=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.6 retrieving revision 1.22 diff -u -p -r1.6 -r1.22 --- mandoc/main.c 2009/03/20 21:29:29 1.6 +++ mandoc/main.c 2009/04/03 11:08:39 1.22 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.6 2009/03/20 21:29:29 kristaps Exp $ */ +/* $Id: main.c,v 1.22 2009/04/03 11:08:39 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -27,89 +27,124 @@ #include #include "mdoc.h" +#include "man.h" +/* Account for FreeBSD and Linux in our declarations. */ + #ifdef __linux__ extern int getsubopt(char **, char * const *, char **); # ifndef __dead # define __dead __attribute__((__noreturn__)) # endif +#elif defined(__dead2) +# ifndef __dead +# define __dead __dead2 +# endif #endif +typedef int (*out_mdoc)(void *, const struct mdoc *); +typedef int (*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_LINT +}; + struct curparse { - const char *file; + const char *file; /* Current parse. */ + int fd; /* Current parse. */ int wflags; #define WARN_WALL 0x03 /* All-warnings mask. */ #define WARN_WCOMPAT (1 << 0) /* Compatibility warnings. */ #define WARN_WSYNTAX (1 << 1) /* Syntax warnings. */ #define WARN_WERR (1 << 2) /* Warnings->errors. */ + int fflags; /* Per-intt flags. */ + enum intt inttype; /* Input parsers. */ + struct man *man; + struct man *lastman; + struct mdoc *mdoc; + struct mdoc *lastmdoc; + enum outt outtype; /* Output devices. */ + out_mdoc outmdoc; + out_man outman; + out_free outfree; + void *outdata; }; -enum outt { - OUTT_ASCII, - OUTT_LATIN1, - OUTT_UTF8, - OUTT_TREE, - OUTT_LINT -}; +#define IGN_SCOPE (1 << 0) /* Ignore scope errors. */ +#define IGN_ESCAPE (1 << 1) /* Ignore bad escapes. */ +#define IGN_MACRO (1 << 2) /* Ignore unknown macros. */ +#define NO_IGN_MACRO (1 << 3) -typedef int (*out_run)(void *, const struct mdoc *); -typedef void (*out_free)(void *); - -extern char *__progname; - extern void *ascii_alloc(void); -extern void *latin1_alloc(void); -extern void *utf8_alloc(void); -extern int terminal_run(void *, const struct mdoc *); -extern int tree_run(void *, const struct mdoc *); +extern int tree_mdoc(void *, const struct mdoc *); +extern int tree_man(void *, const struct man *); +extern int terminal_mdoc(void *, const struct mdoc *); +extern int terminal_man(void *, const struct man *); extern void terminal_free(void *); -__dead static void version(void); -__dead static void usage(void); static int foptions(int *, char *); static int toptions(enum outt *, char *); +static int moptions(enum intt *, char *); static int woptions(int *, char *); static int merr(void *, int, int, const char *); -static int mwarn(void *, int, int, +static int manwarn(void *, int, int, const char *); +static int mdocwarn(void *, int, int, enum mdoc_warn, const char *); -static int file(struct buf *, struct buf *, - const char *, struct mdoc *); +static int fstdin(struct buf *, struct buf *, + struct curparse *); +static int ffile(struct buf *, struct buf *, + const char *, struct curparse *); static int fdesc(struct buf *, struct buf *, - const char *, int, struct mdoc *); + struct curparse *); +static int pset(const char *, int, struct curparse *, + struct man **, struct mdoc **); +static struct man *man_init(struct curparse *); +static struct mdoc *mdoc_init(struct curparse *); +__dead static void version(void); +__dead static void usage(void); +extern char *__progname; + int main(int argc, char *argv[]) { - int c, rc, fflags, wflags; - struct mdoc_cb cb; - struct mdoc *mdoc; - void *outdata; - enum outt outtype; + int c, rc; struct buf ln, blk; - out_run outrun; - out_free outfree; struct curparse curp; - fflags = wflags = 0; - outtype = OUTT_ASCII; - bzero(&curp, sizeof(struct curparse)); + curp.inttype = INTT_AUTO; + curp.outtype = OUTT_ASCII; + /* LINTED */ - while (-1 != (c = getopt(argc, argv, "f:VW:T:"))) + while (-1 != (c = getopt(argc, argv, "f:m:VW:T:"))) switch (c) { case ('f'): - if ( ! foptions(&fflags, optarg)) + if ( ! foptions(&curp.fflags, optarg)) return(0); break; + case ('m'): + if ( ! moptions(&curp.inttype, optarg)) + return(0); + break; case ('T'): - if ( ! toptions(&outtype, optarg)) + if ( ! toptions(&curp.outtype, optarg)) return(0); break; case ('W'): @@ -127,91 +162,44 @@ main(int argc, char *argv[]) argc -= optind; argv += optind; - /* - * Allocate the appropriate front-end. Note that utf8, ascii - * and latin1 all resolve to the terminal front-end with - * different encodings (see terminal.c). Not all frontends have - * cleanup or alloc routines. - */ + /* Configure buffers. */ - switch (outtype) { - case (OUTT_LATIN1): - outdata = latin1_alloc(); - outrun = terminal_run; - outfree = terminal_free; - break; - case (OUTT_UTF8): - outdata = utf8_alloc(); - outrun = terminal_run; - outfree = terminal_free; - break; - case (OUTT_TREE): - outdata = NULL; - outrun = tree_run; - outfree = NULL; - break; - case (OUTT_LINT): - outdata = NULL; - outrun = NULL; - outfree = NULL; - break; - default: - outdata = ascii_alloc(); - outrun = terminal_run; - outfree = terminal_free; - break; - } - - /* - * All callbacks route into here, where we print them onto the - * screen. XXX - for now, no path for debugging messages. - */ - - cb.mdoc_msg = NULL; - cb.mdoc_err = merr; - cb.mdoc_warn = mwarn; - bzero(&ln, sizeof(struct buf)); bzero(&blk, sizeof(struct buf)); - mdoc = mdoc_alloc(&curp, fflags, &cb); + rc = 1; - /* - * Loop around available files. - */ + if (NULL == *argv) + if ( ! fstdin(&blk, &ln, &curp)) + rc = 0; - if (NULL == *argv) { - curp.file = ""; - c = fdesc(&blk, &ln, "stdin", STDIN_FILENO, mdoc); - rc = 0; - if (c && NULL == outrun) - rc = 1; - else if (c && outrun && (*outrun)(outdata, mdoc)) - rc = 1; - } else { - while (*argv) { - curp.file = *argv; - c = file(&blk, &ln, *argv, mdoc); - if ( ! c) - break; - if (outrun && ! (*outrun)(outdata, mdoc)) - break; - /* Reset the parser for another file. */ - mdoc_reset(mdoc); - argv++; + while (rc && *argv) { + if ( ! ffile(&blk, &ln, *argv, &curp)) + rc = 0; + argv++; + if (*argv && rc) { + if (curp.lastman) + if ( ! man_reset(curp.lastman)) + rc = 0; + if (curp.lastmdoc) + if ( ! mdoc_reset(curp.lastmdoc)) + rc = 0; + curp.lastman = NULL; + curp.lastmdoc = NULL; } - rc = NULL == *argv; } if (blk.buf) free(blk.buf); if (ln.buf) free(ln.buf); - if (outfree) - (*outfree)(outdata); + if (curp.outfree) + (*curp.outfree)(curp.outdata); + if (curp.mdoc) + mdoc_free(curp.mdoc); + if (curp.man) + man_free(curp.man); - mdoc_free(mdoc); - return(rc ? EXIT_SUCCESS : EXIT_FAILURE); } @@ -221,8 +209,7 @@ version(void) { (void)printf("%s %s\n", __progname, VERSION); - exit(0); - /* NOTREACHED */ + exit(EXIT_SUCCESS); } @@ -230,78 +217,146 @@ __dead static void usage(void) { - (void)fprintf(stderr, "usage: %s\n", __progname); - exit(1); - /* NOTREACHED */ + (void)fprintf(stderr, "usage: %s [-V] [-foption...] " + "[-mformat] [-Toutput] [-Werr...]\n", + __progname); + exit(EXIT_FAILURE); } +static struct man * +man_init(struct curparse *curp) +{ + int pflags; + struct man *man; + struct man_cb mancb; + + mancb.man_err = merr; + mancb.man_warn = manwarn; + + pflags = MAN_IGN_MACRO; /* XXX */ + + if (curp->fflags & NO_IGN_MACRO) + pflags &= ~MAN_IGN_MACRO; + + if (NULL == (man = man_alloc(curp, pflags, &mancb))) + warnx("memory exhausted"); + + return(man); +} + + +static struct mdoc * +mdoc_init(struct curparse *curp) +{ + int pflags; + struct mdoc *mdoc; + struct mdoc_cb mdoccb; + + mdoccb.mdoc_msg = NULL; + mdoccb.mdoc_err = merr; + mdoccb.mdoc_warn = mdocwarn; + + pflags = 0; /* XXX */ + + if (curp->fflags & IGN_SCOPE) + pflags |= MDOC_IGN_SCOPE; + if (curp->fflags & IGN_ESCAPE) + pflags |= MDOC_IGN_ESCAPE; + if (curp->fflags & IGN_MACRO) + pflags |= MDOC_IGN_MACRO; + + if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb))) + warnx("memory allocated"); + + return(mdoc); +} + + static int -file(struct buf *blk, struct buf *ln, - const char *file, struct mdoc *mdoc) +fstdin(struct buf *blk, struct buf *ln, struct curparse *curp) { - int fd, c; - if (-1 == (fd = open(file, O_RDONLY, 0))) { - warn("%s", file); + curp->file = ""; + curp->fd = STDIN_FILENO; + return(fdesc(blk, ln, curp)); +} + + +static int +ffile(struct buf *blk, struct buf *ln, + const char *file, struct curparse *curp) +{ + int c; + + curp->file = file; + if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { + warn("%s", curp->file); return(0); } - c = fdesc(blk, ln, file, fd, mdoc); + c = fdesc(blk, ln, curp); - if (-1 == close(fd)) - warn("%s", file); + if (-1 == close(curp->fd)) + warn("%s", curp->file); return(c); } static int -fdesc(struct buf *blk, struct buf *ln, - const char *f, int fd, struct mdoc *mdoc) +fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) { size_t sz; ssize_t ssz; struct stat st; int j, i, pos, lnn; + struct man *man; + struct mdoc *mdoc; + sz = BUFSIZ; + man = NULL; + mdoc = NULL; + /* - * Two buffers: ln and buf. buf is the input buffer, optimised - * for each file's block size. ln is a line buffer. Both + * Two buffers: ln and buf. buf is the input buffer optimised + * here for each file's block size. ln is a line buffer. Both * growable, hence passed in by ptr-ptr. */ - sz = BUFSIZ; - - if (-1 == fstat(fd, &st)) - warnx("%s", f); + if (-1 == fstat(curp->fd, &st)) + warnx("%s", curp->file); else if ((size_t)st.st_blksize > sz) sz = st.st_blksize; if (sz > blk->sz) { blk->buf = realloc(blk->buf, sz); - if (NULL == blk->buf) - err(1, "realloc"); + if (NULL == blk->buf) { + warn("realloc"); + return(0); + } blk->sz = sz; } - /* - * Fill buf with file blocksize and parse newlines into ln. - */ + /* Fill buf with file blocksize. */ - for (lnn = 1, pos = 0; ; ) { - if (-1 == (ssz = read(fd, blk->buf, sz))) { - warn("%s", f); + for (lnn = 0, pos = 0; ; ) { + if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { + warn("%s", curp->file); return(0); } else if (0 == ssz) break; + /* Parse the read block into partial or full lines. */ + for (i = 0; i < (int)ssz; i++) { if (pos >= (int)ln->sz) { ln->sz += 256; /* Step-size. */ ln->buf = realloc(ln->buf, ln->sz); - if (NULL == ln->buf) - err(1, "realloc"); + if (NULL == ln->buf) { + warn("realloc"); + return(0); + } } if ('\n' != blk->buf[i]) { @@ -324,27 +379,155 @@ fdesc(struct buf *blk, struct buf *ln, } ln->buf[pos] = 0; - if ( ! mdoc_parseln(mdoc, lnn, ln->buf)) - return(0); lnn++; + + /* + * If no manual parser has been assigned, then + * try to assign one in pset(), which may do + * nothing at all. After this, parse the manual + * line accordingly. + */ + + if ( ! (man || mdoc) && ! pset(ln->buf, + pos, curp, &man, &mdoc)) + return(0); + pos = 0; + + if (man && ! man_parseln(man, lnn, ln->buf)) + return(0); + if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) + return(0); } } - return(mdoc_endparse(mdoc)); + /* Note that a parser may not have been assigned, yet. */ + + if ( ! (man || mdoc)) { + warnx("%s: not a manual", curp->file); + return(0); + } + + if (mdoc && ! mdoc_endparse(mdoc)) + return(0); + if (man && ! man_endparse(man)) + return(0); + + /* + * If an output device hasn't been allocated, see if we should + * do so now. Note that not all outtypes have functions, so + * this switch statement may be superfluous, but it's + * low-overhead enough not to matter very much. + */ + + if ( ! (curp->outman && curp->outmdoc)) { + switch (curp->outtype) { + case (OUTT_TREE): + curp->outman = tree_man; + curp->outmdoc = tree_mdoc; + break; + case (OUTT_LINT): + break; + default: + curp->outdata = ascii_alloc(); + curp->outman = terminal_man; + curp->outmdoc = terminal_mdoc; + curp->outfree = terminal_free; + break; + } + } + + /* Execute the out device, if it exists. */ + + if (man && curp->outman) + if ( ! (*curp->outman)(curp->outdata, man)) + return(0); + if (mdoc && curp->outmdoc) + if ( ! (*curp->outmdoc)(curp->outdata, mdoc)) + return(0); + + return(1); } static int +pset(const char *buf, int pos, struct curparse *curp, + struct man **man, struct mdoc **mdoc) +{ + + /* + * 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 comments, use -mdoc when finding `.Dt', or + * default to -man, which is more lenient. + */ + + if (pos >= 3 && 0 == memcmp(buf, ".\\\"", 3)) + return(1); + + switch (curp->inttype) { + case (INTT_MDOC): + if (NULL == curp->mdoc) + curp->mdoc = mdoc_init(curp); + if (NULL == (*mdoc = curp->mdoc)) + return(0); + curp->lastmdoc = *mdoc; + return(1); + case (INTT_MAN): + if (NULL == curp->man) + curp->man = man_init(curp); + if (NULL == (*man = curp->man)) + return(0); + curp->lastman = *man; + return(1); + default: + break; + } + + if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) { + if (NULL == curp->mdoc) + curp->mdoc = mdoc_init(curp); + if (NULL == (*mdoc = curp->mdoc)) + return(0); + curp->lastmdoc = *mdoc; + return(1); + } + + if (NULL == curp->man) + curp->man = man_init(curp); + if (NULL == (*man = curp->man)) + return(0); + curp->lastman = *man; + return(1); +} + + +static int +moptions(enum intt *tflags, char *arg) +{ + + if (0 == strcmp(arg, "doc")) + *tflags = INTT_MDOC; + else if (0 == strcmp(arg, "andoc")) + *tflags = INTT_AUTO; + else if (0 == strcmp(arg, "an")) + *tflags = INTT_MAN; + else { + warnx("bad argument: -m%s", arg); + return(0); + } + + return(1); +} + + +static int toptions(enum outt *tflags, char *arg) { if (0 == strcmp(arg, "ascii")) *tflags = OUTT_ASCII; - else if (0 == strcmp(arg, "latin1")) - *tflags = OUTT_LATIN1; - else if (0 == strcmp(arg, "utf8")) - *tflags = OUTT_UTF8; else if (0 == strcmp(arg, "lint")) *tflags = OUTT_LINT; else if (0 == strcmp(arg, "tree")) @@ -366,24 +549,28 @@ static int foptions(int *fflags, char *arg) { char *v; - char *toks[4]; + char *toks[5]; toks[0] = "ign-scope"; toks[1] = "ign-escape"; toks[2] = "ign-macro"; - toks[3] = NULL; + toks[3] = "no-ign-macro"; + toks[4] = NULL; while (*arg) switch (getsubopt(&arg, toks, &v)) { case (0): - *fflags |= MDOC_IGN_SCOPE; + *fflags |= IGN_SCOPE; break; case (1): - *fflags |= MDOC_IGN_ESCAPE; + *fflags |= IGN_ESCAPE; break; case (2): - *fflags |= MDOC_IGN_MACRO; + *fflags |= IGN_MACRO; break; + case (3): + *fflags |= NO_IGN_MACRO; + break; default: warnx("bad argument: -f%s", arg); return(0); @@ -447,7 +634,7 @@ merr(void *arg, int line, int col, const char *msg) static int -mwarn(void *arg, int line, int col, +mdocwarn(void *arg, int line, int col, enum mdoc_warn type, const char *msg) { struct curparse *curp; @@ -482,3 +669,23 @@ mwarn(void *arg, int line, int col, } +static int +manwarn(void *arg, int line, int col, const char *msg) +{ + struct curparse *curp; + + curp = (struct curparse *)arg; + + if ( ! (curp->wflags & WARN_WSYNTAX)) + return(1); + + warnx("%s:%d: syntax warning: %s (column %d)", + curp->file, line, msg, col); + + if ( ! (curp->wflags & WARN_WERR)) + return(1); + + warnx("%s: considering warnings as errors", + __progname); + return(0); +}