[BACK]Return to fmt.c CVS log [TXT][DIR] Up to [local] / src / usr.bin / fmt

Annotation of src/usr.bin/fmt/fmt.c, Revision 1.26

1.26    ! tedu        1: /*     $OpenBSD: fmt.c,v 1.25 2006/11/29 21:59:04 jmc Exp $    */
1.1       deraadt     2:
1.10      millert     3: /* Sensible version of fmt
                      4:  *
                      5:  * Syntax: fmt [ options ] [ goal [ max ] ] [ filename ... ]
                      6:  *
                      7:  * Since the documentation for the original fmt is so poor, here
                      8:  * is an accurate description of what this one does. It's usually
                      9:  * the same. The *mechanism* used may differ from that suggested
                     10:  * here. Note that we are *not* entirely compatible with fmt,
                     11:  * because fmt gets so many things wrong.
                     12:  *
                     13:  * 1. Tabs are expanded, assuming 8-space tab stops.
                     14:  *    If the `-t <n>' option is given, we assume <n>-space
                     15:  *    tab stops instead.
                     16:  *    Trailing blanks are removed from all lines.
                     17:  *    x\b == nothing, for any x other than \b.
                     18:  *    Other control characters are simply stripped. This
                     19:  *    includes \r.
                     20:  * 2. Each line is split into leading whitespace and
                     21:  *    everything else. Maximal consecutive sequences of
                     22:  *    lines with the same leading whitespace are considered
                     23:  *    to form paragraphs, except that a blank line is always
                     24:  *    a paragraph to itself.
                     25:  *    If the `-p' option is given then the first line of a
                     26:  *    paragraph is permitted to have indentation different
                     27:  *    from that of the other lines.
                     28:  *    If the `-m' option is given then a line that looks
                     29:  *    like a mail message header, if it is not immediately
                     30:  *    preceded by a non-blank non-message-header line, is
                     31:  *    taken to start a new paragraph, which also contains
                     32:  *    any subsequent lines with non-empty leading whitespace.
1.19      millert    33:  *    Unless the `-n' option is given, lines beginning with
                     34:  *    a . (dot) are not formatted.
1.10      millert    35:  * 3. The "everything else" is split into words; a word
                     36:  *    includes its trailing whitespace, and a word at the
                     37:  *    end of a line is deemed to be followed by a single
                     38:  *    space, or two spaces if it ends with a sentence-end
                     39:  *    character. (See the `-d' option for how to change that.)
                     40:  *    If the `-s' option has been given, then a word's trailing
                     41:  *    whitespace is replaced by what it would have had if it
                     42:  *    had occurred at end of line.
                     43:  * 4. Each paragraph is sent to standard output as follows.
                     44:  *    We output the leading whitespace, and then enough words
                     45:  *    to make the line length as near as possible to the goal
                     46:  *    without exceeding the maximum. (If a single word would
                     47:  *    exceed the maximum, we output that anyway.) Of course
                     48:  *    the trailing whitespace of the last word is ignored.
                     49:  *    We then emit a newline and start again if there are any
                     50:  *    words left.
                     51:  *    Note that for a blank line this translates as "We emit
                     52:  *    a newline".
                     53:  *    If the `-l <n>' option is given, then leading whitespace
                     54:  *    is modified slightly: <n> spaces are replaced by a tab.
                     55:  *    Indented paragraphs (see above under `-p') make matters
                     56:  *    more complicated than this suggests. Actually every paragraph
                     57:  *    has two `leading whitespace' values; the value for the first
                     58:  *    line, and the value for the most recent line. (While processing
                     59:  *    the first line, the two are equal. When `-p' has not been
                     60:  *    given, they are always equal.) The leading whitespace
                     61:  *    actually output is that of the first line (for the first
                     62:  *    line of *output*) or that of the most recent line (for
                     63:  *    all other lines of output).
                     64:  *    When `-m' has been given, message header paragraphs are
                     65:  *    taken as having first-leading-whitespace empty and
                     66:  *    subsequent-leading-whitespace two spaces.
                     67:  *
                     68:  * Multiple input files are formatted one at a time, so that a file
                     69:  * never ends in the middle of a line.
                     70:  *
                     71:  * There's an alternative mode of operation, invoked by giving
                     72:  * the `-c' option. In that case we just center every line,
                     73:  * and most of the other options are ignored. This should
                     74:  * really be in a separate program, but we must stay compatible
                     75:  * with old `fmt'.
                     76:  *
                     77:  * QUERY: Should `-m' also try to do the right thing with quoted text?
                     78:  * QUERY: `-b' to treat backslashed whitespace as old `fmt' does?
                     79:  * QUERY: Option meaning `never join lines'?
                     80:  * QUERY: Option meaning `split in mid-word to avoid overlong lines'?
                     81:  * (Those last two might not be useful, since we have `fold'.)
                     82:  *
                     83:  * Differences from old `fmt':
                     84:  *
                     85:  *   - We have many more options. Options that aren't understood
                     86:  *     generate a lengthy usage message, rather than being
                     87:  *     treated as filenames.
                     88:  *   - Even with `-m', our handling of message headers is
                     89:  *     significantly different. (And much better.)
                     90:  *   - We don't treat `\ ' as non-word-breaking.
                     91:  *   - Downward changes of indentation start new paragraphs
                     92:  *     for us, as well as upward. (I think old `fmt' behaves
                     93:  *     in the way it does in order to allow indented paragraphs,
                     94:  *     but this is a broken way of making indented paragraphs
                     95:  *     behave right.)
                     96:  *   - Given the choice of going over or under |goal_length|
                     97:  *     by the same amount, we go over; old `fmt' goes under.
                     98:  *   - We treat `?' as ending a sentence, and not `:'. Old `fmt'
                     99:  *     does the reverse.
                    100:  *   - We return approved return codes. Old `fmt' returns
                    101:  *     1 for some errors, and *the number of unopenable files*
                    102:  *     when that was all that went wrong.
                    103:  *   - We have fewer crashes and more helpful error messages.
                    104:  *   - We don't turn spaces into tabs at starts of lines unless
                    105:  *     specifically requested.
                    106:  *   - New `fmt' is somewhat smaller and slightly faster than
                    107:  *     old `fmt'.
                    108:  *
                    109:  * Bugs:
                    110:  *
                    111:  *   None known. There probably are some, though.
                    112:  *
                    113:  * Portability:
                    114:  *
                    115:  *   I believe this code to be pretty portable. It does require
                    116:  *   that you have `getopt'. If you need to include "getopt.h"
                    117:  *   for this (e.g., if your system didn't come with `getopt'
                    118:  *   and you installed it yourself) then you should arrange for
                    119:  *   NEED_getopt_h to be #defined.
                    120:  *
                    121:  *   Everything here should work OK even on nasty 16-bit
                    122:  *   machines and nice 64-bit ones. However, it's only really
                    123:  *   been tested on my FreeBSD machine. Your mileage may vary.
                    124:  */
                    125:
                    126: /* Copyright (c) 1997 Gareth McCaughan. All rights reserved.
                    127:  *
                    128:  * Redistribution and use of this code, in source or binary forms,
                    129:  * with or without modification, are permitted subject to the following
                    130:  * conditions:
                    131:  *
                    132:  *  - Redistribution of source code must retain the above copyright
1.1       deraadt   133:  *    notice, this list of conditions and the following disclaimer.
1.10      millert   134:  *
                    135:  *  - If you distribute modified source code it must also include
                    136:  *    a notice saying that it has been modified, and giving a brief
                    137:  *    description of what changes have been made.
                    138:  *
                    139:  * Disclaimer: I am not responsible for the results of using this code.
                    140:  *             If it formats your hard disc, sends obscene messages to
                    141:  *             your boss and kills your children then that's your problem
                    142:  *             not mine. I give absolutely no warranty of any sort as to
                    143:  *             what the program will do, and absolutely refuse to be held
                    144:  *             liable for any consequences of your using it.
                    145:  *             Thank you. Have a nice day.
                    146:  */
                    147:
                    148: /* RCS change log:
                    149:  * Revision 1.5  1998/03/02 18:02:21  gjm11
                    150:  * Minor changes for portability.
                    151:  *
                    152:  * Revision 1.4  1997/10/01 11:51:28  gjm11
                    153:  * Repair broken indented-paragraph handling.
                    154:  * Add mail message header stuff.
                    155:  * Improve comments and layout.
                    156:  * Make usable with non-BSD systems.
                    157:  * Add revision display to usage message.
                    158:  *
                    159:  * Revision 1.3  1997/09/30 16:24:47  gjm11
                    160:  * Add copyright notice, rcsid string and log message.
                    161:  *
                    162:  * Revision 1.2  1997/09/30 16:13:39  gjm11
                    163:  * Add options: -d <chars>, -l <width>, -p, -s, -t <width>, -h .
                    164:  * Parse options with `getopt'. Clean up code generally.
                    165:  * Make comments more accurate.
                    166:  *
                    167:  * Revision 1.1  1997/09/30 11:29:57  gjm11
                    168:  * Initial revision
1.1       deraadt   169:  */
                    170:
                    171: #ifndef lint
1.10      millert   172: static const char rcsid[] =
1.26    ! tedu      173:   "$OpenBSD: fmt.c,v 1.25 2006/11/29 21:59:04 jmc Exp $";
1.10      millert   174: static const char copyright[] =
                    175:   "Copyright (c) 1997 Gareth McCaughan. All rights reserved.\n";
1.1       deraadt   176: #endif /* not lint */
                    177:
1.4       millert   178: #include <ctype.h>
1.18      millert   179: #include <err.h>
                    180: #include <locale.h>
1.1       deraadt   181: #include <stdio.h>
                    182: #include <stdlib.h>
                    183: #include <string.h>
1.18      millert   184: #include <sysexits.h>
                    185: #include <unistd.h>
1.10      millert   186:
                    187: /* Something that, we hope, will never be a genuine line length,
                    188:  * indentation etc.
1.1       deraadt   189:  */
1.10      millert   190: #define SILLY ((size_t)-1)
1.1       deraadt   191:
1.10      millert   192: /* I used to use |strtoul| for this, but (1) not all systems have it
                    193:  * and (2) it's probably better to use |strtol| to detect negative
                    194:  * numbers better.
                    195:  * If |fussyp==0| then we don't complain about non-numbers
                    196:  * (returning 0 instead), but we do complain about bad numbers.
                    197:  */
1.18      millert   198: static size_t
1.21      tedu      199: get_positive(const char *s, const char *err_mess, int fussyP)
                    200: {
                    201:        char *t;
                    202:        long result = strtol(s, &t, 0);
                    203:
                    204:        if (*t) {
                    205:                if (fussyP)
                    206:                        goto Lose;
                    207:                else
                    208:                        return 0;
                    209:        }
                    210:        if (result <= 0) {
                    211: Lose:
                    212:                errx(EX_USAGE, "%s", err_mess);
                    213:        }
                    214:
                    215:        return (size_t) result;
1.1       deraadt   216: }
                    217:
1.10      millert   218: /* Global variables */
                    219:
1.21      tedu      220: static int centerP = 0;                                /* Try to center lines? */
                    221: static size_t goal_length = 0;                 /* Target length for output lines */
                    222: static size_t max_length = 0;                  /* Maximum length for output lines */
                    223: static int coalesce_spaces_P = 0;              /* Coalesce multiple whitespace -> ' ' ? */
                    224: static int allow_indented_paragraphs = 0;      /* Can first line have diff. ind.? */
                    225: static int tab_width = 8;                      /* Number of spaces per tab stop */
                    226: static size_t output_tab_width = 0;            /* Ditto, when squashing leading spaces */
                    227: static const char *sentence_enders = ".?!";    /* Double-space after these */
                    228: static int grok_mail_headers = 0;              /* treat embedded mail headers magically? */
                    229: static int format_troff = 0;                   /* Format troff? */
                    230:
                    231: static int n_errors = 0;                       /* Number of failed files. Return on exit. */
                    232: static char *output_buffer = NULL;             /* Output line will be built here */
                    233: static size_t x;                               /* Horizontal position in output line */
                    234: static size_t x0;                              /* Ditto, ignoring leading whitespace */
                    235: static size_t pending_spaces;                  /* Spaces to add before next word */
                    236: static int output_in_paragraph = 0;            /* Any of current para written out yet? */
1.10      millert   237:
                    238: /* Prototypes */
                    239:
1.21      tedu      240: static void    process_named_file(const char *);
                    241: static void    process_stream(FILE *, const char *);
                    242: static size_t  indent_length(const char *, size_t);
                    243: static int     might_be_header(const unsigned char *);
                    244: static void    new_paragraph(size_t, size_t);
                    245: static void    output_word(size_t, size_t, const char *, size_t, size_t);
                    246: static void    output_indent(size_t);
                    247: static void    center_stream(FILE *, const char *);
                    248: static char    *get_line(FILE *, size_t *);
                    249: static void    *xrealloc(void *, size_t);
                    250: void           usage(void);
1.10      millert   251:
1.21      tedu      252: #define XMALLOC(x) xrealloc(0, x)
1.10      millert   253:
                    254: /* Here is perhaps the right place to mention that this code is
                    255:  * all in top-down order. Hence, |main| comes first.
1.1       deraadt   256:  */
1.4       millert   257: int
1.21      tedu      258: main(int argc, char *argv[])
                    259: {
                    260:        int ch;                 /* used for |getopt| processing */
                    261:
                    262:        (void)setlocale(LC_CTYPE, "");
                    263:
                    264:        /* 1. Grok parameters. */
                    265:        while ((ch = getopt(argc, argv, "0123456789cd:hl:mnpst:w:")) != -1) {
                    266:                switch (ch) {
                    267:                case 'c':
                    268:                        centerP = 1;
                    269:                        break;
                    270:                case 'd':
                    271:                        sentence_enders = optarg;
                    272:                        break;
                    273:                case 'l':
                    274:                        output_tab_width
                    275:                                = get_positive(optarg, "output tab width must be positive", 1);
                    276:                        break;
                    277:                case 'm':
                    278:                        grok_mail_headers = 1;
                    279:                        break;
                    280:                case 'n':
                    281:                        format_troff = 1;
                    282:                        break;
                    283:                case 'p':
                    284:                        allow_indented_paragraphs = 1;
                    285:                        break;
                    286:                case 's':
                    287:                        coalesce_spaces_P = 1;
                    288:                        break;
                    289:                case 't':
                    290:                        tab_width = get_positive(optarg, "tab width must be positive", 1);
                    291:                        break;
                    292:                case 'w':
                    293:                        goal_length = get_positive(optarg, "width must be positive", 1);
                    294:                        max_length = goal_length;
                    295:                        break;
                    296:                case '0': case '1': case '2': case '3': case '4': case '5':
                    297:                case '6': case '7': case '8': case '9':
                    298:                        /* XXX  this is not a stylistically approved use of getopt() */
                    299:                        if (goal_length == 0) {
                    300:                                char *p;
                    301:
                    302:                                p = argv[optind - 1];
                    303:                                if (p[0] == '-' && p[1] == ch && !p[2])
                    304:                                        goal_length = get_positive(++p, "width must be nonzero", 1);
                    305:                                else
                    306:                                        goal_length = get_positive(argv[optind]+1,
                    307:                                                        "width must be nonzero", 1);
                    308:                                max_length = goal_length;
                    309:                        }
                    310:                        break;
                    311:                case 'h':
                    312:                default:
                    313:                        usage();
                    314:                        /* NOT REACHED */
                    315:                }
                    316:        }
                    317:
                    318:        argc -= optind;
                    319:        argv += optind;
                    320:
                    321:        /* [ goal [ maximum ] ] */
                    322:        if (argc > 0 && goal_length == 0 &&
                    323:            (goal_length = get_positive(*argv,"goal length must be positive", 0)) != 0) {
                    324:                --argc;
                    325:                ++argv;
                    326:                if (argc > 0 && (max_length = get_positive(*argv,"max length must be positive", 0)) != 0) {
                    327:                        --argc;
                    328:                        ++argv;
                    329:                        if (max_length < goal_length)
                    330:                                errx(EX_USAGE, "max length must be >= goal length");
                    331:                }
                    332:        }
                    333:
                    334:        if (goal_length == 0)
                    335:                goal_length = 65;
                    336:        if (max_length == 0)
                    337:                max_length = goal_length+10;
                    338:        output_buffer = XMALLOC(max_length+1);  /* really needn't be longer */
                    339:
                    340:        /* 2. Process files. */
                    341:
                    342:        if (argc > 0) {
                    343:                while (argc-- > 0)
                    344:                        process_named_file(*argv++);
                    345:        } else {
                    346:                process_stream(stdin, "standard input");
                    347:        }
1.10      millert   348:
1.21      tedu      349:        /* We're done. */
                    350:        return n_errors ? EX_NOINPUT : 0;
1.10      millert   351:
                    352: }
                    353:
                    354: /* Process a single file, given its name.
                    355:  */
                    356: static void
1.21      tedu      357: process_named_file(const char *name)
                    358: {
                    359:        FILE *f;
                    360:
                    361:        if ((f = fopen(name, "r")) == NULL) {
1.23      cloder    362:                warn("%s", name);
1.21      tedu      363:                ++n_errors;
                    364:        } else {
                    365:                process_stream(f, name);
                    366:                fclose(f);
                    367:        }
1.10      millert   368: }
                    369:
                    370: /* Types of mail header continuation lines:
                    371:  */
                    372: typedef enum {
1.21      tedu      373:        hdr_ParagraphStart      = -1,
                    374:        hdr_NonHeader           = 0,
                    375:        hdr_Header              = 1,
                    376:        hdr_Continuation        = 2
1.10      millert   377: } HdrType;
                    378:
                    379: /* Process a stream. This is where the real work happens,
                    380:  * except that centering is handled separately.
                    381:  */
                    382: static void
1.21      tedu      383: process_stream(FILE *stream, const char *name)
                    384: {
                    385:        size_t n;
                    386:        size_t np;
                    387:        size_t last_indent = SILLY;     /* how many spaces in last indent? */
                    388:        size_t para_line_number = 0;    /* how many lines already read in this para? */
                    389:        size_t first_indent = SILLY;    /* indentation of line 0 of paragraph */
                    390:        HdrType prev_header_type = hdr_ParagraphStart;
                    391:        HdrType header_type;
                    392:
1.10      millert   393:        /* ^-- header_type of previous line; -1 at para start */
1.21      tedu      394:        char *line;
                    395:        size_t length;
1.10      millert   396:
1.21      tedu      397:        if (centerP) {
                    398:                center_stream(stream, name);
                    399:                return;
                    400:        }
                    401:
                    402:        while ((line = get_line(stream, &length)) != NULL) {
                    403:                np = indent_length(line, length);
                    404:                header_type = hdr_NonHeader;
                    405:                if (grok_mail_headers && prev_header_type != hdr_NonHeader) {
                    406:                        if (np == 0 && might_be_header(line))
                    407:                                header_type = hdr_Header;
                    408:                        else if (np > 0 && prev_header_type>hdr_NonHeader)
                    409:                                header_type = hdr_Continuation;
                    410:                }
                    411:
                    412:                /* We need a new paragraph if and only if:
                    413:                 *   this line is blank,
                    414:                 *   OR it's a troff request,
                    415:                 *   OR it's a mail header,
                    416:                 *   OR it's not a mail header AND the last line was one,
                    417:                 *   OR the indentation has changed
                    418:                 *      AND the line isn't a mail header continuation line
                    419:                 *      AND this isn't the second line of an indented paragraph.
                    420:                 */
                    421:                if (length == 0 || (line[0] == '.' && !format_troff) ||
                    422:                    header_type == hdr_Header ||
                    423:                    (header_type == hdr_NonHeader && prev_header_type > hdr_NonHeader) ||
                    424:                    (np != last_indent && header_type != hdr_Continuation &&
                    425:                    (!allow_indented_paragraphs || para_line_number != 1)) ) {
                    426:                        new_paragraph(output_in_paragraph ? last_indent : first_indent, np);
                    427:                        para_line_number = 0;
                    428:                        first_indent = np;
                    429:                        last_indent = np;
                    430:
                    431:                        /* nroff compatibility */
                    432:                        if (length > 0 && line[0] == '.' && !format_troff) {
                    433:                                printf("%.*s\n", (int)length, line);
                    434:                                continue;
                    435:                        }
                    436:                        if (header_type == hdr_Header)
                    437:                                last_indent = 2;        /* for cont. lines */
                    438:                        if (length == 0) {
                    439:                                putchar('\n');
                    440:                                prev_header_type = hdr_ParagraphStart;
                    441:                                continue;
                    442:                        } else {
                    443:                                /* If this is an indented paragraph other than a mail header
                    444:                                 * continuation, set |last_indent|.
                    445:                                 */
                    446:                                if (np != last_indent && header_type != hdr_Continuation)
                    447:                                        last_indent = np;
                    448:                        }
                    449:                        prev_header_type = header_type;
                    450:                }
                    451:
                    452:                n = np;
                    453:                while (n < length) {
                    454:                        /* Find word end and count spaces after it */
                    455:                        size_t word_length = 0, space_length = 0;
                    456:                        while (n+word_length < length && line[n+word_length] != ' ')
                    457:                                ++word_length;
                    458:                        space_length = word_length;
                    459:                        while (n+space_length < length && line[n+space_length] == ' ')
                    460:                                ++space_length;
                    461:                        /* Send the word to the output machinery. */
                    462:                        output_word(first_indent, last_indent,
                    463:                                line+n, word_length, space_length-word_length);
                    464:                        n += space_length;
                    465:                }
                    466:                ++para_line_number;
                    467:        }
                    468:
                    469:        new_paragraph(output_in_paragraph ? last_indent : first_indent, 0);
                    470:        if (ferror(stream)) {
1.23      cloder    471:                warn("%s", name);
1.21      tedu      472:                ++n_errors;
                    473:        }
1.10      millert   474: }
                    475:
                    476: /* How long is the indent on this line?
                    477:  */
                    478: static size_t
1.21      tedu      479: indent_length(const char *line, size_t length)
                    480: {
                    481:        size_t n = 0;
                    482:
                    483:        while (n < length && *line++ == ' ')
                    484:                ++n;
                    485:        return n;
1.10      millert   486: }
                    487:
                    488: /* Might this line be a mail header?
                    489:  * We deem a line to be a possible header if it matches the
                    490:  * Perl regexp /^[A-Z][-A-Za-z0-9]*:\s/. This is *not* the same
                    491:  * as in RFC whatever-number-it-is; we want to be gratuitously
                    492:  * conservative to avoid mangling ordinary civilised text.
                    493:  */
                    494: static int
1.21      tedu      495: might_be_header(const unsigned char *line)
                    496: {
                    497:
                    498:        if (!isupper(*line++))
                    499:                return 0;
1.26    ! tedu      500:        while (isalnum(*line) || *line == '-')
1.21      tedu      501:                ++line;
                    502:        return (*line == ':' && isspace(line[1]));
1.10      millert   503: }
                    504:
                    505: /* Begin a new paragraph with an indent of |indent| spaces.
                    506:  */
                    507: static void
1.21      tedu      508: new_paragraph(size_t old_indent, size_t indent)
                    509: {
                    510:
                    511:        if (x0) {
                    512:                if (old_indent > 0)
                    513:                        output_indent(old_indent);
                    514:                fwrite(output_buffer, 1, x0, stdout);
                    515:                putchar('\n');
                    516:        }
                    517:        x = indent;
                    518:        x0 = 0;
                    519:        pending_spaces = 0;
                    520:        output_in_paragraph = 0;
1.10      millert   521: }
                    522:
                    523: /* Output spaces or tabs for leading indentation.
                    524:  */
                    525: static void
1.21      tedu      526: output_indent(size_t n_spaces)
                    527: {
                    528:
                    529:        if (output_tab_width) {
                    530:                while (n_spaces >= output_tab_width) {
                    531:                        putchar('\t');
                    532:                        n_spaces -= output_tab_width;
                    533:                }
                    534:        }
                    535:        while (n_spaces-- > 0)
                    536:                putchar(' ');
1.10      millert   537: }
                    538:
                    539: /* Output a single word, or add it to the buffer.
                    540:  * indent0 and indent1 are the indents to use on the first and subsequent
                    541:  * lines of a paragraph. They'll often be the same, of course.
                    542:  */
                    543: static void
1.21      tedu      544: output_word(size_t indent0, size_t indent1, const char *word, size_t length, size_t spaces)
                    545: {
                    546:        size_t new_x = x + pending_spaces + length;
                    547:        size_t indent = output_in_paragraph ? indent1 : indent0;
                    548:
                    549:        /* If either |spaces==0| (at end of line) or |coalesce_spaces_P|
                    550:         * (squashing internal whitespace), then add just one space;
                    551:         * except that if the last character was a sentence-ender we
                    552:         * actually add two spaces.
                    553:         */
                    554:        if (coalesce_spaces_P || spaces == 0)
                    555:                spaces = strchr(sentence_enders, word[length-1]) ? 2 : 1;
                    556:
                    557:        if (new_x <= goal_length) {
                    558:                /* After adding the word we still aren't at the goal length,
                    559:                 * so clearly we add it to the buffer rather than outputing it.
                    560:                 */
                    561:                memset(output_buffer+x0, ' ', pending_spaces);
                    562:                x0 += pending_spaces;
                    563:                x += pending_spaces;
                    564:                memcpy(output_buffer+x0, word, length);
                    565:                x0 += length;
                    566:                x += length;
                    567:                pending_spaces = spaces;
                    568:        } else {
                    569:                /* Adding the word takes us past the goal. Print the line-so-far,
                    570:                 * and the word too iff either (1) the lsf is empty or (2) that
                    571:                 * makes us nearer the goal but doesn't take us over the limit,
                    572:                 * or (3) the word on its own takes us over the limit.
                    573:                 * In case (3) we put a newline in between.
                    574:                 */
                    575:                if (indent > 0)
                    576:                        output_indent(indent);
                    577:                fwrite(output_buffer, 1, x0, stdout);
                    578:                if (x0 == 0 || (new_x <= max_length && new_x-goal_length <= goal_length-x)) {
                    579:                        printf("%*s", (int)pending_spaces, "");
                    580:                        goto write_out_word;
                    581:                } else {
                    582:                        /* If the word takes us over the limit on its own, just
                    583:                         * spit it out and don't bother buffering it.
                    584:                         */
                    585:                        if (indent+length > max_length) {
                    586:                                putchar('\n');
                    587:                                if (indent > 0)
                    588:                                        output_indent(indent);
1.10      millert   589: write_out_word:
1.21      tedu      590:                                fwrite(word, 1, length, stdout);
                    591:                                x0 = 0;
                    592:                                x = indent1;
                    593:                                pending_spaces = 0;
                    594:                        } else {
                    595:                                memcpy(output_buffer, word, length);
                    596:                                x0 = length;
                    597:                                x = length+indent1;
                    598:                                pending_spaces = spaces;
                    599:                        }
                    600:                }
                    601:
                    602:                putchar('\n');
                    603:                output_in_paragraph = 1;
                    604:        }
1.10      millert   605: }
                    606:
                    607: /* Process a stream, but just center its lines rather than trying to
                    608:  * format them neatly.
                    609:  */
                    610: static void
1.21      tedu      611: center_stream(FILE *stream, const char *name)
                    612: {
                    613:        char *line;
                    614:        size_t length;
                    615:        size_t l;
                    616:
                    617:        while ((line = get_line(stream, &length)) != 0) {
                    618:                l = length;
                    619:                while (l > 0 && isspace(*line)) {
                    620:                        ++line;
                    621:                        --l;
                    622:                }
                    623:
                    624:                length = l;
                    625:
                    626:                while (l < goal_length) {
                    627:                        putchar(' ');
                    628:                        l += 2;
                    629:                }
                    630:
                    631:                fwrite(line, 1, length, stdout);
                    632:                putchar('\n');
                    633:        }
                    634:
                    635:        if (ferror(stream)) {
1.23      cloder    636:                warn("%s", name);
1.21      tedu      637:                ++n_errors;
                    638:        }
1.10      millert   639: }
                    640:
                    641: /* Get a single line from a stream. Expand tabs, strip control
                    642:  * characters and trailing whitespace, and handle backspaces.
                    643:  * Return the address of the buffer containing the line, and
                    644:  * put the length of the line in |lengthp|.
                    645:  * This can cope with arbitrarily long lines, and with lines
                    646:  * without terminating \n.
                    647:  * If there are no characters left or an error happens, we
                    648:  * return 0.
                    649:  * Don't confuse |spaces_pending| here with the global
                    650:  * |pending_spaces|.
                    651:  */
                    652: static char *
1.21      tedu      653: get_line(FILE *stream, size_t *lengthp)
                    654: {
                    655:        int ch;
                    656:        int troff = 0;
                    657:        static char *buf = NULL;
                    658:        static size_t length = 0;
                    659:        size_t len = 0;
                    660:        size_t spaces_pending = 0;
                    661:
                    662:        if (buf == NULL) {
                    663:                length = 100;
                    664:                buf = XMALLOC(length);
                    665:        }
                    666:
                    667:        while ((ch = getc(stream)) != '\n' && ch != EOF) {
                    668:                if ((len + spaces_pending == 0) && (ch == '.' && !format_troff))
                    669:                        troff = 1;
                    670:                if (ch == ' ') {
                    671:                        ++spaces_pending;
                    672:                } else if (troff || !iscntrl(ch)) {
                    673:                        while (len + spaces_pending >= length) {
                    674:                                length *= 2;
                    675:                                buf = xrealloc(buf, length);
                    676:                        }
                    677:
                    678:                        while (spaces_pending > 0) {
                    679:                                --spaces_pending;
                    680:                                buf[len++] = ' ';
                    681:                        }
                    682:                        buf[len++] = ch;
                    683:                } else if (ch == '\t') {
                    684:                        spaces_pending += tab_width - (len+spaces_pending)%tab_width;
                    685:                } else if (ch == '\b') {
                    686:                        if (len)
                    687:                                --len;
                    688:                }
                    689:        }
                    690:
                    691:        *lengthp = len;
                    692:        return (len > 0 || ch != EOF) ? buf : 0;
1.10      millert   693: }
                    694:
                    695: /* (Re)allocate some memory, exiting with an error if we can't.
                    696:  */
                    697: static void *
1.21      tedu      698: xrealloc(void *ptr, size_t nbytes)
                    699: {
                    700:        void *p;
                    701:
                    702:        p  = realloc(ptr, nbytes);
                    703:        if (p == NULL)
                    704:                errx(EX_OSERR, "out of memory");
                    705:        return p;
                    706: }
                    707:
                    708: void
                    709: usage(void)
                    710: {
1.22      mickey    711:        extern char *__progname;
1.21      tedu      712:
                    713:        fprintf(stderr,
1.25      jmc       714:                "usage: %s [-cmnps] [-d chars] [-l number] [-t number]\n"
                    715:                "\t[goal [maximum] | -width | -w width] [file ...]\n",
                    716:                        __progname);
1.22      mickey    717:        exit (1);
1.1       deraadt   718: }