===================================================================
RCS file: /cvs/mandoc/roff.c,v
retrieving revision 1.144
retrieving revision 1.154
diff -u -p -r1.144 -r1.154
--- mandoc/roff.c	2011/07/08 09:28:33	1.144
+++ mandoc/roff.c	2011/07/27 07:09:41	1.154
@@ -1,4 +1,4 @@
-/*	$Id: roff.c,v 1.144 2011/07/08 09:28:33 kristaps Exp $ */
+/*	$Id: roff.c,v 1.154 2011/07/27 07:09:41 kristaps Exp $ */
 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -71,6 +71,16 @@ enum	roffrule {
 	ROFFRULE_DENY
 };
 
+/*
+ * A single register entity.  If "set" is zero, the value of the
+ * register should be the default one, which is per-register.
+ * Registers are assumed to be unsigned ints for now.
+ */
+struct	reg {
+	int		  set; /* whether set or not */
+	unsigned int	  u; /* unsigned integer */
+};
+
 struct	roffstr {
 	char		*name; /* key of symbol */
 	char		*string; /* current value */
@@ -82,7 +92,7 @@ struct	roff {
 	struct roffnode	*last; /* leaf of stack */
 	enum roffrule	 rstack[RSTACK_MAX]; /* stack of !`ie' rules */
 	int		 rstackpos; /* position in rstack */
-	struct regset	*regs; /* read/writable registers */
+	struct reg	 regs[REG__MAX];
 	struct roffstr	*first_string; /* user-defined strings & macros */
 	const char	*current_string; /* value of last called user macro */
 	struct tbl_node	*first_tbl; /* first table parsed */
@@ -149,7 +159,7 @@ static	const char	*roff_getstrn(const struct roff *, 
 				const char *, size_t);
 static	enum rofferr	 roff_line_ignore(ROFF_ARGS);
 static	enum rofferr	 roff_nr(ROFF_ARGS);
-static	int		 roff_res(struct roff *, 
+static	void		 roff_res(struct roff *, 
 				char **, size_t *, int, int);
 static	enum rofferr	 roff_rm(ROFF_ARGS);
 static	void		 roff_setstr(struct roff *,
@@ -351,6 +361,8 @@ roff_reset(struct roff *r)
 
 	roff_free1(r);
 
+	memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
+
 	for (i = 0; i < PREDEFS_MAX; i++) 
 		roff_setstr(r, predefs[i].name, predefs[i].str, 0);
 }
@@ -366,13 +378,12 @@ roff_free(struct roff *r)
 
 
 struct roff *
-roff_alloc(struct regset *regs, struct mparse *parse)
+roff_alloc(struct mparse *parse)
 {
 	struct roff	*r;
 	int		 i;
 
 	r = mandoc_calloc(1, sizeof(struct roff));
-	r->regs = regs;
 	r->parse = parse;
 	r->rstackpos = -1;
 	
@@ -384,15 +395,16 @@ roff_alloc(struct regset *regs, struct mparse *parse)
 	return(r);
 }
 
-
 /*
  * Pre-filter each and every line for reserved words (one beginning with
  * `\*', e.g., `\*(ab').  These must be handled before the actual line
  * is processed. 
+ * This also checks the syntax of regular escapes.
  */
-static int
+static void
 roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
 {
+	enum mandoc_esc	 esc;
 	const char	*stesc;	/* start of an escape sequence ('\\') */
 	const char	*stnam;	/* start of the name, after "[(*" */
 	const char	*cp;	/* end of the name, e.g. before ']' */
@@ -401,8 +413,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 	size_t		 nsz;
 	char		*n;
 
-	/* Search for a leading backslash and save a pointer to it. */
-
+again:
 	cp = *bufp + pos;
 	while (NULL != (cp = strchr(cp, '\\'))) {
 		stesc = cp++;
@@ -414,10 +425,22 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 		 */
 
 		if ('\0' == *cp)
-			return(1);
-		if ('*' != *cp++)
-			continue;
+			return;
 
+		if ('*' != *cp) {
+			res = cp;
+			esc = mandoc_escape(&cp, NULL, NULL);
+			if (ESCAPE_ERROR != esc)
+				continue;
+			cp = res;
+			mandoc_msg
+				(MANDOCERR_BADESCAPE, r->parse, 
+				 ln, (int)(stesc - *bufp), NULL);
+			return;
+		}
+
+		cp++;
+
 		/*
 		 * The third character decides the length
 		 * of the name of the string.
@@ -426,7 +449,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 
 		switch (*cp) {
 		case ('\0'):
-			return(1);
+			return;
 		case ('('):
 			cp++;
 			maxl = 2;
@@ -444,8 +467,13 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 		/* Advance to the end of the name. */
 
 		for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
-			if ('\0' == *cp)
-				return(1); /* Error. */
+			if ('\0' == *cp) {
+				mandoc_msg
+					(MANDOCERR_BADESCAPE, 
+					 r->parse, ln, 
+					 (int)(stesc - *bufp), NULL);
+				return;
+			}
 			if (0 == maxl && ']' == *cp)
 				break;
 		}
@@ -458,13 +486,16 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 		res = roff_getstrn(r, stnam, (size_t)i);
 
 		if (NULL == res) {
-			/* TODO: keep track of the correct position. */
-			mandoc_msg(MANDOCERR_BADESCAPE, r->parse, ln, pos, NULL);
+			mandoc_msg
+				(MANDOCERR_BADESCAPE, r->parse, 
+				 ln, (int)(stesc - *bufp), NULL);
 			res = "";
 		}
 
 		/* Replace the escape sequence by the string. */
 
+		pos += (stesc - *bufp);
+
 		nsz = *szp + strlen(res) + 1;
 		n = mandoc_malloc(nsz);
 
@@ -476,13 +507,43 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int
 
 		*bufp = n;
 		*szp = nsz;
-		return(0);
+		goto again;
 	}
-
-	return(1);
 }
 
+/*
+ * Process text streams: convert all breakable hyphens into ASCII_HYPH.
+ */
+static enum rofferr
+roff_parsetext(char *p)
+{
+	size_t		 sz;
+	const char	*start;
+	enum mandoc_esc	 esc;
 
+	start = p;
+
+	while ('\0' != *p) {
+		sz = strcspn(p, "-\\");
+		p += sz;
+
+		if ('\\' == *p) {
+			/* Skip over escapes. */
+			p++;
+			esc = mandoc_escape
+				((const char **)&p, NULL, NULL);
+			if (ESCAPE_ERROR == esc)
+				break;
+		} else if ('-' == *p) {
+			if (mandoc_hyph(start, p))
+				*p = ASCII_HYPH;
+			p++;
+		}
+	}
+
+	return(ROFF_CONT);
+}
+
 enum rofferr
 roff_parseln(struct roff *r, int ln, char **bufp, 
 		size_t *szp, int pos, int *offs)
@@ -496,8 +557,7 @@ roff_parseln(struct roff *r, int ln, char **bufp, 
 	 * words to fill in.
 	 */
 
-	if (r->first_string && ! roff_res(r, bufp, szp, ln, pos))
-		return(ROFF_REPARSE);
+	roff_res(r, bufp, szp, ln, pos);
 
 	ppos = pos;
 	ctl = mandoc_getcontrol(*bufp, &pos);
@@ -519,18 +579,18 @@ roff_parseln(struct roff *r, int ln, char **bufp, 
 		if (ROFF_CONT != e)
 			return(e);
 		if (r->eqn)
-			return(eqn_read(&r->eqn, ln, *bufp, pos));
+			return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
 		if (r->tbl)
 			return(tbl_read(r->tbl, ln, *bufp, pos));
-		return(ROFF_CONT);
+		return(roff_parsetext(*bufp + pos));
 	} else if ( ! ctl) {
 		if (r->eqn)
-			return(eqn_read(&r->eqn, ln, *bufp, pos));
+			return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
 		if (r->tbl)
 			return(tbl_read(r->tbl, ln, *bufp, pos));
-		return(ROFF_CONT);
+		return(roff_parsetext(*bufp + pos));
 	} else if (r->eqn)
-		return(eqn_read(&r->eqn, ln, *bufp, ppos));
+		return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
 
 	/*
 	 * If a scope is open, go to the child handler for that macro,
@@ -572,16 +632,14 @@ roff_endparse(struct roff *r)
 
 	if (r->eqn) {
 		mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, 
-				r->eqn->eqn.line, r->eqn->eqn.pos, NULL);
-		eqn_end(r->eqn);
-		r->eqn = NULL;
+				r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
+		eqn_end(&r->eqn);
 	}
 
 	if (r->tbl) {
 		mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, 
 				r->tbl->line, r->tbl->pos, NULL);
-		tbl_end(r->tbl);
-		r->tbl = NULL;
+		tbl_end(&r->tbl);
 	}
 }
 
@@ -1104,7 +1162,27 @@ roff_ds(ROFF_ARGS)
 	return(ROFF_IGN);
 }
 
+int
+roff_regisset(const struct roff *r, enum regs reg)
+{
 
+	return(r->regs[(int)reg].set);
+}
+
+unsigned int
+roff_regget(const struct roff *r, enum regs reg)
+{
+
+	return(r->regs[(int)reg].u);
+}
+
+void
+roff_regunset(struct roff *r, enum regs reg)
+{
+
+	r->regs[(int)reg].set = 0;
+}
+
 /* ARGSUSED */
 static enum rofferr
 roff_nr(ROFF_ARGS)
@@ -1112,18 +1190,16 @@ roff_nr(ROFF_ARGS)
 	const char	*key;
 	char		*val;
 	int		 iv;
-	struct reg	*rg;
 
 	val = *bufp + pos;
 	key = roff_getname(r, &val, ln, pos);
-	rg = r->regs->regs;
 
 	if (0 == strcmp(key, "nS")) {
-		rg[(int)REG_nS].set = 1;
-		if ((iv = mandoc_strntou(val, strlen(val), 10)) >= 0)
-			rg[REG_nS].v.u = (unsigned)iv;
+		r->regs[(int)REG_nS].set = 1;
+		if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
+			r->regs[(int)REG_nS].u = (unsigned)iv;
 		else
-			rg[(int)REG_nS].v.u = 0u;
+			r->regs[(int)REG_nS].u = 0u;
 	}
 
 	return(ROFF_IGN);
@@ -1153,9 +1229,8 @@ roff_TE(ROFF_ARGS)
 	if (NULL == r->tbl)
 		mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
 	else
-		tbl_end(r->tbl);
+		tbl_end(&r->tbl);
 
-	r->tbl = NULL;
 	return(ROFF_IGN);
 }
 
@@ -1172,14 +1247,22 @@ roff_T_(ROFF_ARGS)
 	return(ROFF_IGN);
 }
 
-/* ARGSUSED */
-static enum rofferr
-roff_EQ(ROFF_ARGS)
+int
+roff_closeeqn(struct roff *r)
 {
-	struct eqn_node	*e;
 
+	return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
+}
+
+void
+roff_openeqn(struct roff *r, const char *name, int line, 
+		int offs, const char *buf)
+{
+	struct eqn_node *e;
+	int		 poff;
+
 	assert(NULL == r->eqn);
-	e = eqn_alloc(ppos, ln);
+	e = eqn_alloc(name, offs, line, r->parse);
 
 	if (r->last_eqn)
 		r->last_eqn->next = e;
@@ -1187,6 +1270,19 @@ roff_EQ(ROFF_ARGS)
 		r->first_eqn = r->last_eqn = e;
 
 	r->eqn = r->last_eqn = e;
+
+	if (buf) {
+		poff = 0;
+		eqn_read(&r->eqn, line, buf, offs, &poff);
+	}
+}
+
+/* ARGSUSED */
+static enum rofferr
+roff_EQ(ROFF_ARGS)
+{
+
+	roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
 	return(ROFF_IGN);
 }
 
@@ -1207,7 +1303,7 @@ roff_TS(ROFF_ARGS)
 
 	if (r->tbl) {
 		mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
-		tbl_end(r->tbl);
+		tbl_end(&r->tbl);
 	}
 
 	t = tbl_alloc(ppos, ln, r->parse);
@@ -1442,4 +1538,11 @@ roff_eqn(const struct roff *r)
 {
 	
 	return(r->last_eqn ? &r->last_eqn->eqn : NULL);
+}
+
+char
+roff_eqndelim(const struct roff *r)
+{
+
+	return('\0');
 }