=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.45 retrieving revision 1.66 diff -u -p -r1.45 -r1.66 --- mandoc/main.c 2009/10/13 10:21:24 1.45 +++ mandoc/main.c 2010/05/15 05:11:50 1.66 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.45 2009/10/13 10:21:24 kristaps Exp $ */ +/* $Id: main.c,v 1.66 2010/05/15 05:11:50 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -14,10 +14,14 @@ * 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 @@ -27,21 +31,17 @@ #include "mdoc.h" #include "man.h" +#include "main.h" #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a)) -/* Account for FreeBSD and Linux in our declarations. */ +/* FIXME: Intel's compiler? LLVM? pcc? */ -#ifdef __linux__ -extern int getsubopt(char **, char * const *, char **); -# ifndef __dead -# define __dead __attribute__((__noreturn__)) +#if !defined(__GNUC__) || (__GNUC__ < 2) +# if !defined(lint) +# define __attribute__(x) # endif -#elif defined(__dead2) -# ifndef __dead -# define __dead __dead2 -# endif -#endif +#endif /* !defined(__GNUC__) || (__GNUC__ < 2) */ typedef void (*out_mdoc)(void *, const struct mdoc *); typedef void (*out_man)(void *, const struct man *); @@ -62,6 +62,7 @@ enum outt { OUTT_ASCII = 0, OUTT_TREE, OUTT_HTML, + OUTT_XHTML, OUTT_LINT }; @@ -72,69 +73,61 @@ struct curparse { #define WARN_WALL (1 << 0) /* All-warnings mask. */ #define WARN_WERR (1 << 2) /* Warnings->errors. */ int fflags; -#define IGN_SCOPE (1 << 0) /* Ignore scope errors. */ -#define NO_IGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ -#define NO_IGN_MACRO (1 << 2) /* Don't ignore bad macros. */ -#define NO_IGN_CHARS (1 << 3) /* Don't ignore bad chars. */ -#define IGN_ERRORS (1 << 4) /* Ignore failed parse. */ +#define FL_IGN_SCOPE (1 << 0) /* Ignore scope errors. */ +#define FL_NIGN_ESCAPE (1 << 1) /* Don't ignore bad escapes. */ +#define FL_NIGN_MACRO (1 << 2) /* Don't ignore bad macros. */ +#define FL_IGN_ERRORS (1 << 4) /* Ignore failed parse. */ +#define FL_STRICT FL_NIGN_ESCAPE | \ + FL_NIGN_MACRO 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; - char *outopts; + char outopts[BUFSIZ]; }; -extern void *html_alloc(char *); -extern void html_mdoc(void *, const struct mdoc *); -extern void html_man(void *, const struct man *); -extern void html_free(void *); -extern void *ascii_alloc(void); -extern void tree_mdoc(void *, const struct mdoc *); -extern void tree_man(void *, const struct man *); -extern void terminal_mdoc(void *, const struct mdoc *); -extern void terminal_man(void *, const struct man *); -extern void terminal_free(void *); - +static void fdesc(struct curparse *); +static void ffile(const char *, struct curparse *); static int foptions(int *, char *); -static int toptions(enum outt *, char *); -static int moptions(enum intt *, char *); -static int woptions(int *, char *); +static struct man *man_init(struct curparse *); +static struct mdoc *mdoc_init(struct curparse *); static int merr(void *, int, int, const char *); +static int moptions(enum intt *, char *); static int mwarn(void *, int, int, const char *); -static int ffile(struct buf *, struct buf *, - const char *, struct curparse *); -static int fdesc(struct buf *, struct buf *, - 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); +static int toptions(struct curparse *, char *); +static void usage(void) __attribute__((noreturn)); +static void version(void) __attribute__((noreturn)); +static int woptions(int *, char *); -extern char *__progname; +static const char *progname; +static int with_error; +static int with_warning; - int main(int argc, char *argv[]) { - int c, rc; - struct buf ln, blk; + int c; struct curparse curp; - bzero(&curp, sizeof(struct curparse)); + progname = strrchr(argv[0], '/'); + if (progname == NULL) + progname = argv[0]; + else + ++progname; + memset(&curp, 0, sizeof(struct curparse)); + curp.inttype = INTT_AUTO; curp.outtype = OUTT_ASCII; /* LINTED */ - while (-1 != (c = getopt(argc, argv, "f:m:o:T:VW:"))) + while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:"))) switch (c) { case ('f'): if ( ! foptions(&curp.fflags, optarg)) @@ -144,11 +137,12 @@ main(int argc, char *argv[]) if ( ! moptions(&curp.inttype, optarg)) return(EXIT_FAILURE); break; - case ('o'): - curp.outopts = optarg; + case ('O'): + (void)strlcat(curp.outopts, optarg, BUFSIZ); + (void)strlcat(curp.outopts, ",", BUFSIZ); break; case ('T'): - if ( ! toptions(&curp.outtype, optarg)) + if ( ! toptions(&curp, optarg)) return(EXIT_FAILURE); break; case ('W'): @@ -166,73 +160,45 @@ main(int argc, char *argv[]) argc -= optind; argv += optind; - bzero(&ln, sizeof(struct buf)); - bzero(&blk, sizeof(struct buf)); - - rc = 1; - if (NULL == *argv) { curp.file = ""; curp.fd = STDIN_FILENO; - c = fdesc(&blk, &ln, &curp); - if ( ! (IGN_ERRORS & curp.fflags)) - rc = 1 == c ? 1 : 0; - else - rc = -1 == c ? 0 : 1; + fdesc(&curp); } - while (rc && *argv) { - c = ffile(&blk, &ln, *argv, &curp); - if ( ! (IGN_ERRORS & curp.fflags)) - rc = 1 == c ? 1 : 0; - else - rc = -1 == c ? 0 : 1; + while (*argv) { + ffile(*argv, &curp); - 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; - } + if (with_error && !(curp.fflags & FL_IGN_ERRORS)) + break; + ++argv; } - if (blk.buf) - free(blk.buf); - if (ln.buf) - free(ln.buf); if (curp.outfree) (*curp.outfree)(curp.outdata); - if (curp.mdoc) - mdoc_free(curp.mdoc); - if (curp.man) - man_free(curp.man); - return(rc ? EXIT_SUCCESS : EXIT_FAILURE); + return((with_warning || with_error) ? + EXIT_FAILURE : EXIT_SUCCESS); } -__dead static void +static void version(void) { - (void)printf("%s %s\n", __progname, VERSION); + (void)printf("%s %s\n", progname, VERSION); exit(EXIT_SUCCESS); } -__dead static void +static void usage(void) { - (void)fprintf(stderr, "usage: %s [-V] [-foption...] " - "[-mformat] [-Toutput] [-Werr...]\n", - __progname); + (void)fprintf(stderr, "usage: %s [-V] [-foption] " + "[-mformat] [-Ooption] [-Toutput] " + "[-Werr] [file...]\n", progname); exit(EXIT_FAILURE); } @@ -241,7 +207,6 @@ static struct man * man_init(struct curparse *curp) { int pflags; - struct man *man; struct man_cb mancb; mancb.man_err = merr; @@ -249,19 +214,14 @@ man_init(struct curparse *curp) /* Defaults from mandoc.1. */ - pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS; + pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE; - if (curp->fflags & NO_IGN_MACRO) + if (curp->fflags & FL_NIGN_MACRO) pflags &= ~MAN_IGN_MACRO; - if (curp->fflags & NO_IGN_CHARS) - pflags &= ~MAN_IGN_CHARS; - if (curp->fflags & NO_IGN_ESCAPE) + if (curp->fflags & FL_NIGN_ESCAPE) pflags &= ~MAN_IGN_ESCAPE; - if (NULL == (man = man_alloc(curp, pflags, &mancb))) - warnx("memory exhausted"); - - return(man); + return(man_alloc(curp, pflags, &mancb)); } @@ -269,7 +229,6 @@ static struct mdoc * mdoc_init(struct curparse *curp) { int pflags; - struct mdoc *mdoc; struct mdoc_cb mdoccb; mdoccb.mdoc_err = merr; @@ -277,58 +236,133 @@ mdoc_init(struct curparse *curp) /* Defaults from mandoc.1. */ - pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS; + pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE; - if (curp->fflags & IGN_SCOPE) + if (curp->fflags & FL_IGN_SCOPE) pflags |= MDOC_IGN_SCOPE; - if (curp->fflags & NO_IGN_ESCAPE) + if (curp->fflags & FL_NIGN_ESCAPE) pflags &= ~MDOC_IGN_ESCAPE; - if (curp->fflags & NO_IGN_MACRO) + if (curp->fflags & FL_NIGN_MACRO) pflags &= ~MDOC_IGN_MACRO; - if (curp->fflags & NO_IGN_CHARS) - pflags &= ~MDOC_IGN_CHARS; - if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb))) - warnx("memory exhausted"); - - return(mdoc); + return(mdoc_alloc(curp, pflags, &mdoccb)); } -static int -ffile(struct buf *blk, struct buf *ln, - const char *file, struct curparse *curp) +static void +ffile(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(-1); + perror(curp->file); + with_error = 1; + return; } - c = fdesc(blk, ln, curp); + fdesc(curp); if (-1 == close(curp->fd)) - warn("%s", curp->file); - - return(c); + perror(curp->file); } static int -fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) +read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) { - size_t sz; - ssize_t ssz; struct stat st; - int j, i, pos, lnn, comment; + char *buf; + size_t sz, off; + ssize_t ssz; + + if (-1 == fstat(curp->fd, &st)) { + perror(curp->file); + with_error = 1; + return(0); + } + + /* + * 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 (S_ISREG(st.st_mode)) { + if (st.st_size >= (1U << 31)) { + fprintf(stderr, "%s: input too large\n", + curp->file); + with_error = 1; + return(0); + } + *with_mmap = 1; + fb->sz = st.st_size; + fb->buf = mmap(NULL, fb->sz, PROT_READ, + MAP_FILE, 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. + */ + + *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; + } + if (fb->sz == 0) + sz = 65536; + else + sz = 2 * fb->sz; + buf = realloc(fb->buf, sz); + if (NULL == buf) { + perror(NULL); + break; + } + fb->buf = buf; + fb->sz = sz; + } + ssz = read(curp->fd, fb->buf + off, fb->sz - off); + if (ssz == 0) { + fb->sz = off; + return(1); + } + if (ssz == -1) { + perror(curp->file); + break; + } + off += ssz; + } + + free(fb->buf); + fb->buf = NULL; + with_error = 1; + return(0); +} + + +static void +fdesc(struct curparse *curp) +{ + size_t sz; + struct buf ln, blk; + int j, i, pos, lnn, comment, with_mmap; struct man *man; struct mdoc *mdoc; sz = BUFSIZ; man = NULL; mdoc = NULL; + memset(&ln, 0, sizeof(struct buf)); /* * Two buffers: ln and buf. buf is the input buffer optimised @@ -336,53 +370,33 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse * growable, hence passed in by ptr-ptr. */ - if (-1 == fstat(curp->fd, &st)) - warn("%s", curp->file); - else if ((size_t)st.st_blksize > sz) - sz = st.st_blksize; + if (!read_whole_file(curp, &blk, &with_mmap)) + return; - if (sz > blk->sz) { - blk->buf = realloc(blk->buf, sz); - if (NULL == blk->buf) { - warn("realloc"); - return(-1); - } - blk->sz = sz; - } - /* Fill buf with file blocksize. */ - for (lnn = pos = comment = 0; ; ) { - if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { - warn("%s", curp->file); - return(-1); - } 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) { - warn("realloc"); - return(-1); + for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) { + if (pos >= (int)ln.sz) { + ln.sz += 256; /* Step-size. */ + ln.buf = realloc(ln.buf, ln.sz); + if (NULL == ln.buf) { + perror(NULL); + goto bailout; } } - if ('\n' != blk->buf[i]) { + if ('\n' != blk.buf[i]) { if (comment) continue; - ln->buf[pos++] = blk->buf[i]; + ln.buf[pos++] = blk.buf[i]; /* Handle in-line `\"' comments. */ - if (1 == pos || '\"' != ln->buf[pos - 1]) + if (1 == pos || '\"' != ln.buf[pos - 1]) continue; for (j = pos - 2; j >= 0; j--) - if ('\\' != ln->buf[j]) + if ('\\' != ln.buf[j]) break; if ( ! ((pos - 2 - j) % 2)) @@ -390,15 +404,21 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse comment = 1; pos -= 2; + for (; pos > 0; --pos) { + if (ln.buf[pos - 1] != ' ') + break; + if (pos > 2 && ln.buf[pos - 2] == '\\') + break; + } continue; } /* Handle escaped `\\n' newlines. */ if (pos > 0 && 0 == comment && - '\\' == ln->buf[pos - 1]) { + '\\' == ln.buf[pos - 1]) { for (j = pos - 1; j >= 0; j--) - if ('\\' != ln->buf[j]) + if ('\\' != ln.buf[j]) break; if ( ! ((pos - j) % 2)) { pos--; @@ -407,43 +427,47 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse } } - ln->buf[pos] = 0; + ln.buf[pos] = 0; lnn++; /* If unset, assign parser in pset(). */ - if ( ! (man || mdoc) && ! pset(ln->buf, + if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc)) - return(-1); + goto bailout; pos = comment = 0; /* Pass down into parsers. */ - if (man && ! man_parseln(man, lnn, ln->buf)) - return(0); - if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) - return(0); - } + if (man && ! man_parseln(man, lnn, ln.buf)) + goto bailout; + if (mdoc && ! mdoc_parseln(mdoc, lnn, ln.buf)) + goto bailout; } /* NOTE a parser may not have been assigned, yet. */ if ( ! (man || mdoc)) { - (void)fprintf(stderr, "%s: not a manual\n", - curp->file); - return(0); + fprintf(stderr, "%s: Not a manual\n", curp->file); + goto bailout; } if (mdoc && ! mdoc_endparse(mdoc)) - return(0); + goto bailout; if (man && ! man_endparse(man)) - return(0); + goto bailout; /* 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); + curp->outman = html_man; + curp->outmdoc = html_mdoc; + curp->outfree = html_free; + break; case (OUTT_HTML): curp->outdata = html_alloc(curp->outopts); curp->outman = html_man; @@ -472,7 +496,26 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse if (mdoc && curp->outmdoc) (*curp->outmdoc)(curp->outdata, mdoc); - return(1); + cleanup: + if (curp->mdoc) { + mdoc_free(curp->mdoc); + curp->mdoc = NULL; + } + if (curp->man) { + man_free(curp->man); + curp->man = NULL; + } + if (ln.buf) + free(ln.buf); + if (with_mmap) + munmap(blk.buf, blk.sz); + else + free(blk.buf); + return; + + bailout: + with_error = 1; + goto cleanup; } @@ -504,14 +547,12 @@ pset(const char *buf, int pos, struct curparse *curp, 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; @@ -522,7 +563,6 @@ pset(const char *buf, int pos, struct curparse *curp, curp->mdoc = mdoc_init(curp); if (NULL == (*mdoc = curp->mdoc)) return(0); - curp->lastmdoc = *mdoc; return(1); } @@ -530,7 +570,6 @@ pset(const char *buf, int pos, struct curparse *curp, curp->man = man_init(curp); if (NULL == (*man = curp->man)) return(0); - curp->lastman = *man; return(1); } @@ -546,7 +585,7 @@ moptions(enum intt *tflags, char *arg) else if (0 == strcmp(arg, "an")) *tflags = INTT_MAN; else { - warnx("bad argument: -m%s", arg); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } @@ -555,19 +594,24 @@ moptions(enum intt *tflags, char *arg) static int -toptions(enum outt *tflags, char *arg) +toptions(struct curparse *curp, char *arg) { if (0 == strcmp(arg, "ascii")) - *tflags = OUTT_ASCII; - else if (0 == strcmp(arg, "lint")) - *tflags = OUTT_LINT; + curp->outtype = OUTT_ASCII; + else if (0 == strcmp(arg, "lint")) { + curp->outtype = OUTT_LINT; + curp->wflags |= WARN_WALL; + curp->fflags |= FL_STRICT; + } else if (0 == strcmp(arg, "tree")) - *tflags = OUTT_TREE; + curp->outtype = OUTT_TREE; else if (0 == strcmp(arg, "html")) - *tflags = OUTT_HTML; + curp->outtype = OUTT_HTML; + else if (0 == strcmp(arg, "xhtml")) + curp->outtype = OUTT_XHTML; else { - warnx("bad argument: -T%s", arg); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } @@ -579,40 +623,39 @@ static int foptions(int *fflags, char *arg) { char *v, *o; - const char *toks[7]; + const char *toks[8]; toks[0] = "ign-scope"; toks[1] = "no-ign-escape"; toks[2] = "no-ign-macro"; - toks[3] = "no-ign-chars"; - toks[4] = "ign-errors"; - toks[5] = "strict"; + toks[3] = "ign-errors"; + toks[4] = "strict"; + toks[5] = "ign-escape"; toks[6] = NULL; while (*arg) { o = arg; switch (getsubopt(&arg, UNCONST(toks), &v)) { case (0): - *fflags |= IGN_SCOPE; + *fflags |= FL_IGN_SCOPE; break; case (1): - *fflags |= NO_IGN_ESCAPE; + *fflags |= FL_NIGN_ESCAPE; break; case (2): - *fflags |= NO_IGN_MACRO; + *fflags |= FL_NIGN_MACRO; break; case (3): - *fflags |= NO_IGN_CHARS; + *fflags |= FL_IGN_ERRORS; break; case (4): - *fflags |= IGN_ERRORS; + *fflags |= FL_STRICT; break; case (5): - *fflags |= NO_IGN_ESCAPE | - NO_IGN_MACRO | NO_IGN_CHARS; + *fflags &= ~FL_NIGN_ESCAPE; break; default: - warnx("bad argument: -f%s", o); + fprintf(stderr, "%s: Bad argument\n", o); return(0); } } @@ -641,7 +684,7 @@ woptions(int *wflags, char *arg) *wflags |= WARN_WERR; break; default: - warnx("bad argument: -W%s", o); + fprintf(stderr, "%s: Bad argument\n", o); return(0); } } @@ -661,6 +704,8 @@ merr(void *arg, int line, int col, const char *msg) (void)fprintf(stderr, "%s:%d:%d: error: %s\n", curp->file, line, col + 1, msg); + with_error = 1; + return(0); } @@ -678,9 +723,12 @@ mwarn(void *arg, int line, int col, const char *msg) (void)fprintf(stderr, "%s:%d:%d: warning: %s\n", curp->file, line, col + 1, msg); - if ( ! (curp->wflags & WARN_WERR)) - return(1); - - return(0); + with_warning = 1; + if (curp->wflags & WARN_WERR) { + with_error = 1; + return(0); + } + + return(1); }