text-utils/col.c

   1 /*-
   2  * Copyright (c) 1990 The Regents of the University of California.
   3  * All rights reserved.
   4  *
   5  * This code is derived from software contributed to Berkeley by
   6  * Michael Rendell of the Memorial University of Newfoundland.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  * 3. All advertising materials mentioning features or use of this software
  17  *    must display the following acknowledgement:
  18  *      This product includes software developed by the University of
  19  *      California, Berkeley and its contributors.
  20  * 4. Neither the name of the University nor the names of its contributors
  21  *    may be used to endorse or promote products derived from this software
  22  *    without specific prior written permission.
  23  *
  24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  34  * SUCH DAMAGE.
  35  *
  36  * Wed Jun 22 22:15:41 1994, faith@cs.unc.edu: Added internationalization
  37  *                           patches from Andries.Brouwer@cwi.nl
  38  * Wed Sep 14 22:31:17 1994: patches from Carl Christofferson
  39  *                           (cchris@connected.com)
  40  * 1999-02-22 Arkadiusz Mi¶kiewicz <misiek@pld.ORG.PL>
  41  *      added Native Language Support
  42  * 1999-09-19 Bruno Haible <haible@clisp.cons.org>
  43  *      modified to work correctly in multi-byte locales
  44  *
  45  */
  46
  47 #include <stdlib.h>
  48 #include <errno.h>
  49 #include <ctype.h>
  50 #include <string.h>
  51 #include <stdio.h>
  52 #include <unistd.h>
  53 #include "nls.h"
  54
  55 #include "widechar.h"
  56
  57 #define BS      '\b'            /* backspace */
  58 #define TAB     '\t'            /* tab */
  59 #define SPACE   ' '             /* space */
  60 #define NL      '\n'            /* newline */
  61 #define CR      '\r'            /* carriage return */
  62 #define ESC     '\033'          /* escape */
  63 #define SI      '\017'          /* shift in to normal character set */
  64 #define SO      '\016'          /* shift out to alternate character set */
  65 #define VT      '\013'          /* vertical tab (aka reverse line feed) */
  66 #define RLF     '\007'          /* ESC-07 reverse line feed */
  67 #define RHLF    '\010'          /* ESC-010 reverse half-line feed */
  68 #define FHLF    '\011'          /* ESC-011 forward half-line feed */
  69
  70 /* build up at least this many lines before flushing them out */
  71 #define BUFFER_MARGIN           32
  72
  73 typedef char CSET;
  74
  75 typedef struct char_str {
  76 #define CS_NORMAL       1
  77 #define CS_ALTERNATE    2
  78         short           c_column;       /* column character is in */
  79         CSET            c_set;          /* character set (currently only 2) */
  80         wchar_t         c_char;         /* character in question */
  81         int             c_width;        /* character width */
  82 } CHAR;
  83
  84 typedef struct line_str LINE;
  85 struct line_str {
  86         CHAR    *l_line;                /* characters on the line */
  87         LINE    *l_prev;                /* previous line */
  88         LINE    *l_next;                /* next line */
  89         int     l_lsize;                /* allocated sizeof l_line */
  90         int     l_line_len;             /* strlen(l_line) */
  91         int     l_needs_sort;           /* set if chars went in out of order */
  92         int     l_max_col;              /* max column in the line */
  93 };
  94
  95 void usage(void);
  96 void wrerr(void);
  97 void warn(int);
  98 void free_line(LINE *l);
  99 void flush_line(LINE *l);
 100 void flush_lines(int);
 101 void flush_blanks(void);
 102 void *xmalloc(void *p, size_t size);
 103 LINE *alloc_line(void);
 104
 105 CSET last_set;                  /* char_set of last char printed */
 106 LINE *lines;
 107 int compress_spaces;            /* if doing space -> tab conversion */
 108 int fine;                       /* if `fine' resolution (half lines) */
 109 int max_bufd_lines;             /* max # lines to keep in memory */
 110 int nblank_lines;               /* # blanks after last flushed line */
 111 int no_backspaces;              /* if not to output any backspaces */
 112 int pass_unknown_seqs;          /* whether to pass unknown control sequences */
 113
 114 #define PUTC(ch) \
 115         if (putwchar(ch) == WEOF) \
 116                 wrerr();
 117
 118 int main(int argc, char **argv)
 119 {
 120         register wint_t ch;
 121         CHAR *c;
 122         CSET cur_set;                   /* current character set */
 123         LINE *l;                        /* current line */
 124         int extra_lines;                /* # of lines above first line */
 125         int cur_col;                    /* current column */
 126         int cur_line;                   /* line number of current position */
 127         int max_line;                   /* max value of cur_line */
 128         int this_line;                  /* line l points to */
 129         int nflushd_lines;              /* number of lines that were flushed */
 130         int adjust, opt, warned;
 131         int ret = 0;
 132
 133         setlocale(LC_ALL, "");
 134         bindtextdomain(PACKAGE, LOCALEDIR);
 135         textdomain(PACKAGE);
 136
 137         max_bufd_lines = 128;
 138         compress_spaces = 1;            /* compress spaces into tabs */
 139         pass_unknown_seqs = 0;          /* remove unknown escape sequences */
 140         while ((opt = getopt(argc, argv, "bfhl:px")) != -1)
 141                 switch (opt) {
 142                 case 'b':               /* do not output backspaces */
 143                         no_backspaces = 1;
 144                         break;
 145                 case 'f':               /* allow half forward line feeds */
 146                         fine = 1;
 147                         break;
 148                 case 'h':               /* compress spaces into tabs */
 149                         compress_spaces = 1;
 150                         break;
 151                 case 'l':               /* buffered line count */
 152                         if ((max_bufd_lines = atoi(optarg)) <= 0) {
 153                                 (void)fprintf(stderr,
 154                                     _("col: bad -l argument %s.\n"), optarg);
 155                                 exit(1);
 156                         }
 157                         break;
 158                 case 'p':
 159                         pass_unknown_seqs = 1;
 160                         break;
 161                 case 'x':               /* do not compress spaces into tabs */
 162                         compress_spaces = 0;
 163                         break;
 164                 case '?':
 165                 default:
 166                         usage();
 167                 }
 168
 169         if (optind != argc)
 170                 usage();
 171
 172         /* this value is in half lines */
 173         max_bufd_lines *= 2;
 174
 175         adjust = cur_col = extra_lines = warned = 0;
 176         cur_line = max_line = nflushd_lines = this_line = 0;
 177         cur_set = last_set = CS_NORMAL;
 178         lines = l = alloc_line();
 179
 180         while (feof(stdin)==0) {
 181                 errno = 0;
 182                 if ((ch = getwchar()) == WEOF) {
 183                         if (errno==EILSEQ) {
 184                                 perror("col");
 185                                 ret = 1;
 186                         }
 187                         break;
 188                 }
 189                 if (!iswgraph(ch)) {
 190                         switch (ch) {
 191                         case BS:                /* can't go back further */
 192                                 if (cur_col == 0)
 193                                         continue;
 194                                 --cur_col;
 195                                 continue;
 196                         case CR:
 197                                 cur_col = 0;
 198                                 continue;
 199                         case ESC:               /* just ignore EOF */
 200                                 switch(getwchar()) {
 201                                 case RLF:
 202                                         cur_line -= 2;
 203                                         break;
 204                                 case RHLF:
 205                                         cur_line--;
 206                                         break;
 207                                 case FHLF:
 208                                         cur_line++;
 209                                         if (cur_line > max_line)
 210                                                 max_line = cur_line;
 211                                 }
 212                                 continue;
 213                         case NL:
 214                                 cur_line += 2;
 215                                 if (cur_line > max_line)
 216                                         max_line = cur_line;
 217                                 cur_col = 0;
 218                                 continue;
 219                         case SPACE:
 220                                 ++cur_col;
 221                                 continue;
 222                         case SI:
 223                                 cur_set = CS_NORMAL;
 224                                 continue;
 225                         case SO:
 226                                 cur_set = CS_ALTERNATE;
 227                                 continue;
 228                         case TAB:               /* adjust column */
 229                                 cur_col |= 7;
 230                                 ++cur_col;
 231                                 continue;
 232                         case VT:
 233                                 cur_line -= 2;
 234                                 continue;
 235                         }
 236                         if (iswspace(ch)) {
 237                                 if (wcwidth(ch) > 0)
 238                                         cur_col += wcwidth(ch);
 239                                 continue;
 240                         }
 241                         if (!pass_unknown_seqs)
 242                                 continue;
 243                 }
 244
 245                 /* Must stuff ch in a line - are we at the right one? */
 246                 if (cur_line != this_line - adjust) {
 247                         LINE *lnew;
 248                         int nmove;
 249
 250                         adjust = 0;
 251                         nmove = cur_line - this_line;
 252                         if (!fine) {
 253                                 /* round up to next line */
 254                                 if (cur_line & 1) {
 255                                         adjust = 1;
 256                                         nmove++;
 257                                 }
 258                         }
 259                         if (nmove < 0) {
 260                                 for (; nmove < 0 && l->l_prev; nmove++)
 261                                         l = l->l_prev;
 262                                 if (nmove) {
 263                                         if (nflushd_lines == 0) {
 264                                                 /*
 265                                                  * Allow backup past first
 266                                                  * line if nothing has been
 267                                                  * flushed yet.
 268                                                  */
 269                                                 for (; nmove < 0; nmove++) {
 270                                                         lnew = alloc_line();
 271                                                         l->l_prev = lnew;
 272                                                         lnew->l_next = l;
 273                                                         l = lines = lnew;
 274                                                         extra_lines++;
 275                                                 }
 276                                         } else {
 277                                                 if (!warned++)
 278                                                         warn(cur_line);
 279                                                 cur_line -= nmove;
 280                                         }
 281                                 }
 282                         } else {
 283                                 /* may need to allocate here */
 284                                 for (; nmove > 0 && l->l_next; nmove--)
 285                                         l = l->l_next;
 286                                 for (; nmove > 0; nmove--) {
 287                                         lnew = alloc_line();
 288                                         lnew->l_prev = l;
 289                                         l->l_next = lnew;
 290                                         l = lnew;
 291                                 }
 292                         }
 293                         this_line = cur_line + adjust;
 294                         nmove = this_line - nflushd_lines;
 295                         if (nmove >= max_bufd_lines + BUFFER_MARGIN) {
 296                                 nflushd_lines += nmove - max_bufd_lines;
 297                                 flush_lines(nmove - max_bufd_lines);
 298                         }
 299                 }
 300                 /* grow line's buffer? */
 301                 if (l->l_line_len + 1 >= l->l_lsize) {
 302                         int need;
 303
 304                         need = l->l_lsize ? l->l_lsize * 2 : 90;
 305                         l->l_line = (CHAR *)xmalloc((void *) l->l_line,
 306                             (unsigned) need * sizeof(CHAR));
 307                         l->l_lsize = need;
 308                 }
 309                 c = &l->l_line[l->l_line_len++];
 310                 c->c_char = ch;
 311                 c->c_set = cur_set;
 312                 c->c_column = cur_col;
 313                 c->c_width = wcwidth(ch);
 314                 /*
 315                  * If things are put in out of order, they will need sorting
 316                  * when it is flushed.
 317                  */
 318                 if (cur_col < l->l_max_col)
 319                         l->l_needs_sort = 1;
 320                 else
 321                         l->l_max_col = cur_col;
 322                 if (c->c_width > 0)
 323                         cur_col += c->c_width;
 324         }
 325         /* goto the last line that had a character on it */
 326         for (; l->l_next; l = l->l_next)
 327                 this_line++;
 328         flush_lines(this_line - nflushd_lines + extra_lines + 1);
 329
 330         /* make sure we leave things in a sane state */
 331         if (last_set != CS_NORMAL)
 332                 PUTC('\017');
 333
 334         /* flush out the last few blank lines */
 335         nblank_lines = max_line - this_line;
 336         if (max_line & 1)
 337                 nblank_lines++;
 338         else if (!nblank_lines)
 339                 /* missing a \n on the last line? */
 340                 nblank_lines = 2;
 341         flush_blanks();
 342         if (ferror(stdout) || fclose(stdout))
 343                 return 1;
 344         return ret;
 345 }
 346
 347 void flush_lines(int nflush)
 348 {
 349         LINE *l;
 350
 351         while (--nflush >= 0) {
 352                 l = lines;
 353                 lines = l->l_next;
 354                 if (l->l_line) {
 355                         flush_blanks();
 356                         flush_line(l);
 357                 }
 358                 nblank_lines++;
 359                 if (l->l_line)
 360                         (void)free((void *)l->l_line);
 361                 free_line(l);
 362         }
 363         if (lines)
 364                 lines->l_prev = NULL;
 365 }
 366
 367 /*
 368  * Print a number of newline/half newlines.  If fine flag is set, nblank_lines
 369  * is the number of half line feeds, otherwise it is the number of whole line
 370  * feeds.
 371  */
 372 void flush_blanks()
 373 {
 374         int half, i, nb;
 375
 376         half = 0;
 377         nb = nblank_lines;
 378         if (nb & 1) {
 379                 if (fine)
 380                         half = 1;
 381                 else
 382                         nb++;
 383         }
 384         nb /= 2;
 385         for (i = nb; --i >= 0;)
 386                 PUTC('\n');
 387         if (half) {
 388                 PUTC('\033');
 389                 PUTC('9');
 390                 if (!nb)
 391                         PUTC('\r');
 392         }
 393         nblank_lines = 0;
 394 }
 395
 396 /*
 397  * Write a line to stdout taking care of space to tab conversion (-h flag)
 398  * and character set shifts.
 399  */
 400 void flush_line(LINE *l)
 401 {
 402         CHAR *c, *endc;
 403         int nchars, last_col, this_col;
 404
 405         last_col = 0;
 406         nchars = l->l_line_len;
 407
 408         if (l->l_needs_sort) {
 409                 static CHAR *sorted;
 410                 static int count_size, *count, i, save, sorted_size, tot;
 411
 412                 /*
 413                  * Do an O(n) sort on l->l_line by column being careful to
 414                  * preserve the order of characters in the same column.
 415                  */
 416                 if (l->l_lsize > sorted_size) {
 417                         sorted_size = l->l_lsize;
 418                         sorted = (CHAR *)xmalloc((void *)sorted,
 419                             (unsigned)sizeof(CHAR) * sorted_size);
 420                 }
 421                 if (l->l_max_col >= count_size) {
 422                         count_size = l->l_max_col + 1;
 423                         count = (int *)xmalloc((void *)count,
 424                             (unsigned)sizeof(int) * count_size);
 425                 }
 426                 memset(count, 0, sizeof(int) * l->l_max_col + 1);
 427                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 428                         count[c->c_column]++;
 429
 430                 /*
 431                  * calculate running total (shifted down by 1) to use as
 432                  * indices into new line.
 433                  */
 434                 for (tot = 0, i = 0; i <= l->l_max_col; i++) {
 435                         save = count[i];
 436                         count[i] = tot;
 437                         tot += save;
 438                 }
 439
 440                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 441                         sorted[count[c->c_column]++] = *c;
 442                 c = sorted;
 443         } else
 444                 c = l->l_line;
 445         while (nchars > 0) {
 446                 this_col = c->c_column;
 447                 endc = c;
 448                 do {
 449                         ++endc;
 450                 } while (--nchars > 0 && this_col == endc->c_column);
 451
 452                 /* if -b only print last character */
 453                 if (no_backspaces) {
 454                         c = endc - 1;
 455                         if (nchars > 0 &&
 456                             this_col + c->c_width > endc->c_column)
 457                                 continue;
 458                 }
 459
 460                 if (this_col > last_col) {
 461                         int nspace = this_col - last_col;
 462
 463                         if (compress_spaces && nspace > 1) {
 464                                 int ntabs;
 465
 466                                 ntabs = this_col / 8 - last_col / 8;
 467                                 if (ntabs > 0) {
 468                                         nspace = this_col & 7;
 469                                         while (--ntabs >= 0)
 470                                                 PUTC('\t');
 471                                 }
 472                         }
 473                         while (--nspace >= 0)
 474                                 PUTC(' ');
 475                         last_col = this_col;
 476                 }
 477
 478                 for (;;) {
 479                         if (c->c_set != last_set) {
 480                                 switch (c->c_set) {
 481                                 case CS_NORMAL:
 482                                         PUTC('\017');
 483                                         break;
 484                                 case CS_ALTERNATE:
 485                                         PUTC('\016');
 486                                 }
 487                                 last_set = c->c_set;
 488                         }
 489                         PUTC(c->c_char);
 490                         if ((c+1) < endc) {
 491                                 int i;
 492                                 for (i=0; i < c->c_width; i++)
 493                                         PUTC('\b');
 494                         }
 495                         if (++c >= endc)
 496                                 break;
 497                 }
 498                 last_col += (c-1)->c_width;
 499         }
 500 }
 501
 502 #define NALLOC 64
 503
 504 static LINE *line_freelist;
 505
 506 LINE *
 507 alloc_line()
 508 {
 509         LINE *l;
 510         int i;
 511
 512         if (!line_freelist) {
 513                 l = (LINE *)xmalloc((void *)NULL, sizeof(LINE) * NALLOC);
 514                 line_freelist = l;
 515                 for (i = 1; i < NALLOC; i++, l++)
 516                         l->l_next = l + 1;
 517                 l->l_next = NULL;
 518         }
 519         l = line_freelist;
 520         line_freelist = l->l_next;
 521
 522         memset(l, 0, sizeof(LINE));
 523         return(l);
 524 }
 525
 526 void free_line(LINE *l)
 527 {
 528         l->l_next = line_freelist;
 529         line_freelist = l;
 530 }
 531
 532 void *
 533 xmalloc(void *p, size_t size)
 534 {
 535         if (!(p = (void *)realloc(p, size))) {
 536                 (void)fprintf(stderr, "col: %s.\n", strerror(ENOMEM));
 537                 exit(1);
 538         }
 539         return(p);
 540 }
 541
 542 void usage()
 543 {
 544         (void)fprintf(stderr, _("usage: col [-bfpx] [-l nline]\n"));
 545         exit(1);
 546 }
 547
 548 void wrerr()
 549 {
 550         (void)fprintf(stderr, _("col: write error.\n"));
 551         exit(1);
 552 }
 553
 554 void warn(int line)
 555 {
 556         (void)fprintf(stderr,
 557             _("col: warning: can't back up %s.\n"), line < 0 ?
 558             _("past first line") : _("-- line already flushed"));
 559 }