lib/strutil/strutilascii.c

   1 /*
   2    ASCII strings utilities
   3
   4    Copyright (C) 2007-2018
   5    Free Software Foundation, Inc.
   6
   7    Written by:
   8    Rostislav Benes, 2007
   9
  10    This file is part of the Midnight Commander.
  11
  12    The Midnight Commander is free software: you can redistribute it
  13    and/or modify it under the terms of the GNU General Public License as
  14    published by the Free Software Foundation, either version 3 of the License,
  15    or (at your option) any later version.
  16
  17    The Midnight Commander is distributed in the hope that it will be useful,
  18    but WITHOUT ANY WARRANTY; without even the implied warranty of
  19    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20    GNU General Public License for more details.
  21
  22    You should have received a copy of the GNU General Public License
  23    along with this program.  If not, see <http://www.gnu.org/licenses/>.
  24  */
  25
  26 #include <config.h>
  27
  28 #include <ctype.h>
  29 #include <stdlib.h>
  30
  31 #include "lib/global.h"
  32 #include "lib/strutil.h"
  33
  34 /* using g_ascii function from glib
  35  * on terminal are showed only ascii characters (lower than 0x80)
  36  */
  37
  38 /*** global variables ****************************************************************************/
  39
  40 /*** file scope macro definitions ****************************************************************/
  41
  42 /*** file scope type declarations ****************************************************************/
  43
  44 /*** file scope variables ************************************************************************/
  45
  46 static const char replch = '?';
  47
  48 /* --------------------------------------------------------------------------------------------- */
  49 /*** file scope functions ************************************************************************/
  50 /* --------------------------------------------------------------------------------------------- */
  51
  52 static void
  53 str_ascii_insert_replace_char (GString * buffer)
  54 {
  55     g_string_append_c (buffer, replch);
  56 }
  57
  58 /* --------------------------------------------------------------------------------------------- */
  59
  60 static gboolean
  61 str_ascii_is_valid_string (const char *text)
  62 {
  63     (void) text;
  64     return TRUE;
  65 }
  66
  67 /* --------------------------------------------------------------------------------------------- */
  68
  69 static int
  70 str_ascii_is_valid_char (const char *ch, size_t size)
  71 {
  72     (void) ch;
  73     (void) size;
  74     return 1;
  75 }
  76
  77 /* --------------------------------------------------------------------------------------------- */
  78
  79 static void
  80 str_ascii_cnext_char (const char **text)
  81 {
  82     (*text)++;
  83 }
  84
  85 /* --------------------------------------------------------------------------------------------- */
  86
  87 static void
  88 str_ascii_cprev_char (const char **text)
  89 {
  90     (*text)--;
  91 }
  92
  93 /* --------------------------------------------------------------------------------------------- */
  94
  95 static int
  96 str_ascii_cnext_noncomb_char (const char **text)
  97 {
  98     if (*text[0] == '\0')
  99         return 0;
 100
 101     (*text)++;
 102     return 1;
 103 }
 104
 105 /* --------------------------------------------------------------------------------------------- */
 106
 107 static int
 108 str_ascii_cprev_noncomb_char (const char **text, const char *begin)
 109 {
 110     if ((*text) == begin)
 111         return 0;
 112
 113     (*text)--;
 114     return 1;
 115 }
 116
 117 /* --------------------------------------------------------------------------------------------- */
 118
 119 static gboolean
 120 str_ascii_isspace (const char *text)
 121 {
 122     return g_ascii_isspace ((gchar) text[0]);
 123 }
 124
 125 /* --------------------------------------------------------------------------------------------- */
 126
 127 static gboolean
 128 str_ascii_ispunct (const char *text)
 129 {
 130     return g_ascii_ispunct ((gchar) text[0]);
 131 }
 132
 133 /* --------------------------------------------------------------------------------------------- */
 134
 135 static gboolean
 136 str_ascii_isalnum (const char *text)
 137 {
 138     return g_ascii_isalnum ((gchar) text[0]);
 139 }
 140
 141 /* --------------------------------------------------------------------------------------------- */
 142
 143 static gboolean
 144 str_ascii_isdigit (const char *text)
 145 {
 146     return g_ascii_isdigit ((gchar) text[0]);
 147 }
 148
 149 /* --------------------------------------------------------------------------------------------- */
 150
 151 static gboolean
 152 str_ascii_isprint (const char *text)
 153 {
 154     return g_ascii_isprint ((gchar) text[0]);
 155 }
 156
 157 /* --------------------------------------------------------------------------------------------- */
 158
 159 static gboolean
 160 str_ascii_iscombiningmark (const char *text)
 161 {
 162     (void) text;
 163     return FALSE;
 164 }
 165
 166 /* --------------------------------------------------------------------------------------------- */
 167
 168 static int
 169 str_ascii_toupper (const char *text, char **out, size_t * remain)
 170 {
 171     if (*remain <= 1)
 172         return FALSE;
 173
 174     (*out)[0] = (char) g_ascii_toupper ((gchar) text[0]);
 175     (*out)++;
 176     (*remain)--;
 177     return TRUE;
 178 }
 179
 180 /* --------------------------------------------------------------------------------------------- */
 181
 182 static gboolean
 183 str_ascii_tolower (const char *text, char **out, size_t * remain)
 184 {
 185     if (*remain <= 1)
 186         return FALSE;
 187
 188     (*out)[0] = (char) g_ascii_tolower ((gchar) text[0]);
 189     (*out)++;
 190     (*remain)--;
 191     return TRUE;
 192 }
 193
 194 /* --------------------------------------------------------------------------------------------- */
 195
 196 static int
 197 str_ascii_length (const char *text)
 198 {
 199     return strlen (text);
 200 }
 201
 202 /* --------------------------------------------------------------------------------------------- */
 203
 204 static int
 205 str_ascii_length2 (const char *text, int size)
 206 {
 207     return (size >= 0) ? MIN (strlen (text), (gsize) size) : strlen (text);
 208 }
 209
 210 /* --------------------------------------------------------------------------------------------- */
 211
 212 static gchar *
 213 str_ascii_conv_gerror_message (GError * mcerror, const char *def_msg)
 214 {
 215     /* the same as str_utf8_conv_gerror_message() */
 216     if (mcerror != NULL)
 217         return g_strdup (mcerror->message);
 218
 219     return g_strdup (def_msg != NULL ? def_msg : "");
 220 }
 221
 222 /* --------------------------------------------------------------------------------------------- */
 223
 224 static estr_t
 225 str_ascii_vfs_convert_to (GIConv coder, const char *string, int size, GString * buffer)
 226 {
 227     (void) coder;
 228     g_string_append_len (buffer, string, size);
 229     return ESTR_SUCCESS;
 230 }
 231
 232 /* --------------------------------------------------------------------------------------------- */
 233
 234 static const char *
 235 str_ascii_term_form (const char *text)
 236 {
 237     static char result[BUF_MEDIUM];
 238     char *actual;
 239     size_t remain;
 240     size_t length;
 241     size_t pos = 0;
 242
 243     actual = result;
 244     remain = sizeof (result);
 245     length = strlen (text);
 246
 247     /* go throw all characters and check, if they are ascii and printable */
 248     for (; pos < length && remain > 1; pos++, actual++, remain--)
 249     {
 250         actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 251         actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 252     }
 253
 254     actual[0] = '\0';
 255     return result;
 256 }
 257
 258 /* --------------------------------------------------------------------------------------------- */
 259
 260 static const char *
 261 str_ascii_fit_to_term (const char *text, int width, align_crt_t just_mode)
 262 {
 263     static char result[BUF_MEDIUM];
 264     char *actual;
 265     size_t remain;
 266     int ident = 0;
 267     size_t length;
 268     size_t pos = 0;
 269
 270     length = strlen (text);
 271     actual = result;
 272     remain = sizeof (result);
 273
 274     if ((int) length <= width)
 275     {
 276         switch (HIDE_FIT (just_mode))
 277         {
 278         case J_CENTER_LEFT:
 279         case J_CENTER:
 280             ident = (width - length) / 2;
 281             break;
 282         case J_RIGHT:
 283             ident = width - length;
 284             break;
 285         default:
 286             break;
 287         }
 288
 289         /* add space before text */
 290         if ((int) remain <= ident)
 291             goto finally;
 292         memset (actual, ' ', ident);
 293         actual += ident;
 294         remain -= ident;
 295
 296         /* copy all characters */
 297         for (; pos < (gsize) length && remain > 1; pos++, actual++, remain--)
 298         {
 299             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 300             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 301         }
 302
 303         /* add space after text */
 304         if (width - length - ident > 0)
 305         {
 306             if (remain <= width - length - ident)
 307                 goto finally;
 308             memset (actual, ' ', width - length - ident);
 309             actual += width - length - ident;
 310         }
 311     }
 312     else if (IS_FIT (just_mode))
 313     {
 314         /* copy prefix of text, that is not wider than width / 2 */
 315         for (; pos + 1 <= (gsize) width / 2 && remain > 1; actual++, pos++, remain--)
 316         {
 317             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 318             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 319         }
 320
 321         if (remain <= 1)
 322             goto finally;
 323         actual[0] = '~';
 324         actual++;
 325         remain--;
 326
 327         pos += length - width + 1;
 328
 329         /* copy suffix of text */
 330         for (; pos < length && remain > 1; pos++, actual++, remain--)
 331         {
 332             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 333             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 334         }
 335     }
 336     else
 337     {
 338         switch (HIDE_FIT (just_mode))
 339         {
 340         case J_CENTER:
 341             ident = (length - width) / 2;
 342             break;
 343         case J_RIGHT:
 344             ident = length - width;
 345             break;
 346         default:
 347             break;
 348         }
 349
 350         /* copy substring text, substring start from ident and take width
 351          * characters from text */
 352         pos += ident;
 353         for (; pos < (gsize) (ident + width) && remain > 1; pos++, actual++, remain--)
 354         {
 355             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 356             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 357         }
 358
 359     }
 360
 361   finally:
 362     if (actual >= result + sizeof (result))
 363         actual = result + sizeof (result) - 1;
 364     actual[0] = '\0';
 365     return result;
 366 }
 367
 368 /* --------------------------------------------------------------------------------------------- */
 369
 370 static const char *
 371 str_ascii_term_trim (const char *text, int width)
 372 {
 373     static char result[BUF_MEDIUM];
 374     size_t remain;
 375     char *actual;
 376     size_t length;
 377
 378     length = strlen (text);
 379     actual = result;
 380     remain = sizeof (result);
 381
 382     if (width > 0)
 383     {
 384         size_t pos;
 385
 386         if (width >= (int) length)
 387         {
 388             /* copy all characters */
 389             for (pos = 0; pos < length && remain > 1; pos++, actual++, remain--)
 390             {
 391                 actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 392                 actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 393             }
 394         }
 395         else if (width <= 3)
 396         {
 397             memset (actual, '.', width);
 398             actual += width;
 399         }
 400         else
 401         {
 402             memset (actual, '.', 3);
 403             actual += 3;
 404             remain -= 3;
 405
 406             /* copy suffix of text */
 407             for (pos = length - width + 3; pos < length && remain > 1; pos++, actual++, remain--)
 408             {
 409                 actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 410                 actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 411             }
 412         }
 413     }
 414
 415     actual[0] = '\0';
 416     return result;
 417 }
 418
 419 /* --------------------------------------------------------------------------------------------- */
 420
 421 static int
 422 str_ascii_term_width2 (const char *text, size_t length)
 423 {
 424     return (length != (size_t) (-1)) ? MIN (strlen (text), length) : strlen (text);
 425 }
 426
 427 /* --------------------------------------------------------------------------------------------- */
 428
 429 static int
 430 str_ascii_term_width1 (const char *text)
 431 {
 432     return str_ascii_term_width2 (text, (size_t) (-1));
 433 }
 434
 435 /* --------------------------------------------------------------------------------------------- */
 436
 437 static int
 438 str_ascii_term_char_width (const char *text)
 439 {
 440     (void) text;
 441     return 1;
 442 }
 443
 444 /* --------------------------------------------------------------------------------------------- */
 445
 446 static const char *
 447 str_ascii_term_substring (const char *text, int start, int width)
 448 {
 449     static char result[BUF_MEDIUM];
 450     size_t remain;
 451     char *actual;
 452     size_t length;
 453
 454     actual = result;
 455     remain = sizeof (result);
 456     length = strlen (text);
 457
 458     if (start < (int) length)
 459     {
 460         size_t pos;
 461
 462         /* copy at most width characters from text from start */
 463         for (pos = start; pos < length && width > 0 && remain > 1;
 464              pos++, width--, actual++, remain--)
 465         {
 466             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 467             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 468         }
 469     }
 470
 471     /* if text is shorter then width, add space to the end */
 472     for (; width > 0 && remain > 1; actual++, remain--, width--)
 473         actual[0] = ' ';
 474
 475     actual[0] = '\0';
 476     return result;
 477 }
 478
 479 /* --------------------------------------------------------------------------------------------- */
 480
 481 static const char *
 482 str_ascii_trunc (const char *text, int width)
 483 {
 484     static char result[MC_MAXPATHLEN];
 485     int remain;
 486     char *actual;
 487     size_t pos = 0;
 488     size_t length;
 489
 490     actual = result;
 491     remain = sizeof (result);
 492     length = strlen (text);
 493
 494     if ((int) length > width)
 495     {
 496         /* copy prefix of text */
 497         for (; pos + 1 <= (gsize) width / 2 && remain > 1; actual++, pos++, remain--)
 498         {
 499             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 500             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 501         }
 502
 503         if (remain <= 1)
 504             goto finally;
 505         actual[0] = '~';
 506         actual++;
 507         remain--;
 508
 509         pos += length - width + 1;
 510
 511         /* copy suffix of text */
 512         for (; pos < length && remain > 1; pos++, actual++, remain--)
 513         {
 514             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 515             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 516         }
 517     }
 518     else
 519     {
 520         /* copy all characters */
 521         for (; pos < length && remain > 1; pos++, actual++, remain--)
 522         {
 523             actual[0] = isascii ((unsigned char) text[pos]) ? text[pos] : '?';
 524             actual[0] = g_ascii_isprint ((gchar) actual[0]) ? actual[0] : '.';
 525         }
 526     }
 527
 528   finally:
 529     actual[0] = '\0';
 530     return result;
 531 }
 532
 533 /* --------------------------------------------------------------------------------------------- */
 534
 535 static int
 536 str_ascii_offset_to_pos (const char *text, size_t length)
 537 {
 538     (void) text;
 539     return (int) length;
 540 }
 541
 542 /* --------------------------------------------------------------------------------------------- */
 543
 544 static int
 545 str_ascii_column_to_pos (const char *text, size_t pos)
 546 {
 547     (void) text;
 548     return (int) pos;
 549 }
 550
 551 /* --------------------------------------------------------------------------------------------- */
 552
 553 static char *
 554 str_ascii_create_search_needle (const char *needle, gboolean case_sen)
 555 {
 556     (void) case_sen;
 557     return (char *) needle;
 558 }
 559
 560 /* --------------------------------------------------------------------------------------------- */
 561
 562 static void
 563 str_ascii_release_search_needle (char *needle, gboolean case_sen)
 564 {
 565     (void) case_sen;
 566     (void) needle;
 567
 568 }
 569
 570 /* --------------------------------------------------------------------------------------------- */
 571
 572 static const char *
 573 str_ascii_search_first (const char *text, const char *search, gboolean case_sen)
 574 {
 575     char *fold_text;
 576     char *fold_search;
 577     const char *match;
 578
 579     fold_text = case_sen ? (char *) text : g_ascii_strdown (text, -1);
 580     fold_search = case_sen ? (char *) search : g_ascii_strdown (search, -1);
 581
 582     match = g_strstr_len (fold_text, -1, fold_search);
 583     if (match != NULL)
 584     {
 585         size_t offset;
 586
 587         offset = match - fold_text;
 588         match = text + offset;
 589     }
 590
 591     if (!case_sen)
 592     {
 593         g_free (fold_text);
 594         g_free (fold_search);
 595     }
 596
 597     return match;
 598 }
 599
 600 /* --------------------------------------------------------------------------------------------- */
 601
 602 static const char *
 603 str_ascii_search_last (const char *text, const char *search, gboolean case_sen)
 604 {
 605     char *fold_text;
 606     char *fold_search;
 607     const char *match;
 608
 609     fold_text = case_sen ? (char *) text : g_ascii_strdown (text, -1);
 610     fold_search = case_sen ? (char *) search : g_ascii_strdown (search, -1);
 611
 612     match = g_strrstr_len (fold_text, -1, fold_search);
 613     if (match != NULL)
 614     {
 615         size_t offset;
 616
 617         offset = match - fold_text;
 618         match = text + offset;
 619     }
 620
 621     if (!case_sen)
 622     {
 623         g_free (fold_text);
 624         g_free (fold_search);
 625     }
 626
 627     return match;
 628 }
 629
 630 /* --------------------------------------------------------------------------------------------- */
 631
 632 static int
 633 str_ascii_compare (const char *t1, const char *t2)
 634 {
 635     return strcmp (t1, t2);
 636 }
 637
 638 /* --------------------------------------------------------------------------------------------- */
 639
 640 static int
 641 str_ascii_ncompare (const char *t1, const char *t2)
 642 {
 643     return strncmp (t1, t2, MIN (strlen (t1), strlen (t2)));
 644 }
 645
 646 /* --------------------------------------------------------------------------------------------- */
 647
 648 static int
 649 str_ascii_casecmp (const char *t1, const char *t2)
 650 {
 651     return g_ascii_strcasecmp (t1, t2);
 652 }
 653
 654 /* --------------------------------------------------------------------------------------------- */
 655
 656 static int
 657 str_ascii_ncasecmp (const char *t1, const char *t2)
 658 {
 659     return g_ascii_strncasecmp (t1, t2, MIN (strlen (t1), strlen (t2)));
 660 }
 661
 662 /* --------------------------------------------------------------------------------------------- */
 663
 664 static void
 665 str_ascii_fix_string (char *text)
 666 {
 667     for (; text[0] != '\0'; text++)
 668         text[0] = ((unsigned char) text[0] < 128) ? text[0] : '?';
 669 }
 670
 671 /* --------------------------------------------------------------------------------------------- */
 672
 673 static char *
 674 str_ascii_create_key (const char *text, gboolean case_sen)
 675 {
 676     (void) case_sen;
 677     return (char *) text;
 678 }
 679
 680 /* --------------------------------------------------------------------------------------------- */
 681
 682 static int
 683 str_ascii_key_collate (const char *t1, const char *t2, gboolean case_sen)
 684 {
 685     return case_sen ? strcmp (t1, t2) : g_ascii_strcasecmp (t1, t2);
 686 }
 687
 688 /* --------------------------------------------------------------------------------------------- */
 689
 690 static void
 691 str_ascii_release_key (char *key, gboolean case_sen)
 692 {
 693     (void) key;
 694     (void) case_sen;
 695 }
 696
 697 /* --------------------------------------------------------------------------------------------- */
 698
 699 static int
 700 str_ascii_prefix (const char *text, const char *prefix)
 701 {
 702     int result;
 703
 704     for (result = 0; text[result] != '\0' && prefix[result] != '\0'
 705          && text[result] == prefix[result]; result++);
 706
 707     return result;
 708 }
 709
 710 /* --------------------------------------------------------------------------------------------- */
 711
 712 static int
 713 str_ascii_caseprefix (const char *text, const char *prefix)
 714 {
 715     int result;
 716
 717     for (result = 0; text[result] != '\0' && prefix[result] != '\0'
 718          && g_ascii_toupper (text[result]) == g_ascii_toupper (prefix[result]); result++);
 719
 720     return result;
 721 }
 722
 723 /* --------------------------------------------------------------------------------------------- */
 724 /*** public functions ****************************************************************************/
 725 /* --------------------------------------------------------------------------------------------- */
 726
 727 struct str_class
 728 str_ascii_init (void)
 729 {
 730     struct str_class result;
 731
 732     result.conv_gerror_message = str_ascii_conv_gerror_message;
 733     result.vfs_convert_to = str_ascii_vfs_convert_to;
 734     result.insert_replace_char = str_ascii_insert_replace_char;
 735     result.is_valid_string = str_ascii_is_valid_string;
 736     result.is_valid_char = str_ascii_is_valid_char;
 737     result.cnext_char = str_ascii_cnext_char;
 738     result.cprev_char = str_ascii_cprev_char;
 739     result.cnext_char_safe = str_ascii_cnext_char;
 740     result.cprev_char_safe = str_ascii_cprev_char;
 741     result.cnext_noncomb_char = str_ascii_cnext_noncomb_char;
 742     result.cprev_noncomb_char = str_ascii_cprev_noncomb_char;
 743     result.char_isspace = str_ascii_isspace;
 744     result.char_ispunct = str_ascii_ispunct;
 745     result.char_isalnum = str_ascii_isalnum;
 746     result.char_isdigit = str_ascii_isdigit;
 747     result.char_isprint = str_ascii_isprint;
 748     result.char_iscombiningmark = str_ascii_iscombiningmark;
 749     result.char_toupper = str_ascii_toupper;
 750     result.char_tolower = str_ascii_tolower;
 751     result.length = str_ascii_length;
 752     result.length2 = str_ascii_length2;
 753     result.length_noncomb = str_ascii_length;
 754     result.fix_string = str_ascii_fix_string;
 755     result.term_form = str_ascii_term_form;
 756     result.fit_to_term = str_ascii_fit_to_term;
 757     result.term_trim = str_ascii_term_trim;
 758     result.term_width2 = str_ascii_term_width2;
 759     result.term_width1 = str_ascii_term_width1;
 760     result.term_char_width = str_ascii_term_char_width;
 761     result.term_substring = str_ascii_term_substring;
 762     result.trunc = str_ascii_trunc;
 763     result.offset_to_pos = str_ascii_offset_to_pos;
 764     result.column_to_pos = str_ascii_column_to_pos;
 765     result.create_search_needle = str_ascii_create_search_needle;
 766     result.release_search_needle = str_ascii_release_search_needle;
 767     result.search_first = str_ascii_search_first;
 768     result.search_last = str_ascii_search_last;
 769     result.compare = str_ascii_compare;
 770     result.ncompare = str_ascii_ncompare;
 771     result.casecmp = str_ascii_casecmp;
 772     result.ncasecmp = str_ascii_ncasecmp;
 773     result.prefix = str_ascii_prefix;
 774     result.caseprefix = str_ascii_caseprefix;
 775     result.create_key = str_ascii_create_key;
 776     result.create_key_for_filename = str_ascii_create_key;
 777     result.key_collate = str_ascii_key_collate;
 778     result.release_key = str_ascii_release_key;
 779
 780     return result;
 781 }
 782
 783 /* --------------------------------------------------------------------------------------------- */