===================================================================
RCS file: /cvs/mandoc/Attic/mmain.c,v
retrieving revision 1.4
retrieving revision 1.10
diff -u -p -r1.4 -r1.10
--- mandoc/Attic/mmain.c	2009/02/23 15:38:20	1.4
+++ mandoc/Attic/mmain.c	2009/03/09 13:04:01	1.10
@@ -1,4 +1,4 @@
-	/* $Id: mmain.c,v 1.4 2009/02/23 15:38:20 kristaps Exp $ */
+	/* $Id: mmain.c,v 1.10 2009/03/09 13:04:01 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -30,7 +30,7 @@
 
 #include "mmain.h"
 
-#define	MD_LINE_SZ	(256)		/* Max input line size. */
+#define	MD_LINE_SZ	(256)		/* Input line step-size. */
 
 struct	mmain {
 	int		  warn;		/* Warning flags. */
@@ -41,14 +41,16 @@ struct	mmain {
 	int		  dbg;		/* Debug level. */
 	struct mdoc	 *mdoc;		/* Active parser. */
 	char		 *buf;		/* Input buffer. */
-	u_long		  bufsz;	/* Input buffer size. */
+	size_t		  bufsz;	/* Input buffer size. */
 	char		 *in; 		/* Input file name. */
 	int		  fdin;		/* Input file desc. */
+	int		  pflags;	/* Parse flags. */
 };
 
 extern	char	 	 *__progname;
 
-static	int		  getsopts(struct mmain *, char *);
+static	int		  optswarn(struct mmain *, char *);
+static	int		  optsopt(struct mmain *, char *);
 static	int		  parse(struct mmain *);
 static	void		  msg_msg(void *, int, int, const char *);
 static	int		  msg_err(void *, int, int, const char *);
@@ -69,7 +71,7 @@ void
 mmain_usage(const char *help)
 {
 
-	warnx("usage: %s %s%s[-v] [-Wwarn...] [infile]", __progname, 
+	warnx("usage: %s %s%s[-v] [-foption...] [-Wwarn...] [infile]", __progname, 
 			help ? help : "", help ? " " : "");
 }
 
@@ -109,7 +111,7 @@ mmain_getopt(struct mmain *p, int argc, char *argv[], 
 
 	extern int	 optind;
 
-	sz = strlcpy(opts, "vW:", 32);
+	sz = strlcpy(opts, "VvW:f:", 32);
 	assert(sz < 32);
 
 	if (u) {
@@ -122,21 +124,28 @@ mmain_getopt(struct mmain *p, int argc, char *argv[], 
 	/* LINTED */
 	while (-1 != (c = getopt(argc, argv, opts)))
 		switch (c) {
+		case ('f'):
+			if ( ! optsopt(p, optarg))
+				return(-1);
+			break;
 		case ('v'):
 			p->dbg++;
 			break;
+		case ('V'):
+			(void)printf("%s %s\n", __progname, VERSION);
+			return(0);
 		case ('W'):
-			if ( ! getsopts(p, optarg))
-				return(0);
+			if ( ! optswarn(p, optarg))
+				return(-1);
 			break;
 		case ('?'):
 			mmain_usage(help);
-			return(0);
+			return(-1);
 		default:
 			assert(getopt_cb);
 			if ((*getopt_cb)(arg, c, optarg))
 				break;
-			return(0);
+			return(-1);
 		}
 
 	argv += optind;
@@ -181,7 +190,7 @@ mmain_mdoc(struct mmain *p)
 		warn("%s", p->in);
 		p->bufsz = BUFSIZ;
 	} else 
-		p->bufsz = MAX(st.st_blksize, BUFSIZ);
+		p->bufsz = (size_t)MAX(st.st_blksize, BUFSIZ);
 
 	p->buf = malloc(p->bufsz);
 	if (NULL == p->buf)
@@ -189,7 +198,7 @@ mmain_mdoc(struct mmain *p)
 
 	/* Allocate the parser. */
 
-	p->mdoc = mdoc_alloc(p, &cb);
+	p->mdoc = mdoc_alloc(p, p->pflags, &cb);
 
 	/* Parse the input file. */
 
@@ -205,9 +214,36 @@ mmain_mdoc(struct mmain *p)
 
 
 static int
-getsopts(struct mmain *p, char *arg)
+optsopt(struct mmain *p, char *arg)
 {
 	char		*v;
+	char		*toks[] = { "ign-scope", "ign-escape", 
+				    "ign-macro", NULL };
+
+	while (*arg) 
+		switch (getsubopt(&arg, toks, &v)) {
+		case (0):
+			p->pflags |= MDOC_IGN_SCOPE;
+			break;
+		case (1):
+			p->pflags |= MDOC_IGN_ESCAPE;
+			break;
+		case (2):
+			p->pflags |= MDOC_IGN_MACRO;
+			break;
+		default:
+			warnx("unknown -f argument");
+			return(0);
+		}
+
+	return(1);
+}
+
+
+static int
+optswarn(struct mmain *p, char *arg)
+{
+	char		*v;
 	char		*toks[] = { "all", "compat", 
 				"syntax", "error", NULL };
 
@@ -226,6 +262,7 @@ getsopts(struct mmain *p, char *arg)
 			p->warn |= MD_WARN_ERR;
 			break;
 		default:
+			warnx("unknown -W argument");
 			return(0);
 		}
 
@@ -236,47 +273,57 @@ getsopts(struct mmain *p, char *arg)
 static int
 parse(struct mmain *p)
 {
-	ssize_t		 sz, i;
-	size_t		 pos;
-	char		 line[MD_LINE_SZ];
-	int		 lnn;
+	ssize_t		 sz;
+	int		 j, i, pos, len, lnn;
+	char		*ln;
 
-	/*
-	 * This is a little more complicated than fgets.  TODO: have
-	 * some benchmarks that show it's faster (note that I want to
-	 * check many, many manuals simultaneously, so speed is
-	 * important).  Fill a buffer (sized to the block size) with a
-	 * single read, then parse \n-terminated lines into a line
-	 * buffer, which is passed to the parser.  Hard-code the line
-	 * buffer to a particular size -- a reasonable assumption.
-	 */
-
-	for (lnn = 1, pos = 0; ; ) {
+	for (ln = NULL, lnn = 1, len = pos = 0; ; ) {
 		if (-1 == (sz = read(p->fdin, p->buf, p->bufsz))) {
 			warn("%s", p->in);
 			return(0);
 		} else if (0 == sz) 
 			break;
 
-		for (i = 0; i < sz; i++) {
+		for (i = 0; i < (int)sz; i++) {
+			if (pos >= len) {
+				len += MD_LINE_SZ;
+				ln = realloc(ln, (size_t)len);
+				if (NULL == ln)
+					err(1, "realloc");
+			}
+
 			if ('\n' != p->buf[i]) {
-				if (pos < sizeof(line)) {
-					line[(int)pos++] = p->buf[(int)i];
+				ln[pos++] = p->buf[i];
+				continue;
+			}
+
+			/* Check for escaped newline.  */
+
+			if (pos > 0 && '\\' == ln[pos - 1]) {
+				for (j = pos - 1; j >= 0; j--)
+					if ('\\' != ln[j])
+						break;
+
+				if ( ! ((pos - j) % 2)) {
+					pos--;
+					lnn++;
 					continue;
 				}
-				warnx("%s: line %d too long", p->in, lnn);
-				return(0);
 			}
-	
-			line[(int)pos] = 0;
-			if ( ! mdoc_parseln(p->mdoc, lnn, line))
-				return(0);
 
+			ln[pos] = 0;
+			if ( ! mdoc_parseln(p->mdoc, lnn, ln))
+				return(0);
 			lnn++;
 			pos = 0;
 		}
 	}
 
+	if (pos > 0)
+		warnx("%s: file not eof-terminated", p->in);
+
+	if (ln)
+		free(ln);
 	return(mdoc_endparse(p->mdoc));
 }