=================================================================== RCS file: /cvs/mandoc/main.c,v retrieving revision 1.11 retrieving revision 1.145 diff -u -p -r1.11 -r1.145 --- mandoc/main.c 2009/03/23 15:20:51 1.11 +++ mandoc/main.c 2011/02/09 09:18:15 1.145 @@ -1,144 +1,284 @@ -/* $Id: main.c,v 1.11 2009/03/23 15:20:51 kristaps Exp $ */ +/* $Id: main.c,v 1.145 2011/02/09 09:18:15 kristaps Exp $ */ /* - * Copyright (c) 2008, 2009 Kristaps Dzonsons + * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons + * Copyright (c) 2010, 2011 Ingo Schwarze * * Permission to use, copy, modify, and distribute this software for any - * purpose with or without fee is hereby granted, provided that the - * above copyright notice and this permission notice appear in all - * copies. + * purpose with or without fee is hereby granted, provided that the above + * copyright notice and this permission notice appear in all copies. * - * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL - * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED - * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE - * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL - * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR - * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER - * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR - * PERFORMANCE OF THIS SOFTWARE. + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN + * 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 #include #include +#include "mandoc.h" +#include "main.h" #include "mdoc.h" #include "man.h" +#include "roff.h" -#ifdef __linux__ -extern int getsubopt(char **, char * const *, char **); -# ifndef __dead -# define __dead __attribute__((__noreturn__)) -# endif -#elif defined(__FreeBSD__) -# ifndef __dead -# define __dead __dead2 -# endif +#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) */ + +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; }; -struct curparse { - const char *file; - 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. */ -}; - enum intt { - INTT_MDOC = 0, + INTT_AUTO, + INTT_MDOC, INTT_MAN }; enum outt { OUTT_ASCII = 0, - OUTT_LATIN1, - OUTT_UTF8, OUTT_TREE, - OUTT_LINT + OUTT_HTML, + OUTT_XHTML, + OUTT_LINT, + OUTT_PS, + OUTT_PDF }; -typedef int (*out_run)(void *, const struct man *, - const struct mdoc *); -typedef void (*out_free)(void *); +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 */ + out_mdoc outmdoc; /* mdoc 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 */ +}; -extern char *__progname; +static const char * const mandoclevels[MANDOCLEVEL_MAX] = { + "SUCCESS", + "RESERVED", + "WARNING", + "ERROR", + "FATAL", + "BADARG", + "SYSERR" +}; -extern void *ascii_alloc(void); -extern void *latin1_alloc(void); -extern void *utf8_alloc(void); -extern int terminal_run(void *, const struct man *, - const struct mdoc *); -extern int tree_run(void *, const struct man *, - const struct mdoc *); -extern void terminal_free(void *); +static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = { + MANDOCERR_OK, + MANDOCERR_WARNING, + MANDOCERR_WARNING, + MANDOCERR_ERROR, + MANDOCERR_FATAL, + MANDOCERR_MAX, + MANDOCERR_MAX +}; -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, - enum mdoc_warn, const char *); -static int file(struct buf *, struct buf *, - const char *, - struct man *, struct mdoc *); -static int fdesc(struct buf *, struct buf *, - const char *, int, - struct man *, struct mdoc *); +static const char * const mandocerrs[MANDOCERR_MAX] = { + "ok", -__dead static void version(void); -__dead static void usage(void); + "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", + "skipping no-space 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", + "skipping end of block that is not open", + + /* 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", + + "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", + "data block still open", + "ignoring extra data cells", + + "input stack limit exceeded, infinite loop?", + "skipping bad character", + "escaped character not allowed in a name", + "skipping text before the first section header", + "skipping unknown macro", + "NOT IMPLEMENTED, please use groff: 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 toptions(struct curparse *, char *); +static void usage(void) __attribute__((noreturn)); +static void version(void) __attribute__((noreturn)); +static int woptions(struct curparse *, char *); + +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, rc, fflags; - struct mdoc_cb cb; - struct man *man; - struct mdoc *mdoc; - void *outdata; - enum outt outtype; - enum intt inttype; - struct buf ln, blk; - out_run outrun; - out_free outfree; + int c; struct curparse curp; - fflags = 0; - outtype = OUTT_ASCII; - inttype = INTT_MDOC; + progname = strrchr(argv[0], '/'); + if (progname == NULL) + progname = argv[0]; + else + ++progname; - bzero(&curp, sizeof(struct curparse)); + memset(&curp, 0, sizeof(struct curparse)); + curp.inttype = INTT_AUTO; + curp.outtype = OUTT_ASCII; + curp.wlevel = MANDOCLEVEL_FATAL; + /* LINTED */ - while (-1 != (c = getopt(argc, argv, "f:m:VW:T:"))) + while (-1 != (c = getopt(argc, argv, "m:O:T:VW:"))) switch (c) { - case ('f'): - if ( ! foptions(&fflags, optarg)) - return(0); - break; case ('m'): - if ( ! moptions(&inttype, optarg)) - return(0); + if ( ! moptions(&curp.inttype, optarg)) + return((int)MANDOCLEVEL_BADARG); break; + case ('O'): + (void)strlcat(curp.outopts, optarg, BUFSIZ); + (void)strlcat(curp.outopts, ",", BUFSIZ); + break; case ('T'): - if ( ! toptions(&outtype, optarg)) - return(0); + if ( ! toptions(&curp, optarg)) + return((int)MANDOCLEVEL_BADARG); break; case ('W'): - if ( ! woptions(&curp.wflags, optarg)) - return(0); + if ( ! woptions(&curp, optarg)) + return((int)MANDOCLEVEL_BADARG); break; case ('V'): version(); @@ -151,400 +291,778 @@ 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. - */ + if (NULL == *argv) { + curp.file = ""; + curp.fd = STDIN_FILENO; - 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; + fdesc(&curp); } - /* - * All callbacks route into here, where we print them onto the - * screen. XXX - for now, no path for debugging messages. - */ + while (*argv) { + ffile(*argv, &curp); + if (MANDOCLEVEL_OK != exit_status && curp.wstop) + break; + ++argv; + } - cb.mdoc_msg = NULL; - cb.mdoc_err = merr; - cb.mdoc_warn = mwarn; + 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); - bzero(&ln, sizeof(struct buf)); - bzero(&blk, sizeof(struct buf)); + return((int)exit_status); +} - man = NULL; - mdoc = NULL; - switch (inttype) { - case (INTT_MAN): - man = man_alloc(); - break; - default: - mdoc = mdoc_alloc(&curp, fflags, &cb); - break; - } +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) +{ + /* - * Loop around available files. + * 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(). */ - if (NULL == *argv) { - curp.file = ""; - rc = 0; - c = fdesc(&blk, &ln, "stdin", - STDIN_FILENO, man, mdoc); + curp->file = file; - if (c && NULL == outrun) - rc = 1; - else if (c && outrun && (*outrun)(outdata, man, mdoc)) - rc = 1; - } else { - while (*argv) { - curp.file = *argv; - c = file(&blk, &ln, *argv, man, mdoc); - if ( ! c) - break; - if (outrun && ! (*outrun)(outdata, man, mdoc)) - break; - if (man) - man_reset(man); - if (mdoc) - mdoc_reset(mdoc); - - argv++; - } - rc = NULL == *argv; + if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) { + perror(curp->file); + exit_status = MANDOCLEVEL_SYSERR; + return; } - if (blk.buf) - free(blk.buf); - if (ln.buf) - free(ln.buf); - if (outfree) - (*outfree)(outdata); - if (mdoc) - mdoc_free(mdoc); - if (man) - man_free(man); + fdesc(curp); - return(rc ? EXIT_SUCCESS : EXIT_FAILURE); + if (-1 == close(curp->fd)) + perror(curp->file); } - -__dead static void -version(void) +static int +pfile(const char *file, struct curparse *curp) { + const char *savefile; + int fd, savefd; - (void)printf("%s %s\n", __progname, VERSION); - exit(0); - /* NOTREACHED */ + if (-1 == (fd = open(file, O_RDONLY, 0))) { + perror(file); + file_status = MANDOCLEVEL_SYSERR; + return(0); + } + + savefile = curp->file; + savefd = curp->fd; + + curp->file = file; + curp->fd = fd; + + pdesc(curp); + + curp->file = savefile; + curp->fd = savefd; + + if (-1 == close(fd)) + perror(file); + + return(MANDOCLEVEL_FATAL > file_status ? 1 : 0); } -__dead static void -usage(void) +static void +resize_buf(struct buf *buf, size_t initial) { - (void)fprintf(stderr, "usage: %s\n", __progname); - exit(1); - /* NOTREACHED */ + 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); + } } static int -file(struct buf *blk, struct buf *ln, const char *file, - struct man *man, struct mdoc *mdoc) +read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap) { - int fd, c; + struct stat st; + size_t off; + ssize_t ssz; - if (-1 == (fd = open(file, O_RDONLY, 0))) { - warn("%s", file); + if (-1 == fstat(curp->fd, &st)) { + perror(curp->file); return(0); } - c = fdesc(blk, ln, file, fd, man, mdoc); + /* + * 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 (-1 == close(fd)) - warn("%s", file); + if (S_ISREG(st.st_mode)) { + if (st.st_size >= (1U << 31)) { + fprintf(stderr, "%s: input too large\n", + curp->file); + return(0); + } + *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); + } - return(c); + /* + * 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; + } + 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); + break; + } + off += (size_t)ssz; + } + + free(fb->buf); + fb->buf = NULL; + return(0); } -static int -fdesc(struct buf *blk, struct buf *ln, - const char *f, int fd, - struct man *man, struct mdoc *mdoc) +static void +fdesc(struct curparse *curp) { - size_t sz; - ssize_t ssz; - struct stat st; - int j, i, pos, lnn; - assert( ! (man && mdoc)); + /* + * 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. + */ + /* Zero the parse type. */ + + 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); + } + + /* Fully parse the file. */ + + pdesc(curp); + + if (MANDOCLEVEL_FATAL <= file_status) + goto cleanup; + + /* NOTE a parser may not have been assigned, yet. */ + + if ( ! (curp->man || curp->mdoc)) { + fprintf(stderr, "%s: Not a manual\n", curp->file); + file_status = MANDOCLEVEL_FATAL; + goto cleanup; + } + + /* Clean up the parse routine ASTs. */ + + if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) { + assert(MANDOCLEVEL_FATAL <= file_status); + goto cleanup; + } + + if (curp->man && ! man_endparse(curp->man)) { + assert(MANDOCLEVEL_FATAL <= file_status); + goto cleanup; + } + + assert(curp->roff); + roff_endparse(curp->roff); + /* - * Two buffers: ln and buf. buf is the input buffer, optimised - * for each file's block size. ln is a line buffer. Both - * growable, hence passed in by ptr-ptr. + * With -Wstop and warnings or errors of at least + * the requested level, do not produce output. */ - sz = BUFSIZ; + if (MANDOCLEVEL_OK != file_status && curp->wstop) + goto cleanup; - if (-1 == fstat(fd, &st)) - warnx("%s", f); - else if ((size_t)st.st_blksize > sz) - sz = st.st_blksize; + /* If unset, allocate output dev now (if applicable). */ - if (sz > blk->sz) { - blk->buf = realloc(blk->buf, sz); - if (NULL == blk->buf) - err(1, "realloc"); - blk->sz = sz; + if ( ! (curp->outman && curp->outmdoc)) { + switch (curp->outtype) { + case (OUTT_XHTML): + curp->outdata = xhtml_alloc(curp->outopts); + break; + case (OUTT_HTML): + curp->outdata = html_alloc(curp->outopts); + break; + case (OUTT_ASCII): + curp->outdata = ascii_alloc(curp->outopts); + curp->outfree = ascii_free; + break; + case (OUTT_PDF): + curp->outdata = pdf_alloc(curp->outopts); + curp->outfree = pspdf_free; + break; + case (OUTT_PS): + curp->outdata = ps_alloc(curp->outopts); + curp->outfree = pspdf_free; + break; + default: + break; + } + + switch (curp->outtype) { + case (OUTT_HTML): + /* FALLTHROUGH */ + case (OUTT_XHTML): + curp->outman = html_man; + curp->outmdoc = html_mdoc; + curp->outfree = html_free; + break; + case (OUTT_TREE): + curp->outman = tree_man; + curp->outmdoc = tree_mdoc; + break; + case (OUTT_PDF): + /* FALLTHROUGH */ + case (OUTT_ASCII): + /* FALLTHROUGH */ + case (OUTT_PS): + curp->outman = terminal_man; + curp->outmdoc = terminal_mdoc; + break; + default: + break; + } } + /* 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); + + cleanup: + + memset(&curp->regs, 0, sizeof(struct regset)); + + /* 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; +} + +static void +pdesc(struct curparse *curp) +{ + struct buf blk; + int with_mmap; + /* - * Fill buf with file blocksize and parse newlines into ln. + * 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. */ - for (lnn = 1, pos = 0; ; ) { - if (-1 == (ssz = read(fd, blk->buf, sz))) { - warn("%s", f); - return(0); - } else if (0 == ssz) + if ( ! read_whole_file(curp, &blk, &with_mmap)) { + file_status = MANDOCLEVEL_SYSERR; + return; + } + + /* 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) +{ + const struct tbl_span *span; + 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; - 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 (start) { + curp->line = lnn; + curp->reparse_count = 0; + } + + while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) { + + /* + * When finding an unescaped newline character, + * leave the character loop to process the line. + * Skip a preceding carriage return, if any. + */ + + if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz && + '\n' == blk.buf[i + 1]) + ++i; + if ('\n' == blk.buf[i]) { + ++i; + ++lnn; + break; } - if ('\n' != blk->buf[i]) { - ln->buf[pos++] = blk->buf[i]; + /* + * 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; } - /* Check for CPP-escaped newline. */ + /* Trailing backslash = a plain char. */ - if (pos > 0 && '\\' == ln->buf[pos - 1]) { - for (j = pos - 1; j >= 0; j--) - if ('\\' != ln->buf[j]) + 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 and at least one other character. + * When it's a newline character, skip it. + * When there is a carriage return in between, + * skip that one as well. + */ + + if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz && + '\n' == blk.buf[i + 2]) + ++i; + if ('\n' == blk.buf[i + 1]) { + i += 2; + ++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; + } + } - if ( ! ((pos - j) % 2)) { - pos--; - lnn++; - continue; + /* Backout trailing whitespaces */ + for (; pos > 0; --pos) { + if (ln.buf[pos - 1] != ' ') + break; + if (pos > 2 && ln.buf[pos - 2] == '\\') + break; } + break; } - ln->buf[pos] = 0; - if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf)) - return(0); - if (man && ! man_parseln(man, lnn, ln->buf)) - return(0); - lnn++; + /* 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; + continue; + } else + break; + default: + break; } - } - if (mdoc) - return(mdoc_endparse(mdoc)); + /* + * If we encounter errors in the recursive parsebuf() + * call, make sure we don't continue parsing. + */ - return(man_endparse(man)); + 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); + while (NULL != (span = roff_span(curp->roff))) { + if (curp->man) + man_addspan(curp->man, span); + else + mdoc_addspan(curp->mdoc, span); + } + } else if (ROFF_EQN == rr) { + assert(curp->man || curp->mdoc); + assert(roff_eqn(curp->roff)); + rc = curp->mdoc ? + mdoc_addeqn(curp->mdoc, + roff_eqn(curp->roff)) : + man_addeqn(curp->man, + roff_eqn(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; + } + } + + /* Temporary buffers typically are not full. */ + + if (0 == start && '\0' == blk.buf[i]) + break; + + /* Start the next input line. */ + + pos = 0; + } + + free(ln.buf); } +static void +pset(const char *buf, int pos, struct curparse *curp) +{ + 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; + } + + 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; +} + static int moptions(enum intt *tflags, char *arg) { - if (0 == strcmp(arg, "mdoc")) + if (0 == strcmp(arg, "doc")) *tflags = INTT_MDOC; - else if (0 == strcmp(arg, "man")) + 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); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } return(1); } - 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, "latin1")) - *tflags = OUTT_LATIN1; - else if (0 == strcmp(arg, "utf8")) - *tflags = OUTT_UTF8; - else if (0 == strcmp(arg, "lint")) - *tflags = OUTT_LINT; + curp->outtype = OUTT_ASCII; + else if (0 == strcmp(arg, "lint")) { + curp->outtype = OUTT_LINT; + curp->wlevel = MANDOCLEVEL_WARNING; + } else if (0 == strcmp(arg, "tree")) - *tflags = OUTT_TREE; + curp->outtype = OUTT_TREE; + else if (0 == strcmp(arg, "html")) + curp->outtype = OUTT_HTML; + else if (0 == strcmp(arg, "xhtml")) + curp->outtype = OUTT_XHTML; + else if (0 == strcmp(arg, "ps")) + curp->outtype = OUTT_PS; + else if (0 == strcmp(arg, "pdf")) + curp->outtype = OUTT_PDF; else { - warnx("bad argument: -T%s", arg); + fprintf(stderr, "%s: Bad argument\n", arg); return(0); } return(1); } - -/* - * Parse out the options for [-fopt...] setting compiler options. These - * can be comma-delimited or called again. - */ static int -foptions(int *fflags, char *arg) +woptions(struct curparse *curp, char *arg) { - char *v; - char *toks[4]; + char *v, *o; + const char *toks[6]; - toks[0] = "ign-scope"; - toks[1] = "ign-escape"; - toks[2] = "ign-macro"; - toks[3] = NULL; - - while (*arg) - switch (getsubopt(&arg, toks, &v)) { - case (0): - *fflags |= MDOC_IGN_SCOPE; - break; - case (1): - *fflags |= MDOC_IGN_ESCAPE; - break; - case (2): - *fflags |= MDOC_IGN_MACRO; - break; - default: - warnx("bad argument: -f%s", arg); - return(0); - } - - return(1); -} - - -/* - * Parse out the options for [-Werr...], which sets warning modes. - * These can be comma-delimited or called again. - */ -static int -woptions(int *wflags, char *arg) -{ - char *v; - char *toks[5]; - - toks[0] = "all"; - toks[1] = "compat"; - toks[2] = "syntax"; + toks[0] = "stop"; + toks[1] = "all"; + toks[2] = "warning"; toks[3] = "error"; - toks[4] = NULL; + toks[4] = "fatal"; + toks[5] = NULL; - while (*arg) - switch (getsubopt(&arg, toks, &v)) { + while (*arg) { + o = arg; + switch (getsubopt(&arg, UNCONST(toks), &v)) { case (0): - *wflags |= WARN_WALL; + curp->wstop = 1; break; case (1): - *wflags |= WARN_WCOMPAT; - break; + /* FALLTHROUGH */ case (2): - *wflags |= WARN_WSYNTAX; + curp->wlevel = MANDOCLEVEL_WARNING; break; case (3): - *wflags |= WARN_WERR; + curp->wlevel = MANDOCLEVEL_ERROR; break; + case (4): + curp->wlevel = MANDOCLEVEL_FATAL; + break; default: - warnx("bad argument: -W%s", arg); + fprintf(stderr, "-W%s: Bad argument\n", o); return(0); } + } return(1); } - -/* ARGSUSED */ static int -merr(void *arg, int line, int col, const char *msg) +mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg) { - struct curparse *curp; + struct curparse *cp; + enum mandoclevel level; - curp = (struct curparse *)arg; + level = MANDOCLEVEL_FATAL; + while (t < mandoclimits[level]) + /* LINTED */ + level--; - warnx("%s:%d: error: %s (column %d)", - curp->file, line, msg, col); - return(0); -} - - -static int -mwarn(void *arg, int line, int col, - enum mdoc_warn type, const char *msg) -{ - struct curparse *curp; - char *wtype; - - curp = (struct curparse *)arg; - wtype = NULL; - - switch (type) { - case (WARN_COMPAT): - wtype = "compat"; - if (curp->wflags & WARN_WCOMPAT) - break; + cp = (struct curparse *)arg; + if (level < cp->wlevel) return(1); - case (WARN_SYNTAX): - wtype = "syntax"; - if (curp->wflags & WARN_WSYNTAX) - break; - return(1); - } - assert(wtype); - warnx("%s:%d: %s warning: %s (column %d)", - curp->file, line, wtype, msg, col); + fprintf(stderr, "%s:%d:%d: %s: %s", + cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]); + if (msg) + fprintf(stderr, ": %s", msg); + fputc('\n', stderr); - if ( ! (curp->wflags & WARN_WERR)) - return(1); - - warnx("%s: considering warnings as errors", - __progname); - return(0); + if (file_status < level) + file_status = level; + + return(level < MANDOCLEVEL_FATAL); } - -