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> 176e026a48cSDavid E. O'Brien __FBSDID("$FreeBSD$"); 1779b50d902SRodney W. Grimes 17863ffb113SPhilippe Charnier #include <err.h> 179a2229407SJuli Mallett #include <limits.h> 1808bbd9072SAndrey A. Chernov #include <locale.h> 18163ffb113SPhilippe Charnier #include <stdio.h> 182e6c267f1SJoerg Wunsch #include <stdlib.h> 183821df508SXin LI #include <string.h> 1841438aefcSRuslan Ermilov #include <sysexits.h> 1851438aefcSRuslan Ermilov #include <unistd.h> 18628d92b74STim J. Robbins #include <wchar.h> 18728d92b74STim J. Robbins #include <wctype.h> 1889b50d902SRodney W. Grimes 1891438aefcSRuslan Ermilov /* Something that, we hope, will never be a genuine line length, 1901438aefcSRuslan Ermilov * indentation etc. 1919b50d902SRodney W. Grimes */ 1921438aefcSRuslan Ermilov #define SILLY ((size_t)-1) 1939b50d902SRodney W. Grimes 1941438aefcSRuslan Ermilov /* I used to use |strtoul| for this, but (1) not all systems have it 1951438aefcSRuslan Ermilov * and (2) it's probably better to use |strtol| to detect negative 1961438aefcSRuslan Ermilov * numbers better. 1971438aefcSRuslan Ermilov * If |fussyp==0| then we don't complain about non-numbers 1981438aefcSRuslan Ermilov * (returning 0 instead), but we do complain about bad numbers. 1999b50d902SRodney W. Grimes */ 2001438aefcSRuslan Ermilov static size_t 20186aba0f2SPedro F. Giffuni get_positive(const char *s, const char *err_mess, int fussyP) 20286aba0f2SPedro F. Giffuni { 2031438aefcSRuslan Ermilov char *t; 2041438aefcSRuslan Ermilov long result = strtol(s, &t, 0); 20586aba0f2SPedro F. Giffuni 20686aba0f2SPedro F. Giffuni if (*t) { 20786aba0f2SPedro F. Giffuni if (fussyP) 20886aba0f2SPedro F. Giffuni goto Lose; 20986aba0f2SPedro F. Giffuni else 21086aba0f2SPedro F. Giffuni return 0; 21186aba0f2SPedro F. Giffuni } 21286aba0f2SPedro F. Giffuni if (result <= 0) { 21386aba0f2SPedro F. Giffuni Lose: errx(EX_USAGE, "%s", err_mess); 21486aba0f2SPedro F. Giffuni } 2151438aefcSRuslan Ermilov return (size_t)result; 2161438aefcSRuslan Ermilov } 2179b50d902SRodney W. Grimes 218ccb8bea4SRuslan Ermilov static size_t 21986aba0f2SPedro F. Giffuni get_nonnegative(const char *s, const char *err_mess, int fussyP) 22086aba0f2SPedro F. Giffuni { 221ccb8bea4SRuslan Ermilov char *t; 222ccb8bea4SRuslan Ermilov long result = strtol(s, &t, 0); 22386aba0f2SPedro F. Giffuni 22486aba0f2SPedro F. Giffuni if (*t) { 22586aba0f2SPedro F. Giffuni if (fussyP) 22686aba0f2SPedro F. Giffuni goto Lose; 22786aba0f2SPedro F. Giffuni else 22886aba0f2SPedro F. Giffuni return 0; 22986aba0f2SPedro F. Giffuni } 23086aba0f2SPedro F. Giffuni if (result < 0) { 23186aba0f2SPedro F. Giffuni Lose: errx(EX_USAGE, "%s", err_mess); 23286aba0f2SPedro F. Giffuni } 233ccb8bea4SRuslan Ermilov return (size_t)result; 234ccb8bea4SRuslan Ermilov } 235ccb8bea4SRuslan Ermilov 2361438aefcSRuslan Ermilov /* Global variables */ 2379b50d902SRodney W. Grimes 2381438aefcSRuslan Ermilov static int centerP = 0; /* Try to center lines? */ 2391438aefcSRuslan Ermilov static size_t goal_length = 0; /* Target length for output lines */ 2401438aefcSRuslan Ermilov static size_t max_length = 0; /* Maximum length for output lines */ 2411438aefcSRuslan Ermilov static int coalesce_spaces_P = 0; /* Coalesce multiple whitespace -> ' ' ? */ 2421438aefcSRuslan Ermilov static int allow_indented_paragraphs = 0; /* Can first line have diff. ind.? */ 2431438aefcSRuslan Ermilov static int tab_width = 8; /* Number of spaces per tab stop */ 244ccb8bea4SRuslan Ermilov static size_t output_tab_width = 8; /* Ditto, when squashing leading spaces */ 24528d92b74STim J. Robbins static const wchar_t *sentence_enders = L".?!"; /* Double-space after these */ 2461438aefcSRuslan Ermilov static int grok_mail_headers = 0; /* treat embedded mail headers magically? */ 247deba2451SRuslan Ermilov static int format_troff = 0; /* Format troff? */ 2489b50d902SRodney W. Grimes 2491438aefcSRuslan Ermilov static int n_errors = 0; /* Number of failed files. Return on exit. */ 250*074e77e6SPedro F. Giffuni static wchar_t *output_buffer = NULL; /* Output line will be built here */ 2511438aefcSRuslan Ermilov static size_t x; /* Horizontal position in output line */ 2521438aefcSRuslan Ermilov static size_t x0; /* Ditto, ignoring leading whitespace */ 25328d92b74STim J. Robbins static size_t output_buffer_length = 0; 2541438aefcSRuslan Ermilov static size_t pending_spaces; /* Spaces to add before next word */ 2551438aefcSRuslan Ermilov static int output_in_paragraph = 0; /* Any of current para written out yet? */ 25663ffb113SPhilippe Charnier 2571438aefcSRuslan Ermilov /* Prototypes */ 2581438aefcSRuslan Ermilov 2591438aefcSRuslan Ermilov static void process_named_file(const char *); 2601438aefcSRuslan Ermilov static void process_stream(FILE *, const char *); 26128d92b74STim J. Robbins static size_t indent_length(const wchar_t *, size_t); 26228d92b74STim J. Robbins static int might_be_header(const wchar_t *); 2631438aefcSRuslan Ermilov static void new_paragraph(size_t, size_t); 26486aba0f2SPedro F. Giffuni static void output_word(size_t, size_t, const wchar_t *, size_t, size_t); 2651438aefcSRuslan Ermilov static void output_indent(size_t); 2661438aefcSRuslan Ermilov static void center_stream(FILE *, const char *); 26728d92b74STim J. Robbins static wchar_t *get_line(FILE *, size_t *); 2681438aefcSRuslan Ermilov static void *xrealloc(void *, size_t); 2691438aefcSRuslan Ermilov 2701438aefcSRuslan Ermilov #define XMALLOC(x) xrealloc(0,x) 2711438aefcSRuslan Ermilov 2721438aefcSRuslan Ermilov /* Here is perhaps the right place to mention that this code is 2731438aefcSRuslan Ermilov * all in top-down order. Hence, |main| comes first. 2749b50d902SRodney W. Grimes */ 27563ffb113SPhilippe Charnier int 27686aba0f2SPedro F. Giffuni main(int argc, char *argv[]) 27786aba0f2SPedro F. Giffuni { 2781438aefcSRuslan Ermilov int ch; /* used for |getopt| processing */ 27928d92b74STim J. Robbins wchar_t *tmp; 28028d92b74STim J. Robbins size_t len; 28128d92b74STim J. Robbins const char *src; 2829b50d902SRodney W. Grimes 2838bbd9072SAndrey A. Chernov (void)setlocale(LC_CTYPE, ""); 2848bbd9072SAndrey A. Chernov 2851438aefcSRuslan Ermilov /* 1. Grok parameters. */ 2861438aefcSRuslan Ermilov 287deba2451SRuslan Ermilov while ((ch = getopt(argc, argv, "0123456789cd:hl:mnpst:w:")) != -1) 2881438aefcSRuslan Ermilov switch (ch) { 2891438aefcSRuslan Ermilov case 'c': 2901438aefcSRuslan Ermilov centerP = 1; 291deba2451SRuslan Ermilov format_troff = 1; 2921438aefcSRuslan Ermilov continue; 2931438aefcSRuslan Ermilov case 'd': 29428d92b74STim J. Robbins src = optarg; 29528d92b74STim J. Robbins len = mbsrtowcs(NULL, &src, 0, NULL); 29628d92b74STim J. Robbins if (len == (size_t)-1) 29728d92b74STim J. Robbins err(EX_USAGE, "bad sentence-ending character set"); 29828d92b74STim J. Robbins tmp = XMALLOC((len + 1) * sizeof(wchar_t)); 29928d92b74STim J. Robbins mbsrtowcs(tmp, &src, len + 1, NULL); 30028d92b74STim J. Robbins sentence_enders = tmp; 3011438aefcSRuslan Ermilov continue; 3021438aefcSRuslan Ermilov case 'l': 3031438aefcSRuslan Ermilov output_tab_width 304ccb8bea4SRuslan Ermilov = get_nonnegative(optarg, "output tab width must be non-negative", 1); 3051438aefcSRuslan Ermilov continue; 3061438aefcSRuslan Ermilov case 'm': 3071438aefcSRuslan Ermilov grok_mail_headers = 1; 3081438aefcSRuslan Ermilov continue; 309deba2451SRuslan Ermilov case 'n': 310deba2451SRuslan Ermilov format_troff = 1; 311deba2451SRuslan Ermilov continue; 3121438aefcSRuslan Ermilov case 'p': 3131438aefcSRuslan Ermilov allow_indented_paragraphs = 1; 3141438aefcSRuslan Ermilov continue; 3151438aefcSRuslan Ermilov case 's': 3161438aefcSRuslan Ermilov coalesce_spaces_P = 1; 3171438aefcSRuslan Ermilov continue; 3181438aefcSRuslan Ermilov case 't': 3191438aefcSRuslan Ermilov tab_width = get_positive(optarg, "tab width must be positive", 1); 3201438aefcSRuslan Ermilov continue; 3211438aefcSRuslan Ermilov case 'w': 3221438aefcSRuslan Ermilov goal_length = get_positive(optarg, "width must be positive", 1); 3231438aefcSRuslan Ermilov max_length = goal_length; 3241438aefcSRuslan Ermilov continue; 3251438aefcSRuslan Ermilov case '0': case '1': case '2': case '3': case '4': case '5': 3261438aefcSRuslan Ermilov case '6': case '7': case '8': case '9': 32786aba0f2SPedro F. Giffuni /* 32886aba0f2SPedro F. Giffuni * XXX this is not a stylistically approved use of 32986aba0f2SPedro F. Giffuni * getopt() 33086aba0f2SPedro F. Giffuni */ 3311438aefcSRuslan Ermilov if (goal_length == 0) { 3321438aefcSRuslan Ermilov char *p; 33386aba0f2SPedro F. Giffuni 3341438aefcSRuslan Ermilov p = argv[optind - 1]; 3351438aefcSRuslan Ermilov if (p[0] == '-' && p[1] == ch && !p[2]) 3361438aefcSRuslan Ermilov goal_length = get_positive(++p, "width must be nonzero", 1); 3371438aefcSRuslan Ermilov else 3381438aefcSRuslan Ermilov goal_length = get_positive(argv[optind] + 1, 3391438aefcSRuslan Ermilov "width must be nonzero", 1); 3401438aefcSRuslan Ermilov max_length = goal_length; 3411438aefcSRuslan Ermilov } 3421438aefcSRuslan Ermilov continue; 34386aba0f2SPedro F. Giffuni case 'h': 34486aba0f2SPedro F. Giffuni default: 3451438aefcSRuslan Ermilov fprintf(stderr, 346d3974088SDag-Erling Smørgrav "usage: fmt [-cmps] [-d chars] [-l num] [-t num]\n" 3471438aefcSRuslan Ermilov " [-w width | -width | goal [maximum]] [file ...]\n" 3481438aefcSRuslan Ermilov "Options: -c center each line instead of formatting\n" 3491438aefcSRuslan Ermilov " -d <chars> double-space after <chars> at line end\n" 3501438aefcSRuslan Ermilov " -l <n> turn each <n> spaces at start of line into a tab\n" 3511438aefcSRuslan Ermilov " -m try to make sure mail header lines stay separate\n" 352deba2451SRuslan Ermilov " -n format lines beginning with a dot\n" 3531438aefcSRuslan Ermilov " -p allow indented paragraphs\n" 3541438aefcSRuslan Ermilov " -s coalesce whitespace inside lines\n" 3551438aefcSRuslan Ermilov " -t <n> have tabs every <n> columns\n" 3561438aefcSRuslan Ermilov " -w <n> set maximum width to <n>\n" 3571438aefcSRuslan Ermilov " goal set target width to goal\n"); 3581438aefcSRuslan Ermilov exit(ch == 'h' ? 0 : EX_USAGE); 3591438aefcSRuslan Ermilov } 36086aba0f2SPedro F. Giffuni argc -= optind; 36186aba0f2SPedro F. Giffuni argv += optind; 3621438aefcSRuslan Ermilov 3631438aefcSRuslan Ermilov /* [ goal [ maximum ] ] */ 3641438aefcSRuslan Ermilov 3651438aefcSRuslan Ermilov if (argc > 0 && goal_length == 0 3661438aefcSRuslan Ermilov && (goal_length = get_positive(*argv, "goal length must be positive", 0)) 3671438aefcSRuslan Ermilov != 0) { 36886aba0f2SPedro F. Giffuni --argc; 36986aba0f2SPedro F. Giffuni ++argv; 3701438aefcSRuslan Ermilov if (argc > 0 3711438aefcSRuslan Ermilov && (max_length = get_positive(*argv, "max length must be positive", 0)) 3721438aefcSRuslan Ermilov != 0) { 37386aba0f2SPedro F. Giffuni --argc; 37486aba0f2SPedro F. Giffuni ++argv; 3751438aefcSRuslan Ermilov if (max_length < goal_length) 3761438aefcSRuslan Ermilov errx(EX_USAGE, "max length must be >= goal length"); 3771438aefcSRuslan Ermilov } 3781438aefcSRuslan Ermilov } 37986aba0f2SPedro F. Giffuni if (goal_length == 0) 38086aba0f2SPedro F. Giffuni goal_length = 65; 38186aba0f2SPedro F. Giffuni if (max_length == 0) 38286aba0f2SPedro F. Giffuni max_length = goal_length + 10; 38386aba0f2SPedro F. Giffuni if (max_length >= SIZE_T_MAX / sizeof(wchar_t)) 38486aba0f2SPedro F. Giffuni errx(EX_USAGE, "max length too large"); 38528d92b74STim J. Robbins /* really needn't be longer */ 38628d92b74STim J. Robbins output_buffer = XMALLOC((max_length + 1) * sizeof(wchar_t)); 3871438aefcSRuslan Ermilov 3881438aefcSRuslan Ermilov /* 2. Process files. */ 3891438aefcSRuslan Ermilov 3901438aefcSRuslan Ermilov if (argc > 0) { 39186aba0f2SPedro F. Giffuni while (argc-- > 0) 39286aba0f2SPedro F. Giffuni process_named_file(*argv++); 39386aba0f2SPedro F. Giffuni } else { 3941438aefcSRuslan Ermilov process_stream(stdin, "standard input"); 3951438aefcSRuslan Ermilov } 3961438aefcSRuslan Ermilov 3971438aefcSRuslan Ermilov /* We're done. */ 3981438aefcSRuslan Ermilov 3991438aefcSRuslan Ermilov return n_errors ? EX_NOINPUT : 0; 4001438aefcSRuslan Ermilov 4011438aefcSRuslan Ermilov } 4021438aefcSRuslan Ermilov 4031438aefcSRuslan Ermilov /* Process a single file, given its name. 4049b50d902SRodney W. Grimes */ 4051438aefcSRuslan Ermilov static void 40686aba0f2SPedro F. Giffuni process_named_file(const char *name) 40786aba0f2SPedro F. Giffuni { 4081438aefcSRuslan Ermilov FILE *f = fopen(name, "r"); 40986aba0f2SPedro F. Giffuni 41086aba0f2SPedro F. Giffuni if (!f) { 41186aba0f2SPedro F. Giffuni warn("%s", name); 41286aba0f2SPedro F. Giffuni ++n_errors; 41386aba0f2SPedro F. Giffuni } else { 4141438aefcSRuslan Ermilov process_stream(f, name); 41586aba0f2SPedro F. Giffuni if (ferror(f)) { 41686aba0f2SPedro F. Giffuni warn("%s", name); 41786aba0f2SPedro F. Giffuni ++n_errors; 41886aba0f2SPedro F. Giffuni } 4191438aefcSRuslan Ermilov fclose(f); 4209b50d902SRodney W. Grimes } 4219b50d902SRodney W. Grimes } 4221438aefcSRuslan Ermilov 4231438aefcSRuslan Ermilov /* Types of mail header continuation lines: 4241438aefcSRuslan Ermilov */ 4251438aefcSRuslan Ermilov typedef enum { 4261438aefcSRuslan Ermilov hdr_ParagraphStart = -1, 4271438aefcSRuslan Ermilov hdr_NonHeader = 0, 4281438aefcSRuslan Ermilov hdr_Header = 1, 4291438aefcSRuslan Ermilov hdr_Continuation = 2 4301438aefcSRuslan Ermilov } HdrType; 4311438aefcSRuslan Ermilov 4321438aefcSRuslan Ermilov /* Process a stream. This is where the real work happens, 4331438aefcSRuslan Ermilov * except that centering is handled separately. 4341438aefcSRuslan Ermilov */ 4351438aefcSRuslan Ermilov static void 43686aba0f2SPedro F. Giffuni process_stream(FILE *stream, const char *name) 43786aba0f2SPedro F. Giffuni { 4381438aefcSRuslan Ermilov size_t last_indent = SILLY; /* how many spaces in last indent? */ 4391438aefcSRuslan Ermilov size_t para_line_number = 0; /* how many lines already read in this para? */ 4401438aefcSRuslan Ermilov size_t first_indent = SILLY; /* indentation of line 0 of paragraph */ 4411438aefcSRuslan Ermilov HdrType prev_header_type = hdr_ParagraphStart; 44286aba0f2SPedro F. Giffuni 4431438aefcSRuslan Ermilov /* ^-- header_type of previous line; -1 at para start */ 44428d92b74STim J. Robbins wchar_t *line; 4451438aefcSRuslan Ermilov size_t length; 4461438aefcSRuslan Ermilov 44786aba0f2SPedro F. Giffuni if (centerP) { 44886aba0f2SPedro F. Giffuni center_stream(stream, name); 44986aba0f2SPedro F. Giffuni return; 45086aba0f2SPedro F. Giffuni } 4511438aefcSRuslan Ermilov while ((line = get_line(stream, &length)) != NULL) { 4521438aefcSRuslan Ermilov size_t np = indent_length(line, length); 45386aba0f2SPedro F. Giffuni 45486aba0f2SPedro F. Giffuni { 45586aba0f2SPedro F. Giffuni HdrType header_type = hdr_NonHeader; 45686aba0f2SPedro F. Giffuni 4571438aefcSRuslan Ermilov if (grok_mail_headers && prev_header_type != hdr_NonHeader) { 4581438aefcSRuslan Ermilov if (np == 0 && might_be_header(line)) 4591438aefcSRuslan Ermilov header_type = hdr_Header; 4601438aefcSRuslan Ermilov else if (np > 0 && prev_header_type > hdr_NonHeader) 4611438aefcSRuslan Ermilov header_type = hdr_Continuation; 4629b50d902SRodney W. Grimes } 46386aba0f2SPedro F. Giffuni /* 46486aba0f2SPedro F. Giffuni * We need a new paragraph if and only if: this line 46586aba0f2SPedro F. Giffuni * is blank, OR it's a troff request (and we don't 46686aba0f2SPedro F. Giffuni * format troff), OR it's a mail header, OR it's not 46786aba0f2SPedro F. Giffuni * a mail header AND the last line was one, OR the 46886aba0f2SPedro F. Giffuni * indentation has changed AND the line isn't a mail 46986aba0f2SPedro F. Giffuni * header continuation line AND this isn't the 47086aba0f2SPedro F. Giffuni * second line of an indented paragraph. 4711438aefcSRuslan Ermilov */ 4721438aefcSRuslan Ermilov if (length == 0 473deba2451SRuslan Ermilov || (line[0] == '.' && !format_troff) 4741438aefcSRuslan Ermilov || header_type == hdr_Header 4751438aefcSRuslan Ermilov || (header_type == hdr_NonHeader && prev_header_type > hdr_NonHeader) 4761438aefcSRuslan Ermilov || (np != last_indent 4771438aefcSRuslan Ermilov && header_type != hdr_Continuation 4781438aefcSRuslan Ermilov && (!allow_indented_paragraphs || para_line_number != 1))) { 4791438aefcSRuslan Ermilov new_paragraph(output_in_paragraph ? last_indent : first_indent, np); 4801438aefcSRuslan Ermilov para_line_number = 0; 4811438aefcSRuslan Ermilov first_indent = np; 4821438aefcSRuslan Ermilov last_indent = np; 48386aba0f2SPedro F. Giffuni if (header_type == hdr_Header) 48486aba0f2SPedro F. Giffuni last_indent = 2; /* for cont. lines */ 485deba2451SRuslan Ermilov if (length == 0 || (line[0] == '.' && !format_troff)) { 486deba2451SRuslan Ermilov if (length == 0) 48728d92b74STim J. Robbins putwchar('\n'); 488deba2451SRuslan Ermilov else 48986aba0f2SPedro F. Giffuni wprintf(L"%.*ls\n", (int)length, 49086aba0f2SPedro F. Giffuni line); 4911438aefcSRuslan Ermilov prev_header_type = hdr_ParagraphStart; 4929b50d902SRodney W. Grimes continue; 4939b50d902SRodney W. Grimes } 49486aba0f2SPedro F. Giffuni } else { 49586aba0f2SPedro F. Giffuni /* 49686aba0f2SPedro F. Giffuni * If this is an indented paragraph other 49786aba0f2SPedro F. Giffuni * than a mail header continuation, set 49886aba0f2SPedro F. Giffuni * |last_indent|. 4999b50d902SRodney W. Grimes */ 50086aba0f2SPedro F. Giffuni if (np != last_indent && 50186aba0f2SPedro F. Giffuni header_type != hdr_Continuation) 5021438aefcSRuslan Ermilov last_indent = np; 5031438aefcSRuslan Ermilov } 5041438aefcSRuslan Ermilov prev_header_type = header_type; 5051438aefcSRuslan Ermilov } 506dc3001cfSJonathan Lemon 50786aba0f2SPedro F. Giffuni { 50886aba0f2SPedro F. Giffuni size_t n = np; 50986aba0f2SPedro F. Giffuni 5101438aefcSRuslan Ermilov while (n < length) { 5111438aefcSRuslan Ermilov /* Find word end and count spaces after it */ 5121438aefcSRuslan Ermilov size_t word_length = 0, space_length = 0; 51386aba0f2SPedro F. Giffuni 51486aba0f2SPedro F. Giffuni while (n + word_length < length && 51586aba0f2SPedro F. Giffuni line[n + word_length] != ' ') 5161438aefcSRuslan Ermilov ++word_length; 5171438aefcSRuslan Ermilov space_length = word_length; 51886aba0f2SPedro F. Giffuni while (n + space_length < length && 51986aba0f2SPedro F. Giffuni line[n + space_length] == ' ') 5201438aefcSRuslan Ermilov ++space_length; 5211438aefcSRuslan Ermilov /* Send the word to the output machinery. */ 5221438aefcSRuslan Ermilov output_word(first_indent, last_indent, 52386aba0f2SPedro F. Giffuni line + n, word_length, 52486aba0f2SPedro F. Giffuni space_length - word_length); 5251438aefcSRuslan Ermilov n += space_length; 5261438aefcSRuslan Ermilov } 5271438aefcSRuslan Ermilov } 5281438aefcSRuslan Ermilov ++para_line_number; 5291438aefcSRuslan Ermilov } 5301438aefcSRuslan Ermilov new_paragraph(output_in_paragraph ? last_indent : first_indent, 0); 53186aba0f2SPedro F. Giffuni if (ferror(stream)) { 53286aba0f2SPedro F. Giffuni warn("%s", name); 53386aba0f2SPedro F. Giffuni ++n_errors; 53486aba0f2SPedro F. Giffuni } 5351438aefcSRuslan Ermilov } 5369b50d902SRodney W. Grimes 5371438aefcSRuslan Ermilov /* How long is the indent on this line? 5381438aefcSRuslan Ermilov */ 5391438aefcSRuslan Ermilov static size_t 54086aba0f2SPedro F. Giffuni indent_length(const wchar_t *line, size_t length) 54186aba0f2SPedro F. Giffuni { 5421438aefcSRuslan Ermilov size_t n = 0; 54386aba0f2SPedro F. Giffuni 54486aba0f2SPedro F. Giffuni while (n < length && *line++ == ' ') 54586aba0f2SPedro F. Giffuni ++n; 5461438aefcSRuslan Ermilov return n; 5471438aefcSRuslan Ermilov } 5481438aefcSRuslan Ermilov 5491438aefcSRuslan Ermilov /* Might this line be a mail header? 5501438aefcSRuslan Ermilov * We deem a line to be a possible header if it matches the 5511438aefcSRuslan Ermilov * Perl regexp /^[A-Z][-A-Za-z0-9]*:\s/. This is *not* the same 5521438aefcSRuslan Ermilov * as in RFC whatever-number-it-is; we want to be gratuitously 5531438aefcSRuslan Ermilov * conservative to avoid mangling ordinary civilised text. 5541438aefcSRuslan Ermilov */ 5551438aefcSRuslan Ermilov static int 55686aba0f2SPedro F. Giffuni might_be_header(const wchar_t *line) 55786aba0f2SPedro F. Giffuni { 55886aba0f2SPedro F. Giffuni if (!iswupper(*line++)) 55986aba0f2SPedro F. Giffuni return 0; 56086aba0f2SPedro F. Giffuni while (*line && (iswalnum(*line) || *line == '-')) 56186aba0f2SPedro F. Giffuni ++line; 56228d92b74STim J. Robbins return (*line == ':' && iswspace(line[1])); 5631438aefcSRuslan Ermilov } 5641438aefcSRuslan Ermilov 5651438aefcSRuslan Ermilov /* Begin a new paragraph with an indent of |indent| spaces. 5661438aefcSRuslan Ermilov */ 5671438aefcSRuslan Ermilov static void 56886aba0f2SPedro F. Giffuni new_paragraph(size_t old_indent, size_t indent) 56986aba0f2SPedro F. Giffuni { 57028d92b74STim J. Robbins if (output_buffer_length) { 57186aba0f2SPedro F. Giffuni if (old_indent > 0) 57286aba0f2SPedro F. Giffuni output_indent(old_indent); 57328d92b74STim J. Robbins wprintf(L"%.*ls\n", (int)output_buffer_length, output_buffer); 5743f8da92bSPoul-Henning Kamp } 57586aba0f2SPedro F. Giffuni x = indent; 57686aba0f2SPedro F. Giffuni x0 = 0; 57786aba0f2SPedro F. Giffuni output_buffer_length = 0; 57886aba0f2SPedro F. Giffuni pending_spaces = 0; 5791438aefcSRuslan Ermilov output_in_paragraph = 0; 5809b50d902SRodney W. Grimes } 5819b50d902SRodney W. Grimes 5821438aefcSRuslan Ermilov /* Output spaces or tabs for leading indentation. 5839b50d902SRodney W. Grimes */ 5841438aefcSRuslan Ermilov static void 58586aba0f2SPedro F. Giffuni output_indent(size_t n_spaces) 58686aba0f2SPedro F. Giffuni { 5871438aefcSRuslan Ermilov if (output_tab_width) { 5881438aefcSRuslan Ermilov while (n_spaces >= output_tab_width) { 58928d92b74STim J. Robbins putwchar('\t'); 5901438aefcSRuslan Ermilov n_spaces -= output_tab_width; 5911438aefcSRuslan Ermilov } 5921438aefcSRuslan Ermilov } 59386aba0f2SPedro F. Giffuni while (n_spaces-- > 0) 59486aba0f2SPedro F. Giffuni putwchar(' '); 5951438aefcSRuslan Ermilov } 5969b50d902SRodney W. Grimes 5971438aefcSRuslan Ermilov /* Output a single word, or add it to the buffer. 5981438aefcSRuslan Ermilov * indent0 and indent1 are the indents to use on the first and subsequent 5991438aefcSRuslan Ermilov * lines of a paragraph. They'll often be the same, of course. 6001438aefcSRuslan Ermilov */ 6011438aefcSRuslan Ermilov static void 60286aba0f2SPedro F. Giffuni output_word(size_t indent0, size_t indent1, const wchar_t *word, size_t length, size_t spaces) 60386aba0f2SPedro F. Giffuni { 60428d92b74STim J. Robbins size_t new_x; 6051438aefcSRuslan Ermilov size_t indent = output_in_paragraph ? indent1 : indent0; 60628d92b74STim J. Robbins size_t width; 60728d92b74STim J. Robbins const wchar_t *p; 60828d92b74STim J. Robbins int cwidth; 60928d92b74STim J. Robbins 61028d92b74STim J. Robbins for (p = word, width = 0; p < &word[length]; p++) 61128d92b74STim J. Robbins width += (cwidth = wcwidth(*p)) > 0 ? cwidth : 1; 61228d92b74STim J. Robbins 61328d92b74STim J. Robbins new_x = x + pending_spaces + width; 6141438aefcSRuslan Ermilov 61586aba0f2SPedro F. Giffuni /* 61686aba0f2SPedro F. Giffuni * If either |spaces==0| (at end of line) or |coalesce_spaces_P| 61786aba0f2SPedro F. Giffuni * (squashing internal whitespace), then add just one space; except 61886aba0f2SPedro F. Giffuni * that if the last character was a sentence-ender we actually add 61986aba0f2SPedro F. Giffuni * two spaces. 6201438aefcSRuslan Ermilov */ 6211438aefcSRuslan Ermilov if (coalesce_spaces_P || spaces == 0) 62228d92b74STim J. Robbins spaces = wcschr(sentence_enders, word[length - 1]) ? 2 : 1; 6231438aefcSRuslan Ermilov 6241438aefcSRuslan Ermilov if (new_x <= goal_length) { 62586aba0f2SPedro F. Giffuni /* 62686aba0f2SPedro F. Giffuni * After adding the word we still aren't at the goal length, 62786aba0f2SPedro F. Giffuni * so clearly we add it to the buffer rather than outputing 62886aba0f2SPedro F. Giffuni * it. 6291438aefcSRuslan Ermilov */ 63086aba0f2SPedro F. Giffuni wmemset(output_buffer + output_buffer_length, L' ', 63186aba0f2SPedro F. Giffuni pending_spaces); 63286aba0f2SPedro F. Giffuni x0 += pending_spaces; 63386aba0f2SPedro F. Giffuni x += pending_spaces; 63428d92b74STim J. Robbins output_buffer_length += pending_spaces; 63528d92b74STim J. Robbins wmemcpy(output_buffer + output_buffer_length, word, length); 63686aba0f2SPedro F. Giffuni x0 += width; 63786aba0f2SPedro F. Giffuni x += width; 63886aba0f2SPedro F. Giffuni output_buffer_length += length; 6391438aefcSRuslan Ermilov pending_spaces = spaces; 64086aba0f2SPedro F. Giffuni } else { 64186aba0f2SPedro F. Giffuni /* 64286aba0f2SPedro F. Giffuni * Adding the word takes us past the goal. Print the 64386aba0f2SPedro F. Giffuni * line-so-far, and the word too iff either (1) the lsf is 64486aba0f2SPedro F. Giffuni * empty or (2) that makes us nearer the goal but doesn't 64586aba0f2SPedro F. Giffuni * take us over the limit, or (3) the word on its own takes 64686aba0f2SPedro F. Giffuni * us over the limit. In case (3) we put a newline in 64786aba0f2SPedro F. Giffuni * between. 6481438aefcSRuslan Ermilov */ 64986aba0f2SPedro F. Giffuni if (indent > 0) 65086aba0f2SPedro F. Giffuni output_indent(indent); 65128d92b74STim J. Robbins wprintf(L"%.*ls", (int)output_buffer_length, output_buffer); 65286aba0f2SPedro F. Giffuni if (x0 == 0 || (new_x <= max_length && 65386aba0f2SPedro F. Giffuni new_x - goal_length <= goal_length - x)) { 65428d92b74STim J. Robbins wprintf(L"%*ls", (int)pending_spaces, L""); 6551438aefcSRuslan Ermilov goto write_out_word; 65686aba0f2SPedro F. Giffuni } else { 65786aba0f2SPedro F. Giffuni /* 65886aba0f2SPedro F. Giffuni * If the word takes us over the limit on its own, 65986aba0f2SPedro F. Giffuni * just spit it out and don't bother buffering it. 6601438aefcSRuslan Ermilov */ 66128d92b74STim J. Robbins if (indent + width > max_length) { 66228d92b74STim J. Robbins putwchar('\n'); 66386aba0f2SPedro F. Giffuni if (indent > 0) 66486aba0f2SPedro F. Giffuni output_indent(indent); 6651438aefcSRuslan Ermilov write_out_word: 66628d92b74STim J. Robbins wprintf(L"%.*ls", (int)length, word); 66786aba0f2SPedro F. Giffuni x0 = 0; 66886aba0f2SPedro F. Giffuni x = indent1; 66986aba0f2SPedro F. Giffuni pending_spaces = 0; 67028d92b74STim J. Robbins output_buffer_length = 0; 67186aba0f2SPedro F. Giffuni } else { 67228d92b74STim J. Robbins wmemcpy(output_buffer, word, length); 67386aba0f2SPedro F. Giffuni x0 = width; 67486aba0f2SPedro F. Giffuni x = width + indent1; 67586aba0f2SPedro F. Giffuni pending_spaces = spaces; 67628d92b74STim J. Robbins output_buffer_length = length; 6779b50d902SRodney W. Grimes } 6789b50d902SRodney W. Grimes } 67928d92b74STim J. Robbins putwchar('\n'); 6801438aefcSRuslan Ermilov output_in_paragraph = 1; 6819c61e111SJoerg Wunsch } 6829c61e111SJoerg Wunsch } 6839c61e111SJoerg Wunsch 6841438aefcSRuslan Ermilov /* Process a stream, but just center its lines rather than trying to 6851438aefcSRuslan Ermilov * format them neatly. 6869b50d902SRodney W. Grimes */ 6871438aefcSRuslan Ermilov static void 68886aba0f2SPedro F. Giffuni center_stream(FILE *stream, const char *name) 68986aba0f2SPedro F. Giffuni { 69028d92b74STim J. Robbins wchar_t *line, *p; 6911438aefcSRuslan Ermilov size_t length; 69228d92b74STim J. Robbins size_t width; 69328d92b74STim J. Robbins int cwidth; 69486aba0f2SPedro F. Giffuni 695*074e77e6SPedro F. Giffuni while ((line = get_line(stream, &length)) != NULL) { 6961438aefcSRuslan Ermilov size_t l = length; 69786aba0f2SPedro F. Giffuni 69886aba0f2SPedro F. Giffuni while (l > 0 && iswspace(*line)) { 69986aba0f2SPedro F. Giffuni ++line; 70086aba0f2SPedro F. Giffuni --l; 70186aba0f2SPedro F. Giffuni } 7021438aefcSRuslan Ermilov length = l; 70328d92b74STim J. Robbins for (p = line, width = 0; p < &line[length]; p++) 70428d92b74STim J. Robbins width += (cwidth = wcwidth(*p)) > 0 ? cwidth : 1; 70528d92b74STim J. Robbins l = width; 70686aba0f2SPedro F. Giffuni while (l < goal_length) { 70786aba0f2SPedro F. Giffuni putwchar(' '); 70886aba0f2SPedro F. Giffuni l += 2; 70986aba0f2SPedro F. Giffuni } 71028d92b74STim J. Robbins wprintf(L"%.*ls\n", (int)length, line); 7111438aefcSRuslan Ermilov } 71286aba0f2SPedro F. Giffuni if (ferror(stream)) { 71386aba0f2SPedro F. Giffuni warn("%s", name); 71486aba0f2SPedro F. Giffuni ++n_errors; 71586aba0f2SPedro F. Giffuni } 7169b50d902SRodney W. Grimes } 7179b50d902SRodney W. Grimes 7181438aefcSRuslan Ermilov /* Get a single line from a stream. Expand tabs, strip control 7191438aefcSRuslan Ermilov * characters and trailing whitespace, and handle backspaces. 7201438aefcSRuslan Ermilov * Return the address of the buffer containing the line, and 7211438aefcSRuslan Ermilov * put the length of the line in |lengthp|. 7221438aefcSRuslan Ermilov * This can cope with arbitrarily long lines, and with lines 7231438aefcSRuslan Ermilov * without terminating \n. 7241438aefcSRuslan Ermilov * If there are no characters left or an error happens, we 7251438aefcSRuslan Ermilov * return 0. 7261438aefcSRuslan Ermilov * Don't confuse |spaces_pending| here with the global 7271438aefcSRuslan Ermilov * |pending_spaces|. 7289b50d902SRodney W. Grimes */ 72928d92b74STim J. Robbins static wchar_t * 73086aba0f2SPedro F. Giffuni get_line(FILE *stream, size_t *lengthp) 73186aba0f2SPedro F. Giffuni { 73228d92b74STim J. Robbins static wchar_t *buf = NULL; 7331438aefcSRuslan Ermilov static size_t length = 0; 7341438aefcSRuslan Ermilov size_t len = 0; 73528d92b74STim J. Robbins wint_t ch; 7361438aefcSRuslan Ermilov size_t spaces_pending = 0; 737deba2451SRuslan Ermilov int troff = 0; 73828d92b74STim J. Robbins size_t col = 0; 73928d92b74STim J. Robbins int cwidth; 7401438aefcSRuslan Ermilov 74186aba0f2SPedro F. Giffuni if (buf == NULL) { 74286aba0f2SPedro F. Giffuni length = 100; 74386aba0f2SPedro F. Giffuni buf = XMALLOC(length * sizeof(wchar_t)); 74486aba0f2SPedro F. Giffuni } 74528d92b74STim J. Robbins while ((ch = getwc(stream)) != '\n' && ch != WEOF) { 74686aba0f2SPedro F. Giffuni if (len + spaces_pending == 0 && ch == '.' && !format_troff) 74786aba0f2SPedro F. Giffuni troff = 1; 74886aba0f2SPedro F. Giffuni if (ch == ' ') 74986aba0f2SPedro F. Giffuni ++spaces_pending; 75028d92b74STim J. Robbins else if (troff || iswprint(ch)) { 7511438aefcSRuslan Ermilov while (len + spaces_pending >= length) { 75286aba0f2SPedro F. Giffuni length *= 2; 75386aba0f2SPedro F. Giffuni buf = xrealloc(buf, length * sizeof(wchar_t)); 7549b50d902SRodney W. Grimes } 75586aba0f2SPedro F. Giffuni while (spaces_pending > 0) { 75686aba0f2SPedro F. Giffuni --spaces_pending; 75786aba0f2SPedro F. Giffuni buf[len++] = ' '; 75886aba0f2SPedro F. Giffuni col++; 75986aba0f2SPedro F. Giffuni } 7601438aefcSRuslan Ermilov buf[len++] = ch; 76128d92b74STim J. Robbins col += (cwidth = wcwidth(ch)) > 0 ? cwidth : 1; 76286aba0f2SPedro F. Giffuni } else if (ch == '\t') 76386aba0f2SPedro F. Giffuni spaces_pending += tab_width - 76486aba0f2SPedro F. Giffuni (col + spaces_pending) % tab_width; 76586aba0f2SPedro F. Giffuni else if (ch == '\b') { 76686aba0f2SPedro F. Giffuni if (len) 76786aba0f2SPedro F. Giffuni --len; 76886aba0f2SPedro F. Giffuni if (col) 76986aba0f2SPedro F. Giffuni --col; 7709b50d902SRodney W. Grimes } 7711438aefcSRuslan Ermilov } 7721438aefcSRuslan Ermilov *lengthp = len; 77328d92b74STim J. Robbins return (len > 0 || ch != WEOF) ? buf : 0; 7749b50d902SRodney W. Grimes } 7759b50d902SRodney W. Grimes 7761438aefcSRuslan Ermilov /* (Re)allocate some memory, exiting with an error if we can't. 7779b50d902SRodney W. Grimes */ 7781438aefcSRuslan Ermilov static void * 77986aba0f2SPedro F. Giffuni xrealloc(void *ptr, size_t nbytes) 78086aba0f2SPedro F. Giffuni { 7811438aefcSRuslan Ermilov void *p = realloc(ptr, nbytes); 78286aba0f2SPedro F. Giffuni 78386aba0f2SPedro F. Giffuni if (p == NULL) 78486aba0f2SPedro F. Giffuni errx(EX_OSERR, "out of memory"); 7851438aefcSRuslan Ermilov return p; 7869b50d902SRodney W. Grimes } 787