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 const char copyright[] =
  39 "@(#) Copyright (c) 1990, 1993, 1994\n\
  40         The Regents of the University of California.  All rights reserved.\n";
  41 #endif
  42
  43 #if 0
  44 #ifndef lint
  45 static char sccsid[] = "@(#)col.c       8.5 (Berkeley) 5/4/95";
  46 #endif
  47 #endif
  48
  49 #include <sys/cdefs.h>
  50 __FBSDID("$FreeBSD$");
  51
  52 #include <ctype.h>
  53 #include <err.h>
  54 #include <string.h>
  55 #include <stdio.h>
  56 #include <stdlib.h>
  57 #include <unistd.h>
  58 #include <locale.h>
  59
  60 #define BS      '\b'            /* backspace */
  61 #define TAB     '\t'            /* tab */
  62 #define SPACE   ' '             /* space */
  63 #define NL      '\n'            /* newline */
  64 #define CR      '\r'            /* carriage return */
  65 #define ESC     '\033'          /* escape */
  66 #define SI      '\017'          /* shift in to normal character set */
  67 #define SO      '\016'          /* shift out to alternate character set */
  68 #define VT      '\013'          /* vertical tab (aka reverse line feed) */
  69 #define RLF     '\007'          /* ESC-07 reverse line feed */
  70 #define RHLF    '\010'          /* ESC-010 reverse half-line feed */
  71 #define FHLF    '\011'          /* ESC-011 forward half-line feed */
  72
  73 /* build up at least this many lines before flushing them out */
  74 #define BUFFER_MARGIN           32
  75
  76 typedef char CSET;
  77
  78 typedef struct char_str {
  79 #define CS_NORMAL       1
  80 #define CS_ALTERNATE    2
  81         short           c_column;       /* column character is in */
  82         CSET            c_set;          /* character set (currently only 2) */
  83         char            c_char;         /* character in question */
  84 } CHAR;
  85
  86 typedef struct line_str LINE;
  87 struct line_str {
  88         CHAR    *l_line;                /* characters on the line */
  89         LINE    *l_prev;                /* previous line */
  90         LINE    *l_next;                /* next line */
  91         int     l_lsize;                /* allocated sizeof l_line */
  92         int     l_line_len;             /* strlen(l_line) */
  93         int     l_needs_sort;           /* set if chars went in out of order */
  94         int     l_max_col;              /* max column in the line */
  95 };
  96
  97 LINE   *alloc_line(void);
  98 void    dowarn(int);
  99 void    flush_line(LINE *);
 100 void    flush_lines(int);
 101 void    flush_blanks(void);
 102 void    free_line(LINE *);
 103 int     main(int, char **);
 104 void    usage(void);
 105
 106 CSET    last_set;               /* char_set of last char printed */
 107 LINE   *lines;
 108 int     compress_spaces;        /* if doing space -> tab conversion */
 109 int     fine;                   /* if `fine' resolution (half lines) */
 110 int     max_bufd_lines;         /* max # lines to keep in memory */
 111 int     nblank_lines;           /* # blanks after last flushed line */
 112 int     no_backspaces;          /* if not to output any backspaces */
 113 int     pass_unknown_seqs;      /* pass unknown control sequences */
 114
 115 #define PUTC(ch) \
 116         do {                                    \
 117                 if (putchar(ch) == EOF)         \
 118                         errx(1, "write error"); \
 119         } while (0)
 120
 121 int
 122 main(argc, argv)
 123         int argc;
 124         char **argv;
 125 {
 126         int ch;
 127         CHAR *c;
 128         CSET cur_set;                   /* current character set */
 129         LINE *l;                        /* current line */
 130         int extra_lines;                /* # of lines above first line */
 131         int cur_col;                    /* current column */
 132         int cur_line;                   /* line number of current position */
 133         int max_line;                   /* max value of cur_line */
 134         int this_line;                  /* line l points to */
 135         int nflushd_lines;              /* number of lines that were flushed */
 136         int adjust, opt, warned;
 137
 138         (void)setlocale(LC_CTYPE, "");
 139
 140         max_bufd_lines = 128;
 141         compress_spaces = 1;            /* compress spaces into tabs */
 142         while ((opt = getopt(argc, argv, "bfhl:px")) != -1)
 143                 switch (opt) {
 144                 case 'b':               /* do not output backspaces */
 145                         no_backspaces = 1;
 146                         break;
 147                 case 'f':               /* allow half forward line feeds */
 148                         fine = 1;
 149                         break;
 150                 case 'h':               /* compress spaces into tabs */
 151                         compress_spaces = 1;
 152                         break;
 153                 case 'l':               /* buffered line count */
 154                         if ((max_bufd_lines = atoi(optarg)) <= 0)
 155                                 errx(1, "bad -l argument %s", optarg);
 156                         break;
 157                 case 'p':               /* pass unknown control sequences */
 158                         pass_unknown_seqs = 1;
 159                         break;
 160                 case 'x':               /* do not compress spaces into tabs */
 161                         compress_spaces = 0;
 162                         break;
 163                 case '?':
 164                 default:
 165                         usage();
 166                 }
 167
 168         if (optind != argc)
 169                 usage();
 170
 171         /* this value is in half lines */
 172         max_bufd_lines *= 2;
 173
 174         adjust = cur_col = extra_lines = warned = 0;
 175         cur_line = max_line = nflushd_lines = this_line = 0;
 176         cur_set = last_set = CS_NORMAL;
 177         lines = l = alloc_line();
 178
 179         while ((ch = getchar()) != EOF) {
 180                 if (!isgraph(ch)) {
 181                         switch (ch) {
 182                         case BS:                /* can't go back further */
 183                                 if (cur_col == 0)
 184                                         continue;
 185                                 --cur_col;
 186                                 continue;
 187                         case CR:
 188                                 cur_col = 0;
 189                                 continue;
 190                         case ESC:               /* just ignore EOF */
 191                                 switch(getchar()) {
 192                                 case RLF:
 193                                         cur_line -= 2;
 194                                         break;
 195                                 case RHLF:
 196                                         cur_line--;
 197                                         break;
 198                                 case FHLF:
 199                                         cur_line++;
 200                                         if (cur_line > max_line)
 201                                                 max_line = cur_line;
 202                                 }
 203                                 continue;
 204                         case NL:
 205                                 cur_line += 2;
 206                                 if (cur_line > max_line)
 207                                         max_line = cur_line;
 208                                 cur_col = 0;
 209                                 continue;
 210                         case SPACE:
 211                                 ++cur_col;
 212                                 continue;
 213                         case SI:
 214                                 cur_set = CS_NORMAL;
 215                                 continue;
 216                         case SO:
 217                                 cur_set = CS_ALTERNATE;
 218                                 continue;
 219                         case TAB:               /* adjust column */
 220                                 cur_col |= 7;
 221                                 ++cur_col;
 222                                 continue;
 223                         case VT:
 224                                 cur_line -= 2;
 225                                 continue;
 226                         }
 227                         if (!pass_unknown_seqs)
 228                                 continue;
 229                 }
 230
 231                 /* Must stuff ch in a line - are we at the right one? */
 232                 if (cur_line != this_line - adjust) {
 233                         LINE *lnew;
 234                         int nmove;
 235
 236                         adjust = 0;
 237                         nmove = cur_line - this_line;
 238                         if (!fine) {
 239                                 /* round up to next line */
 240                                 if (cur_line & 1) {
 241                                         adjust = 1;
 242                                         nmove++;
 243                                 }
 244                         }
 245                         if (nmove < 0) {
 246                                 for (; nmove < 0 && l->l_prev; nmove++)
 247                                         l = l->l_prev;
 248                                 if (nmove) {
 249                                         if (nflushd_lines == 0) {
 250                                                 /*
 251                                                  * Allow backup past first
 252                                                  * line if nothing has been
 253                                                  * flushed yet.
 254                                                  */
 255                                                 for (; nmove < 0; nmove++) {
 256                                                         lnew = alloc_line();
 257                                                         l->l_prev = lnew;
 258                                                         lnew->l_next = l;
 259                                                         l = lines = lnew;
 260                                                         extra_lines++;
 261                                                 }
 262                                         } else {
 263                                                 if (!warned++)
 264                                                         dowarn(cur_line);
 265                                                 cur_line -= nmove;
 266                                         }
 267                                 }
 268                         } else {
 269                                 /* may need to allocate here */
 270                                 for (; nmove > 0 && l->l_next; nmove--)
 271                                         l = l->l_next;
 272                                 for (; nmove > 0; nmove--) {
 273                                         lnew = alloc_line();
 274                                         lnew->l_prev = l;
 275                                         l->l_next = lnew;
 276                                         l = lnew;
 277                                 }
 278                         }
 279                         this_line = cur_line + adjust;
 280                         nmove = this_line - nflushd_lines;
 281                         if (nmove >= max_bufd_lines + BUFFER_MARGIN) {
 282                                 nflushd_lines += nmove - max_bufd_lines;
 283                                 flush_lines(nmove - max_bufd_lines);
 284                         }
 285                 }
 286                 /* grow line's buffer? */
 287                 if (l->l_line_len + 1 >= l->l_lsize) {
 288                         int need;
 289
 290                         need = l->l_lsize ? l->l_lsize * 2 : 90;
 291                         if ((l->l_line = realloc(l->l_line,
 292                             (unsigned)need * sizeof(CHAR))) == NULL)
 293                                 err(1, (char *)NULL);
 294                         l->l_lsize = need;
 295                 }
 296                 c = &l->l_line[l->l_line_len++];
 297                 c->c_char = ch;
 298                 c->c_set = cur_set;
 299                 c->c_column = cur_col;
 300                 /*
 301                  * If things are put in out of order, they will need sorting
 302                  * when it is flushed.
 303                  */
 304                 if (cur_col < l->l_max_col)
 305                         l->l_needs_sort = 1;
 306                 else
 307                         l->l_max_col = cur_col;
 308                 cur_col++;
 309         }
 310         if (max_line == 0)
 311                 exit(0);        /* no lines, so just exit */
 312
 313         /* goto the last line that had a character on it */
 314         for (; l->l_next; l = l->l_next)
 315                 this_line++;
 316         flush_lines(this_line - nflushd_lines + extra_lines + 1);
 317
 318         /* make sure we leave things in a sane state */
 319         if (last_set != CS_NORMAL)
 320                 PUTC('\017');
 321
 322         /* flush out the last few blank lines */
 323         nblank_lines = max_line - this_line;
 324         if (max_line & 1)
 325                 nblank_lines++;
 326         else if (!nblank_lines)
 327                 /* missing a \n on the last line? */
 328                 nblank_lines = 2;
 329         flush_blanks();
 330         exit(0);
 331 }
 332
 333 void
 334 flush_lines(nflush)
 335         int nflush;
 336 {
 337         LINE *l;
 338
 339         while (--nflush >= 0) {
 340                 l = lines;
 341                 lines = l->l_next;
 342                 if (l->l_line) {
 343                         flush_blanks();
 344                         flush_line(l);
 345                 }
 346                 nblank_lines++;
 347                 if (l->l_line)
 348                         (void)free(l->l_line);
 349                 free_line(l);
 350         }
 351         if (lines)
 352                 lines->l_prev = NULL;
 353 }
 354
 355 /*
 356  * Print a number of newline/half newlines.  If fine flag is set, nblank_lines
 357  * is the number of half line feeds, otherwise it is the number of whole line
 358  * feeds.
 359  */
 360 void
 361 flush_blanks()
 362 {
 363         int half, i, nb;
 364
 365         half = 0;
 366         nb = nblank_lines;
 367         if (nb & 1) {
 368                 if (fine)
 369                         half = 1;
 370                 else
 371                         nb++;
 372         }
 373         nb /= 2;
 374         for (i = nb; --i >= 0;)
 375                 PUTC('\n');
 376         if (half) {
 377                 PUTC('\033');
 378                 PUTC('9');
 379                 if (!nb)
 380                         PUTC('\r');
 381         }
 382         nblank_lines = 0;
 383 }
 384
 385 /*
 386  * Write a line to stdout taking care of space to tab conversion (-h flag)
 387  * and character set shifts.
 388  */
 389 void
 390 flush_line(l)
 391         LINE *l;
 392 {
 393         CHAR *c, *endc;
 394         int nchars, last_col, this_col;
 395
 396         last_col = 0;
 397         nchars = l->l_line_len;
 398
 399         if (l->l_needs_sort) {
 400                 static CHAR *sorted;
 401                 static int count_size, *count, i, save, sorted_size, tot;
 402
 403                 /*
 404                  * Do an O(n) sort on l->l_line by column being careful to
 405                  * preserve the order of characters in the same column.
 406                  */
 407                 if (l->l_lsize > sorted_size) {
 408                         sorted_size = l->l_lsize;
 409                         if ((sorted = realloc(sorted,
 410                             (unsigned)sizeof(CHAR) * sorted_size)) == NULL)
 411                                 err(1, (char *)NULL);
 412                 }
 413                 if (l->l_max_col >= count_size) {
 414                         count_size = l->l_max_col + 1;
 415                         if ((count = realloc(count,
 416                             (unsigned)sizeof(int) * count_size)) == NULL)
 417                                 err(1, (char *)NULL);
 418                 }
 419                 memset(count, 0, sizeof(int) * l->l_max_col + 1);
 420                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 421                         count[c->c_column]++;
 422
 423                 /*
 424                  * calculate running total (shifted down by 1) to use as
 425                  * indices into new line.
 426                  */
 427                 for (tot = 0, i = 0; i <= l->l_max_col; i++) {
 428                         save = count[i];
 429                         count[i] = tot;
 430                         tot += save;
 431                 }
 432
 433                 for (i = nchars, c = l->l_line; --i >= 0; c++)
 434                         sorted[count[c->c_column]++] = *c;
 435                 c = sorted;
 436         } else
 437                 c = l->l_line;
 438         while (nchars > 0) {
 439                 this_col = c->c_column;
 440                 endc = c;
 441                 do {
 442                         ++endc;
 443                 } while (--nchars > 0 && this_col == endc->c_column);
 444
 445                 /* if -b only print last character */
 446                 if (no_backspaces)
 447                         c = endc - 1;
 448
 449                 if (this_col > last_col) {
 450                         int nspace = this_col - last_col;
 451
 452                         if (compress_spaces && nspace > 1) {
 453                                 while (1) {
 454                                         int tab_col, tab_size;;
 455
 456                                         tab_col = (last_col + 8) & ~7;
 457                                         if (tab_col > this_col)
 458                                                 break;
 459                                         tab_size = tab_col - last_col;
 460                                         if (tab_size == 1)
 461                                                 PUTC(' ');
 462                                         else
 463                                                 PUTC('\t');
 464                                         nspace -= tab_size;
 465                                         last_col = tab_col;
 466                                 }
 467                         }
 468                         while (--nspace >= 0)
 469                                 PUTC(' ');
 470                         last_col = this_col;
 471                 }
 472                 last_col++;
 473
 474                 for (;;) {
 475                         if (c->c_set != last_set) {
 476                                 switch (c->c_set) {
 477                                 case CS_NORMAL:
 478                                         PUTC('\017');
 479                                         break;
 480                                 case CS_ALTERNATE:
 481                                         PUTC('\016');
 482                                 }
 483                                 last_set = c->c_set;
 484                         }
 485                         PUTC(c->c_char);
 486                         if (++c >= endc)
 487                                 break;
 488                         PUTC('\b');
 489                 }
 490         }
 491 }
 492
 493 #define NALLOC 64
 494
 495 static LINE *line_freelist;
 496
 497 LINE *
 498 alloc_line()
 499 {
 500         LINE *l;
 501         int i;
 502
 503         if (!line_freelist) {
 504                 if ((l = realloc(NULL, sizeof(LINE) * NALLOC)) == NULL)
 505                         err(1, (char *)NULL);
 506                 line_freelist = l;
 507                 for (i = 1; i < NALLOC; i++, l++)
 508                         l->l_next = l + 1;
 509                 l->l_next = NULL;
 510         }
 511         l = line_freelist;
 512         line_freelist = l->l_next;
 513
 514         memset(l, 0, sizeof(LINE));
 515         return (l);
 516 }
 517
 518 void
 519 free_line(l)
 520         LINE *l;
 521 {
 522
 523         l->l_next = line_freelist;
 524         line_freelist = l;
 525 }
 526
 527 void
 528 usage()
 529 {
 530
 531         (void)fprintf(stderr, "usage: col [-bfhpx] [-l nline]\n");
 532         exit(1);
 533 }
 534
 535 void
 536 dowarn(line)
 537         int line;
 538 {
 539
 540         warnx("warning: can't back up %s",
 541                 line < 0 ? "past first line" : "-- line already flushed");
 542 }