=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.147 retrieving revision 1.273.2.5 diff -u -p -r1.147 -r1.273.2.5 --- mandoc/main.c 2011/03/07 01:35:51 1.147 +++ mandoc/main.c 2017/01/27 14:41:40 1.273.2.5 @@ -1,977 +1,952 @@ -/* $Id: main.c,v 1.147 2011/03/07 01:35:51 schwarze Exp $ */ +/* $Id: main.c,v 1.273.2.5 2017/01/27 14:41:40 schwarze Exp $ */ /* - * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons - * Copyright (c) 2010, 2011 Ingo Schwarze + * Copyright (c) 2008-2012 Kristaps Dzonsons + * Copyright (c) 2010-2012, 2014-2017 Ingo Schwarze + * Copyright (c) 2010 Joerg Sonnenberger * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * - * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#ifdef HAVE_CONFIG_H #include "config.h" -#endif -#include -#include +#include +#include /* MACHINE */ +#include #include #include +#if HAVE_ERR +#include +#endif +#include #include +#include +#if HAVE_SANDBOX_INIT +#include +#endif +#include #include #include #include #include +#include #include +#include "mandoc_aux.h" #include "mandoc.h" -#include "main.h" +#include "roff.h" #include "mdoc.h" #include "man.h" -#include "roff.h" +#include "tag.h" +#include "main.h" +#include "manconf.h" +#include "mansearch.h" -#ifndef MAP_FILE -#define MAP_FILE 0 -#endif - -#define REPARSE_LIMIT 1000 -#define UNCONST(a) ((void *)(uintptr_t)(const void *)(a)) - -/* FIXME: Intel's compiler? LLVM? pcc? */ - -#if !defined(__GNUC__) || (__GNUC__ < 2) -# if !defined(lint) -# define __attribute__(x) -# endif -#endif /* !defined(__GNUC__) || (__GNUC__ < 2) */ - -typedef void (*out_mdoc)(void *, const struct mdoc *); -typedef void (*out_man)(void *, const struct man *); -typedef void (*out_free)(void *); - -struct buf { - char *buf; - size_t sz; +enum outmode { + OUTMODE_DEF = 0, + OUTMODE_FLN, + OUTMODE_LST, + OUTMODE_ALL, + OUTMODE_INT, + OUTMODE_ONE }; -enum intt { - INTT_AUTO, - INTT_MDOC, - INTT_MAN -}; - enum outt { - OUTT_ASCII = 0, - OUTT_TREE, - OUTT_HTML, - OUTT_XHTML, - OUTT_LINT, - OUTT_PS, - OUTT_PDF + OUTT_ASCII = 0, /* -Tascii */ + OUTT_LOCALE, /* -Tlocale */ + OUTT_UTF8, /* -Tutf8 */ + OUTT_TREE, /* -Ttree */ + OUTT_MAN, /* -Tman */ + OUTT_HTML, /* -Thtml */ + OUTT_LINT, /* -Tlint */ + OUTT_PS, /* -Tps */ + OUTT_PDF /* -Tpdf */ }; struct curparse { - const char *file; /* Current parse. */ - int fd; /* Current parse. */ - int line; /* Line number in the file. */ - enum mandoclevel wlevel; /* Ignore messages below this. */ - int wstop; /* Stop after a file with a warning. */ - enum intt inttype; /* which parser to use */ - struct man *pman; /* persistent man parser */ - struct mdoc *pmdoc; /* persistent mdoc parser */ - struct man *man; /* man parser */ - struct mdoc *mdoc; /* mdoc parser */ - struct roff *roff; /* roff parser (!NULL) */ - struct regset regs; /* roff registers */ - int reparse_count; /* finite interpolation stack */ - enum outt outtype; /* which output to use */ - out_mdoc outmdoc; /* mdoc output ptr */ - out_man outman; /* man output ptr */ - out_free outfree; /* free output ptr */ + struct mparse *mp; + enum mandoclevel wlevel; /* ignore messages below this */ + int wstop; /* stop after a file with a warning */ + enum outt outtype; /* which output to use */ void *outdata; /* data for output */ - char outopts[BUFSIZ]; /* buf of output opts */ + struct manoutput *outopts; /* output options */ }; -static const char * const mandoclevels[MANDOCLEVEL_MAX] = { - "SUCCESS", - "RESERVED", - "WARNING", - "ERROR", - "FATAL", - "BADARG", - "SYSERR" -}; -static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = { - MANDOCERR_OK, - MANDOCERR_WARNING, - MANDOCERR_WARNING, - MANDOCERR_ERROR, - MANDOCERR_FATAL, - MANDOCERR_MAX, - MANDOCERR_MAX -}; +#if HAVE_SQLITE3 +int mandocdb(int, char *[]); +#endif -static const char * const mandocerrs[MANDOCERR_MAX] = { - "ok", - - "generic warning", - - /* related to the prologue */ - "no title in document", - "document title should be all caps", - "unknown manual section", - "date missing, using today's date", - "cannot parse date, using it verbatim", - "prologue macros out of order", - "duplicate prologue macro", - "macro not allowed in prologue", - "macro not allowed in body", - - /* related to document structure */ - ".so is fragile, better use ln(1)", - "NAME section must come first", - "bad NAME section contents", - "manual name not yet set", - "sections out of conventional order", - "duplicate section name", - "section not in conventional manual section", - - /* related to macros and nesting */ - "skipping obsolete macro", - "skipping paragraph macro", - "skipping no-space macro", - "blocks badly nested", - "child violates parent syntax", - "nested displays are not portable", - "already in literal mode", - - /* related to missing macro arguments */ - "skipping empty macro", - "argument count wrong", - "missing display type", - "list type must come first", - "tag lists require a width argument", - "missing font type", - "skipping end of block that is not open", - - /* related to bad macro arguments */ - "skipping argument", - "duplicate argument", - "duplicate display type", - "duplicate list type", - "unknown AT&T UNIX version", - "bad Boolean value", - "unknown font", - "unknown standard specifier", - "bad width argument", - - /* related to plain text */ - "blank line in non-literal context", - "tab in non-literal context", - "end of line whitespace", - "bad comment style", - "unknown escape sequence", - "unterminated quoted string", - - "generic error", - - /* related to tables */ - "bad table syntax", - "bad table option", - "bad table layout", - "no table layout cells specified", - "no table data cells specified", - "ignore data in cell", - "data block still open", - "ignoring extra data cells", - - "input stack limit exceeded, infinite loop?", - "skipping bad character", - "escaped character not allowed in a name", - "skipping text before the first section header", - "skipping unknown macro", - "NOT IMPLEMENTED, please use groff: skipping request", - "line scope broken", - "argument count wrong", - "skipping end of block that is not open", - "missing end of block", - "scope open on exit", - "uname(3) system call failed", - "macro requires line argument(s)", - "macro requires body argument(s)", - "macro requires argument(s)", - "missing list type", - "line argument(s) will be lost", - "body argument(s) will be lost", - - "generic fatal error", - - "column syntax is inconsistent", - "NOT IMPLEMENTED: .Bd -file", - "line scope broken, syntax violated", - "argument count wrong, violates syntax", - "child violates parent syntax", - "argument count wrong, violates syntax", - "NOT IMPLEMENTED: .so with absolute path or \"..\"", - "no document body", - "no document prologue", - "static buffer exhausted", -}; - -static void parsebuf(struct curparse *, struct buf, int); -static void pdesc(struct curparse *); -static void fdesc(struct curparse *); -static void ffile(const char *, struct curparse *); -static int pfile(const char *, struct curparse *); -static int moptions(enum intt *, char *); -static int mmsg(enum mandocerr, void *, - int, int, const char *); -static void pset(const char *, int, struct curparse *); +static int fs_lookup(const struct manpaths *, + size_t ipath, const char *, + const char *, const char *, + struct manpage **, size_t *); +static void fs_search(const struct mansearch *, + const struct manpaths *, int, char**, + struct manpage **, size_t *); +static int koptions(int *, char *); +static int moptions(int *, char *); +static void mmsg(enum mandocerr, enum mandoclevel, + const char *, int, int, const char *); +static void outdata_alloc(struct curparse *); +static void parse(struct curparse *, int, const char *); +static void passthrough(const char *, int, int); +static pid_t spawn_pager(struct tag_files *); static int toptions(struct curparse *, char *); -static void usage(void) __attribute__((noreturn)); -static void version(void) __attribute__((noreturn)); +static void usage(enum argmode) __attribute__((noreturn)); static int woptions(struct curparse *, char *); -static const char *progname; -static enum mandoclevel file_status = MANDOCLEVEL_OK; -static enum mandoclevel exit_status = MANDOCLEVEL_OK; +static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9}; +static char help_arg[] = "help"; +static char *help_argv[] = {help_arg, NULL}; +static enum mandoclevel rc; + int main(int argc, char *argv[]) { - int c; + struct manconf conf; struct curparse curp; + struct mansearch search; + struct tag_files *tag_files; + const char *progname; + char *auxpaths; + char *defos; + unsigned char *uc; + struct manpage *res, *resp; + char *conf_file, *defpaths; + const char *sec; + const char *thisarg; + size_t i, sz; + int prio, best_prio; + enum outmode outmode; + int fd; + int show_usage; + int options; + int use_pager; + int status, signum; + int c; + pid_t pager_pid, tc_pgid, man_pgid, pid; - progname = strrchr(argv[0], '/'); - if (progname == NULL) +#if HAVE_PROGNAME + progname = getprogname(); +#else + if (argc < 1) + progname = mandoc_strdup("mandoc"); + else if ((progname = strrchr(argv[0], '/')) == NULL) progname = argv[0]; else ++progname; + setprogname(progname); +#endif +#if HAVE_SQLITE3 + if (strncmp(progname, "mandocdb", 8) == 0 || + strcmp(progname, BINM_MAKEWHATIS) == 0) + return mandocdb(argc, argv); +#endif + +#if HAVE_PLEDGE + if (pledge("stdio rpath tmppath tty proc exec flock", NULL) == -1) + err((int)MANDOCLEVEL_SYSERR, "pledge"); +#endif + +#if HAVE_SANDBOX_INIT + if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) + errx((int)MANDOCLEVEL_SYSERR, "sandbox_init"); +#endif + + /* Search options. */ + + memset(&conf, 0, sizeof(conf)); + conf_file = defpaths = NULL; + auxpaths = NULL; + + memset(&search, 0, sizeof(struct mansearch)); + search.outkey = "Nd"; + + if (strcmp(progname, BINM_MAN) == 0) + search.argmode = ARG_NAME; + else if (strcmp(progname, BINM_APROPOS) == 0) + search.argmode = ARG_EXPR; + else if (strcmp(progname, BINM_WHATIS) == 0) + search.argmode = ARG_WORD; + else if (strncmp(progname, "help", 4) == 0) + search.argmode = ARG_NAME; + else + search.argmode = ARG_FILE; + + /* Parser and formatter options. */ + memset(&curp, 0, sizeof(struct curparse)); + curp.outtype = OUTT_LOCALE; + curp.wlevel = MANDOCLEVEL_BADARG; + curp.outopts = &conf.output; + options = MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1; + defos = NULL; - curp.inttype = INTT_AUTO; - curp.outtype = OUTT_ASCII; - curp.wlevel = MANDOCLEVEL_FATAL; + use_pager = 1; + tag_files = NULL; + show_usage = 0; + outmode = OUTMODE_DEF; - /* LINTED */ - while (-1 != (c = getopt(argc, argv, "m:O:T:VW:"))) + while (-1 != (c = getopt(argc, argv, + "aC:cfhI:iK:klM:m:O:S:s:T:VW:w"))) { switch (c) { - case ('m'): - if ( ! moptions(&curp.inttype, optarg)) - return((int)MANDOCLEVEL_BADARG); + case 'a': + outmode = OUTMODE_ALL; break; - case ('O'): - (void)strlcat(curp.outopts, optarg, BUFSIZ); - (void)strlcat(curp.outopts, ",", BUFSIZ); + case 'C': + conf_file = optarg; break; - case ('T'): + case 'c': + use_pager = 0; + break; + case 'f': + search.argmode = ARG_WORD; + break; + case 'h': + conf.output.synopsisonly = 1; + use_pager = 0; + outmode = OUTMODE_ALL; + break; + case 'I': + if (strncmp(optarg, "os=", 3)) { + warnx("-I %s: Bad argument", optarg); + return (int)MANDOCLEVEL_BADARG; + } + if (defos) { + warnx("-I %s: Duplicate argument", optarg); + return (int)MANDOCLEVEL_BADARG; + } + defos = mandoc_strdup(optarg + 3); + break; + case 'i': + outmode = OUTMODE_INT; + break; + case 'K': + if ( ! koptions(&options, optarg)) + return (int)MANDOCLEVEL_BADARG; + break; + case 'k': + search.argmode = ARG_EXPR; + break; + case 'l': + search.argmode = ARG_FILE; + outmode = OUTMODE_ALL; + break; + case 'M': + defpaths = optarg; + break; + case 'm': + auxpaths = optarg; + break; + case 'O': + search.outkey = optarg; + while (optarg != NULL) { + thisarg = optarg; + if (manconf_output(&conf.output, + strsep(&optarg, ","), 0) == 0) + continue; + warnx("-O %s: Bad argument", thisarg); + return (int)MANDOCLEVEL_BADARG; + } + break; + case 'S': + search.arch = optarg; + break; + case 's': + search.sec = optarg; + break; + case 'T': if ( ! toptions(&curp, optarg)) - return((int)MANDOCLEVEL_BADARG); + return (int)MANDOCLEVEL_BADARG; break; - case ('W'): + case 'W': if ( ! woptions(&curp, optarg)) - return((int)MANDOCLEVEL_BADARG); + return (int)MANDOCLEVEL_BADARG; break; - case ('V'): - version(); - /* NOTREACHED */ + case 'w': + outmode = OUTMODE_FLN; + break; default: - usage(); - /* NOTREACHED */ + show_usage = 1; + break; } + } - argc -= optind; - argv += optind; + if (show_usage) + usage(search.argmode); - if (NULL == *argv) { - curp.file = ""; - curp.fd = STDIN_FILENO; + /* Postprocess options. */ - fdesc(&curp); - } - - while (*argv) { - ffile(*argv, &curp); - if (MANDOCLEVEL_OK != exit_status && curp.wstop) + if (outmode == OUTMODE_DEF) { + switch (search.argmode) { + case ARG_FILE: + outmode = OUTMODE_ALL; + use_pager = 0; break; - ++argv; + case ARG_NAME: + outmode = OUTMODE_ONE; + break; + default: + outmode = OUTMODE_LST; + break; + } } - if (curp.outfree) - (*curp.outfree)(curp.outdata); - if (curp.pmdoc) - mdoc_free(curp.pmdoc); - if (curp.pman) - man_free(curp.pman); - if (curp.roff) - roff_free(curp.roff); + if (outmode == OUTMODE_FLN || + outmode == OUTMODE_LST || + !isatty(STDOUT_FILENO)) + use_pager = 0; - return((int)exit_status); -} +#if HAVE_PLEDGE + if (!use_pager) + if (pledge("stdio rpath flock", NULL) == -1) + err((int)MANDOCLEVEL_SYSERR, "pledge"); +#endif + /* Parse arguments. */ -static void -version(void) -{ + if (argc > 0) { + argc -= optind; + argv += optind; + } + resp = NULL; - (void)printf("%s %s\n", progname, VERSION); - exit((int)MANDOCLEVEL_OK); -} + /* + * Quirks for help(1) + * and for a man(1) section argument without -s. + */ + if (search.argmode == ARG_NAME) { + if (*progname == 'h') { + if (argc == 0) { + argv = help_argv; + argc = 1; + } + } else if (argc > 1 && + ((uc = (unsigned char *)argv[0]) != NULL) && + ((isdigit(uc[0]) && (uc[1] == '\0' || + (isalpha(uc[1]) && uc[2] == '\0'))) || + (uc[0] == 'n' && uc[1] == '\0'))) { + search.sec = (char *)uc; + argv++; + argc--; + } + if (search.arch == NULL) + search.arch = getenv("MACHINE"); +#ifdef MACHINE + if (search.arch == NULL) + search.arch = MACHINE; +#endif + } -static void -usage(void) -{ + rc = MANDOCLEVEL_OK; - (void)fprintf(stderr, "usage: %s " - "[-V] " - "[-foption] " - "[-mformat] " - "[-Ooption] " - "[-Toutput] " - "[-Werr] " - "[file...]\n", - progname); + /* man(1), whatis(1), apropos(1) */ - exit((int)MANDOCLEVEL_BADARG); -} + if (search.argmode != ARG_FILE) { + if (argc == 0) + usage(search.argmode); -static void -ffile(const char *file, struct curparse *curp) -{ + if (search.argmode == ARG_NAME && + outmode == OUTMODE_ONE) + search.firstmatch = 1; - /* - * Called once per input file. Get the file ready for reading, - * pass it through to the parser-driver, then close it out. - * XXX: don't do anything special as this is only called for - * files; stdin goes directly to fdesc(). - */ + /* Access the mandoc database. */ - curp->file = file; + manconf_parse(&conf, conf_file, defpaths, auxpaths); +#if HAVE_SQLITE3 + mansearch_setup(1); + if ( ! mansearch(&search, &conf.manpath, + argc, argv, &res, &sz)) + usage(search.argmode); +#else + if (search.argmode != ARG_NAME) { + fputs("mandoc: database support not compiled in\n", + stderr); + return (int)MANDOCLEVEL_BADARG; + } + sz = 0; +#endif - if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { - perror(curp->file); - exit_status = MANDOCLEVEL_SYSERR; - return; - } + if (sz == 0) { + if (search.argmode == ARG_NAME) + fs_search(&search, &conf.manpath, + argc, argv, &res, &sz); + else + warnx("nothing appropriate"); + } - fdesc(curp); + if (sz == 0) { + rc = MANDOCLEVEL_BADARG; + goto out; + } - if (-1 == close(curp->fd)) - perror(curp->file); -} + /* + * For standard man(1) and -a output mode, + * prepare for copying filename pointers + * into the program parameter array. + */ -static int -pfile(const char *file, struct curparse *curp) -{ - const char *savefile; - int fd, savefd; + if (outmode == OUTMODE_ONE) { + argc = 1; + best_prio = 20; + } else if (outmode == OUTMODE_ALL) + argc = (int)sz; - if (-1 == (fd = open(file, O_RDONLY, 0))) { - perror(file); - file_status = MANDOCLEVEL_SYSERR; - return(0); - } + /* Iterate all matching manuals. */ - savefile = curp->file; - savefd = curp->fd; + resp = res; + for (i = 0; i < sz; i++) { + if (outmode == OUTMODE_FLN) + puts(res[i].file); + else if (outmode == OUTMODE_LST) + printf("%s - %s\n", res[i].names, + res[i].output == NULL ? "" : + res[i].output); + else if (outmode == OUTMODE_ONE) { + /* Search for the best section. */ + sec = res[i].file; + sec += strcspn(sec, "123456789"); + if (sec[0] == '\0') + continue; + prio = sec_prios[sec[0] - '1']; + if (sec[1] != '/') + prio += 10; + if (prio >= best_prio) + continue; + best_prio = prio; + resp = res + i; + } + } - curp->file = file; - curp->fd = fd; + /* + * For man(1), -a and -i output mode, fall through + * to the main mandoc(1) code iterating files + * and running the parsers on each of them. + */ - pdesc(curp); + if (outmode == OUTMODE_FLN || outmode == OUTMODE_LST) + goto out; + } - curp->file = savefile; - curp->fd = savefd; + /* mandoc(1) */ - if (-1 == close(fd)) - perror(file); +#if HAVE_PLEDGE + if (use_pager) { + if (pledge("stdio rpath tmppath tty proc exec", NULL) == -1) + err((int)MANDOCLEVEL_SYSERR, "pledge"); + } else { + if (pledge("stdio rpath", NULL) == -1) + err((int)MANDOCLEVEL_SYSERR, "pledge"); + } +#endif - return(MANDOCLEVEL_FATAL > file_status ? 1 : 0); -} + if (search.argmode == ARG_FILE && ! moptions(&options, auxpaths)) + return (int)MANDOCLEVEL_BADARG; + mchars_alloc(); + curp.mp = mparse_alloc(options, curp.wlevel, mmsg, defos); -static void -resize_buf(struct buf *buf, size_t initial) -{ + /* + * Conditionally start up the lookaside buffer before parsing. + */ + if (OUTT_MAN == curp.outtype) + mparse_keep(curp.mp); - buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial; - buf->buf = realloc(buf->buf, buf->sz); - if (NULL == buf->buf) { - perror(NULL); - exit((int)MANDOCLEVEL_SYSERR); + if (argc < 1) { + if (use_pager) + tag_files = tag_init(); + parse(&curp, STDIN_FILENO, ""); } -} + while (argc > 0) { + fd = mparse_open(curp.mp, resp != NULL ? resp->file : *argv); + if (fd != -1) { + if (use_pager) { + tag_files = tag_init(); + use_pager = 0; + } -static int -read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) -{ - struct stat st; - size_t off; - ssize_t ssz; + if (resp == NULL) + parse(&curp, fd, *argv); + else if (resp->form & FORM_SRC) { + /* For .so only; ignore failure. */ + chdir(conf.manpath.paths[resp->ipath]); + parse(&curp, fd, resp->file); + } else + passthrough(resp->file, fd, + conf.output.synopsisonly); - if (-1 == fstat(curp->fd, &st)) { - perror(curp->file); - return(0); - } + if (argc > 1 && curp.outtype <= OUTT_UTF8) { + if (curp.outdata == NULL) + outdata_alloc(&curp); + terminal_sepline(curp.outdata); + } + } else if (rc < MANDOCLEVEL_ERROR) + rc = MANDOCLEVEL_ERROR; - /* - * If we're a regular file, try just reading in the whole entry - * via mmap(). This is faster than reading it into blocks, and - * since each file is only a few bytes to begin with, I'm not - * concerned that this is going to tank any machines. - */ + if (MANDOCLEVEL_OK != rc && curp.wstop) + break; - if (S_ISREG(st.st_mode)) { - if (st.st_size >= (1U << 31)) { - fprintf(stderr, "%s: input too large\n", - curp->file); - return(0); + if (resp != NULL) + resp++; + else + argv++; + if (--argc) + mparse_reset(curp.mp); + } + + if (curp.outdata != NULL) { + switch (curp.outtype) { + case OUTT_HTML: + html_free(curp.outdata); + break; + case OUTT_UTF8: + case OUTT_LOCALE: + case OUTT_ASCII: + ascii_free(curp.outdata); + break; + case OUTT_PDF: + case OUTT_PS: + pspdf_free(curp.outdata); + break; + default: + break; } - *with_mmap = 1; - fb->sz = (size_t)st.st_size; - fb->buf = mmap(NULL, fb->sz, PROT_READ, - MAP_FILE|MAP_SHARED, curp->fd, 0); - if (fb->buf != MAP_FAILED) - return(1); } + mparse_free(curp.mp); + mchars_free(); +out: + if (search.argmode != ARG_FILE) { + manconf_free(&conf); +#if HAVE_SQLITE3 + mansearch_free(res, sz); + mansearch_setup(0); +#endif + } + + free(defos); + /* - * If this isn't a regular file (like, say, stdin), then we must - * go the old way and just read things in bit by bit. + * When using a pager, finish writing both temporary files, + * fork it, wait for the user to close it, and clean up. */ - *with_mmap = 0; - off = 0; - fb->sz = 0; - fb->buf = NULL; - for (;;) { - if (off == fb->sz) { - if (fb->sz == (1U << 31)) { - fprintf(stderr, "%s: input too large\n", - curp->file); + if (tag_files != NULL) { + fclose(stdout); + tag_write(); + man_pgid = getpgid(0); + tag_files->tcpgid = man_pgid == getpid() ? + getpgid(getppid()) : man_pgid; + pager_pid = 0; + signum = SIGSTOP; + for (;;) { + + /* Stop here until moved to the foreground. */ + + tc_pgid = tcgetpgrp(tag_files->ofd); + if (tc_pgid != man_pgid) { + if (tc_pgid == pager_pid) { + (void)tcsetpgrp(tag_files->ofd, + man_pgid); + if (signum == SIGTTIN) + continue; + } else + tag_files->tcpgid = tc_pgid; + kill(0, signum); + continue; + } + + /* Once in the foreground, activate the pager. */ + + if (pager_pid) { + (void)tcsetpgrp(tag_files->ofd, pager_pid); + kill(pager_pid, SIGCONT); + } else + pager_pid = spawn_pager(tag_files); + + /* Wait for the pager to stop or exit. */ + + while ((pid = waitpid(pager_pid, &status, + WUNTRACED)) == -1 && errno == EINTR) + continue; + + if (pid == -1) { + warn("wait"); + rc = MANDOCLEVEL_SYSERR; break; } - resize_buf(fb, 65536); + if (!WIFSTOPPED(status)) + break; + + signum = WSTOPSIG(status); } - ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off); - if (ssz == 0) { - fb->sz = off; - return(1); - } - if (ssz == -1) { - perror(curp->file); - break; - } - off += (size_t)ssz; + tag_unlink(); } - free(fb->buf); - fb->buf = NULL; - return(0); + return (int)rc; } - static void -fdesc(struct curparse *curp) +usage(enum argmode argmode) { - /* - * Called once per file with an opened file descriptor. All - * pre-file-parse operations (whether stdin or a file) should go - * here. - * - * This calls down into the nested parser, which drills down and - * fully parses a file and all its dependences (i.e., `so'). It - * then runs the cleanup validators and pushes to output. - */ + switch (argmode) { + case ARG_FILE: + fputs("usage: mandoc [-acfhkl] [-I os=name] " + "[-K encoding] [-mformat] [-O option]\n" + "\t [-T output] [-W level] [file ...]\n", stderr); + break; + case ARG_NAME: + fputs("usage: man [-acfhklw] [-C file] [-I os=name] " + "[-K encoding] [-M path] [-m path]\n" + "\t [-O option=value] [-S subsection] [-s section] " + "[-T output] [-W level]\n" + "\t [section] name ...\n", stderr); + break; + case ARG_WORD: + fputs("usage: whatis [-acfhklw] [-C file] " + "[-M path] [-m path] [-O outkey] [-S arch]\n" + "\t [-s section] name ...\n", stderr); + break; + case ARG_EXPR: + fputs("usage: apropos [-acfhklw] [-C file] " + "[-M path] [-m path] [-O outkey] [-S arch]\n" + "\t [-s section] expression ...\n", stderr); + break; + } + exit((int)MANDOCLEVEL_BADARG); +} - /* Zero the parse type. */ +static int +fs_lookup(const struct manpaths *paths, size_t ipath, + const char *sec, const char *arch, const char *name, + struct manpage **res, size_t *ressz) +{ + glob_t globinfo; + struct manpage *page; + char *file; + int form, globres; - curp->mdoc = NULL; - curp->man = NULL; - file_status = MANDOCLEVEL_OK; + form = FORM_SRC; + mandoc_asprintf(&file, "%s/man%s/%s.%s", + paths->paths[ipath], sec, name, sec); + if (access(file, R_OK) != -1) + goto found; + free(file); - /* Make sure the mandotory roff parser is initialised. */ + mandoc_asprintf(&file, "%s/cat%s/%s.0", + paths->paths[ipath], sec, name); + if (access(file, R_OK) != -1) { + form = FORM_CAT; + goto found; + } + free(file); - if (NULL == curp->roff) { - curp->roff = roff_alloc(&curp->regs, curp, mmsg); - assert(curp->roff); + if (arch != NULL) { + mandoc_asprintf(&file, "%s/man%s/%s/%s.%s", + paths->paths[ipath], sec, arch, name, sec); + if (access(file, R_OK) != -1) + goto found; + free(file); } - /* Fully parse the file. */ + mandoc_asprintf(&file, "%s/man%s/%s.[01-9]*", + paths->paths[ipath], sec, name); + globres = glob(file, 0, NULL, &globinfo); + if (globres != 0 && globres != GLOB_NOMATCH) + warn("%s: glob", file); + free(file); + if (globres == 0) + file = mandoc_strdup(*globinfo.gl_pathv); + globfree(&globinfo); + if (globres != 0) + return 0; - pdesc(curp); +found: +#if HAVE_SQLITE3 + warnx("outdated mandoc.db lacks %s(%s) entry, run %s %s", + name, sec, BINM_MAKEWHATIS, paths->paths[ipath]); +#endif + *res = mandoc_reallocarray(*res, ++*ressz, sizeof(struct manpage)); + page = *res + (*ressz - 1); + page->file = file; + page->names = NULL; + page->output = NULL; + page->ipath = ipath; + page->bits = NAME_FILE & NAME_MASK; + page->sec = (*sec >= '1' && *sec <= '9') ? *sec - '1' + 1 : 10; + page->form = form; + return 1; +} - if (MANDOCLEVEL_FATAL <= file_status) - goto cleanup; +static void +fs_search(const struct mansearch *cfg, const struct manpaths *paths, + int argc, char **argv, struct manpage **res, size_t *ressz) +{ + const char *const sections[] = + {"1", "8", "6", "2", "3", "5", "7", "4", "9", "3p"}; + const size_t nsec = sizeof(sections)/sizeof(sections[0]); - /* NOTE a parser may not have been assigned, yet. */ + size_t ipath, isec, lastsz; - if ( ! (curp->man || curp->mdoc)) { - fprintf(stderr, "%s: Not a manual\n", curp->file); - file_status = MANDOCLEVEL_FATAL; - goto cleanup; + assert(cfg->argmode == ARG_NAME); + + *res = NULL; + *ressz = lastsz = 0; + while (argc) { + for (ipath = 0; ipath < paths->sz; ipath++) { + if (cfg->sec != NULL) { + if (fs_lookup(paths, ipath, cfg->sec, + cfg->arch, *argv, res, ressz) && + cfg->firstmatch) + return; + } else for (isec = 0; isec < nsec; isec++) + if (fs_lookup(paths, ipath, sections[isec], + cfg->arch, *argv, res, ressz) && + cfg->firstmatch) + return; + } + if (*ressz == lastsz) + warnx("No entry for %s in the manual.", *argv); + lastsz = *ressz; + argv++; + argc--; } +} - /* Clean up the parse routine ASTs. */ +static void +parse(struct curparse *curp, int fd, const char *file) +{ + enum mandoclevel rctmp; + struct roff_man *man; - if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) { - assert(MANDOCLEVEL_FATAL <= file_status); - goto cleanup; - } + /* Begin by parsing the file itself. */ - if (curp->man && ! man_endparse(curp->man)) { - assert(MANDOCLEVEL_FATAL <= file_status); - goto cleanup; - } + assert(file); + assert(fd >= 0); - assert(curp->roff); - roff_endparse(curp->roff); + rctmp = mparse_readfd(curp->mp, fd, file); + if (fd != STDIN_FILENO) + close(fd); + if (rc < rctmp) + rc = rctmp; /* - * With -Wstop and warnings or errors of at least - * the requested level, do not produce output. + * With -Wstop and warnings or errors of at least the requested + * level, do not produce output. */ - if (MANDOCLEVEL_OK != file_status && curp->wstop) - goto cleanup; + if (rctmp != MANDOCLEVEL_OK && curp->wstop) + return; - /* If unset, allocate output dev now (if applicable). */ + if (curp->outdata == NULL) + outdata_alloc(curp); - if ( ! (curp->outman && curp->outmdoc)) { + mparse_result(curp->mp, &man, NULL); + + /* Execute the out device, if it exists. */ + + if (man == NULL) + return; + if (man->macroset == MACROSET_MDOC) { + mdoc_validate(man); switch (curp->outtype) { - case (OUTT_XHTML): - curp->outdata = xhtml_alloc(curp->outopts); + case OUTT_HTML: + html_mdoc(curp->outdata, man); break; - case (OUTT_HTML): - curp->outdata = html_alloc(curp->outopts); + case OUTT_TREE: + tree_mdoc(curp->outdata, man); break; - case (OUTT_ASCII): - curp->outdata = ascii_alloc(curp->outopts); - curp->outfree = ascii_free; + case OUTT_MAN: + man_mdoc(curp->outdata, man); break; - case (OUTT_PDF): - curp->outdata = pdf_alloc(curp->outopts); - curp->outfree = pspdf_free; + case OUTT_PDF: + case OUTT_ASCII: + case OUTT_UTF8: + case OUTT_LOCALE: + case OUTT_PS: + terminal_mdoc(curp->outdata, man); break; - case (OUTT_PS): - curp->outdata = ps_alloc(curp->outopts); - curp->outfree = pspdf_free; - break; default: break; } - + } + if (man->macroset == MACROSET_MAN) { + man_validate(man); switch (curp->outtype) { - case (OUTT_HTML): - /* FALLTHROUGH */ - case (OUTT_XHTML): - curp->outman = html_man; - curp->outmdoc = html_mdoc; - curp->outfree = html_free; + case OUTT_HTML: + html_man(curp->outdata, man); break; - case (OUTT_TREE): - curp->outman = tree_man; - curp->outmdoc = tree_mdoc; + case OUTT_TREE: + tree_man(curp->outdata, man); break; - case (OUTT_PDF): - /* FALLTHROUGH */ - case (OUTT_ASCII): - /* FALLTHROUGH */ - case (OUTT_PS): - curp->outman = terminal_man; - curp->outmdoc = terminal_mdoc; + case OUTT_MAN: + man_man(curp->outdata, man); break; + case OUTT_PDF: + case OUTT_ASCII: + case OUTT_UTF8: + case OUTT_LOCALE: + case OUTT_PS: + terminal_man(curp->outdata, man); + break; default: break; } } - - /* Execute the out device, if it exists. */ - - if (curp->man && curp->outman) - (*curp->outman)(curp->outdata, curp->man); - if (curp->mdoc && curp->outmdoc) - (*curp->outmdoc)(curp->outdata, curp->mdoc); - - cleanup: - - memset(&curp->regs, 0, sizeof(struct regset)); - - /* Reset the current-parse compilers. */ - - if (curp->mdoc) - mdoc_reset(curp->mdoc); - if (curp->man) - man_reset(curp->man); - - assert(curp->roff); - roff_reset(curp->roff); - - if (exit_status < file_status) - exit_status = file_status; - - return; + mparse_updaterc(curp->mp, &rc); } static void -pdesc(struct curparse *curp) +outdata_alloc(struct curparse *curp) { - struct buf blk; - int with_mmap; - - /* - * Run for each opened file; may be called more than once for - * each full parse sequence if the opened file is nested (i.e., - * from `so'). Simply sucks in the whole file and moves into - * the parse phase for the file. - */ - - if ( ! read_whole_file(curp, &blk, &with_mmap)) { - file_status = MANDOCLEVEL_SYSERR; - return; + switch (curp->outtype) { + case OUTT_HTML: + curp->outdata = html_alloc(curp->outopts); + break; + case OUTT_UTF8: + curp->outdata = utf8_alloc(curp->outopts); + break; + case OUTT_LOCALE: + curp->outdata = locale_alloc(curp->outopts); + break; + case OUTT_ASCII: + curp->outdata = ascii_alloc(curp->outopts); + break; + case OUTT_PDF: + curp->outdata = pdf_alloc(curp->outopts); + break; + case OUTT_PS: + curp->outdata = ps_alloc(curp->outopts); + break; + default: + break; } - - /* Line number is per-file. */ - - curp->line = 1; - - parsebuf(curp, blk, 1); - - if (with_mmap) - munmap(blk.buf, blk.sz); - else - free(blk.buf); } static void -parsebuf(struct curparse *curp, struct buf blk, int start) +passthrough(const char *file, int fd, int synopsis_only) { - const struct tbl_span *span; - struct buf ln; - enum rofferr rr; - int i, of, rc; - int pos; /* byte number in the ln buffer */ - int lnn; /* line number in the real file */ - unsigned char c; + const char synb[] = "S\bSY\bYN\bNO\bOP\bPS\bSI\bIS\bS"; + const char synr[] = "SYNOPSIS"; - /* - * Main parse routine for an opened file. This is called for - * each opened file and simply loops around the full input file, - * possibly nesting (i.e., with `so'). - */ + FILE *stream; + const char *syscall; + char *line, *cp; + size_t linesz; + ssize_t len, written; + int print; - memset(&ln, 0, sizeof(struct buf)); + line = NULL; + linesz = 0; - lnn = curp->line; - pos = 0; + if (fflush(stdout) == EOF) { + syscall = "fflush"; + goto fail; + } - for (i = 0; i < (int)blk.sz; ) { - if (0 == pos && '\0' == blk.buf[i]) - break; + if ((stream = fdopen(fd, "r")) == NULL) { + close(fd); + syscall = "fdopen"; + goto fail; + } - if (start) { - curp->line = lnn; - curp->reparse_count = 0; - } - - while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) { - - /* - * When finding an unescaped newline character, - * leave the character loop to process the line. - * Skip a preceding carriage return, if any. - */ - - if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz && - '\n' == blk.buf[i + 1]) - ++i; - if ('\n' == blk.buf[i]) { - ++i; - ++lnn; - break; - } - - /* - * Warn about bogus characters. If you're using - * non-ASCII encoding, you're screwing your - * readers. Since I'd rather this not happen, - * I'll be helpful and drop these characters so - * we don't display gibberish. Note to manual - * writers: use special characters. - */ - - c = (unsigned char) blk.buf[i]; - - if ( ! (isascii(c) && - (isgraph(c) || isblank(c)))) { - mmsg(MANDOCERR_BADCHAR, curp, - curp->line, pos, "ignoring byte"); - i++; + print = 0; + while ((len = getline(&line, &linesz, stream)) != -1) { + cp = line; + if (synopsis_only) { + if (print) { + if ( ! isspace((unsigned char)*cp)) + goto done; + while (isspace((unsigned char)*cp)) { + cp++; + len--; + } + } else { + if (strcmp(cp, synb) == 0 || + strcmp(cp, synr) == 0) + print = 1; continue; } - - /* Trailing backslash = a plain char. */ - - if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) { - if (pos >= (int)ln.sz) - resize_buf(&ln, 256); - ln.buf[pos++] = blk.buf[i++]; - continue; - } - - /* - * Found escape and at least one other character. - * When it's a newline character, skip it. - * When there is a carriage return in between, - * skip that one as well. - */ - - if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz && - '\n' == blk.buf[i + 2]) - ++i; - if ('\n' == blk.buf[i + 1]) { - i += 2; - ++lnn; - continue; - } - - if ('"' == blk.buf[i + 1]) { - i += 2; - /* Comment, skip to end of line */ - for (; i < (int)blk.sz; ++i) { - if ('\n' == blk.buf[i]) { - ++i; - ++lnn; - break; - } - } - - /* Backout trailing whitespaces */ - for (; pos > 0; --pos) { - if (ln.buf[pos - 1] != ' ') - break; - if (pos > 2 && ln.buf[pos - 2] == '\\') - break; - } - break; - } - - /* Some other escape sequence, copy & cont. */ - - if (pos + 1 >= (int)ln.sz) - resize_buf(&ln, 256); - - ln.buf[pos++] = blk.buf[i++]; - ln.buf[pos++] = blk.buf[i++]; } - - if (pos >= (int)ln.sz) - resize_buf(&ln, 256); - - ln.buf[pos] = '\0'; - - /* - * A significant amount of complexity is contained by - * the roff preprocessor. It's line-oriented but can be - * expressed on one line, so we need at times to - * readjust our starting point and re-run it. The roff - * preprocessor can also readjust the buffers with new - * data, so we pass them in wholesale. - */ - - of = 0; - -rerun: - rr = roff_parseln - (curp->roff, curp->line, - &ln.buf, &ln.sz, of, &of); - - switch (rr) { - case (ROFF_REPARSE): - if (REPARSE_LIMIT >= ++curp->reparse_count) - parsebuf(curp, ln, 0); - else - mmsg(MANDOCERR_ROFFLOOP, curp, - curp->line, pos, NULL); - pos = 0; - continue; - case (ROFF_APPEND): - pos = strlen(ln.buf); - continue; - case (ROFF_RERUN): - goto rerun; - case (ROFF_IGN): - pos = 0; - continue; - case (ROFF_ERR): - assert(MANDOCLEVEL_FATAL <= file_status); - break; - case (ROFF_SO): - if (pfile(ln.buf + of, curp)) { - pos = 0; + for (; len > 0; len -= written) { + if ((written = write(STDOUT_FILENO, cp, len)) != -1) continue; - } else - break; - default: - break; + fclose(stream); + syscall = "write"; + goto fail; } + } - /* - * If we encounter errors in the recursive parsebuf() - * call, make sure we don't continue parsing. - */ - - if (MANDOCLEVEL_FATAL <= file_status) - break; - - /* - * If input parsers have not been allocated, do so now. - * We keep these instanced betwen parsers, but set them - * locally per parse routine since we can use different - * parsers with each one. - */ - - if ( ! (curp->man || curp->mdoc)) - pset(ln.buf + of, pos - of, curp); - - /* - * Lastly, push down into the parsers themselves. One - * of these will have already been set in the pset() - * routine. - * If libroff returns ROFF_TBL, then add it to the - * currently open parse. Since we only get here if - * there does exist data (see tbl_data.c), we're - * guaranteed that something's been allocated. - * Do the same for ROFF_EQN. - */ - - rc = -1; - - if (ROFF_TBL == rr) - while (NULL != (span = roff_span(curp->roff))) { - rc = curp->man ? - man_addspan(curp->man, span) : - mdoc_addspan(curp->mdoc, span); - if (0 == rc) - break; - } - else if (ROFF_EQN == rr) - rc = curp->mdoc ? - mdoc_addeqn(curp->mdoc, - roff_eqn(curp->roff)) : - man_addeqn(curp->man, - roff_eqn(curp->roff)); - else if (curp->man || curp->mdoc) - rc = curp->man ? - man_parseln(curp->man, - curp->line, ln.buf, of) : - mdoc_parseln(curp->mdoc, - curp->line, ln.buf, of); - - if (0 == rc) { - assert(MANDOCLEVEL_FATAL <= file_status); - break; - } - - /* Temporary buffers typically are not full. */ - - if (0 == start && '\0' == blk.buf[i]) - break; - - /* Start the next input line. */ - - pos = 0; + if (ferror(stream)) { + fclose(stream); + syscall = "getline"; + goto fail; } - free(ln.buf); +done: + free(line); + fclose(stream); + return; + +fail: + free(line); + warn("%s: SYSERR: %s", file, syscall); + if (rc < MANDOCLEVEL_SYSERR) + rc = MANDOCLEVEL_SYSERR; } -static void -pset(const char *buf, int pos, struct curparse *curp) +static int +koptions(int *options, char *arg) { - int i; - /* - * Try to intuit which kind of manual parser should be used. If - * passed in by command-line (-man, -mdoc), then use that - * explicitly. If passed as -mandoc, then try to guess from the - * line: either skip dot-lines, use -mdoc when finding `.Dt', or - * default to -man, which is more lenient. - * - * Separate out pmdoc/pman from mdoc/man: the first persists - * through all parsers, while the latter is used per-parse. - */ - - if ('.' == buf[0] || '\'' == buf[0]) { - for (i = 1; buf[i]; i++) - if (' ' != buf[i] && '\t' != buf[i]) - break; - if ('\0' == buf[i]) - return; + if ( ! strcmp(arg, "utf-8")) { + *options |= MPARSE_UTF8; + *options &= ~MPARSE_LATIN1; + } else if ( ! strcmp(arg, "iso-8859-1")) { + *options |= MPARSE_LATIN1; + *options &= ~MPARSE_UTF8; + } else if ( ! strcmp(arg, "us-ascii")) { + *options &= ~(MPARSE_UTF8 | MPARSE_LATIN1); + } else { + warnx("-K %s: Bad argument", arg); + return 0; } - - switch (curp->inttype) { - case (INTT_MDOC): - if (NULL == curp->pmdoc) - curp->pmdoc = mdoc_alloc - (&curp->regs, curp, mmsg); - assert(curp->pmdoc); - curp->mdoc = curp->pmdoc; - return; - case (INTT_MAN): - if (NULL == curp->pman) - curp->pman = man_alloc - (&curp->regs, curp, mmsg); - assert(curp->pman); - curp->man = curp->pman; - return; - default: - break; - } - - if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) { - if (NULL == curp->pmdoc) - curp->pmdoc = mdoc_alloc - (&curp->regs, curp, mmsg); - assert(curp->pmdoc); - curp->mdoc = curp->pmdoc; - return; - } - - if (NULL == curp->pman) - curp->pman = man_alloc(&curp->regs, curp, mmsg); - assert(curp->pman); - curp->man = curp->pman; + return 1; } static int -moptions(enum intt *tflags, char *arg) +moptions(int *options, char *arg) { - if (0 == strcmp(arg, "doc")) - *tflags = INTT_MDOC; + if (arg == NULL) + /* nothing to do */; + else if (0 == strcmp(arg, "doc")) + *options |= MPARSE_MDOC; else if (0 == strcmp(arg, "andoc")) - *tflags = INTT_AUTO; + /* nothing to do */; else if (0 == strcmp(arg, "an")) - *tflags = INTT_MAN; + *options |= MPARSE_MAN; else { - fprintf(stderr, "%s: Bad argument\n", arg); - return(0); + warnx("-m %s: Bad argument", arg); + return 0; } - return(1); + return 1; } static int @@ -983,87 +958,172 @@ toptions(struct curparse *curp, char *arg) else if (0 == strcmp(arg, "lint")) { curp->outtype = OUTT_LINT; curp->wlevel = MANDOCLEVEL_WARNING; - } - else if (0 == strcmp(arg, "tree")) + } else if (0 == strcmp(arg, "tree")) curp->outtype = OUTT_TREE; + else if (0 == strcmp(arg, "man")) + curp->outtype = OUTT_MAN; else if (0 == strcmp(arg, "html")) curp->outtype = OUTT_HTML; + else if (0 == strcmp(arg, "utf8")) + curp->outtype = OUTT_UTF8; + else if (0 == strcmp(arg, "locale")) + curp->outtype = OUTT_LOCALE; else if (0 == strcmp(arg, "xhtml")) - curp->outtype = OUTT_XHTML; + curp->outtype = OUTT_HTML; else if (0 == strcmp(arg, "ps")) curp->outtype = OUTT_PS; else if (0 == strcmp(arg, "pdf")) curp->outtype = OUTT_PDF; else { - fprintf(stderr, "%s: Bad argument\n", arg); - return(0); + warnx("-T %s: Bad argument", arg); + return 0; } - return(1); + return 1; } static int woptions(struct curparse *curp, char *arg) { char *v, *o; - const char *toks[6]; + const char *toks[7]; toks[0] = "stop"; toks[1] = "all"; toks[2] = "warning"; toks[3] = "error"; - toks[4] = "fatal"; - toks[5] = NULL; + toks[4] = "unsupp"; + toks[5] = "fatal"; + toks[6] = NULL; while (*arg) { o = arg; - switch (getsubopt(&arg, UNCONST(toks), &v)) { - case (0): + switch (getsubopt(&arg, (char * const *)toks, &v)) { + case 0: curp->wstop = 1; break; - case (1): - /* FALLTHROUGH */ - case (2): + case 1: + case 2: curp->wlevel = MANDOCLEVEL_WARNING; break; - case (3): + case 3: curp->wlevel = MANDOCLEVEL_ERROR; break; - case (4): - curp->wlevel = MANDOCLEVEL_FATAL; + case 4: + curp->wlevel = MANDOCLEVEL_UNSUPP; break; + case 5: + curp->wlevel = MANDOCLEVEL_BADARG; + break; default: - fprintf(stderr, "-W%s: Bad argument\n", o); - return(0); + warnx("-W %s: Bad argument", o); + return 0; } } - return(1); + return 1; } -static int -mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) +static void +mmsg(enum mandocerr t, enum mandoclevel lvl, + const char *file, int line, int col, const char *msg) { - struct curparse *cp; - enum mandoclevel level; + const char *mparse_msg; - level = MANDOCLEVEL_FATAL; - while (t < mandoclimits[level]) - /* LINTED */ - level--; + fprintf(stderr, "%s: %s:", getprogname(), + file == NULL ? "" : file); - cp = (struct curparse *)arg; - if (level < cp->wlevel) - return(1); + if (line) + fprintf(stderr, "%d:%d:", line, col + 1); - fprintf(stderr, "%s:%d:%d: %s: %s", - cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]); + fprintf(stderr, " %s", mparse_strlevel(lvl)); + + if (NULL != (mparse_msg = mparse_strerror(t))) + fprintf(stderr, ": %s", mparse_msg); + if (msg) fprintf(stderr, ": %s", msg); + fputc('\n', stderr); +} - if (file_status < level) - file_status = level; - - return(level < MANDOCLEVEL_FATAL); +static pid_t +spawn_pager(struct tag_files *tag_files) +{ + const struct timespec timeout = { 0, 100000000 }; /* 0.1s */ +#define MAX_PAGER_ARGS 16 + char *argv[MAX_PAGER_ARGS]; + const char *pager; + char *cp; + size_t cmdlen; + int argc; + pid_t pager_pid; + + pager = getenv("MANPAGER"); + if (pager == NULL || *pager == '\0') + pager = getenv("PAGER"); + if (pager == NULL || *pager == '\0') + pager = "more -s"; + cp = mandoc_strdup(pager); + + /* + * Parse the pager command into words. + * Intentionally do not do anything fancy here. + */ + + argc = 0; + while (argc + 4 < MAX_PAGER_ARGS) { + argv[argc++] = cp; + cp = strchr(cp, ' '); + if (cp == NULL) + break; + *cp++ = '\0'; + while (*cp == ' ') + cp++; + if (*cp == '\0') + break; + } + + /* For less(1), use the tag file. */ + + if ((cmdlen = strlen(argv[0])) >= 4) { + cp = argv[0] + cmdlen - 4; + if (strcmp(cp, "less") == 0) { + argv[argc++] = mandoc_strdup("-T"); + argv[argc++] = tag_files->tfn; + } + } + argv[argc++] = tag_files->ofn; + argv[argc] = NULL; + + switch (pager_pid = fork()) { + case -1: + err((int)MANDOCLEVEL_SYSERR, "fork"); + case 0: + break; + default: + (void)setpgid(pager_pid, 0); + (void)tcsetpgrp(tag_files->ofd, pager_pid); +#if HAVE_PLEDGE + if (pledge("stdio rpath tmppath tty proc", NULL) == -1) + err((int)MANDOCLEVEL_SYSERR, "pledge"); +#endif + tag_files->pager_pid = pager_pid; + return pager_pid; + } + + /* The child process becomes the pager. */ + + if (dup2(tag_files->ofd, STDOUT_FILENO) == -1) + err((int)MANDOCLEVEL_SYSERR, "pager stdout"); + close(tag_files->ofd); + close(tag_files->tfd); + + /* Do not start the pager before controlling the terminal. */ + + while (tcgetpgrp(STDOUT_FILENO) != getpid()) + nanosleep(&timeout, NULL); + + execvp(argv[0], argv); + err((int)MANDOCLEVEL_SYSERR, "exec %s", argv[0]); }