=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.64 retrieving revision 1.68 diff -u -p -r1.64 -r1.68 --- mandoc/main.c 2010/05/14 17:59:07 1.64 +++ mandoc/main.c 2010/05/15 09:46:31 1.68 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.64 2010/05/14 17:59:07 kristaps Exp $ */ +/* $Id: main.c,v 1.68 2010/05/15 09:46:31 joerg Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -18,6 +18,7 @@ #include "config.h" #endif +#include #include #include @@ -76,11 +77,11 @@ struct curparse { #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; @@ -89,34 +90,29 @@ struct curparse { char outopts[BUFSIZ]; }; -#define FL_STRICT FL_NIGN_ESCAPE | \ - FL_NIGN_MACRO - +static void fdesc(struct curparse *); +static void ffile(const char *, struct curparse *); static int foptions(int *, char *); -static int toptions(struct curparse *, 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 void ffile(struct buf *, struct buf *, - const char *, struct curparse *); -static void 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 *); -static void version(void) __attribute__((noreturn)); +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, with_warning; +static int with_error; +static int with_warning; int main(int argc, char *argv[]) { int c; - struct buf ln, blk; struct curparse curp; progname = strrchr(argv[0], '/'); @@ -164,32 +160,26 @@ main(int argc, char *argv[]) argc -= optind; argv += optind; - memset(&ln, 0, sizeof(struct buf)); - memset(&blk, 0, sizeof(struct buf)); - if (NULL == *argv) { curp.file = ""; curp.fd = STDIN_FILENO; - fdesc(&blk, &ln, &curp); + fdesc(&curp); } while (*argv) { - ffile(&blk, &ln, *argv, &curp); + ffile(*argv, &curp); 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); - return((with_warning || with_error) ? EXIT_FAILURE : EXIT_SUCCESS ); + return((with_warning || with_error) ? + EXIT_FAILURE : EXIT_SUCCESS); } @@ -260,8 +250,7 @@ mdoc_init(struct curparse *curp) static void -ffile(struct buf *blk, struct buf *ln, - const char *file, struct curparse *curp) +ffile(const char *file, struct curparse *curp) { curp->file = file; @@ -271,134 +260,193 @@ ffile(struct buf *blk, struct buf *ln, return; } - fdesc(blk, ln, curp); + fdesc(curp); if (-1 == close(curp->fd)) perror(curp->file); } -static void -fdesc(struct buf *blk, struct buf *ln, struct curparse *curp) +static int +resize_buf(struct buf *buf, size_t initial) { - size_t sz; - ssize_t ssz; - struct stat st; - int j, i, pos, lnn, comment; - struct man *man; - struct mdoc *mdoc; + void *tmp; + size_t sz; - sz = BUFSIZ; - man = NULL; - mdoc = NULL; + 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); +} - /* - * 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. - */ +static int +read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) +{ + struct stat st; + size_t off; + ssize_t ssz; + if (-1 == fstat(curp->fd, &st)) { perror(curp->file); with_error = 1; - return; + return(0); } - if ((size_t)st.st_blksize > sz) - sz = st.st_blksize; - if (sz > blk->sz) { - void *buf = realloc(blk->buf, sz); + /* + * 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 (NULL == buf) { - perror(NULL); + 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; + return(0); } - blk->buf = buf; - blk->sz = sz; + *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); } - /* Fill buf with file blocksize. */ + /* + * 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. + */ - for (lnn = pos = comment = 0; ; ) { - if (-1 == (ssz = read(curp->fd, blk->buf, sz))) { + *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); - goto bailout; - } else if (0 == ssz) break; + } + off += ssz; + } - /* Parse the read block into partial or full lines. */ + free(fb->buf); + fb->buf = NULL; + with_error = 1; + return(0); +} - 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); - goto bailout; - } - } - if ('\n' != blk->buf[i]) { - if (comment) - continue; - ln->buf[pos++] = blk->buf[i]; +static void +fdesc(struct curparse *curp) +{ + struct buf ln, blk; + int j, i, pos, lnn, comment, with_mmap; + struct man *man; + struct mdoc *mdoc; - /* Handle in-line `\"' comments. */ + man = NULL; + mdoc = NULL; + memset(&ln, 0, sizeof(struct buf)); - if (1 == pos || '\"' != ln->buf[pos - 1]) - continue; + /* + * 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. + */ - for (j = pos - 2; j >= 0; j--) - if ('\\' != ln->buf[j]) - break; + if (!read_whole_file(curp, &blk, &with_mmap)) + return; - if ( ! ((pos - 2 - j) % 2)) - continue; + /* Fill buf with file blocksize. */ - comment = 1; - pos -= 2; - for (; pos > 0; --pos) { - if (ln->buf[pos - 1] != ' ') - break; - if (pos > 2 && ln->buf[pos - 2] == '\\') - break; - } + for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) { + if (pos >= (int)ln.sz) { + if (! resize_buf(&ln, 256)) + goto bailout; + } + + if ('\n' != blk.buf[i]) { + if (comment) continue; - } + ln.buf[pos++] = blk.buf[i]; - /* Handle escaped `\\n' newlines. */ + /* Handle in-line `\"' comments. */ - 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; - } + if (1 == pos || '\"' != ln.buf[pos - 1]) + continue; + + for (j = pos - 2; j >= 0; j--) + if ('\\' != ln.buf[j]) + break; + + if ( ! ((pos - 2 - j) % 2)) + continue; + + comment = 1; + pos -= 2; + for (; pos > 0; --pos) { + if (ln.buf[pos - 1] != ' ') + break; + if (pos > 2 && ln.buf[pos - 2] == '\\') + break; } + continue; + } - ln->buf[pos] = 0; - lnn++; + /* Handle escaped `\\n' newlines. */ - /* If unset, assign parser in pset(). */ + 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; + } + } - if ( ! (man || mdoc) && ! pset(ln->buf, - pos, curp, &man, &mdoc)) - goto bailout; + ln.buf[pos] = 0; + lnn++; - pos = comment = 0; + /* If unset, assign parser in pset(). */ - /* Pass down into parsers. */ + if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc)) + goto bailout; - if (man && ! man_parseln(man, lnn, ln->buf)) - goto bailout; - if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) - goto bailout; - } + pos = comment = 0; + + /* Pass down into parsers. */ + + 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. */ @@ -460,6 +508,12 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse 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: @@ -496,14 +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; @@ -514,7 +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); } @@ -522,7 +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); }