=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.134 retrieving revision 1.210 diff -u -p -r1.134 -r1.210 --- mandoc/main.c 2011/01/04 12:06:21 1.134 +++ mandoc/main.c 2014/12/31 16:52:39 1.210 @@ -1,7 +1,8 @@ -/* $Id: main.c,v 1.134 2011/01/04 12:06:21 kristaps Exp $ */ +/* $Id: main.c,v 1.210 2014/12/31 16:52:39 schwarze Exp $ */ /* - * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons - * Copyright (c) 2010 Ingo Schwarze + * Copyright (c) 2008-2012 Kristaps Dzonsons + * Copyright (c) 2010, 2011, 2012, 2014 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 @@ -15,15 +16,13 @@ * 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 #include +#include #include #include #include @@ -32,220 +31,104 @@ #include #include "mandoc.h" +#include "mandoc_aux.h" #include "main.h" #include "mdoc.h" #include "man.h" -#include "roff.h" +#include "manpath.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) */ +enum outmode { + OUTMODE_DEF = 0, + OUTMODE_FLN, + OUTMODE_LST, + OUTMODE_ALL, + OUTMODE_INT, + OUTMODE_ONE +}; + 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 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 */ + struct mparse *mp; + struct mchars *mchars; /* character table */ + enum mandoclevel wlevel; /* ignore messages below this */ + int wstop; /* stop after a file with a warning */ + enum outt outtype; /* which output to use */ out_mdoc outmdoc; /* mdoc output ptr */ - out_man outman; /* man output ptr */ + out_man outman; /* man output ptr */ out_free outfree; /* free output ptr */ void *outdata; /* data for output */ char outopts[BUFSIZ]; /* buf of output opts */ }; -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 -}; - -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", - "cannot parse date argument", - "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", - "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", - - /* 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", - - /* related to tables */ - "extra data cells", - - "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", - - "input stack limit exceeded, infinite loop?", - "skipping bad character", - "skipping text before the first section header", - "skipping unknown macro", - "NOT IMPLEMENTED: 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 koptions(int *, char *); +#if HAVE_SQLITE3 +int mandocdb(int, char**); +#endif +static int moptions(int *, char *); +static void mmsg(enum mandocerr, enum mandoclevel, + const char *, int, int, const char *); +static void parse(struct curparse *, int, + const char *, enum mandoclevel *); +#if HAVE_SQLITE3 +static enum mandoclevel passthrough(const char *, int, int); +#endif +static void spawn_pager(void); static int toptions(struct curparse *, char *); -static void usage(void) __attribute__((noreturn)); +static void usage(enum argmode) __attribute__((noreturn)); static void version(void) __attribute__((noreturn)); static int woptions(struct curparse *, char *); +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 const char *progname; -static enum mandoclevel file_status = MANDOCLEVEL_OK; -static enum mandoclevel exit_status = MANDOCLEVEL_OK; + int main(int argc, char *argv[]) { - int c; struct curparse curp; + struct mansearch search; + struct manpaths paths; + char *auxpaths; + char *defos; + unsigned char *uc; +#if HAVE_SQLITE3 + struct manpage *res, *resp; + char *conf_file, *defpaths; + size_t isec, i, sz; + int prio, best_prio, synopsis_only; + char sec; +#endif + enum mandoclevel rc; + enum outmode outmode; + int fd; + int show_usage; + int use_pager; + int options; + int c; progname = strrchr(argv[0], '/'); if (progname == NULL) @@ -253,317 +136,440 @@ main(int argc, char *argv[]) else ++progname; - memset(&curp, 0, sizeof(struct curparse)); +#if HAVE_SQLITE3 + if (strcmp(progname, BINM_MAKEWHATIS) == 0) + return(mandocdb(argc, argv)); +#endif - curp.inttype = INTT_AUTO; - curp.outtype = OUTT_ASCII; + /* Search options. */ + + memset(&paths, 0, sizeof(struct manpaths)); +#if HAVE_SQLITE3 + conf_file = defpaths = NULL; +#endif + 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_FATAL; + options = MPARSE_SO | MPARSE_UTF8 | MPARSE_LATIN1; + defos = NULL; - /* LINTED */ - while (-1 != (c = getopt(argc, argv, "m:O:T:VW:"))) + use_pager = 1; + show_usage = 0; +#if HAVE_SQLITE3 + synopsis_only = 0; +#endif + outmode = OUTMODE_DEF; + + 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)) + case 'a': + outmode = OUTMODE_ALL; + break; + case 'C': +#if HAVE_SQLITE3 + conf_file = optarg; +#endif + break; + case 'c': + use_pager = 0; + break; + case 'f': + search.argmode = ARG_WORD; + break; + case 'h': + (void)strlcat(curp.outopts, "synopsis,", BUFSIZ); +#if HAVE_SQLITE3 + synopsis_only = 1; +#endif + use_pager = 0; + outmode = OUTMODE_ALL; + break; + case 'I': + if (strncmp(optarg, "os=", 3)) { + fprintf(stderr, + "%s: -I %s: Bad argument\n", + progname, optarg); return((int)MANDOCLEVEL_BADARG); + } + if (defos) { + fprintf(stderr, + "%s: -I %s: Duplicate argument\n", + progname, optarg); + return((int)MANDOCLEVEL_BADARG); + } + defos = mandoc_strdup(optarg + 3); break; - case ('O'): + 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': +#if HAVE_SQLITE3 + defpaths = optarg; +#endif + break; + case 'm': + auxpaths = optarg; + break; + case 'O': + search.outkey = optarg; (void)strlcat(curp.outopts, optarg, BUFSIZ); (void)strlcat(curp.outopts, ",", BUFSIZ); break; - case ('T'): + case 'S': + search.arch = optarg; + break; + case 's': + search.sec = optarg; + break; + case 'T': if ( ! toptions(&curp, optarg)) return((int)MANDOCLEVEL_BADARG); break; - case ('W'): + case 'W': if ( ! woptions(&curp, optarg)) return((int)MANDOCLEVEL_BADARG); break; - case ('V'): + case 'w': + outmode = OUTMODE_FLN; + break; + case 'V': version(); /* NOTREACHED */ 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); + /* Parse arguments. */ - return((int)exit_status); -} + argc -= optind; + argv += optind; +#if HAVE_SQLITE3 + resp = NULL; +#endif - -static void -version(void) -{ - - (void)printf("%s %s\n", progname, VERSION); - exit((int)MANDOCLEVEL_OK); -} - - -static void -usage(void) -{ - - (void)fprintf(stderr, "usage: %s " - "[-V] " - "[-foption] " - "[-mformat] " - "[-Ooption] " - "[-Toutput] " - "[-Werr] " - "[file...]\n", - progname); - - exit((int)MANDOCLEVEL_BADARG); -} - -static void -ffile(const char *file, struct curparse *curp) -{ - /* - * 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(). + * Quirks for help(1) + * and for a man(1) section argument without -s. */ - curp->file = file; - - if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { - perror(curp->file); - exit_status = MANDOCLEVEL_SYSERR; - return; + if (search.argmode == ARG_NAME) { + if (*progname == 'h') { + if (argc == 0) { + argv = help_argv; + argc = 1; + } + } else if (((uc = argv[0]) != NULL) && + ((isdigit(uc[0]) && (uc[1] == '\0' || + (isalpha(uc[1]) && uc[2] == '\0'))) || + (uc[0] == 'n' && uc[1] == '\0'))) { + search.sec = uc; + argv++; + argc--; + } } - fdesc(curp); + rc = MANDOCLEVEL_OK; - if (-1 == close(curp->fd)) - perror(curp->file); -} + /* man(1), whatis(1), apropos(1) */ -static int -pfile(const char *file, struct curparse *curp) -{ - const char *savefile; - int fd, savefd; + if (search.argmode != ARG_FILE) { +#if HAVE_SQLITE3 + if (argc == 0) + usage(search.argmode); - if (-1 == (fd = open(file, O_RDONLY, 0))) { - perror(file); - file_status = MANDOCLEVEL_SYSERR; - return(0); - } + if (search.argmode == ARG_NAME && + outmode == OUTMODE_ONE) + search.firstmatch = 1; - savefile = curp->file; - savefd = curp->fd; + /* Access the mandoc database. */ - curp->file = file; - curp->fd = fd; + manpath_parse(&paths, conf_file, defpaths, auxpaths); + mansearch_setup(1); + if( ! mansearch(&search, &paths, argc, argv, &res, &sz)) + usage(search.argmode); + resp = res; - pdesc(curp); + if (sz == 0) { + if (search.argmode == ARG_NAME) + fprintf(stderr, "%s: No entry for %s " + "in the manual.\n", progname, argv[0]); + rc = MANDOCLEVEL_BADARG; + goto out; + } - curp->file = savefile; - curp->fd = savefd; + /* + * For standard man(1) and -a output mode, + * prepare for copying filename pointers + * into the program parameter array. + */ - if (-1 == close(fd)) - perror(file); + if (outmode == OUTMODE_ONE) { + argc = 1; + best_prio = 10; + } else if (outmode == OUTMODE_ALL) + argc = (int)sz; - return(MANDOCLEVEL_FATAL > file_status ? 1 : 0); -} + /* Iterate all matching manuals. */ + 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. */ + isec = strcspn(res[i].file, "123456789"); + sec = res[i].file[isec]; + if ('\0' == sec) + continue; + prio = sec_prios[sec - '1']; + if (prio >= best_prio) + continue; + best_prio = prio; + resp = res + i; + } + } -static void -resize_buf(struct buf *buf, size_t initial) -{ + /* + * 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. + */ - 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 (outmode == OUTMODE_FLN || outmode == OUTMODE_LST) + goto out; +#else + fputs("mandoc: database support not compiled in\n", + stderr); + return((int)MANDOCLEVEL_BADARG); +#endif } -} + /* mandoc(1) */ -static int -read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) -{ - struct stat st; - size_t off; - ssize_t ssz; + if (search.argmode == ARG_FILE && ! moptions(&options, auxpaths)) + return((int)MANDOCLEVEL_BADARG); - if (-1 == fstat(curp->fd, &st)) { - perror(curp->file); - return(0); - } + if (use_pager && isatty(STDOUT_FILENO)) + spawn_pager(); + curp.mchars = mchars_alloc(); + curp.mp = mparse_alloc(options, curp.wlevel, mmsg, + curp.mchars, defos); + /* - * 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. + * Conditionally start up the lookaside buffer before parsing. */ + if (OUTT_MAN == curp.outtype) + mparse_keep(curp.mp); - if (S_ISREG(st.st_mode)) { - if (st.st_size >= (1U << 31)) { - fprintf(stderr, "%s: input too large\n", - curp->file); - return(0); + if (argc == 0) + parse(&curp, STDIN_FILENO, "", &rc); + + while (argc) { +#if HAVE_SQLITE3 + if (resp != NULL) { + rc = mparse_open(curp.mp, &fd, resp->file); + if (fd == -1) + /* nothing */; + else if (resp->form & FORM_SRC) { + /* For .so only; ignore failure. */ + chdir(paths.paths[resp->ipath]); + parse(&curp, fd, resp->file, &rc); + } else + rc = passthrough(resp->file, fd, + synopsis_only); + resp++; + } else +#endif + { + rc = mparse_open(curp.mp, &fd, *argv++); + if (fd != -1) + parse(&curp, fd, argv[-1], &rc); } - *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); - } - /* - * 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. - */ + if (mparse_wait(curp.mp) != MANDOCLEVEL_OK) + rc = MANDOCLEVEL_SYSERR; - *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); - break; - } - resize_buf(fb, 65536); - } - 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); + if (MANDOCLEVEL_OK != rc && curp.wstop) break; - } - off += (size_t)ssz; + + if (--argc && curp.outtype <= OUTT_UTF8) + ascii_sepline(curp.outdata); } - free(fb->buf); - fb->buf = NULL; - return(0); -} + if (curp.outfree) + (*curp.outfree)(curp.outdata); + mparse_free(curp.mp); + mchars_free(curp.mchars); +#if HAVE_SQLITE3 +out: + if (search.argmode != ARG_FILE) { + manpath_free(&paths); + mansearch_free(res, sz); + mansearch_setup(0); + } +#endif + free(defos); + + return((int)rc); +} + static void -fdesc(struct curparse *curp) +version(void) { - /* - * 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. - */ + printf("mandoc %s\n", VERSION); + exit((int)MANDOCLEVEL_OK); +} - /* Zero the parse type. */ +static void +usage(enum argmode argmode) +{ - curp->mdoc = NULL; - curp->man = NULL; - file_status = MANDOCLEVEL_OK; - - /* Make sure the mandotory roff parser is initialised. */ - - if (NULL == curp->roff) { - curp->roff = roff_alloc(&curp->regs, curp, mmsg); - assert(curp->roff); + switch (argmode) { + case ARG_FILE: + fputs("usage: mandoc [-acfhklV] [-Ios=name] " + "[-Kencoding] [-mformat] [-Ooption]\n" + "\t [-Toutput] [-Wlevel] [file ...]\n", stderr); + break; + case ARG_NAME: + fputs("usage: man [-acfhklVw] [-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 [-acfhklVw] [-C file] " + "[-M path] [-m path] [-O outkey] [-S arch]\n" + "\t [-s section] name ...\n", stderr); + break; + case ARG_EXPR: + fputs("usage: apropos [-acfhklVw] [-C file] " + "[-M path] [-m path] [-O outkey] [-S arch]\n" + "\t [-s section] expression ...\n", stderr); + break; } + exit((int)MANDOCLEVEL_BADARG); +} - /* Fully parse the file. */ +static void +parse(struct curparse *curp, int fd, const char *file, + enum mandoclevel *level) +{ + enum mandoclevel rc; + struct mdoc *mdoc; + struct man *man; - pdesc(curp); + /* Begin by parsing the file itself. */ - if (MANDOCLEVEL_FATAL <= file_status) - goto cleanup; + assert(file); + assert(fd >= -1); - /* NOTE a parser may not have been assigned, yet. */ + rc = mparse_readfd(curp->mp, fd, file); - if ( ! (curp->man || curp->mdoc)) { - fprintf(stderr, "%s: Not a manual\n", curp->file); - file_status = MANDOCLEVEL_FATAL; - goto cleanup; - } + /* Stop immediately if the parse has failed. */ - /* Clean up the parse routine ASTs. */ - - if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) { - assert(MANDOCLEVEL_FATAL <= file_status); + if (MANDOCLEVEL_FATAL <= rc) goto cleanup; - } - if (curp->man && ! man_endparse(curp->man)) { - assert(MANDOCLEVEL_FATAL <= file_status); - goto cleanup; - } - - assert(curp->roff); - roff_endparse(curp->roff); - /* - * 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) + if (MANDOCLEVEL_OK != rc && curp->wstop) goto cleanup; /* If unset, allocate output dev now (if applicable). */ if ( ! (curp->outman && curp->outmdoc)) { switch (curp->outtype) { - case (OUTT_XHTML): - curp->outdata = xhtml_alloc(curp->outopts); + case OUTT_HTML: + curp->outdata = html_alloc(curp->mchars, + curp->outopts); + curp->outfree = html_free; break; - case (OUTT_HTML): - curp->outdata = html_alloc(curp->outopts); + case OUTT_UTF8: + curp->outdata = utf8_alloc(curp->mchars, + curp->outopts); + curp->outfree = ascii_free; break; - case (OUTT_ASCII): - curp->outdata = ascii_alloc(curp->outopts); + case OUTT_LOCALE: + curp->outdata = locale_alloc(curp->mchars, + curp->outopts); curp->outfree = ascii_free; break; - case (OUTT_PDF): - curp->outdata = pdf_alloc(curp->outopts); + case OUTT_ASCII: + curp->outdata = ascii_alloc(curp->mchars, + curp->outopts); + curp->outfree = ascii_free; + break; + case OUTT_PDF: + curp->outdata = pdf_alloc(curp->mchars, + curp->outopts); curp->outfree = pspdf_free; break; - case (OUTT_PS): - curp->outdata = ps_alloc(curp->outopts); + case OUTT_PS: + curp->outdata = ps_alloc(curp->mchars, + curp->outopts); curp->outfree = pspdf_free; break; default: @@ -571,22 +577,27 @@ fdesc(struct curparse *curp) } switch (curp->outtype) { - case (OUTT_HTML): - /* FALLTHROUGH */ - case (OUTT_XHTML): + case OUTT_HTML: curp->outman = html_man; curp->outmdoc = html_mdoc; - curp->outfree = html_free; break; - case (OUTT_TREE): + case OUTT_TREE: curp->outman = tree_man; curp->outmdoc = tree_mdoc; break; - case (OUTT_PDF): + case OUTT_MAN: + curp->outmdoc = man_mdoc; + curp->outman = man_man; + break; + case OUTT_PDF: /* FALLTHROUGH */ - case (OUTT_ASCII): + case OUTT_ASCII: /* FALLTHROUGH */ - case (OUTT_PS): + case OUTT_UTF8: + /* FALLTHROUGH */ + case OUTT_LOCALE: + /* FALLTHROUGH */ + case OUTT_PS: curp->outman = terminal_man; curp->outmdoc = terminal_mdoc; break; @@ -595,349 +606,126 @@ fdesc(struct curparse *curp) } } + mparse_result(curp->mp, &mdoc, &man, NULL); + /* 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); + if (man && curp->outman) + (*curp->outman)(curp->outdata, man); + if (mdoc && curp->outmdoc) + (*curp->outmdoc)(curp->outdata, mdoc); cleanup: - memset(&curp->regs, 0, sizeof(struct regset)); + mparse_reset(curp->mp); - /* 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; + if (*level < rc) + *level = rc; } -static void -pdesc(struct curparse *curp) +#if HAVE_SQLITE3 +static enum mandoclevel +passthrough(const char *file, int fd, int synopsis_only) { - struct buf blk; - int with_mmap; + const char synb[] = "S\bSY\bYN\bNO\bOP\bPS\bSI\bIS\bS"; + const char synr[] = "SYNOPSIS"; - /* - * 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. - */ + FILE *stream; + const char *syscall; + char *line; + size_t len, off; + ssize_t nw; + int print; - if ( ! read_whole_file(curp, &blk, &with_mmap)) { - file_status = MANDOCLEVEL_SYSERR; - return; + fflush(stdout); + + if ((stream = fdopen(fd, "r")) == NULL) { + close(fd); + syscall = "fdopen"; + goto fail; } - /* 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) -{ - 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; - - /* - * 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'). - */ - - memset(&ln, 0, sizeof(struct buf)); - - lnn = curp->line; - pos = 0; - - for (i = 0; i < (int)blk.sz; ) { - if (0 == pos && '\0' == blk.buf[i]) - break; - - if (start) { - curp->line = lnn; - curp->reparse_count = 0; - } - - while (i < (int)blk.sz && (start || '\0' != blk.buf[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++; - 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 & at least one other char. */ - - if ('\n' == blk.buf[i + 1]) { - i += 2; - /* Escaped newlines are skipped over */ - ++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; - } + print = 0; + while ((line = fgetln(stream, &len)) != NULL) { + if (synopsis_only) { + if (print) { + if ( ! isspace((unsigned char)*line)) + goto done; + while (len && + isspace((unsigned char)*line)) { + line++; + len--; } - - /* 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; + } else { + if ((len == sizeof(synb) && + ! strncmp(line, synb, len - 1)) || + (len == sizeof(synr) && + ! strncmp(line, synr, len - 1))) + print = 1; continue; - } else - break; - default: - break; - } - - /* - * 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. - */ - - if (ROFF_TBL == rr) { - assert(curp->man || curp->mdoc); - if (curp->man) - man_addspan(curp->man, roff_span(curp->roff)); - else - mdoc_addspan(curp->mdoc, roff_span(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 ( ! rc) { - assert(MANDOCLEVEL_FATAL <= file_status); - break; } } + for (off = 0; off < len; off += nw) + if ((nw = write(STDOUT_FILENO, line + off, + len - off)) == -1 || nw == 0) { + fclose(stream); + syscall = "write"; + goto fail; + } + } - /* 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 = "fgetln"; + goto fail; } - free(ln.buf); +done: + fclose(stream); + return(MANDOCLEVEL_OK); + +fail: + fprintf(stderr, "%s: %s: SYSERR: %s: %s", + progname, file, syscall, strerror(errno)); + return(MANDOCLEVEL_SYSERR); } +#endif -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 { + fprintf(stderr, "%s: -K %s: Bad argument\n", + progname, 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); + fprintf(stderr, "%s: -m %s: Bad argument\n", + progname, arg); return(0); } @@ -953,19 +741,25 @@ 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); + fprintf(stderr, "%s: -T %s: Bad argument\n", + progname, arg); return(0); } @@ -976,7 +770,7 @@ static int woptions(struct curparse *curp, char *arg) { char *v, *o; - const char *toks[6]; + const char *toks[6]; toks[0] = "stop"; toks[1] = "all"; @@ -988,22 +782,23 @@ woptions(struct curparse *curp, char *arg) while (*arg) { o = arg; switch (getsubopt(&arg, UNCONST(toks), &v)) { - case (0): + case 0: curp->wstop = 1; break; - case (1): + case 1: /* FALLTHROUGH */ - case (2): + case 2: curp->wlevel = MANDOCLEVEL_WARNING; break; - case (3): + case 3: curp->wlevel = MANDOCLEVEL_ERROR; break; - case (4): + case 4: curp->wlevel = MANDOCLEVEL_FATAL; break; default: - fprintf(stderr, "-W%s: Bad argument\n", o); + fprintf(stderr, "%s: -W %s: Bad argument\n", + progname, o); return(0); } } @@ -1011,29 +806,100 @@ woptions(struct curparse *curp, char *arg) 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:", progname, 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 void +spawn_pager(void) +{ +#define MAX_PAGER_ARGS 16 + char *argv[MAX_PAGER_ARGS]; + const char *pager; + char *cp; + int fildes[2]; + int argc; + + if (pipe(fildes) == -1) { + fprintf(stderr, "%s: pipe: %s\n", + progname, strerror(errno)); + return; + } + + switch (fork()) { + case -1: + fprintf(stderr, "%s: fork: %s\n", + progname, strerror(errno)); + exit((int)MANDOCLEVEL_SYSERR); + case 0: + close(fildes[0]); + if (dup2(fildes[1], STDOUT_FILENO) == -1) { + fprintf(stderr, "%s: dup output: %s\n", + progname, strerror(errno)); + exit((int)MANDOCLEVEL_SYSERR); + } + return; + default: + break; + } + + /* The original process becomes the pager. */ + + close(fildes[1]); + if (dup2(fildes[0], STDIN_FILENO) == -1) { + fprintf(stderr, "%s: dup input: %s\n", + progname, strerror(errno)); + exit((int)MANDOCLEVEL_SYSERR); + } + + pager = getenv("MANPAGER"); + if (pager == NULL || *pager == '\0') + pager = getenv("PAGER"); + if (pager == NULL || *pager == '\0') + pager = "/usr/bin/more -s"; + cp = mandoc_strdup(pager); + + /* + * Parse the pager command into words. + * Intentionally do not do anything fancy here. + */ + + argc = 0; + while (argc + 1 < MAX_PAGER_ARGS) { + argv[argc++] = cp; + cp = strchr(cp, ' '); + if (cp == NULL) + break; + *cp++ = '\0'; + while (*cp == ' ') + cp++; + if (*cp == '\0') + break; + } + argv[argc] = NULL; + + /* Hand over to the pager. */ + + execvp(argv[0], argv); + fprintf(stderr, "%s: exec: %s\n", + progname, strerror(errno)); + exit((int)MANDOCLEVEL_SYSERR); }