usr.bin/col/col.c

   1 /*-
   2  * Copyright (c) 1990, 1993, 1994
   3  *      The Regents of the University of California.  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
  37 #ifndef lint
  38 static char copyright[] =
  39 "@(#) Copyright (c) 1990, 1993, 1994\n\
  40         The Regents of the University of California.  All rights reserved.\n";
  41 #endif /* not lint */
  42
  43 #ifndef lint
  44 static char sccsid[] = "@(#)col.c       8.3 (Berkeley) 4/2/94";
  45 #endif /* not lint */
  46
  47 #include <ctype.h>
  48 #include <err.h>
  49 #include <string.h>
  50 #include <stdio.h>
  51 #include <stdlib.h>
  52
  53 #define BS      '\b'            /* backspace */
  54 #define TAB     '\t'            /* tab */
  55 #define SPACE   ' '             /* space */
  56 #define NL      '\n'            /* newline */
  57 #define CR      '\r'            /* carriage return */
  58 #define ESC     '\033'          /* escape */
  59 #define SI      '\017'          /* shift in to normal character set */
  60 #define SO      '\016'          /* shift out to alternate character set */
  61 #define VT      '\013'          /* vertical tab (aka reverse line feed) */
  62 #define RLF     '\007'          /* ESC-07 reverse line feed */
  63 #define RHLF    '\010'          /* ESC-010 reverse half-line feed */
  64 #define FHLF    '\011'          /* ESC-011 forward half-line feed */
  65
  66 /* build up at least this many lines before flushing them out */
  67 #define BUFFER_MARGIN           32
  68
  69 typedef char CSET;
  70
  71 typedef struct char_str {
  72 #define CS_NORMAL       1
  73 #define CS_ALTERNATE    2
  74         short           c_column;       /* column character is in */
  75         CSET            c_set;          /* character set (currently only 2) */
  76         char            c_char;         /* character in question */
  77 } CHAR;
  78
  79 typedef struct line_str LINE;
  80 struct line_str {
  81         CHAR    *l_line;                /* characters on the line */
  82         LINE    *l_prev;                /* previous line */
  83         LINE    *l_next;                /* next line */
  84         int     l_lsize;                /* allocated sizeof l_line */
  85         int     l_line_len;             /* strlen(l_line) */
  86         int     l_needs_sort;           /* set if chars went in out of order */
  87         int     l_max_col;              /* max column in the line */
  88 };
  89
  90 LINE   *alloc_line __P((void));
  91 void    dowarn __P((int));
  92 void    flush_line __P((LINE *));
  93 void    flush_lines __P((int));
  94 void    flush_blanks __P((void));
  95 void    free_line __P((LINE *));
  96 void    usage __P((void));
  97 void    wrerr __P((void));
  98 void   *xmalloc __P((void *, size_t));
  99
 100 CSET    last_set;               /* char_set of last char printed */
 101 LINE   *lines;
 102 int     compress_spaces;        /* if doing space -> tab conversion */
 103 int     fine;                   /* if `fine' resolution (half lines) */
 104 int     max_bufd_lines;         /* max # lines to keep in memory */
 105 int     nblank_lines;           /* # blanks after last flushed line */
 106 int     no_backspaces;          /* if not to output any backspaces */
 107
 108 #define PUTC(ch) \
 109         if (putchar(ch) == EOF) \
 110                 wrerr();
 111
 112 int
 113 main(argc, argv)
 114         int argc;
 115         char **argv;
 116 {
 117         int ch;
 118         CHAR *c;
 119         CSET cur_set;                   /* current character set */
 120         LINE *l;                        /* current line */
 121         int extra_lines;                /* # of lines above first line */
 122         int cur_col;                    /* current column */
 123         int cur_line;                   /* line number of current position */
 124         int max_line;                   /* max value of cur_line */
 125         int this_line;                  /* line l points to */
 126         int nflushd_lines;              /* number of lines that were flushed */
 127         int adjust, opt, warned;
 128
 129         max_bufd_lines = 128;
 130         compress_spaces = 1;            /* compress spaces into tabs */
 131         while ((opt = getopt(argc, argv, "bfhl:x")) != EOF)
 132                 switch (opt) {
 133                 case 'b':               /* do not output backspaces */
 134                         no_backspaces = 1;
 135                         break;
 136                 case 'f':               /* allow half forward line feeds */
 137                         fine = 1;
 138                         break;
 139                 case 'h':               /* compress spaces into tabs */
 140                         compress_spaces = 1;
 141                         break;
 142                 case 'l':               /* buffered line count */
 143                         if ((max_bufd_lines = atoi(optarg)) <= 0) {
 144                                 (void)fprintf(stderr,
 145                                     "col: bad -l argument %s.\n", optarg);
 146                                 exit(1);
 147                         }
 148                         break;
 149                 case 'x':               /* do not compress spaces into tabs */
 150                         compress_spaces = 0;
 151                         break;
 152                 case '?':
 153                 default:
 154                         usage();
 155                 }
 156
 157         if (optind != argc)
 158                 usage();
 159
 160         /* this value is in half lines */
 161         max_bufd_lines *= 2;
 162
 163         adjust = cur_col = extra_lines = warned = 0;
 164         cur_line = max_line = nflushd_lines = this_line = 0;
 165         cur_set = last_set = CS_NORMAL;
 166         lines = l = alloc_line();
 167
 168         while ((ch = getchar()) != EOF) {
 169                 if (!isgraph(ch)) {
 170                         switch (ch) {
 171                         case BS:                /* can't go back further */
 172                                 if (cur_col == 0)
 173                                         continue;
 174                                 --cur_col;
 175                                 continue;
 176                         case CR:
 177                                 cur_col = 0;
 178                                 continue;
 179                         case ESC:               /* just ignore EOF */
 180                                 switch(getchar()) {
 181                                 case RLF:
 182                                         cur_line -= 2;
 183                                         break;
 184                                 case RHLF:
 185                                         cur_line--;
 186                                         break;
 187                                 case FHLF:
 188                                         cur_line++;
 189                                         if (cur_line > max_line)
 190                                                 max_line = cur_line;
 191                                 }
 192                                 continue;
 193                         case NL:
 194                                 cur_line += 2;
 195                                 if (cur_line > max_line)
 196                                         max_line = cur_line;
 197                                 cur_col = 0;
 198                                 continue;
 199                         case SPACE:
 200                                 ++cur_col;
 201                                 continue;
 202                         case SI:
 203                                 cur_set = CS_NORMAL;
 204                                 continue;
 205                         case SO:
 206                                 cur_set = CS_ALTERNATE;
 207                                 continue;
 208                         case TAB:               /* adjust column */
 209                                 cur_col |= 7;
 210                                 ++cur_col;
 211                                 continue;
 212                         case VT:
 213                                 cur_line -= 2;
 214                                 continue;
 215                         }
 216                         continue;
 217                 }
 218
 219                 /* Must stuff ch in a line - are we at the right one? */
 220                 if (cur_line != this_line - adjust) {
 221                         LINE *lnew;
 222                         int nmove;
 223
 224                         adjust = 0;
 225                         nmove = cur_line - this_line;
 226                         if (!fine) {
 227                                 /* round up to next line */
 228                                 if (cur_line & 1) {
 229                                         adjust = 1;
 230                                         nmove++;
 231                                 }
 232                         }
 233                         if (nmove < 0) {
 234                                 for (; nmove < 0 && l->l_prev; nmove++)
 235                                         l = l->l_prev;
 236                                 if (nmove) {
 237                                         if (nflushd_lines == 0) {
 238                                                 /*
 239                                                  * Allow backup past first
 240                                                  * line if nothing has been
 241                                                  * flushed yet.
 242                                                  */
 243                                                 for (; nmove < 0; nmove++) {
 244                                                         lnew = alloc_line();
 245                                                         l->l_prev = lnew;
 246                                                         lnew->l_next = l;
 247                                                         l = lines = lnew;
 248                                                         extra_lines++;
 249                                                 }
 250                                         } else {
 251                                                 if (!warned++)
 252                                                         dowarn(cur_line);
 253                                                 cur_line -= nmove;
 254                                         }
 255                                 }
 256                         } else {
 257                                 /* may need to allocate here */
 258                                 for (; nmove > 0 && l->l_next; nmove--)
 259                                         l = l->l_next;
 260                                 for (; nmove > 0; nmove--) {
 261                                         lnew = alloc_line();
 262                                         lnew->l_prev = l;
 263                                         l->l_next = lnew;
 264                                         l = lnew;
 265                                 }
 266                         }
 267                         this_line = cur_line + adjust;
 268                         nmove = this_line - nflushd_lines;
 269                         if (nmove >= max_bufd_lines + BUFFER_MARGIN) {
 270                                 nflushd_lines += nmove - max_bufd_lines;
 271                                 flush_lines(nmove - max_bufd_lines);
 272                         }
 273                 }
 274                 /* grow line's buffer? */
 275                 if (l->l_line_len + 1 >= l->l_lsize) {
 276                         int need;
 277
 278                         need = l->l_lsize ? l->l_lsize * 2 : 90;
 279                         l->l_line = (CHAR *)xmalloc((void *) l->l_line,
 280                             (unsigned) need * sizeof(CHAR));
 281                         l->l_lsize = need;
 282                 }
 283                 c = &l->l_line[l->l_line_len++];
 284                 c->c_char = ch;
 285                 c->c_set = cur_set;
 286                 c->c_column = cur_col;
 287                 /*
 288                  * If things are put in out of order, they will need sorting
 289                  * when it is flushed.
 290                  */
 291                 if (cur_col < l->l_max_col)
 292                         l->l_needs_sort = 1;
 293                 else
 294                         l->l_max_col = cur_col;
 295                 cur_col++;
 296         }
 297         /* goto the last line that had a character on it */
 298         for (; l->l_next; l = l->l_next)
 299                 this_line++;
 300         flush_lines(this_line - nflushd_lines + extra_lines + 1);
 301
 302         /* make sure we leave things in a sane state */
 303         if (last_set != CS_NORMAL)
 304                 PUTC('\017');
 305
 306         /* flush out the last few blank lines */
 307         nblank_lines = max_line - this_line;
 308         if (max_line & 1)
 309                 nblank_lines++;
 310         else if (!nblank_lines)
 311                 /* missing a \n on the last line? */
 312                 nblank_lines = 2;
 313         flush_blanks();
 314         exit(0);
 315 }
 316
 317 void
 318 flush_lines(nflush)
 319         int nflush;
 320 {
 321         LINE *l;
 322
 323         while (--nflush >= 0) {
 324                 l = lines;
 325                 lines = l->l_next;
 326                 if (l->l_line) {
 327                         flush_blanks();
 328                         flush_line(l);
 329                 }
 330                 nblank_lines++;
 331                 if (l->l_line)
 332                         (void)free((void *)l->l_line);
 333                 free_line(l);
 334         }
 335         if (lines)
 336                 lines->l_prev = NULL;
 337 }
 338
 339 /*
 340  * Print a number of newline/half newlines.  If fine flag is set, nblank_lines
 341  * is the number of half line feeds, otherwise it is the number of whole line
 342  * feeds.
 343  */
 344 void
 345 flush_blanks()
 346 {
 347         int half, i, nb;
 348
 349         half = 0;
 350         nb = nblank_lines;
 351         if (nb & 1) {
 352                 if (fine)
 353                         half = 1;
 354                 else
 355                         nb++;
 356         }
 357         nb /= 2;
 358         for (i = nb; --i >= 0;)
 359                 PUTC('\n');
 360         if (half) {
 361                 PUTC('\033');
 362                 PUTC('9');
 363                 if (!nb)
 364                         PUTC('\r');
 365         }
 366         nblank_lines = 0;
 367 }
 368
 369 /*
 370  * Write a line to stdout taking care of space to tab conversion (-h flag)
 371  * and character set shifts.
 372  */
 373 void
 374 flush_line(l)
 375         LINE *l;
 376 {
 377         CHAR *c, *endc;
 378         int nchars, last_col, this_col;
 379
 380         last_col = 0;
 381         nchars = l->l_line_len;
 382
 383         if (l->l_needs_sort) {
 384                 static CHAR *sorted;
 385                 static int count_size, *count, i, save, sorted_size, tot;
 386
 387                 /*
 388                  * Do an O(n) sort on l->l_line by column being careful to
 389                  * preserve the order of characters in the same column.
 390                  */
 391                 if (l->l_lsize > sorted_size) {
 392                         sorted_size = l->l_lsize;
 393                         sorted = (CHAR *)xmalloc((void *)sorted,
 394                             (unsigned)sizeof(CHAR) * sorted_size);
 395                 }
 396                 if (l->l_max_col >= count_size) {
 397                         count_size = l->l_max_col + 1;
 398                         count = (int *)xmalloc((void *)count,
 399                             (unsigned)sizeof(int) * count_size);
 400                 }
 401                 memset((char *)count, 0, sizeof(int) * l->l_max_col + 1);
 402                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 403                         count[c->c_column]++;
 404
 405                 /*
 406                  * calculate running total (shifted down by 1) to use as
 407                  * indices into new line.
 408                  */
 409                 for (tot = 0, i = 0; i <= l->l_max_col; i++) {
 410                         save = count[i];
 411                         count[i] = tot;
 412                         tot += save;
 413                 }
 414
 415                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 416                         sorted[count[c->c_column]++] = *c;
 417                 c = sorted;
 418         } else
 419                 c = l->l_line;
 420         while (nchars > 0) {
 421                 this_col = c->c_column;
 422                 endc = c;
 423                 do {
 424                         ++endc;
 425                 } while (--nchars > 0 && this_col == endc->c_column);
 426
 427                 /* if -b only print last character */
 428                 if (no_backspaces)
 429                         c = endc - 1;
 430
 431                 if (this_col > last_col) {
 432                         int nspace = this_col - last_col;
 433
 434                         if (compress_spaces && nspace > 1) {
 435                                 int ntabs;
 436
 437                                 ntabs = this_col / 8 - last_col / 8;
 438                                 nspace -= ntabs * 8;
 439                                 while (--ntabs >= 0)
 440                                         PUTC('\t');
 441                         }
 442                         while (--nspace >= 0)
 443                                 PUTC(' ');
 444                         last_col = this_col;
 445                 }
 446                 last_col++;
 447
 448                 for (;;) {
 449                         if (c->c_set != last_set) {
 450                                 switch (c->c_set) {
 451                                 case CS_NORMAL:
 452                                         PUTC('\017');
 453                                         break;
 454                                 case CS_ALTERNATE:
 455                                         PUTC('\016');
 456                                 }
 457                                 last_set = c->c_set;
 458                         }
 459                         PUTC(c->c_char);
 460                         if (++c >= endc)
 461                                 break;
 462                         PUTC('\b');
 463                 }
 464         }
 465 }
 466
 467 #define NALLOC 64
 468
 469 static LINE *line_freelist;
 470
 471 LINE *
 472 alloc_line()
 473 {
 474         LINE *l;
 475         int i;
 476
 477         if (!line_freelist) {
 478                 l = (LINE *)xmalloc((void *)NULL, sizeof(LINE) * NALLOC);
 479                 line_freelist = l;
 480                 for (i = 1; i < NALLOC; i++, l++)
 481                         l->l_next = l + 1;
 482                 l->l_next = NULL;
 483         }
 484         l = line_freelist;
 485         line_freelist = l->l_next;
 486
 487         memset(l, 0, sizeof(LINE));
 488         return (l);
 489 }
 490
 491 void
 492 free_line(l)
 493         LINE *l;
 494 {
 495
 496         l->l_next = line_freelist;
 497         line_freelist = l;
 498 }
 499
 500 void *
 501 xmalloc(p, size)
 502         void *p;
 503         size_t size;
 504 {
 505
 506         if (!(p = (void *)realloc(p, size)))
 507                 err(1, NULL);
 508         return (p);
 509 }
 510
 511 void
 512 usage()
 513 {
 514
 515         (void)fprintf(stderr, "usage: col [-bfx] [-l nline]\n");
 516         exit(1);
 517 }
 518
 519 void
 520 wrerr()
 521 {
 522
 523         (void)fprintf(stderr, "col: write error.\n");
 524         exit(1);
 525 }
 526
 527 void
 528 dowarn(line)
 529         int line;
 530 {
 531
 532         warnx("warning: can't back up %s",
 533                 line < 0 ? "past first line" : "-- line already flushed");
 534 }