186aba0f2SPedro F. Giffuni /* $OpenBSD: fmt.c,v 1.21 2004/04/01 23:14:19 tedu Exp $ */
21438aefcSRuslan Ermilov
31438aefcSRuslan Ermilov /* Sensible version of fmt
49b50d902SRodney W. Grimes *
51438aefcSRuslan Ermilov * Syntax: fmt [ options ] [ goal [ max ] ] [ filename ... ]
61438aefcSRuslan Ermilov *
71438aefcSRuslan Ermilov * Since the documentation for the original fmt is so poor, here
81438aefcSRuslan Ermilov * is an accurate description of what this one does. It's usually
91438aefcSRuslan Ermilov * the same. The *mechanism* used may differ from that suggested
101438aefcSRuslan Ermilov * here. Note that we are *not* entirely compatible with fmt,
111438aefcSRuslan Ermilov * because fmt gets so many things wrong.
121438aefcSRuslan Ermilov *
131438aefcSRuslan Ermilov * 1. Tabs are expanded, assuming 8-space tab stops.
141438aefcSRuslan Ermilov * If the `-t <n>' option is given, we assume <n>-space
151438aefcSRuslan Ermilov * tab stops instead.
161438aefcSRuslan Ermilov * Trailing blanks are removed from all lines.
171438aefcSRuslan Ermilov * x\b == nothing, for any x other than \b.
181438aefcSRuslan Ermilov * Other control characters are simply stripped. This
191438aefcSRuslan Ermilov * includes \r.
201438aefcSRuslan Ermilov * 2. Each line is split into leading whitespace and
211438aefcSRuslan Ermilov * everything else. Maximal consecutive sequences of
221438aefcSRuslan Ermilov * lines with the same leading whitespace are considered
231438aefcSRuslan Ermilov * to form paragraphs, except that a blank line is always
241438aefcSRuslan Ermilov * a paragraph to itself.
251438aefcSRuslan Ermilov * If the `-p' option is given then the first line of a
261438aefcSRuslan Ermilov * paragraph is permitted to have indentation different
271438aefcSRuslan Ermilov * from that of the other lines.
281438aefcSRuslan Ermilov * If the `-m' option is given then a line that looks
291438aefcSRuslan Ermilov * like a mail message header, if it is not immediately
301438aefcSRuslan Ermilov * preceded by a non-blank non-message-header line, is
311438aefcSRuslan Ermilov * taken to start a new paragraph, which also contains
321438aefcSRuslan Ermilov * any subsequent lines with non-empty leading whitespace.
33deba2451SRuslan Ermilov * Unless the `-n' option is given, lines beginning with
34deba2451SRuslan Ermilov * a . (dot) are not formatted.
351438aefcSRuslan Ermilov * 3. The "everything else" is split into words; a word
361438aefcSRuslan Ermilov * includes its trailing whitespace, and a word at the
371438aefcSRuslan Ermilov * end of a line is deemed to be followed by a single
381438aefcSRuslan Ermilov * space, or two spaces if it ends with a sentence-end
391438aefcSRuslan Ermilov * character. (See the `-d' option for how to change that.)
401438aefcSRuslan Ermilov * If the `-s' option has been given, then a word's trailing
411438aefcSRuslan Ermilov * whitespace is replaced by what it would have had if it
421438aefcSRuslan Ermilov * had occurred at end of line.
431438aefcSRuslan Ermilov * 4. Each paragraph is sent to standard output as follows.
441438aefcSRuslan Ermilov * We output the leading whitespace, and then enough words
451438aefcSRuslan Ermilov * to make the line length as near as possible to the goal
461438aefcSRuslan Ermilov * without exceeding the maximum. (If a single word would
471438aefcSRuslan Ermilov * exceed the maximum, we output that anyway.) Of course
481438aefcSRuslan Ermilov * the trailing whitespace of the last word is ignored.
491438aefcSRuslan Ermilov * We then emit a newline and start again if there are any
501438aefcSRuslan Ermilov * words left.
511438aefcSRuslan Ermilov * Note that for a blank line this translates as "We emit
521438aefcSRuslan Ermilov * a newline".
531438aefcSRuslan Ermilov * If the `-l <n>' option is given, then leading whitespace
541438aefcSRuslan Ermilov * is modified slightly: <n> spaces are replaced by a tab.
551438aefcSRuslan Ermilov * Indented paragraphs (see above under `-p') make matters
561438aefcSRuslan Ermilov * more complicated than this suggests. Actually every paragraph
571438aefcSRuslan Ermilov * has two `leading whitespace' values; the value for the first
581438aefcSRuslan Ermilov * line, and the value for the most recent line. (While processing
591438aefcSRuslan Ermilov * the first line, the two are equal. When `-p' has not been
601438aefcSRuslan Ermilov * given, they are always equal.) The leading whitespace
611438aefcSRuslan Ermilov * actually output is that of the first line (for the first
621438aefcSRuslan Ermilov * line of *output*) or that of the most recent line (for
631438aefcSRuslan Ermilov * all other lines of output).
641438aefcSRuslan Ermilov * When `-m' has been given, message header paragraphs are
651438aefcSRuslan Ermilov * taken as having first-leading-whitespace empty and
661438aefcSRuslan Ermilov * subsequent-leading-whitespace two spaces.
671438aefcSRuslan Ermilov *
681438aefcSRuslan Ermilov * Multiple input files are formatted one at a time, so that a file
691438aefcSRuslan Ermilov * never ends in the middle of a line.
701438aefcSRuslan Ermilov *
711438aefcSRuslan Ermilov * There's an alternative mode of operation, invoked by giving
721438aefcSRuslan Ermilov * the `-c' option. In that case we just center every line,
731438aefcSRuslan Ermilov * and most of the other options are ignored. This should
741438aefcSRuslan Ermilov * really be in a separate program, but we must stay compatible
751438aefcSRuslan Ermilov * with old `fmt'.
761438aefcSRuslan Ermilov *
771438aefcSRuslan Ermilov * QUERY: Should `-m' also try to do the right thing with quoted text?
781438aefcSRuslan Ermilov * QUERY: `-b' to treat backslashed whitespace as old `fmt' does?
791438aefcSRuslan Ermilov * QUERY: Option meaning `never join lines'?
801438aefcSRuslan Ermilov * QUERY: Option meaning `split in mid-word to avoid overlong lines'?
811438aefcSRuslan Ermilov * (Those last two might not be useful, since we have `fold'.)
821438aefcSRuslan Ermilov *
831438aefcSRuslan Ermilov * Differences from old `fmt':
841438aefcSRuslan Ermilov *
851438aefcSRuslan Ermilov * - We have many more options. Options that aren't understood
861438aefcSRuslan Ermilov * generate a lengthy usage message, rather than being
871438aefcSRuslan Ermilov * treated as filenames.
881438aefcSRuslan Ermilov * - Even with `-m', our handling of message headers is
891438aefcSRuslan Ermilov * significantly different. (And much better.)
901438aefcSRuslan Ermilov * - We don't treat `\ ' as non-word-breaking.
911438aefcSRuslan Ermilov * - Downward changes of indentation start new paragraphs
921438aefcSRuslan Ermilov * for us, as well as upward. (I think old `fmt' behaves
931438aefcSRuslan Ermilov * in the way it does in order to allow indented paragraphs,
941438aefcSRuslan Ermilov * but this is a broken way of making indented paragraphs
951438aefcSRuslan Ermilov * behave right.)
961438aefcSRuslan Ermilov * - Given the choice of going over or under |goal_length|
971438aefcSRuslan Ermilov * by the same amount, we go over; old `fmt' goes under.
981438aefcSRuslan Ermilov * - We treat `?' as ending a sentence, and not `:'. Old `fmt'
991438aefcSRuslan Ermilov * does the reverse.
1001438aefcSRuslan Ermilov * - We return approved return codes. Old `fmt' returns
1011438aefcSRuslan Ermilov * 1 for some errors, and *the number of unopenable files*
1021438aefcSRuslan Ermilov * when that was all that went wrong.
1031438aefcSRuslan Ermilov * - We have fewer crashes and more helpful error messages.
1041438aefcSRuslan Ermilov * - We don't turn spaces into tabs at starts of lines unless
1051438aefcSRuslan Ermilov * specifically requested.
1061438aefcSRuslan Ermilov * - New `fmt' is somewhat smaller and slightly faster than
1071438aefcSRuslan Ermilov * old `fmt'.
1081438aefcSRuslan Ermilov *
1091438aefcSRuslan Ermilov * Bugs:
1101438aefcSRuslan Ermilov *
1111438aefcSRuslan Ermilov * None known. There probably are some, though.
1121438aefcSRuslan Ermilov *
1131438aefcSRuslan Ermilov * Portability:
1141438aefcSRuslan Ermilov *
1151438aefcSRuslan Ermilov * I believe this code to be pretty portable. It does require
1161438aefcSRuslan Ermilov * that you have `getopt'. If you need to include "getopt.h"
1171438aefcSRuslan Ermilov * for this (e.g., if your system didn't come with `getopt'
1181438aefcSRuslan Ermilov * and you installed it yourself) then you should arrange for
1191438aefcSRuslan Ermilov * NEED_getopt_h to be #defined.
1201438aefcSRuslan Ermilov *
1211438aefcSRuslan Ermilov * Everything here should work OK even on nasty 16-bit
1221438aefcSRuslan Ermilov * machines and nice 64-bit ones. However, it's only really
1231438aefcSRuslan Ermilov * been tested on my FreeBSD machine. Your mileage may vary.
1241438aefcSRuslan Ermilov */
1251438aefcSRuslan Ermilov
1261438aefcSRuslan Ermilov /* Copyright (c) 1997 Gareth McCaughan. All rights reserved.
1271438aefcSRuslan Ermilov *
1281438aefcSRuslan Ermilov * Redistribution and use of this code, in source or binary forms,
1291438aefcSRuslan Ermilov * with or without modification, are permitted subject to the following
1301438aefcSRuslan Ermilov * conditions:
1311438aefcSRuslan Ermilov *
1321438aefcSRuslan Ermilov * - Redistribution of source code must retain the above copyright
1339b50d902SRodney W. Grimes * notice, this list of conditions and the following disclaimer.
1349b50d902SRodney W. Grimes *
1351438aefcSRuslan Ermilov * - If you distribute modified source code it must also include
1361438aefcSRuslan Ermilov * a notice saying that it has been modified, and giving a brief
1371438aefcSRuslan Ermilov * description of what changes have been made.
1381438aefcSRuslan Ermilov *
1391438aefcSRuslan Ermilov * Disclaimer: I am not responsible for the results of using this code.
1401438aefcSRuslan Ermilov * If it formats your hard disc, sends obscene messages to
1411438aefcSRuslan Ermilov * your boss and kills your children then that's your problem
1421438aefcSRuslan Ermilov * not mine. I give absolutely no warranty of any sort as to
1431438aefcSRuslan Ermilov * what the program will do, and absolutely refuse to be held
1441438aefcSRuslan Ermilov * liable for any consequences of your using it.
1451438aefcSRuslan Ermilov * Thank you. Have a nice day.
1461438aefcSRuslan Ermilov */
1471438aefcSRuslan Ermilov
1481438aefcSRuslan Ermilov /* RCS change log:
1491438aefcSRuslan Ermilov * Revision 1.5 1998/03/02 18:02:21 gjm11
1501438aefcSRuslan Ermilov * Minor changes for portability.
1511438aefcSRuslan Ermilov *
1521438aefcSRuslan Ermilov * Revision 1.4 1997/10/01 11:51:28 gjm11
1531438aefcSRuslan Ermilov * Repair broken indented-paragraph handling.
1541438aefcSRuslan Ermilov * Add mail message header stuff.
1551438aefcSRuslan Ermilov * Improve comments and layout.
1561438aefcSRuslan Ermilov * Make usable with non-BSD systems.
1571438aefcSRuslan Ermilov * Add revision display to usage message.
1581438aefcSRuslan Ermilov *
1591438aefcSRuslan Ermilov * Revision 1.3 1997/09/30 16:24:47 gjm11
1601438aefcSRuslan Ermilov * Add copyright notice, rcsid string and log message.
1611438aefcSRuslan Ermilov *
1621438aefcSRuslan Ermilov * Revision 1.2 1997/09/30 16:13:39 gjm11
1631438aefcSRuslan Ermilov * Add options: -d <chars>, -l <width>, -p, -s, -t <width>, -h .
1641438aefcSRuslan Ermilov * Parse options with `getopt'. Clean up code generally.
1651438aefcSRuslan Ermilov * Make comments more accurate.
1661438aefcSRuslan Ermilov *
1671438aefcSRuslan Ermilov * Revision 1.1 1997/09/30 11:29:57 gjm11
1681438aefcSRuslan Ermilov * Initial revision
1699b50d902SRodney W. Grimes */
1709b50d902SRodney W. Grimes
1719b50d902SRodney W. Grimes #ifndef lint
1721438aefcSRuslan Ermilov static const char copyright[] =
1731438aefcSRuslan Ermilov "Copyright (c) 1997 Gareth McCaughan. All rights reserved.\n";
1749b50d902SRodney W. Grimes #endif /* not lint */
175e026a48cSDavid E. O'Brien #include <sys/cdefs.h>
17663ffb113SPhilippe Charnier #include <err.h>
177a2229407SJuli Mallett #include <limits.h>
1788bbd9072SAndrey A. Chernov #include <locale.h>
17963ffb113SPhilippe Charnier #include <stdio.h>
180e6c267f1SJoerg Wunsch #include <stdlib.h>
181821df508SXin LI #include <string.h>
1821438aefcSRuslan Ermilov #include <sysexits.h>
1831438aefcSRuslan Ermilov #include <unistd.h>
18428d92b74STim J. Robbins #include <wchar.h>
18528d92b74STim J. Robbins #include <wctype.h>
1869b50d902SRodney W. Grimes
1871438aefcSRuslan Ermilov /* Something that, we hope, will never be a genuine line length,
1881438aefcSRuslan Ermilov * indentation etc.
1899b50d902SRodney W. Grimes */
1901438aefcSRuslan Ermilov #define SILLY ((size_t)-1)
1919b50d902SRodney W. Grimes
1921438aefcSRuslan Ermilov /* I used to use |strtoul| for this, but (1) not all systems have it
1931438aefcSRuslan Ermilov * and (2) it's probably better to use |strtol| to detect negative
1941438aefcSRuslan Ermilov * numbers better.
1951438aefcSRuslan Ermilov * If |fussyp==0| then we don't complain about non-numbers
1961438aefcSRuslan Ermilov * (returning 0 instead), but we do complain about bad numbers.
1979b50d902SRodney W. Grimes */
1981438aefcSRuslan Ermilov static size_t
get_positive(const char * s,const char * err_mess,int fussyP)19986aba0f2SPedro F. Giffuni get_positive(const char *s, const char *err_mess, int fussyP)
20086aba0f2SPedro F. Giffuni {
2011438aefcSRuslan Ermilov char *t;
2021438aefcSRuslan Ermilov long result = strtol(s, &t, 0);
20386aba0f2SPedro F. Giffuni
20486aba0f2SPedro F. Giffuni if (*t) {
20586aba0f2SPedro F. Giffuni if (fussyP)
20686aba0f2SPedro F. Giffuni goto Lose;
20786aba0f2SPedro F. Giffuni else
20886aba0f2SPedro F. Giffuni return 0;
20986aba0f2SPedro F. Giffuni }
21086aba0f2SPedro F. Giffuni if (result <= 0) {
21186aba0f2SPedro F. Giffuni Lose: errx(EX_USAGE, "%s", err_mess);
21286aba0f2SPedro F. Giffuni }
2131438aefcSRuslan Ermilov return (size_t)result;
2141438aefcSRuslan Ermilov }
2159b50d902SRodney W. Grimes
216ccb8bea4SRuslan Ermilov static size_t
get_nonnegative(const char * s,const char * err_mess,int fussyP)21786aba0f2SPedro F. Giffuni get_nonnegative(const char *s, const char *err_mess, int fussyP)
21886aba0f2SPedro F. Giffuni {
219ccb8bea4SRuslan Ermilov char *t;
220ccb8bea4SRuslan Ermilov long result = strtol(s, &t, 0);
22186aba0f2SPedro F. Giffuni
22286aba0f2SPedro F. Giffuni if (*t) {
22386aba0f2SPedro F. Giffuni if (fussyP)
22486aba0f2SPedro F. Giffuni goto Lose;
22586aba0f2SPedro F. Giffuni else
22686aba0f2SPedro F. Giffuni return 0;
22786aba0f2SPedro F. Giffuni }
22886aba0f2SPedro F. Giffuni if (result < 0) {
22986aba0f2SPedro F. Giffuni Lose: errx(EX_USAGE, "%s", err_mess);
23086aba0f2SPedro F. Giffuni }
231ccb8bea4SRuslan Ermilov return (size_t)result;
232ccb8bea4SRuslan Ermilov }
233ccb8bea4SRuslan Ermilov
2341438aefcSRuslan Ermilov /* Global variables */
2359b50d902SRodney W. Grimes
2361438aefcSRuslan Ermilov static int centerP = 0; /* Try to center lines? */
2371438aefcSRuslan Ermilov static size_t goal_length = 0; /* Target length for output lines */
2381438aefcSRuslan Ermilov static size_t max_length = 0; /* Maximum length for output lines */
2391438aefcSRuslan Ermilov static int coalesce_spaces_P = 0; /* Coalesce multiple whitespace -> ' ' ? */
2401438aefcSRuslan Ermilov static int allow_indented_paragraphs = 0; /* Can first line have diff. ind.? */
2411438aefcSRuslan Ermilov static int tab_width = 8; /* Number of spaces per tab stop */
242ccb8bea4SRuslan Ermilov static size_t output_tab_width = 8; /* Ditto, when squashing leading spaces */
24328d92b74STim J. Robbins static const wchar_t *sentence_enders = L".?!"; /* Double-space after these */
2441438aefcSRuslan Ermilov static int grok_mail_headers = 0; /* treat embedded mail headers magically? */
245deba2451SRuslan Ermilov static int format_troff = 0; /* Format troff? */
2469b50d902SRodney W. Grimes
2471438aefcSRuslan Ermilov static int n_errors = 0; /* Number of failed files. Return on exit. */
248074e77e6SPedro F. Giffuni static wchar_t *output_buffer = NULL; /* Output line will be built here */
2491438aefcSRuslan Ermilov static size_t x; /* Horizontal position in output line */
2501438aefcSRuslan Ermilov static size_t x0; /* Ditto, ignoring leading whitespace */
25128d92b74STim J. Robbins static size_t output_buffer_length = 0;
2521438aefcSRuslan Ermilov static size_t pending_spaces; /* Spaces to add before next word */
2531438aefcSRuslan Ermilov static int output_in_paragraph = 0; /* Any of current para written out yet? */
25463ffb113SPhilippe Charnier
2551438aefcSRuslan Ermilov /* Prototypes */
2561438aefcSRuslan Ermilov
2571438aefcSRuslan Ermilov static void process_named_file(const char *);
2581438aefcSRuslan Ermilov static void process_stream(FILE *, const char *);
25928d92b74STim J. Robbins static size_t indent_length(const wchar_t *, size_t);
26028d92b74STim J. Robbins static int might_be_header(const wchar_t *);
2611438aefcSRuslan Ermilov static void new_paragraph(size_t, size_t);
26286aba0f2SPedro F. Giffuni static void output_word(size_t, size_t, const wchar_t *, size_t, size_t);
2631438aefcSRuslan Ermilov static void output_indent(size_t);
2641438aefcSRuslan Ermilov static void center_stream(FILE *, const char *);
26528d92b74STim J. Robbins static wchar_t *get_line(FILE *, size_t *);
2661438aefcSRuslan Ermilov static void *xrealloc(void *, size_t);
2671438aefcSRuslan Ermilov
2681438aefcSRuslan Ermilov #define XMALLOC(x) xrealloc(0,x)
2691438aefcSRuslan Ermilov
2701438aefcSRuslan Ermilov /* Here is perhaps the right place to mention that this code is
2711438aefcSRuslan Ermilov * all in top-down order. Hence, |main| comes first.
2729b50d902SRodney W. Grimes */
27363ffb113SPhilippe Charnier int
main(int argc,char * argv[])27486aba0f2SPedro F. Giffuni main(int argc, char *argv[])
27586aba0f2SPedro F. Giffuni {
2761438aefcSRuslan Ermilov int ch; /* used for |getopt| processing */
27728d92b74STim J. Robbins wchar_t *tmp;
27828d92b74STim J. Robbins size_t len;
27928d92b74STim J. Robbins const char *src;
2809b50d902SRodney W. Grimes
2818bbd9072SAndrey A. Chernov (void)setlocale(LC_CTYPE, "");
2828bbd9072SAndrey A. Chernov
2831438aefcSRuslan Ermilov /* 1. Grok parameters. */
2841438aefcSRuslan Ermilov
285deba2451SRuslan Ermilov while ((ch = getopt(argc, argv, "0123456789cd:hl:mnpst:w:")) != -1)
2861438aefcSRuslan Ermilov switch (ch) {
2871438aefcSRuslan Ermilov case 'c':
2881438aefcSRuslan Ermilov centerP = 1;
289deba2451SRuslan Ermilov format_troff = 1;
2901438aefcSRuslan Ermilov continue;
2911438aefcSRuslan Ermilov case 'd':
29228d92b74STim J. Robbins src = optarg;
29328d92b74STim J. Robbins len = mbsrtowcs(NULL, &src, 0, NULL);
29428d92b74STim J. Robbins if (len == (size_t)-1)
29528d92b74STim J. Robbins err(EX_USAGE, "bad sentence-ending character set");
29628d92b74STim J. Robbins tmp = XMALLOC((len + 1) * sizeof(wchar_t));
29728d92b74STim J. Robbins mbsrtowcs(tmp, &src, len + 1, NULL);
29828d92b74STim J. Robbins sentence_enders = tmp;
2991438aefcSRuslan Ermilov continue;
3001438aefcSRuslan Ermilov case 'l':
3011438aefcSRuslan Ermilov output_tab_width
302ccb8bea4SRuslan Ermilov = get_nonnegative(optarg, "output tab width must be non-negative", 1);
3031438aefcSRuslan Ermilov continue;
3041438aefcSRuslan Ermilov case 'm':
3051438aefcSRuslan Ermilov grok_mail_headers = 1;
3061438aefcSRuslan Ermilov continue;
307deba2451SRuslan Ermilov case 'n':
308deba2451SRuslan Ermilov format_troff = 1;
309deba2451SRuslan Ermilov continue;
3101438aefcSRuslan Ermilov case 'p':
3111438aefcSRuslan Ermilov allow_indented_paragraphs = 1;
3121438aefcSRuslan Ermilov continue;
3131438aefcSRuslan Ermilov case 's':
3141438aefcSRuslan Ermilov coalesce_spaces_P = 1;
3151438aefcSRuslan Ermilov continue;
3161438aefcSRuslan Ermilov case 't':
3171438aefcSRuslan Ermilov tab_width = get_positive(optarg, "tab width must be positive", 1);
3181438aefcSRuslan Ermilov continue;
3191438aefcSRuslan Ermilov case 'w':
3201438aefcSRuslan Ermilov goal_length = get_positive(optarg, "width must be positive", 1);
3211438aefcSRuslan Ermilov max_length = goal_length;
3221438aefcSRuslan Ermilov continue;
3231438aefcSRuslan Ermilov case '0': case '1': case '2': case '3': case '4': case '5':
3241438aefcSRuslan Ermilov case '6': case '7': case '8': case '9':
32586aba0f2SPedro F. Giffuni /*
32686aba0f2SPedro F. Giffuni * XXX this is not a stylistically approved use of
32786aba0f2SPedro F. Giffuni * getopt()
32886aba0f2SPedro F. Giffuni */
3291438aefcSRuslan Ermilov if (goal_length == 0) {
3301438aefcSRuslan Ermilov char *p;
33186aba0f2SPedro F. Giffuni
3321438aefcSRuslan Ermilov p = argv[optind - 1];
3331438aefcSRuslan Ermilov if (p[0] == '-' && p[1] == ch && !p[2])
3341438aefcSRuslan Ermilov goal_length = get_positive(++p, "width must be nonzero", 1);
3351438aefcSRuslan Ermilov else
3361438aefcSRuslan Ermilov goal_length = get_positive(argv[optind] + 1,
3371438aefcSRuslan Ermilov "width must be nonzero", 1);
3381438aefcSRuslan Ermilov max_length = goal_length;
3391438aefcSRuslan Ermilov }
3401438aefcSRuslan Ermilov continue;
34186aba0f2SPedro F. Giffuni case 'h':
34286aba0f2SPedro F. Giffuni default:
3431438aefcSRuslan Ermilov fprintf(stderr,
344d3974088SDag-Erling Smørgrav "usage: fmt [-cmps] [-d chars] [-l num] [-t num]\n"
3451438aefcSRuslan Ermilov " [-w width | -width | goal [maximum]] [file ...]\n"
3461438aefcSRuslan Ermilov "Options: -c center each line instead of formatting\n"
3471438aefcSRuslan Ermilov " -d <chars> double-space after <chars> at line end\n"
3481438aefcSRuslan Ermilov " -l <n> turn each <n> spaces at start of line into a tab\n"
3491438aefcSRuslan Ermilov " -m try to make sure mail header lines stay separate\n"
350deba2451SRuslan Ermilov " -n format lines beginning with a dot\n"
3511438aefcSRuslan Ermilov " -p allow indented paragraphs\n"
3521438aefcSRuslan Ermilov " -s coalesce whitespace inside lines\n"
3531438aefcSRuslan Ermilov " -t <n> have tabs every <n> columns\n"
3541438aefcSRuslan Ermilov " -w <n> set maximum width to <n>\n"
3551438aefcSRuslan Ermilov " goal set target width to goal\n");
3561438aefcSRuslan Ermilov exit(ch == 'h' ? 0 : EX_USAGE);
3571438aefcSRuslan Ermilov }
35886aba0f2SPedro F. Giffuni argc -= optind;
35986aba0f2SPedro F. Giffuni argv += optind;
3601438aefcSRuslan Ermilov
3611438aefcSRuslan Ermilov /* [ goal [ maximum ] ] */
3621438aefcSRuslan Ermilov
3631438aefcSRuslan Ermilov if (argc > 0 && goal_length == 0
3641438aefcSRuslan Ermilov && (goal_length = get_positive(*argv, "goal length must be positive", 0))
3651438aefcSRuslan Ermilov != 0) {
36686aba0f2SPedro F. Giffuni --argc;
36786aba0f2SPedro F. Giffuni ++argv;
3681438aefcSRuslan Ermilov if (argc > 0
3691438aefcSRuslan Ermilov && (max_length = get_positive(*argv, "max length must be positive", 0))
3701438aefcSRuslan Ermilov != 0) {
37186aba0f2SPedro F. Giffuni --argc;
37286aba0f2SPedro F. Giffuni ++argv;
3731438aefcSRuslan Ermilov if (max_length < goal_length)
3741438aefcSRuslan Ermilov errx(EX_USAGE, "max length must be >= goal length");
3751438aefcSRuslan Ermilov }
3761438aefcSRuslan Ermilov }
37786aba0f2SPedro F. Giffuni if (goal_length == 0)
37886aba0f2SPedro F. Giffuni goal_length = 65;
37986aba0f2SPedro F. Giffuni if (max_length == 0)
38086aba0f2SPedro F. Giffuni max_length = goal_length + 10;
38186aba0f2SPedro F. Giffuni if (max_length >= SIZE_T_MAX / sizeof(wchar_t))
38286aba0f2SPedro F. Giffuni errx(EX_USAGE, "max length too large");
38328d92b74STim J. Robbins /* really needn't be longer */
38428d92b74STim J. Robbins output_buffer = XMALLOC((max_length + 1) * sizeof(wchar_t));
3851438aefcSRuslan Ermilov
3861438aefcSRuslan Ermilov /* 2. Process files. */
3871438aefcSRuslan Ermilov
3881438aefcSRuslan Ermilov if (argc > 0) {
38986aba0f2SPedro F. Giffuni while (argc-- > 0)
39086aba0f2SPedro F. Giffuni process_named_file(*argv++);
39186aba0f2SPedro F. Giffuni } else {
3921438aefcSRuslan Ermilov process_stream(stdin, "standard input");
3931438aefcSRuslan Ermilov }
3941438aefcSRuslan Ermilov
3951438aefcSRuslan Ermilov /* We're done. */
3961438aefcSRuslan Ermilov
3971438aefcSRuslan Ermilov return n_errors ? EX_NOINPUT : 0;
3981438aefcSRuslan Ermilov
3991438aefcSRuslan Ermilov }
4001438aefcSRuslan Ermilov
4011438aefcSRuslan Ermilov /* Process a single file, given its name.
4029b50d902SRodney W. Grimes */
4031438aefcSRuslan Ermilov static void
process_named_file(const char * name)40486aba0f2SPedro F. Giffuni process_named_file(const char *name)
40586aba0f2SPedro F. Giffuni {
4061438aefcSRuslan Ermilov FILE *f = fopen(name, "r");
40786aba0f2SPedro F. Giffuni
40886aba0f2SPedro F. Giffuni if (!f) {
40986aba0f2SPedro F. Giffuni warn("%s", name);
41086aba0f2SPedro F. Giffuni ++n_errors;
41186aba0f2SPedro F. Giffuni } else {
4121438aefcSRuslan Ermilov process_stream(f, name);
41386aba0f2SPedro F. Giffuni if (ferror(f)) {
41486aba0f2SPedro F. Giffuni warn("%s", name);
41586aba0f2SPedro F. Giffuni ++n_errors;
41686aba0f2SPedro F. Giffuni }
4171438aefcSRuslan Ermilov fclose(f);
4189b50d902SRodney W. Grimes }
4199b50d902SRodney W. Grimes }
4201438aefcSRuslan Ermilov
4211438aefcSRuslan Ermilov /* Types of mail header continuation lines:
4221438aefcSRuslan Ermilov */
4231438aefcSRuslan Ermilov typedef enum {
4241438aefcSRuslan Ermilov hdr_ParagraphStart = -1,
4251438aefcSRuslan Ermilov hdr_NonHeader = 0,
4261438aefcSRuslan Ermilov hdr_Header = 1,
4271438aefcSRuslan Ermilov hdr_Continuation = 2
4281438aefcSRuslan Ermilov } HdrType;
4291438aefcSRuslan Ermilov
4301438aefcSRuslan Ermilov /* Process a stream. This is where the real work happens,
4311438aefcSRuslan Ermilov * except that centering is handled separately.
4321438aefcSRuslan Ermilov */
4331438aefcSRuslan Ermilov static void
process_stream(FILE * stream,const char * name)43486aba0f2SPedro F. Giffuni process_stream(FILE *stream, const char *name)
43586aba0f2SPedro F. Giffuni {
4361438aefcSRuslan Ermilov size_t last_indent = SILLY; /* how many spaces in last indent? */
4371438aefcSRuslan Ermilov size_t para_line_number = 0; /* how many lines already read in this para? */
4381438aefcSRuslan Ermilov size_t first_indent = SILLY; /* indentation of line 0 of paragraph */
4391438aefcSRuslan Ermilov HdrType prev_header_type = hdr_ParagraphStart;
44086aba0f2SPedro F. Giffuni
4411438aefcSRuslan Ermilov /* ^-- header_type of previous line; -1 at para start */
44228d92b74STim J. Robbins wchar_t *line;
4431438aefcSRuslan Ermilov size_t length;
4441438aefcSRuslan Ermilov
44586aba0f2SPedro F. Giffuni if (centerP) {
44686aba0f2SPedro F. Giffuni center_stream(stream, name);
44786aba0f2SPedro F. Giffuni return;
44886aba0f2SPedro F. Giffuni }
4491438aefcSRuslan Ermilov while ((line = get_line(stream, &length)) != NULL) {
4501438aefcSRuslan Ermilov size_t np = indent_length(line, length);
45186aba0f2SPedro F. Giffuni
45286aba0f2SPedro F. Giffuni {
45386aba0f2SPedro F. Giffuni HdrType header_type = hdr_NonHeader;
45486aba0f2SPedro F. Giffuni
4551438aefcSRuslan Ermilov if (grok_mail_headers && prev_header_type != hdr_NonHeader) {
4561438aefcSRuslan Ermilov if (np == 0 && might_be_header(line))
4571438aefcSRuslan Ermilov header_type = hdr_Header;
4581438aefcSRuslan Ermilov else if (np > 0 && prev_header_type > hdr_NonHeader)
4591438aefcSRuslan Ermilov header_type = hdr_Continuation;
4609b50d902SRodney W. Grimes }
46186aba0f2SPedro F. Giffuni /*
46286aba0f2SPedro F. Giffuni * We need a new paragraph if and only if: this line
46386aba0f2SPedro F. Giffuni * is blank, OR it's a troff request (and we don't
46486aba0f2SPedro F. Giffuni * format troff), OR it's a mail header, OR it's not
46586aba0f2SPedro F. Giffuni * a mail header AND the last line was one, OR the
46686aba0f2SPedro F. Giffuni * indentation has changed AND the line isn't a mail
46786aba0f2SPedro F. Giffuni * header continuation line AND this isn't the
46886aba0f2SPedro F. Giffuni * second line of an indented paragraph.
4691438aefcSRuslan Ermilov */
4701438aefcSRuslan Ermilov if (length == 0
471deba2451SRuslan Ermilov || (line[0] == '.' && !format_troff)
4721438aefcSRuslan Ermilov || header_type == hdr_Header
4731438aefcSRuslan Ermilov || (header_type == hdr_NonHeader && prev_header_type > hdr_NonHeader)
4741438aefcSRuslan Ermilov || (np != last_indent
4751438aefcSRuslan Ermilov && header_type != hdr_Continuation
4761438aefcSRuslan Ermilov && (!allow_indented_paragraphs || para_line_number != 1))) {
4771438aefcSRuslan Ermilov new_paragraph(output_in_paragraph ? last_indent : first_indent, np);
4781438aefcSRuslan Ermilov para_line_number = 0;
4791438aefcSRuslan Ermilov first_indent = np;
4801438aefcSRuslan Ermilov last_indent = np;
48186aba0f2SPedro F. Giffuni if (header_type == hdr_Header)
48286aba0f2SPedro F. Giffuni last_indent = 2; /* for cont. lines */
483deba2451SRuslan Ermilov if (length == 0 || (line[0] == '.' && !format_troff)) {
484deba2451SRuslan Ermilov if (length == 0)
48528d92b74STim J. Robbins putwchar('\n');
486deba2451SRuslan Ermilov else
48786aba0f2SPedro F. Giffuni wprintf(L"%.*ls\n", (int)length,
48886aba0f2SPedro F. Giffuni line);
4891438aefcSRuslan Ermilov prev_header_type = hdr_ParagraphStart;
4909b50d902SRodney W. Grimes continue;
4919b50d902SRodney W. Grimes }
49286aba0f2SPedro F. Giffuni } else {
49386aba0f2SPedro F. Giffuni /*
49486aba0f2SPedro F. Giffuni * If this is an indented paragraph other
49586aba0f2SPedro F. Giffuni * than a mail header continuation, set
49686aba0f2SPedro F. Giffuni * |last_indent|.
4979b50d902SRodney W. Grimes */
49886aba0f2SPedro F. Giffuni if (np != last_indent &&
49986aba0f2SPedro F. Giffuni header_type != hdr_Continuation)
5001438aefcSRuslan Ermilov last_indent = np;
5011438aefcSRuslan Ermilov }
5021438aefcSRuslan Ermilov prev_header_type = header_type;
5031438aefcSRuslan Ermilov }
504dc3001cfSJonathan Lemon
50586aba0f2SPedro F. Giffuni {
50686aba0f2SPedro F. Giffuni size_t n = np;
50786aba0f2SPedro F. Giffuni
5081438aefcSRuslan Ermilov while (n < length) {
5091438aefcSRuslan Ermilov /* Find word end and count spaces after it */
5101438aefcSRuslan Ermilov size_t word_length = 0, space_length = 0;
51186aba0f2SPedro F. Giffuni
51286aba0f2SPedro F. Giffuni while (n + word_length < length &&
51386aba0f2SPedro F. Giffuni line[n + word_length] != ' ')
5141438aefcSRuslan Ermilov ++word_length;
5151438aefcSRuslan Ermilov space_length = word_length;
51686aba0f2SPedro F. Giffuni while (n + space_length < length &&
51786aba0f2SPedro F. Giffuni line[n + space_length] == ' ')
5181438aefcSRuslan Ermilov ++space_length;
5191438aefcSRuslan Ermilov /* Send the word to the output machinery. */
5201438aefcSRuslan Ermilov output_word(first_indent, last_indent,
52186aba0f2SPedro F. Giffuni line + n, word_length,
52286aba0f2SPedro F. Giffuni space_length - word_length);
5231438aefcSRuslan Ermilov n += space_length;
5241438aefcSRuslan Ermilov }
5251438aefcSRuslan Ermilov }
5261438aefcSRuslan Ermilov ++para_line_number;
5271438aefcSRuslan Ermilov }
5281438aefcSRuslan Ermilov new_paragraph(output_in_paragraph ? last_indent : first_indent, 0);
52986aba0f2SPedro F. Giffuni if (ferror(stream)) {
53086aba0f2SPedro F. Giffuni warn("%s", name);
53186aba0f2SPedro F. Giffuni ++n_errors;
53286aba0f2SPedro F. Giffuni }
5331438aefcSRuslan Ermilov }
5349b50d902SRodney W. Grimes
5351438aefcSRuslan Ermilov /* How long is the indent on this line?
5361438aefcSRuslan Ermilov */
5371438aefcSRuslan Ermilov static size_t
indent_length(const wchar_t * line,size_t length)53886aba0f2SPedro F. Giffuni indent_length(const wchar_t *line, size_t length)
53986aba0f2SPedro F. Giffuni {
5401438aefcSRuslan Ermilov size_t n = 0;
54186aba0f2SPedro F. Giffuni
54286aba0f2SPedro F. Giffuni while (n < length && *line++ == ' ')
54386aba0f2SPedro F. Giffuni ++n;
5441438aefcSRuslan Ermilov return n;
5451438aefcSRuslan Ermilov }
5461438aefcSRuslan Ermilov
5471438aefcSRuslan Ermilov /* Might this line be a mail header?
5481438aefcSRuslan Ermilov * We deem a line to be a possible header if it matches the
5491438aefcSRuslan Ermilov * Perl regexp /^[A-Z][-A-Za-z0-9]*:\s/. This is *not* the same
5501438aefcSRuslan Ermilov * as in RFC whatever-number-it-is; we want to be gratuitously
5511438aefcSRuslan Ermilov * conservative to avoid mangling ordinary civilised text.
5521438aefcSRuslan Ermilov */
5531438aefcSRuslan Ermilov static int
might_be_header(const wchar_t * line)55486aba0f2SPedro F. Giffuni might_be_header(const wchar_t *line)
55586aba0f2SPedro F. Giffuni {
55686aba0f2SPedro F. Giffuni if (!iswupper(*line++))
55786aba0f2SPedro F. Giffuni return 0;
55886aba0f2SPedro F. Giffuni while (*line && (iswalnum(*line) || *line == '-'))
55986aba0f2SPedro F. Giffuni ++line;
56028d92b74STim J. Robbins return (*line == ':' && iswspace(line[1]));
5611438aefcSRuslan Ermilov }
5621438aefcSRuslan Ermilov
5631438aefcSRuslan Ermilov /* Begin a new paragraph with an indent of |indent| spaces.
5641438aefcSRuslan Ermilov */
5651438aefcSRuslan Ermilov static void
new_paragraph(size_t old_indent,size_t indent)56686aba0f2SPedro F. Giffuni new_paragraph(size_t old_indent, size_t indent)
56786aba0f2SPedro F. Giffuni {
56828d92b74STim J. Robbins if (output_buffer_length) {
56986aba0f2SPedro F. Giffuni if (old_indent > 0)
57086aba0f2SPedro F. Giffuni output_indent(old_indent);
57128d92b74STim J. Robbins wprintf(L"%.*ls\n", (int)output_buffer_length, output_buffer);
5723f8da92bSPoul-Henning Kamp }
57386aba0f2SPedro F. Giffuni x = indent;
57486aba0f2SPedro F. Giffuni x0 = 0;
57586aba0f2SPedro F. Giffuni output_buffer_length = 0;
57686aba0f2SPedro F. Giffuni pending_spaces = 0;
5771438aefcSRuslan Ermilov output_in_paragraph = 0;
5789b50d902SRodney W. Grimes }
5799b50d902SRodney W. Grimes
5801438aefcSRuslan Ermilov /* Output spaces or tabs for leading indentation.
5819b50d902SRodney W. Grimes */
5821438aefcSRuslan Ermilov static void
output_indent(size_t n_spaces)58386aba0f2SPedro F. Giffuni output_indent(size_t n_spaces)
58486aba0f2SPedro F. Giffuni {
5851438aefcSRuslan Ermilov if (output_tab_width) {
5861438aefcSRuslan Ermilov while (n_spaces >= output_tab_width) {
58728d92b74STim J. Robbins putwchar('\t');
5881438aefcSRuslan Ermilov n_spaces -= output_tab_width;
5891438aefcSRuslan Ermilov }
5901438aefcSRuslan Ermilov }
59186aba0f2SPedro F. Giffuni while (n_spaces-- > 0)
59286aba0f2SPedro F. Giffuni putwchar(' ');
5931438aefcSRuslan Ermilov }
5949b50d902SRodney W. Grimes
5951438aefcSRuslan Ermilov /* Output a single word, or add it to the buffer.
5961438aefcSRuslan Ermilov * indent0 and indent1 are the indents to use on the first and subsequent
5971438aefcSRuslan Ermilov * lines of a paragraph. They'll often be the same, of course.
5981438aefcSRuslan Ermilov */
5991438aefcSRuslan Ermilov static void
output_word(size_t indent0,size_t indent1,const wchar_t * word,size_t length,size_t spaces)60086aba0f2SPedro F. Giffuni output_word(size_t indent0, size_t indent1, const wchar_t *word, size_t length, size_t spaces)
60186aba0f2SPedro F. Giffuni {
60228d92b74STim J. Robbins size_t new_x;
6031438aefcSRuslan Ermilov size_t indent = output_in_paragraph ? indent1 : indent0;
60428d92b74STim J. Robbins size_t width;
60528d92b74STim J. Robbins const wchar_t *p;
60628d92b74STim J. Robbins int cwidth;
60728d92b74STim J. Robbins
60828d92b74STim J. Robbins for (p = word, width = 0; p < &word[length]; p++)
60928d92b74STim J. Robbins width += (cwidth = wcwidth(*p)) > 0 ? cwidth : 1;
61028d92b74STim J. Robbins
61128d92b74STim J. Robbins new_x = x + pending_spaces + width;
6121438aefcSRuslan Ermilov
61386aba0f2SPedro F. Giffuni /*
61486aba0f2SPedro F. Giffuni * If either |spaces==0| (at end of line) or |coalesce_spaces_P|
61586aba0f2SPedro F. Giffuni * (squashing internal whitespace), then add just one space; except
61686aba0f2SPedro F. Giffuni * that if the last character was a sentence-ender we actually add
61786aba0f2SPedro F. Giffuni * two spaces.
6181438aefcSRuslan Ermilov */
6191438aefcSRuslan Ermilov if (coalesce_spaces_P || spaces == 0)
62028d92b74STim J. Robbins spaces = wcschr(sentence_enders, word[length - 1]) ? 2 : 1;
6211438aefcSRuslan Ermilov
6221438aefcSRuslan Ermilov if (new_x <= goal_length) {
62386aba0f2SPedro F. Giffuni /*
62486aba0f2SPedro F. Giffuni * After adding the word we still aren't at the goal length,
625*5666643aSGordon Bergling * so clearly we add it to the buffer rather than outputting
62686aba0f2SPedro F. Giffuni * it.
6271438aefcSRuslan Ermilov */
62886aba0f2SPedro F. Giffuni wmemset(output_buffer + output_buffer_length, L' ',
62986aba0f2SPedro F. Giffuni pending_spaces);
63086aba0f2SPedro F. Giffuni x0 += pending_spaces;
63186aba0f2SPedro F. Giffuni x += pending_spaces;
63228d92b74STim J. Robbins output_buffer_length += pending_spaces;
63328d92b74STim J. Robbins wmemcpy(output_buffer + output_buffer_length, word, length);
63486aba0f2SPedro F. Giffuni x0 += width;
63586aba0f2SPedro F. Giffuni x += width;
63686aba0f2SPedro F. Giffuni output_buffer_length += length;
6371438aefcSRuslan Ermilov pending_spaces = spaces;
63886aba0f2SPedro F. Giffuni } else {
63986aba0f2SPedro F. Giffuni /*
64086aba0f2SPedro F. Giffuni * Adding the word takes us past the goal. Print the
64186aba0f2SPedro F. Giffuni * line-so-far, and the word too iff either (1) the lsf is
64286aba0f2SPedro F. Giffuni * empty or (2) that makes us nearer the goal but doesn't
64386aba0f2SPedro F. Giffuni * take us over the limit, or (3) the word on its own takes
64486aba0f2SPedro F. Giffuni * us over the limit. In case (3) we put a newline in
64586aba0f2SPedro F. Giffuni * between.
6461438aefcSRuslan Ermilov */
64786aba0f2SPedro F. Giffuni if (indent > 0)
64886aba0f2SPedro F. Giffuni output_indent(indent);
64928d92b74STim J. Robbins wprintf(L"%.*ls", (int)output_buffer_length, output_buffer);
65086aba0f2SPedro F. Giffuni if (x0 == 0 || (new_x <= max_length &&
65186aba0f2SPedro F. Giffuni new_x - goal_length <= goal_length - x)) {
65228d92b74STim J. Robbins wprintf(L"%*ls", (int)pending_spaces, L"");
6531438aefcSRuslan Ermilov goto write_out_word;
65486aba0f2SPedro F. Giffuni } else {
65586aba0f2SPedro F. Giffuni /*
65686aba0f2SPedro F. Giffuni * If the word takes us over the limit on its own,
65786aba0f2SPedro F. Giffuni * just spit it out and don't bother buffering it.
6581438aefcSRuslan Ermilov */
65928d92b74STim J. Robbins if (indent + width > max_length) {
66028d92b74STim J. Robbins putwchar('\n');
66186aba0f2SPedro F. Giffuni if (indent > 0)
66286aba0f2SPedro F. Giffuni output_indent(indent);
6631438aefcSRuslan Ermilov write_out_word:
66428d92b74STim J. Robbins wprintf(L"%.*ls", (int)length, word);
66586aba0f2SPedro F. Giffuni x0 = 0;
66686aba0f2SPedro F. Giffuni x = indent1;
66786aba0f2SPedro F. Giffuni pending_spaces = 0;
66828d92b74STim J. Robbins output_buffer_length = 0;
66986aba0f2SPedro F. Giffuni } else {
67028d92b74STim J. Robbins wmemcpy(output_buffer, word, length);
67186aba0f2SPedro F. Giffuni x0 = width;
67286aba0f2SPedro F. Giffuni x = width + indent1;
67386aba0f2SPedro F. Giffuni pending_spaces = spaces;
67428d92b74STim J. Robbins output_buffer_length = length;
6759b50d902SRodney W. Grimes }
6769b50d902SRodney W. Grimes }
67728d92b74STim J. Robbins putwchar('\n');
6781438aefcSRuslan Ermilov output_in_paragraph = 1;
6799c61e111SJoerg Wunsch }
6809c61e111SJoerg Wunsch }
6819c61e111SJoerg Wunsch
6821438aefcSRuslan Ermilov /* Process a stream, but just center its lines rather than trying to
6831438aefcSRuslan Ermilov * format them neatly.
6849b50d902SRodney W. Grimes */
6851438aefcSRuslan Ermilov static void
center_stream(FILE * stream,const char * name)68686aba0f2SPedro F. Giffuni center_stream(FILE *stream, const char *name)
68786aba0f2SPedro F. Giffuni {
68828d92b74STim J. Robbins wchar_t *line, *p;
6891438aefcSRuslan Ermilov size_t length;
69028d92b74STim J. Robbins size_t width;
69128d92b74STim J. Robbins int cwidth;
69286aba0f2SPedro F. Giffuni
693074e77e6SPedro F. Giffuni while ((line = get_line(stream, &length)) != NULL) {
6941438aefcSRuslan Ermilov size_t l = length;
69586aba0f2SPedro F. Giffuni
69686aba0f2SPedro F. Giffuni while (l > 0 && iswspace(*line)) {
69786aba0f2SPedro F. Giffuni ++line;
69886aba0f2SPedro F. Giffuni --l;
69986aba0f2SPedro F. Giffuni }
7001438aefcSRuslan Ermilov length = l;
70128d92b74STim J. Robbins for (p = line, width = 0; p < &line[length]; p++)
70228d92b74STim J. Robbins width += (cwidth = wcwidth(*p)) > 0 ? cwidth : 1;
70328d92b74STim J. Robbins l = width;
70486aba0f2SPedro F. Giffuni while (l < goal_length) {
70586aba0f2SPedro F. Giffuni putwchar(' ');
70686aba0f2SPedro F. Giffuni l += 2;
70786aba0f2SPedro F. Giffuni }
70828d92b74STim J. Robbins wprintf(L"%.*ls\n", (int)length, line);
7091438aefcSRuslan Ermilov }
71086aba0f2SPedro F. Giffuni if (ferror(stream)) {
71186aba0f2SPedro F. Giffuni warn("%s", name);
71286aba0f2SPedro F. Giffuni ++n_errors;
71386aba0f2SPedro F. Giffuni }
7149b50d902SRodney W. Grimes }
7159b50d902SRodney W. Grimes
7161438aefcSRuslan Ermilov /* Get a single line from a stream. Expand tabs, strip control
7171438aefcSRuslan Ermilov * characters and trailing whitespace, and handle backspaces.
7181438aefcSRuslan Ermilov * Return the address of the buffer containing the line, and
7191438aefcSRuslan Ermilov * put the length of the line in |lengthp|.
7201438aefcSRuslan Ermilov * This can cope with arbitrarily long lines, and with lines
7211438aefcSRuslan Ermilov * without terminating \n.
7221438aefcSRuslan Ermilov * If there are no characters left or an error happens, we
7231438aefcSRuslan Ermilov * return 0.
7241438aefcSRuslan Ermilov * Don't confuse |spaces_pending| here with the global
7251438aefcSRuslan Ermilov * |pending_spaces|.
7269b50d902SRodney W. Grimes */
72728d92b74STim J. Robbins static wchar_t *
get_line(FILE * stream,size_t * lengthp)72886aba0f2SPedro F. Giffuni get_line(FILE *stream, size_t *lengthp)
72986aba0f2SPedro F. Giffuni {
73028d92b74STim J. Robbins static wchar_t *buf = NULL;
7311438aefcSRuslan Ermilov static size_t length = 0;
7321438aefcSRuslan Ermilov size_t len = 0;
73328d92b74STim J. Robbins wint_t ch;
7341438aefcSRuslan Ermilov size_t spaces_pending = 0;
735deba2451SRuslan Ermilov int troff = 0;
73628d92b74STim J. Robbins size_t col = 0;
73728d92b74STim J. Robbins int cwidth;
7381438aefcSRuslan Ermilov
73986aba0f2SPedro F. Giffuni if (buf == NULL) {
74086aba0f2SPedro F. Giffuni length = 100;
74186aba0f2SPedro F. Giffuni buf = XMALLOC(length * sizeof(wchar_t));
74286aba0f2SPedro F. Giffuni }
74328d92b74STim J. Robbins while ((ch = getwc(stream)) != '\n' && ch != WEOF) {
74486aba0f2SPedro F. Giffuni if (len + spaces_pending == 0 && ch == '.' && !format_troff)
74586aba0f2SPedro F. Giffuni troff = 1;
74686aba0f2SPedro F. Giffuni if (ch == ' ')
74786aba0f2SPedro F. Giffuni ++spaces_pending;
74828d92b74STim J. Robbins else if (troff || iswprint(ch)) {
7491438aefcSRuslan Ermilov while (len + spaces_pending >= length) {
75086aba0f2SPedro F. Giffuni length *= 2;
75186aba0f2SPedro F. Giffuni buf = xrealloc(buf, length * sizeof(wchar_t));
7529b50d902SRodney W. Grimes }
75386aba0f2SPedro F. Giffuni while (spaces_pending > 0) {
75486aba0f2SPedro F. Giffuni --spaces_pending;
75586aba0f2SPedro F. Giffuni buf[len++] = ' ';
75686aba0f2SPedro F. Giffuni col++;
75786aba0f2SPedro F. Giffuni }
7581438aefcSRuslan Ermilov buf[len++] = ch;
75928d92b74STim J. Robbins col += (cwidth = wcwidth(ch)) > 0 ? cwidth : 1;
76086aba0f2SPedro F. Giffuni } else if (ch == '\t')
76186aba0f2SPedro F. Giffuni spaces_pending += tab_width -
76286aba0f2SPedro F. Giffuni (col + spaces_pending) % tab_width;
76386aba0f2SPedro F. Giffuni else if (ch == '\b') {
76486aba0f2SPedro F. Giffuni if (len)
76586aba0f2SPedro F. Giffuni --len;
76686aba0f2SPedro F. Giffuni if (col)
76786aba0f2SPedro F. Giffuni --col;
7689b50d902SRodney W. Grimes }
7691438aefcSRuslan Ermilov }
7701438aefcSRuslan Ermilov *lengthp = len;
77128d92b74STim J. Robbins return (len > 0 || ch != WEOF) ? buf : 0;
7729b50d902SRodney W. Grimes }
7739b50d902SRodney W. Grimes
7741438aefcSRuslan Ermilov /* (Re)allocate some memory, exiting with an error if we can't.
7759b50d902SRodney W. Grimes */
7761438aefcSRuslan Ermilov static void *
xrealloc(void * ptr,size_t nbytes)77786aba0f2SPedro F. Giffuni xrealloc(void *ptr, size_t nbytes)
77886aba0f2SPedro F. Giffuni {
7791438aefcSRuslan Ermilov void *p = realloc(ptr, nbytes);
78086aba0f2SPedro F. Giffuni
78186aba0f2SPedro F. Giffuni if (p == NULL)
78286aba0f2SPedro F. Giffuni errx(EX_OSERR, "out of memory");
7831438aefcSRuslan Ermilov return p;
7849b50d902SRodney W. Grimes }
785