll((wchar_t *) v1->data, (wchar_t *) v2->data);
 422         else
 423         if (MPDM_IS_ARRAY(v1) && MPDM_IS_ARRAY(v2)) {
 424                 /* compare first the sizes */
 425                 if ((r = mpdm_size(v1) - mpdm_size(v2)) == 0) {
 426                         int n;
 427
 428                         /* they have the same size;
 429                            compare each pair of elements */
 430                         for (n = 0; n < mpdm_size(v1); n++) {
 431                                 if ((r = mpdm_cmp(mpdm_aget(v1, n),
 432                                         mpdm_aget(v2, n))) != 0)
 433                                         break;
 434                         }
 435                 }
 436         }
 437         else
 438                 /* in any other case, compare just pointers */
 439                 r = (int) ((char *)v1->data - (char *)v2->data);
 440
 441         return r;
 442 }
 443
 444
 445 /**
 446  * mpdm_cmp_s - Compares two values (string version).
 447  * @v1: the first value
 448  * @v2: the second value
 449  *
 450  * Compares two values. Compares both values using wcscoll()
 451  * if the first one is a string, or returns 1 otherwise.
 452  */
 453 int mpdm_cmp_s(const mpdm_t v1, const wchar_t *v2)
 454 {
 455         int r;
 456
 457         if (MPDM_IS_STRING(v1))
 458                 r = wcscoll((wchar_t *) v1->data, v2);
 459         else
 460                 r = (int) ((wchar_t *)v1->data - v2);
 461
 462         return r;
 463 }
 464
 465
 466 /**
 467  * mpdm_splice - Creates a new string value from another.
 468  * @v: the original value
 469  * @i: the value to be inserted
 470  * @offset: offset where the substring is to be inserted
 471  * @del: number of characters to delete
 472  *
 473  * Creates a new string value from @v, deleting @del chars at @offset
 474  * and substituting them by @i. If @del is 0, no deletion is done.
 475  * both @offset and @del can be negative; if this is the case, it's
 476  * assumed as counting from the end of @v. If @v is NULL, @i will become
 477  * the new string, and both @offset and @del will be ignored. If @v is
 478  * not NULL and @i is, no insertion process is done (only deletion, if
 479  * applicable).
 480  *
 481  * Returns a two element array, with the new string in the first
 482  * element and the deleted string in the second (with a NULL value
 483  * if @del is 0).
 484  * [Strings]
 485  */
 486 mpdm_t mpdm_splice(const mpdm_t v, const mpdm_t i, int offset, int del)
 487 {
 488         mpdm_t w;
 489         mpdm_t n = NULL;
 490         mpdm_t d = NULL;
 491         int os, ns, r;
 492         int ins = 0;
 493         wchar_t *ptr;
 494
 495         mpdm_ref(v);
 496         mpdm_ref(i);
 497
 498         if (v != NULL) {
 499                 os = mpdm_size(v);
 500
 501                 /* negative offsets start from the end */
 502                 if (offset < 0)
 503                         offset = os + 1 - offset;
 504
 505                 /* never add further the end */
 506                 if (offset > os)
 507                         offset = os;
 508
 509                 /* negative del counts as 'characters left' */
 510                 if (del < 0)
 511                         del = os + 1 - offset + del;
 512
 513                 /* something to delete? */
 514                 if (del > 0) {
 515                         /* never delete further the end */
 516                         if (offset + del > os)
 517                                 del = os - offset;
 518
 519                         /* deleted string */
 520                         d = MPDM_NS(((wchar_t *) v->data) + offset, del);
 521                 }
 522                 else
 523                         del = 0;
 524
 525                 /* something to insert? */
 526                 ins = mpdm_size(i);
 527
 528                 /* new size and remainder */
 529                 ns = os + ins - del;
 530                 r = offset + del;
 531
 532                 n = MPDM_NS(NULL, ns);
 533
 534                 ptr = (wchar_t *)n->data;
 535
 536                 /* copy the beginning */
 537                 if (offset > 0) {
 538                         wcsncpy(ptr, v->data, offset);
 539                         ptr += offset;
 540                 }
 541
 542                 /* copy the text to be inserted */
 543                 if (ins > 0) {
 544                         wcsncpy(ptr, i->data, ins);
 545                         ptr += ins;
 546                 }
 547
 548                 /* copy the remaining */
 549                 os -= r;
 550                 if (os > 0) {
 551                         wcsncpy(ptr, ((wchar_t *) v->data) + r, os);
 552                         ptr += os;
 553                 }
 554
 555                 /* null terminate */
 556                 *ptr = L'\0';
 557         }
 558         else
 559                 n = i;
 560
 561         /* creates the output array */
 562         w = MPDM_A(2);
 563
 564         mpdm_aset(w, n, 0);
 565         mpdm_aset(w, d, 1);
 566
 567         mpdm_unref(i);
 568         mpdm_unref(v);
 569
 570         return w;
 571 }
 572
 573
 574 /**
 575  * mpdm_strcat_sn - Concatenates two strings (string with size version).
 576  * @s1: the first string
 577  * @s2: the second string
 578  * @size: the size of the second string
 579  *
 580  * Returns a new string formed by the concatenation of @s1 and @s2.
 581  * [Strings]
 582  */
 583 mpdm_t mpdm_strcat_sn(const mpdm_t s1, const wchar_t *s2, int size)
 584 {
 585         wchar_t *ptr = NULL;
 586         int s = 0;
 587
 588         if (s1 == NULL && s2 == NULL)
 589                 return NULL;
 590
 591         ptr = mpdm_pokev(ptr, &s, s1);
 592         ptr = mpdm_pokewsn(ptr, &s, s2, size);
 593
 594         ptr = mpdm_poke(ptr, &s, L"", 1, sizeof(wchar_t));
 595         return MPDM_ENS(ptr, s - 1);
 596 }
 597
 598
 599 /**
 600  * mpdm_strcat_s - Concatenates two strings (string version).
 601  * @s1: the first string
 602  * @s2: the second string
 603  *
 604  * Returns a new string formed by the concatenation of @s1 and @s2.
 605  * [Strings]
 606  */
 607 mpdm_t mpdm_strcat_s(const mpdm_t s1, const wchar_t *s2)
 608 {
 609         return mpdm_strcat_sn(s1, s2, s2 ? wcslen(s2) : 0);
 610 }
 611
 612
 613 /**
 614  * mpdm_strcat - Concatenates two strings.
 615  * @s1: the first string
 616  * @s2: the second string
 617  *
 618  * Returns a new string formed by the concatenation of @s1 and @s2.
 619  * [Strings]
 620  */
 621 mpdm_t mpdm_strcat(const mpdm_t s1, const mpdm_t s2)
 622 {
 623         return mpdm_strcat_s(s1, s2 ? mpdm_string(s2) : NULL);
 624 }
 625
 626
 627 /**
 628  * mpdm_ival - Returns a value's data as an integer.
 629  * @v: the value
 630  *
 631  * Returns a value's data as an integer. If the value is a string,
 632  * it's converted via sscanf and returned; non-string values have all
 633  * an ival of 0. The converted integer is cached, so costly string
 634  * conversions are only done once. Values created with the MPDM_IVAL
 635  * flag set have its ival cached from the beginning.
 636  * [Strings]
 637  * [Value Management]
 638  */
 639 int mpdm_ival(mpdm_t v)
 640 {
 641         if (v == NULL)
 642                 return 0;
 643
 644         /* if there is no cached integer, calculate it */
 645         if (!(v->flags & MPDM_IVAL)) {
 646                 int i = 0;
 647
 648                 /* if it's a string, calculate it; other
 649                    values will have an ival of 0 */
 650                 if (v->flags & MPDM_STRING) {
 651                         char tmp[32];
 652                         char *fmt = "%i";
 653
 654                         wcstombs(tmp, (wchar_t *) v->data, sizeof(tmp));
 655                         tmp[sizeof(tmp) - 1] = '\0';
 656
 657                         /* workaround for mingw32: as it doesn't
 658                            correctly parse octal and hexadecimal
 659                            numbers, they are tried as special cases */
 660                         if (tmp[0] == '0') {
 661                                 if (tmp[1] == 'b' || tmp[1] == 'B') {
 662                                         /* binary number */
 663                                         fmt = NULL;
 664                                         char *ptr = &tmp[2];
 665
 666                                         while (*ptr == '0' || *ptr == '1') {
 667                                                 i <<= 1;
 668
 669                                                 if (*ptr == '1')
 670                                                         i |= 1;
 671
 672                                                 ptr++;
 673                                         }
 674                                 }
 675                                 else
 676                                 if (tmp[1] == 'x' || tmp[1] == 'X')
 677                                         fmt = "%x";
 678                                 else
 679                                         fmt = "%o";
 680                         }
 681
 682                         if (fmt != NULL)
 683                                 sscanf(tmp, fmt, &i);
 684                 }
 685
 686                 mpdm_set_ival(v, i);
 687         }
 688
 689         return v->ival;
 690 }
 691
 692
 693 /**
 694  * mpdm_rval - Returns a value's data as a real number (double).
 695  * @v: the value
 696  *
 697  * Returns a value's data as a real number (double float). If the value
 698  * is a string, it's converted via sscanf and returned; non-string values
 699  * have all an rval of 0. The converted double is cached, so costly string
 700  * conversions are only done once. Values created with the MPDM_RVAL
 701  * flag set have its rval cached from the beginning.
 702  * [Strings]
 703  * [Value Management]
 704  */
 705 double mpdm_rval(mpdm_t v)
 706 {
 707         if (v == NULL)
 708                 return 0;
 709
 710         /* if there is no cached double, calculate it */
 711         if (!(v->flags & MPDM_RVAL)) {
 712                 double r = 0.0;
 713
 714                 /* if it's a string, calculate it; other
 715                    values will have an rval of 0.0 */
 716                 if (v->flags & MPDM_STRING) {
 717                         char tmp[128];
 718                         char *prev_locale;
 719
 720                         wcstombs(tmp, (wchar_t *) v->data, sizeof(tmp));
 721                         tmp[sizeof(tmp) - 1] = '\0';
 722
 723                         /* if the number starts with 0, it's
 724                            an octal or hexadecimal number; just
 725                            take the integer value and cast it */
 726                         if (tmp[0] == '0' && tmp[1] != '.')
 727                                 r = (double) mpdm_ival(v);
 728                         else {
 729                                 /* set locale to C for non locale-dependent
 730                                    floating point conversion */
 731                                 prev_locale = setlocale(LC_NUMERIC, "C");
 732
 733                                 /* read */
 734                                 sscanf(tmp, "%lf", &r);
 735
 736                                 /* set previous locale */
 737                                 setlocale(LC_NUMERIC, prev_locale);
 738                         }
 739                 }
 740
 741                 mpdm_set_rval(v, r);
 742         }
 743
 744         return v->rval;
 745 }
 746
 747
 748 /**
 749  * mpdm_gettext - Translates a string to the current language.
 750  * @str: the string
 751  *
 752  * Translates the @str string to the current language.
 753  *
 754  * This function can still be used even if there is no real gettext
 755  * support() by manually filling the __I18N__ hash.
 756  *
 757  * If the string is found in the current table, the translation is
 758  * returned; otherwise, the same @str value is returned.
 759  * [Strings]
 760  * [Localization]
 761  */
 762 mpdm_t mpdm_gettext(const mpdm_t str)
 763 {
 764         mpdm_t v;
 765         mpdm_t i18n = NULL;
 766
 767         /* gets the cache, if any */
 768         if ((i18n = mpdm_hget_s(mpdm_root(), L"__I18N__")) == NULL)
 769                 return str;
 770
 771         /* try first the cache */
 772         if ((v = mpdm_hget(i18n, str)) == NULL) {
 773 #ifdef CONFOPT_GETTEXT
 774                 char *s;
 775                 mpdm_t t;
 776
 777                 /* convert to mbs */
 778                 t = mpdm_ref(MPDM_2MBS(str->data));
 779
 780                 /* ask gettext for it */
 781                 s = gettext((char *) t->data);
 782
 783                 /* create new value only if it's different */
 784                 if (s != t->data) {
 785                         v = MPDM_MBS(s);
 786
 787                         /* store in the cache */
 788                         mpdm_hset(i18n, str, v);
 789                 }
 790                 else
 791                         v = str;
 792
 793                 mpdm_unref(t);
 794
 795 #else                           /* CONFOPT_GETTEXT */
 796
 797                 v = str;
 798
 799 #endif                          /* CONFOPT_GETTEXT */
 800         }
 801
 802         return v;
 803 }
 804
 805
 806 /**
 807  * mpdm_gettext_domain - Sets domain and data directory for translations.
 808  * @dom: the domain (application name)
 809  * @data: directory contaning the .mo files
 810  *
 811  * Sets the domain (application name) and translation data for translating
 812  * strings that will be returned by mpdm_gettext().@data must point to a
 813  * directory containing the .mo (compiled .po) files.
 814  *
 815  * If there is no gettext support, returns 0, or 1 otherwise.
 816  * [Strings]
 817  * [Localization]
 818  */
 819 int mpdm_gettext_domain(const mpdm_t dom, const mpdm_t data)
 820 {
 821         int ret = 0;
 822
 823 #ifdef CONFOPT_GETTEXT
 824
 825         mpdm_t dm;
 826         mpdm_t dt;
 827
 828         /* convert both to mbs,s */
 829         dm = mpdm_ref(MPDM_2MBS(dom->data));
 830         dt = mpdm_ref(MPDM_2MBS(data->data));
 831
 832         /* bind and set domain */
 833         bindtextdomain((char *) dm->data, (char *) dt->data);
 834         textdomain((char *) dm->data);
 835
 836         mpdm_hset_s(mpdm_root(), L"__I18N__", MPDM_H(0));
 837
 838         mpdm_unref(dt);
 839         mpdm_unref(dm);
 840
 841         ret = 1;
 842
 843 #endif                          /* CONFOPT_GETTEXT */
 844
 845 #ifdef CONFOPT_WIN32
 846
 847         mpdm_t v;
 848
 849         if ((v = mpdm_hget_s(mpdm_root(), L"ENV")) != NULL &&
 850                 mpdm_hget_s(v, L"LANG") == NULL) {
 851                 wchar_t *wptr = L"en";
 852
 853                 /* MS Windows crappy language constants... */
 854
 855                 switch((GetSystemDefaultLangID() & 0x00ff)) {
 856                 case 0x01: wptr = L"ar"; break; /* arabic */
 857                 case 0x02: wptr = L"bg"; break; /* bulgarian */
 858                 case 0x03: wptr = L"ca"; break; /* catalan */
 859                 case 0x04: wptr = L"zh"; break; /* chinese */
 860                 case 0x05: wptr = L"cz"; break; /* czech */
 861                 case 0x06: wptr = L"da"; break; /* danish */
 862                 case 0x07: wptr = L"de"; break; /* german */
 863                 case 0x08: wptr = L"el"; break; /* greek */
 864                 case 0x09: wptr = L"en"; break; /* english */
 865                 case 0x0a: wptr = L"es"; break; /* spanish */
 866                 case 0x0b: wptr = L"fi"; break; /* finnish */
 867                 case 0x0c: wptr = L"fr"; break; /* french */
 868                 case 0x0d: wptr = L"he"; break; /* hebrew */
 869                 case 0x0e: wptr = L"hu"; break; /* hungarian */
 870                 case 0x0f: wptr = L"is"; break; /* icelandic */
 871                 case 0x10: wptr = L"it"; break; /* italian */
 872                 case 0x11: wptr = L"jp"; break; /* japanese */
 873                 case 0x12: wptr = L"ko"; break; /* korean */
 874                 case 0x13: wptr = L"nl"; break; /* dutch */
 875                 case 0x14: wptr = L"no"; break; /* norwegian */
 876                 case 0x15: wptr = L"po"; break; /* polish */
 877                 case 0x16: wptr = L"pt"; break; /* portuguese */
 878                 case 0x17: wptr = L"rm"; break; /* romansh (switzerland) */
 879                 case 0x18: wptr = L"ro"; break; /* romanian */
 880                 case 0x19: wptr = L"ru"; break; /* russian */
 881                 case 0x1a: wptr = L"sr"; break; /* serbian */
 882                 case 0x1b: wptr = L"sk"; break; /* slovak */
 883                 case 0x1c: wptr = L"sq"; break; /* albanian */
 884                 case 0x1d: wptr = L"sv"; break; /* swedish */
 885                 }
 886
 887                 mpdm_hset_s(v, L"LANG", MPDM_S(wptr));
 888         }
 889
 890 #endif                          /* CONFOPT_WIN32 */
 891
 892         return ret;
 893 }
 894
 895
 896 #ifdef CONFOPT_WCWIDTH
 897
 898 int wcwidth(wchar_t);
 899
 900 int mpdm_wcwidth(wchar_t c)
 901 {
 902         return wcwidth(c);
 903 }
 904
 905 #else                           /* CONFOPT_WCWIDTH */
 906
 907 #include "wcwidth.c"
 908
 909 int mpdm_wcwidth(wchar_t c)
 910 {
 911         return mk_wcwidth(c);
 912 }
 913
 914 #endif                          /* CONFOPT_WCWIDTH */
 915
 916
 917 /**
 918  * mpdm_sprintf - Formats a sprintf()-like string.
 919  * @fmt: the string format
 920  * @args: an array of values
 921  *
 922  * Formats a string using the sprintf() format taking the values from @args.
 923  * [Strings]
 924  */
 925 mpdm_t mpdm_sprintf(const mpdm_t fmt, const mpdm_t args)
 926 {
 927         const wchar_t *i = fmt->data;
 928         wchar_t *o = NULL;
 929         int l = 0, n = 0;
 930         wchar_t c;
 931
 932         /* loop all characters */
 933         while ((c = *i++) != L'\0') {
 934                 int m = 0;
 935                 wchar_t *tptr = NULL;
 936                 wchar_t *wptr = NULL;
 937
 938                 if (c == L'%') {
 939                         /* format directive */
 940                         char t_fmt[128];
 941                         char tmp[1024];
 942                         mpdm_t v;
 943                         char *ptr = NULL;
 944
 945                         /* transfer the % */
 946                         t_fmt[m++] = '%';
 947
 948                         /* transform the format to mbs */
 949                         while (*i != L'\0' &&
 950                                 m < (int)(sizeof(t_fmt) - MB_CUR_MAX - 1) &&
 951                                 wcschr(L"-.0123456789", *i) != NULL)
 952                                 m += wctomb(&t_fmt[m], *i++);
 953
 954                         /* transfer the directive */
 955                         m += wctomb(&t_fmt[m], *i++);
 956
 957                         t_fmt[m] = '\0';
 958
 959                         /* by default, copies the format */
 960                         strcpy(tmp, t_fmt);
 961
 962                         /* pick next value */
 963                         v = mpdm_aget(args, n++);
 964
 965                         switch (t_fmt[m - 1]) {
 966                         case 'd':
 967                         case 'i':
 968                         case 'u':
 969                         case 'x':
 970                         case 'X':
 971                         case 'o':
 972
 973                                 /* integer value */
 974                                 snprintf(tmp, sizeof(tmp) - 1,
 975                                         t_fmt, mpdm_ival(v));
 976                                 break;
 977
 978                         case 'f':
 979
 980                                 /* float (real) value */
 981                                 snprintf(tmp, sizeof(tmp) - 1,
 982                                         t_fmt, mpdm_rval(v));
 983                                 break;
 984
 985                         case 's':
 986
 987                                 /* string value */
 988                                 ptr = mpdm_wcstombs(mpdm_string(v), NULL);
 989                                 snprintf(tmp, sizeof(tmp) - 1, t_fmt, ptr);
 990                                 free(ptr);
 991
 992                                 break;
 993
 994                         case 'c':
 995
 996                                 /* char */
 997                                 m = 1;
 998                                 wptr = &c;
 999                                 c = mpdm_ival(v);
1000                                 break;
1001
1002                         case 'b':
1003
1004                                 ptr = tmp;
1005                                 unsigned int mask;
1006                                 int p = 0;
1007
1008                                 mask = 1 << ((sizeof(int) * 8) - 1);
1009                                 while (mask) {
1010                                         if (mask & (unsigned int) mpdm_ival(v)) {
1011                                                 *ptr++ = '1';
1012                                                 p = 1;
1013                                         }
1014                                         else
1015                                         if (p)
1016                                                 *ptr++ = '0';
1017
1018                                         mask >>= 1;
1019                                 }
1020
1021                                 if (ptr == tmp)
1022                                         *ptr++ = '0';
1023
1024                                 *ptr = '\0';
1025                                 break;
1026
1027                         case '%':
1028
1029                                 /* percent sign */
1030                                 m = 1;
1031                                 wptr = &c;
1032                                 break;
1033                         }
1034
1035                         /* transfer */
1036                         if (wptr == NULL)
1037                                 wptr = tptr = mpdm_mbstowcs(tmp, &m, -1);
1038                 }
1039                 else {
1040                         /* raw character */
1041                         m = 1;
1042                         wptr = &c;
1043                 }
1044
1045                 /* transfer */
1046                 o = mpdm_poke(o, &l, wptr, m, sizeof(wchar_t));
1047
1048                 /* free the temporary buffer, if any */
1049                 if (tptr != NULL)
1050                         free(tptr);
1051         }
1052
1053         if (o == NULL)
1054                 return NULL;
1055
1056         /* null-terminate */
1057         o = mpdm_poke(o, &l, L"", 1, sizeof(wchar_t));
1058
1059         return MPDM_ENS(o, l - 1);
1060 }
1061
1062
1063 /**
1064  * mpdm_ulc - Converts a string to uppercase or lowecase.
1065  * @s: the string
1066  * @u: convert to uppercase (1) or to lowercase (0).
1067  *
1068  * Converts @s to uppercase (for @u == 1) or to lowercase (@u == 0).
1069  * [Strings]
1070  */
1071 mpdm_t mpdm_ulc(const mpdm_t s, int u)
1072 {
1073         mpdm_t r = NULL;
1074         wchar_t *optr;
1075         int i = mpdm_size(s);
1076
1077         if ((optr = malloc((i + 1) * sizeof(wchar_t))) != NULL) {
1078                 wchar_t *iptr = mpdm_string(s);
1079                 int n;
1080
1081                 for (n = 0; n < i; n++)
1082                         optr[n] = u ? towupper(iptr[n]) : towlower(iptr[n]);
1083
1084                 optr[n] = L'\0';
1085                 r = MPDM_ENS(optr, i);
1086         }
1087
1088         return r;
1089 }
1090
1091
1092 /* scanf working buffers */
1093 #define SCANF_BUF_SIZE 1024
1094 static wchar_t scanf_yset[SCANF_BUF_SIZE];
1095 static wchar_t scanf_nset[SCANF_BUF_SIZE];
1096 static wchar_t scanf_mark[SCANF_BUF_SIZE];
1097
1098 struct {
1099         wchar_t cmd;
1100         wchar_t *yset;
1101         wchar_t *nset;
1102 } scanf_sets[] = {
1103         { L's',         L"",                            L" \t" },
1104         { L'u',         L"0123456789",                  L"" },
1105         { L'd',         L"-0123456789",                 L"" },
1106         { L'i',         L"-0123456789",                 L"" },
1107         { L'f',         L"-0123456789.",                L"" },
1108         { L'x',         L"-0123456789xabcdefABCDEF",    L"" },
1109         { L'\0',        NULL,                           NULL },
1110 };
1111
1112 /**
1113  * mpdm_sscanf - Extracts data like sscanf().
1114  * @fmt: the string format
1115  * @str: the string to be parsed
1116  * @offset: the character offset to start scanning
1117  *
1118  * Extracts data from a string using a special format pattern, very
1119  * much like the scanf() series of functions in the C library. Apart
1120  * from the standard percent-sign-commands (s, u, d, i, f, x,
1121  * n, [, with optional size and * to ignore), it implements S,
1122  * to match a string of characters upto what follows in the format
1123  * string. Also, the [ set of characters can include other % formats.
1124  *
1125  * Returns an array with the extracted values. If %n is used, the
1126  * position in the scanned string is returned as the value.
1127  * [Strings]
1128  */
1129 mpdm_t mpdm_sscanf(const mpdm_t fmt, const mpdm_t str, int offset)
1130 {
1131         wchar_t *i = (wchar_t *)str->data;
1132         wchar_t *f = (wchar_t *)fmt->data;
1133         mpdm_t r;
1134
1135         i += offset;
1136         r = MPDM_A(0);
1137
1138         while (*f) {
1139                 if (*f == L'%') {
1140                         wchar_t *ptr = NULL;
1141                         int size = 0;
1142                         wchar_t cmd;
1143                         int vsize = 0;
1144                         int ignore = 0;
1145                         int msize = 0;
1146
1147                         /* empty all buffers */
1148                         scanf_yset[0] = scanf_nset[0] = scanf_mark[0] = L'\0';
1149
1150                         f++;
1151
1152                         /* an asterisk? don't return next value */
1153                         if (*f == L'*') {
1154                                 ignore = 1;
1155                                 f++;
1156                         }
1157
1158                         /* does it have a size? */
1159                         while (wcschr(L"0123456789", *f)) {
1160                                 vsize *= 10;
1161                                 vsize += *f - L'0';
1162                                 f++;
1163                         }
1164
1165                         /* if no size, set it to an arbitrary big limit */
1166                         if (!vsize)
1167                                 vsize = 0xfffffff;
1168
1169                         /* now *f should contain a command */
1170                         cmd = *f;
1171                         f++;
1172
1173                         /* is it a verbatim percent sign? */
1174                         if (cmd == L'%') {
1175                                 vsize = 1;
1176                                 ignore = 1;
1177                                 wcscpy(scanf_yset, L"%");
1178                         }
1179                         else
1180                         /* a position? */
1181                         if (cmd == L'n') {
1182                                 vsize = 0;
1183                                 ignore = 1;
1184                                 mpdm_push(r, MPDM_I(i - (wchar_t *)str->data));
1185                         }
1186                         else
1187                         /* string upto a mark */
1188                         if (cmd == L'S') {
1189                                 wchar_t *tmp = f;
1190
1191                                 /* fill the mark upto another command */
1192                                 while (*tmp) {
1193                                         if (*tmp == L'%') {
1194                                                 tmp++;
1195
1196                                                 /* is it an 'n'? ignore and go on */
1197                                                 if (*tmp == L'n') {
1198                                                         tmp++;
1199                                                         continue;
1200                                                 }
1201                                                 else
1202                                                 if (*tmp == L'%')
1203                                                         scanf_mark[msize++] = *tmp;
1204                                                 else
1205                                                         break;
1206                                         }
1207                                         else
1208                                                 scanf_mark[msize++] = *tmp;
1209
1210                                         tmp++;
1211                                 }
1212
1213                                 scanf_mark[msize] = L'\0';
1214                         }
1215                         else
1216                         /* raw set */
1217                         if (cmd == L'[') {
1218                                 int n = 0;
1219                                 wchar_t *set = scanf_yset;
1220
1221                                 /* is it an inverse set? */
1222                                 if (*f == L'^') {
1223                                         set = scanf_nset;
1224                                         f++;
1225                                 }
1226
1227                                 /* first one is a ]? add it */
1228                                 if (*f == L']') {
1229                                         set[n++] = *f;
1230                                         f++;
1231                                 }
1232
1233                                 /* now build the set */
1234                                 for (; n < SCANF_BUF_SIZE - 1 && *f && *f != L']'; f++) {
1235                                         /* is it a range? */
1236                                         if (*f == L'-') {
1237                                                 f++;
1238
1239                                                 /* start or end? hyphen itself */
1240                                                 if (n == 0 || *f == L']')
1241                                                         set[n++] = L'-';
1242                                                 else {
1243                                                         /* pick previous char */
1244                                                         wchar_t c = set[n - 1];
1245
1246                                                         /* fill */
1247                                                         while (n < SCANF_BUF_SIZE - 1 && c < *f)
1248                                                                 set[n++] = ++c;
1249                                                 }
1250                                         }
1251                                         else
1252                                         /* is it another command? */
1253                                         if (*f == L'%') {
1254                                                 int i;
1255
1256                                                 f++;
1257                                                 for (i = 0; scanf_sets[i].cmd; i++) {
1258                                                         if (*f == scanf_sets[i].cmd) {
1259                                                                 set[n] = L'\0';
1260                                                                 wcscat(set, scanf_sets[i].yset);
1261                                                                 n += wcslen(scanf_sets[i].yset);
1262                                                                 break;
1263                                                         }
1264                                                 }
1265                                         }
1266                                         else
1267                                                 set[n++] = *f;
1268                                 }
1269
1270                                 /* skip the ] */
1271                                 f++;
1272
1273                                 set[n] = L'\0';
1274                         }
1275                         else
1276                         /* a standard set? */
1277                         {
1278                                 int n;
1279
1280                                 for (n = 0; scanf_sets[n].cmd != L'\0'; n++) {
1281                                         if (cmd == scanf_sets[n].cmd) {
1282                                                 wcscpy(scanf_yset, scanf_sets[n].yset);
1283                                                 wcscpy(scanf_nset, scanf_sets[n].nset);
1284                                                 break;
1285                                         }
1286                                 }
1287                         }
1288
1289                         /* now fill the dynamic string */
1290                         while (vsize &&
1291                                !wcschr(scanf_nset, *i) &&
1292                                (scanf_yset[0] == L'\0' || wcschr(scanf_yset, *i)) &&
1293                                (msize == 0 || wcsncmp(i, scanf_mark, msize) != 0)) {
1294
1295                                 /* only add if not being ignored */
1296                                 if (!ignore)
1297                                         ptr = mpdm_poke(ptr, &size, i, 1, sizeof(wchar_t));
1298
1299                                 i++;
1300                                 vsize--;
1301                         }
1302
1303                         if (!ignore && size) {
1304                                 /* null terminate and push */
1305                                 ptr = mpdm_poke(ptr, &size, L"", 1, sizeof(wchar_t));
1306                                 mpdm_push(r, MPDM_ENS(ptr, size - 1));
1307                         }
1308                 }
1309                 else
1310                 if (*f == L' ' || *f == L'\t') {
1311                         /* if it's a blank, sync to next non-blank */
1312                         f++;
1313
1314                         while (*i == L' ' || *i == L'\t')
1315                                 i++;
1316                 }
1317                 else
1318                 /* test for literals in the format string */
1319                 if (*i == *f) {
1320                         i++;
1321                         f++;
1322                 }
1323                 else
1324                         break;
1325         }
1326
1327         return r;
1328 }