===================================================================
RCS file: /cvs/mandoc/main.c,v
retrieving revision 1.54
retrieving revision 1.65
diff -u -p -r1.54 -r1.65
--- mandoc/main.c	2009/10/31 06:17:19	1.54
+++ mandoc/main.c	2010/05/15 04:46:10	1.65
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.54 2009/10/31 06:17:19 kristaps Exp $ */
+/*	$Id: main.c,v 1.65 2010/05/15 04:46:10 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -14,6 +14,10 @@
  * 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 <sys/stat.h>
 
 #include <assert.h>
@@ -30,19 +34,13 @@
 
 #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 +61,7 @@ enum	outt {
 	OUTT_ASCII = 0,
 	OUTT_TREE,
 	OUTT_HTML,
+	OUTT_XHTML,
 	OUTT_LINT
 };
 
@@ -73,16 +72,13 @@ 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. */
 	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;
@@ -91,30 +87,33 @@ struct	curparse {
 	char		  outopts[BUFSIZ];
 };
 
+#define	FL_STRICT	  FL_NIGN_ESCAPE | \
+			  FL_NIGN_MACRO
+
 static	int		  foptions(int *, char *);
-static	int		  toptions(enum outt *, char *);
+static	int		  toptions(struct curparse *, 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, const char *);
-static	int		  ffile(struct buf *, struct buf *, 
+static	void		  ffile(struct buf *, struct buf *, 
 				const char *, struct curparse *);
-static	int		  fdesc(struct buf *, struct buf *,
+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 *);
-__dead	static void	  version(void);
-__dead	static void	  usage(void);
+static	void		  version(void) __attribute__((noreturn));
+static	void		  usage(void) __attribute__((noreturn));
 
 static	const char	 *progname;
+static 	int		  with_error, with_warning;
 
-
 int
 main(int argc, char *argv[])
 {
-	int		 c, rc;
+	int		 c;
 	struct buf	 ln, blk;
 	struct curparse	 curp;
 
@@ -145,7 +144,7 @@ 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'):
@@ -166,35 +165,19 @@ main(int argc, char *argv[])
 	memset(&ln, 0, sizeof(struct buf));
 	memset(&blk, 0, sizeof(struct buf));
 
-	rc = 1;
-
 	if (NULL == *argv) {
 		curp.file = "<stdin>";
 		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(&blk, &ln, &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(&blk, &ln, *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)
@@ -203,16 +186,12 @@ main(int argc, char *argv[])
 		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 +200,13 @@ 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 +222,11 @@ 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));
@@ -267,43 +244,39 @@ 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
+static void
 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))) {
 		perror(curp->file);
-		return(-1);
+		with_error = 1;
+		return;
 	}
 
-	c = fdesc(blk, ln, curp);
+	fdesc(blk, ln, curp);
 
 	if (-1 == close(curp->fd))
 		perror(curp->file);
-
-	return(c);
 }
 
 
-static int
+static void
 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 {
 	size_t		 sz;
@@ -323,17 +296,23 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse
 	 * growable, hence passed in by ptr-ptr.
 	 */
 
-	if (-1 == fstat(curp->fd, &st))
+	if (-1 == fstat(curp->fd, &st)) {
 		perror(curp->file);
-	else if ((size_t)st.st_blksize > sz)
+		with_error = 1;
+		return;
+	}
+	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) {
+		void *buf = realloc(blk->buf, sz);
+
+		if (NULL == buf) {
 			perror(NULL);
-			exit(EXIT_FAILURE);
+			with_error = 1;
+			return;
 		}
+		blk->buf = buf;
 		blk->sz = sz;
 	}
 
@@ -342,7 +321,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse
 	for (lnn = pos = comment = 0; ; ) {
 		if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 			perror(curp->file);
-			return(-1);
+			goto bailout;
 		} else if (0 == ssz) 
 			break;
 
@@ -354,7 +333,7 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse
 				ln->buf = realloc(ln->buf, ln->sz);
 				if (NULL == ln->buf) {
 					perror(NULL);
-					return(EXIT_FAILURE);
+					goto bailout;
 				}
 			}
 
@@ -377,6 +356,12 @@ 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;
 			} 
 
@@ -401,16 +386,16 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse
 
 			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);
+				goto bailout;
 			if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
-				return(0);
+				goto bailout;
 		}
 	}
 
@@ -418,18 +403,24 @@ fdesc(struct buf *blk, struct buf *ln, struct curparse
 
 	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 +449,20 @@ 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;
+	}
+	return;
+
+ bailout:
+	with_error = 1;
+	goto cleanup;
 }
 
 
@@ -490,14 +494,12 @@ 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 +510,6 @@ 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 +517,6 @@ 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 +532,7 @@ 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 +541,24 @@ 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 +575,34 @@ 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 +631,7 @@ 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 +651,8 @@ 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 +670,12 @@ 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);
 }