Use the freshly improved roff_getname() function

for the main roff request parsing routine, roff_parse().

In request or macro invocations, escape sequences now terminate the
request or macro name; what follows is treated as arguments.  Besides,
the names of user-defined macros can now contain backslashes (eek!).

4 files changed, 83 insertions, 19 deletions

diff --git a/regress/usr.bin/mandoc/roff/de/Makefile b/regress/usr.bin/mandoc/roff/de/Makefile
index 72c0788535f..390c004849c 100644
--- a/regress/usr.bin/mandoc/roff/de/Makefile
+++ b/regress/usr.bin/mandoc/roff/de/Makefile
@@ -1,6 +1,6 @@
-# $OpenBSD: Makefile,v 1.2 2014/03/30 19:43:19 schwarze Exp $
+# $OpenBSD: Makefile,v 1.3 2014/06/29 23:23:16 schwarze Exp $
 
-REGRESS_TARGETS=TH Dd
+REGRESS_TARGETS = escname TH Dd
 
 .include <bsd.regress.mk>
 
diff --git a/regress/usr.bin/mandoc/roff/de/escname.in b/regress/usr.bin/mandoc/roff/de/escname.in
new file mode 100644
index 00000000000..db5b0506254
--- /dev/null
+++ b/regress/usr.bin/mandoc/roff/de/escname.in
@@ -0,0 +1,42 @@
+.Dd June 29, 2014
+.Dt DE-ESCNAME 1
+.Os OpenBSD
+.Sh NAME
+.Nm de-escname
+.Nd escape sequences in macro names
+.Sh DESCRIPTION
+initial text
+.Pp
+define second = val2
+.de second
+val2
+..
+.Pp
+define first\esecond = val3
+.de first\\second end3
+val3
+.end3
+.Pp
+define first = val1
+.de first\esecond
+val1
+..
+.Pp
+Values (first, second, first\esecond):
+.first
+.second
+.first\\second
+.Pp
+Remove all but second:
+.rm first\\second first\esecond second
+.first
+.second
+.first\\second
+.Pp
+macro seperated from argument by an escape sequence:
+.de witharg
+.Dq \\$1
+..
+.witharg\(enargument
+.Pp
+final text
diff --git a/regress/usr.bin/mandoc/roff/de/escname.out_ascii b/regress/usr.bin/mandoc/roff/de/escname.out_ascii
new file mode 100644
index 00000000000..d4a08f59253
--- /dev/null
+++ b/regress/usr.bin/mandoc/roff/de/escname.out_ascii
@@ -0,0 +1,23 @@
+DE-ESCNAME(1)              OpenBSD Reference Manual              DE-ESCNAME(1)
+
+NNAAMMEE
+     ddee--eessccnnaammee - escape sequences in macro names
+
+DDEESSCCRRIIPPTTIIOONN
+     initial text
+
+     define second = val2
+
+     define first\second = val3
+
+     define first = val1
+
+     Values (first, second, first\second): val1 val2 val3
+
+     Remove all but second: val2
+
+     macro seperated from argument by an escape sequence: ``argument''
+
+     final text
+
+OpenBSD                          June 29, 2014                         OpenBSD
diff --git a/usr.bin/mandoc/roff.c b/usr.bin/mandoc/roff.c
index 79e771780d4..707efed3332 100644
--- a/usr.bin/mandoc/roff.c
+++ b/usr.bin/mandoc/roff.c
@@ -1,4 +1,4 @@
-/*	$Id: roff.c,v 1.86 2014/06/29 22:38:41 schwarze Exp $ */
+/*	$Id: roff.c,v 1.87 2014/06/29 23:23:16 schwarze Exp $ */
 /*
  * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
@@ -196,7 +196,8 @@ static	enum rofferr	 roff_line_ignore(ROFF_ARGS);
 static	enum rofferr	 roff_nr(ROFF_ARGS);
 static	void		 roff_openeqn(struct roff *, const char *,
 				int, int, const char *);
-static	enum rofft	 roff_parse(struct roff *, const char *, int *);
+static	enum rofft	 roff_parse(struct roff *, char *, int *,
+				int, int);
 static	enum rofferr	 roff_parsetext(char **, size_t *, int, int *);
 static	enum rofferr	 roff_res(struct roff *,
 				char **, size_t *, int, int);
@@ -760,7 +761,7 @@ roff_parseln(struct roff *r, int ln, char **bufp,
 	 * the compilers handle it.
 	 */
 
-	if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
+	if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
 		return(ROFF_CONT);
 
 	assert(roffs[t].proc);
@@ -793,28 +794,26 @@ roff_endparse(struct roff *r)
  * form of ".foo xxx" in the usual way.
  */
 static enum rofft
-roff_parse(struct roff *r, const char *buf, int *pos)
+roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
 {
+	char		*cp;
 	const char	*mac;
 	size_t		 maclen;
 	enum rofft	 t;
 
-	if ('\0' == buf[*pos] || '"' == buf[*pos] ||
-	    '\t' == buf[*pos] || ' ' == buf[*pos])
-		return(ROFF_MAX);
+	cp = buf + *pos;
 
-	/* We stop the macro parse at an escape, tab, space, or nil. */
+	if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
+		return(ROFF_MAX);
 
-	mac = buf + *pos;
-	maclen = strcspn(mac, " \\\t\0");
+	mac = cp;
+	maclen = roff_getname(r, &cp, ln, ppos);
 
 	t = (r->current_string = roff_getstrn(r, mac, maclen))
 	    ? ROFF_USERDEF : roffhash_find(mac, maclen);
 
-	*pos += (int)maclen;
-
-	while (buf[*pos] && ' ' == buf[*pos])
-		(*pos)++;
+	if (ROFF_MAX != t)
+		*pos = cp - buf;
 
 	return(t);
 }
@@ -992,7 +991,7 @@ roff_block_sub(ROFF_ARGS)
 				i++;
 
 			pos = i;
-			if (ROFF_MAX != roff_parse(r, *bufp, &pos))
+			if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
 				return(ROFF_RERUN);
 			return(ROFF_IGN);
 		}
@@ -1003,7 +1002,7 @@ roff_block_sub(ROFF_ARGS)
 	 * pulling it out of the hashtable.
 	 */
 
-	t = roff_parse(r, *bufp, &pos);
+	t = roff_parse(r, *bufp, &pos, ln, ppos);
 
 	/*
 	 * Macros other than block-end are only significant
@@ -1038,7 +1037,7 @@ roff_cond_sub(ROFF_ARGS)
 
 	rr = r->last->rule;
 	roffnode_cleanscope(r);
-	t = roff_parse(r, *bufp, &pos);
+	t = roff_parse(r, *bufp, &pos, ln, ppos);
 
 	/*
 	 * Fully handle known macros when they are structurally