=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.54 retrieving revision 1.68 diff -u -p -r1.54 -r1.68 --- mandoc/main.c 2009/10/31 06:17:19 1.54 +++ mandoc/main.c 2010/05/15 09:46:31 1.68 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.54 2009/10/31 06:17:19 kristaps Exp $ */ +/* $Id: main.c,v 1.68 2010/05/15 09:46:31 joerg Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -14,6 +14,11 @@ * 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 @@ -30,19 +35,13 @@ #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 **); -extern size_t strlcat(char *, const char *, size_t); -# 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 *); @@ -63,6 +62,7 @@ enum outt { OUTT_ASCII = 0, OUTT_TREE, OUTT_HTML, + OUTT_XHTML, OUTT_LINT }; @@ -73,16 +73,15 @@ 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; @@ -91,31 +90,29 @@ struct curparse { char outopts[BUFSIZ]; }; +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 *); 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; progname = strrchr(argv[0], '/'); @@ -145,7 +142,7 @@ main(int argc, char *argv[]) (void)strlcat(curp.outopts, ",", BUFSIZ); break; case ('T'): - if ( ! toptions(&curp.outtype, optarg)) + if ( ! toptions(&curp, optarg)) return(EXIT_FAILURE); break; case ('W'): @@ -163,56 +160,30 @@ main(int argc, char *argv[]) argc -= optind; argv += optind; - memset(&ln, 0, sizeof(struct buf)); - memset(&blk, 0, 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) - man_reset(curp.lastman); - if (curp.lastmdoc) - mdoc_reset(curp.lastmdoc); - 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) { @@ -221,13 +192,13 @@ version(void) } -__dead static void +static void usage(void) { - (void)fprintf(stderr, "usage: %s [-V] [-foption...] " + (void)fprintf(stderr, "usage: %s [-V] [-foption] " "[-mformat] [-Ooption] [-Toutput] " - "[-Werr...]\n", progname); + "[-Werr] [file...]\n", progname); exit(EXIT_FAILURE); } @@ -243,13 +214,11 @@ 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; return(man_alloc(curp, pflags, &mancb)); @@ -267,55 +236,142 @@ 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; 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))) { perror(curp->file); - return(-1); + with_error = 1; + return; } - c = fdesc(blk, ln, curp); + fdesc(curp); if (-1 == close(curp->fd)) perror(curp->file); +} - return(c); + +static int +resize_buf(struct buf *buf, size_t initial) +{ + void *tmp; + size_t sz; + + if (buf->sz == 0) + sz = initial; + else + sz = 2 * buf->sz; + tmp = realloc(buf->buf, sz); + if (NULL == tmp) { + perror(NULL); + return(0); + } + buf->buf = tmp; + buf->sz = sz; + return(1); } 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; + size_t 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 (! resize_buf(fb, 65536)) + break; + } + 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) +{ + 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 @@ -323,113 +379,98 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse * growable, hence passed in by ptr-ptr. */ - if (-1 == fstat(curp->fd, &st)) - perror(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) { - perror(NULL); - exit(EXIT_FAILURE); - } - blk->sz = sz; - } - /* Fill buf with file blocksize. */ - for (lnn = pos = comment = 0; ; ) { - if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { - perror(curp->file); - return(-1); - } else if (0 == ssz) - break; + for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) { + if (pos >= (int)ln.sz) { + if (! resize_buf(&ln, 256)) + goto bailout; + } - /* Parse the read block into partial or full lines. */ + if ('\n' != blk.buf[i]) { + if (comment) + continue; + ln.buf[pos++] = blk.buf[i]; - 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) { - perror(NULL); - return(EXIT_FAILURE); - } - } + /* Handle in-line `\"' comments. */ - if ('\n' != blk->buf[i]) { - if (comment) - continue; - ln->buf[pos++] = blk->buf[i]; + if (1 == pos || '\"' != ln.buf[pos - 1]) + continue; - /* Handle in-line `\"' comments. */ + for (j = pos - 2; j >= 0; j--) + if ('\\' != ln.buf[j]) + break; - if (1 == pos || '\"' != ln->buf[pos - 1]) - continue; + if ( ! ((pos - 2 - j) % 2)) + continue; - for (j = pos - 2; j >= 0; j--) - if ('\\' != ln->buf[j]) - break; + comment = 1; + pos -= 2; + for (; pos > 0; --pos) { + if (ln.buf[pos - 1] != ' ') + break; + if (pos > 2 && ln.buf[pos - 2] == '\\') + break; + } + continue; + } - if ( ! ((pos - 2 - j) % 2)) - continue; + /* Handle escaped `\\n' newlines. */ - comment = 1; - pos -= 2; + if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) { + for (j = pos - 1; j >= 0; j--) + if ('\\' != ln.buf[j]) + break; + if ( ! ((pos - j) % 2)) { + pos--; + lnn++; continue; - } - - /* Handle escaped `\\n' newlines. */ - - if (pos > 0 && 0 == comment && - '\\' == ln->buf[pos - 1]) { - for (j = pos - 1; j >= 0; j--) - if ('\\' != ln->buf[j]) - break; - if ( ! ((pos - j) % 2)) { - pos--; - lnn++; - continue; - } } + } - ln->buf[pos] = 0; - lnn++; + ln.buf[pos] = 0; + lnn++; - /* If unset, assign parser in pset(). */ + /* If unset, assign parser in pset(). */ - if ( ! (man || mdoc) && ! pset(ln->buf, - pos, curp, &man, &mdoc)) - return(-1); + if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc)) + goto bailout; - pos = comment = 0; + pos = comment = 0; - /* Pass down into parsers. */ + /* 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)) { fprintf(stderr, "%s: Not a manual\n", curp->file); - return(0); + 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; @@ -458,7 +499,26 @@ 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; } @@ -490,14 +550,12 @@ 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; @@ -508,7 +566,6 @@ 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); } @@ -516,7 +573,6 @@ 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); } @@ -532,7 +588,7 @@ moptions(enum intt *tflags, char *arg) else if (0 == strcmp(arg, "an")) *tflags = INTT_MAN; else { - fprintf(stderr, "%s: Bad argument", arg); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } @@ -541,19 +597,24 @@ 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 { - fprintf(stderr, "%s: Bad argument", arg); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } @@ -570,39 +631,34 @@ foptions(int *fflags, char *arg) 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[6] = "ign-escape"; - toks[7] = NULL; + 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; - case (6): - *fflags &= ~NO_IGN_ESCAPE; - break; default: - fprintf(stderr, "%s: Bad argument", o); + fprintf(stderr, "%s: Bad argument\n", o); return(0); } } @@ -631,7 +687,7 @@ woptions(int *wflags, char *arg) *wflags |= WARN_WERR; break; default: - fprintf(stderr, "%s: Bad argument", o); + fprintf(stderr, "%s: Bad argument\n", o); return(0); } } @@ -651,6 +707,8 @@ 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); } @@ -668,9 +726,12 @@ 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); }