===================================================================
RCS file: /cvs/mandoc/mdoc_macro.c,v
retrieving revision 1.13
retrieving revision 1.28
diff -u -p -r1.13 -r1.28
--- mandoc/mdoc_macro.c	2009/06/16 19:45:51	1.13
+++ mandoc/mdoc_macro.c	2009/07/29 08:46:06	1.28
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_macro.c,v 1.13 2009/06/16 19:45:51 kristaps Exp $ */
+/*	$Id: mdoc_macro.c,v 1.28 2009/07/29 08:46:06 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -22,22 +22,6 @@
 
 #include "libmdoc.h"
 
-/* FIXME: .Fl, .Ar, .Cd handling of `|'. */
-
-enum	mwarn {
-	WIGNE,
-	WIMPBRK,
-	WMACPARM,
-	WOBS
-};
-
-enum	merr {
-	EOPEN,
-	EQUOT,
-	ENOCTX,
-	ENOPARMS
-};
-
 #define	REWIND_REWIND	(1 << 0)
 #define	REWIND_NOHALT	(1 << 1)
 #define	REWIND_HALT	(1 << 2)
@@ -63,14 +47,11 @@ static	int	  rew_subblock(enum mdoc_type, 
 			struct mdoc *, int, int, int);
 static	int	  rew_last(struct mdoc *, struct mdoc_node *);
 static	int	  append_delims(struct mdoc *, int, int *, char *);
-static	int	  lookup(struct mdoc *, int, int, int, const char *);
-static	int	  pwarn(struct mdoc *, int, int, enum mwarn);
-static	int	  perr(struct mdoc *, int, int, enum merr);
+static	int	  lookup(struct mdoc *, int, const char *);
+static	int	  lookup_raw(struct mdoc *, const char *);
 static	int	  swarn(struct mdoc *, enum mdoc_type, int, int, 
 			const struct mdoc_node *);
 
-#define	nerr(m, n, t) perr((m), (n)->line, (n)->pos, (t))
-
 /* Central table of library: who gets parsed how. */
 
 const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
@@ -80,7 +61,7 @@ const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
 	{ in_line_eoln, MDOC_PROLOGUE }, /* Os */
 	{ blk_full, 0 }, /* Sh */
 	{ blk_full, 0 }, /* Ss */ 
-	{ in_line, 0 }, /* Pp */ 
+	{ in_line_eoln, 0 }, /* Pp */ 
 	{ blk_part_imp, MDOC_PARSED }, /* D1 */
 	{ blk_part_imp, MDOC_PARSED }, /* Dl */
 	{ blk_full, MDOC_EXPLICIT }, /* Bd */
@@ -91,7 +72,7 @@ const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
-	{ in_line_eoln, MDOC_CALLABLE }, /* Cd */
+	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 
@@ -103,9 +84,9 @@ const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */ 
-	{ in_line_eoln, 0 }, /* In */  /* FIXME: historic usage! */
+	{ in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */
-	{ in_line_eoln, 0 }, /* Nd */ 
+	{ blk_full, 0 }, /* Nd */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 
 	{ blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
 	{ obsolete, 0 }, /* Ot */
@@ -181,7 +162,7 @@ const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
 	{ obsolete, 0 }, /* Fr */
 	{ in_line_eoln, 0 }, /* Ud */
 	{ in_line_eoln, 0 }, /* Lb */
-	{ in_line, 0 }, /* Lp */ 
+	{ in_line_eoln, 0 }, /* Lp */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 
 	{ in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 
 	{ blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */
@@ -192,62 +173,14 @@ const	struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
 	{ obsolete, 0 }, /* En */
 	{ in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
 	{ in_line_eoln, 0 }, /* %Q */
+	{ in_line_eoln, 0 }, /* br */
+	{ in_line_eoln, 0 }, /* sp */
 };
 
 const	struct mdoc_macro * const mdoc_macros = __mdoc_macros;
 
 
 static int
-perr(struct mdoc *mdoc, int line, int pos, enum merr type)
-{
-	char		*p;
-
-	p = NULL;
-	switch (type) {
-	case (EOPEN):
-		p = "explicit scope still open on exit";
-		break;
-	case (EQUOT):
-		p = "unterminated quotation";
-		break;
-	case (ENOCTX):
-		p = "closure has no prior context";
-		break;
-	case (ENOPARMS):
-		p = "unexpect line arguments";
-		break;
-	}
-	assert(p);
-	return(mdoc_perr(mdoc, line, pos, p));
-}
-
-
-static int
-pwarn(struct mdoc *mdoc, int line, int pos, enum mwarn type)
-{
-	char		*p;
-
-	p = NULL;
-	switch (type) {
-	case (WIGNE):
-		p = "ignoring empty element";
-		break;
-	case (WIMPBRK):
-		p = "crufty end-of-line scope violation";
-		break;
-	case (WMACPARM):
-		p = "macro-like parameter";
-		break;
-	case (WOBS):
-		p = "macro marked obsolete";
-		break;
-	}
-	assert(p);
-	return(mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, p));
-}
-
-
-static int
 swarn(struct mdoc *mdoc, enum mdoc_type type, 
 		int line, int pos, const struct mdoc_node *p)
 {
@@ -285,10 +218,10 @@ swarn(struct mdoc *mdoc, enum mdoc_type type, 
 	}
 
 	if ( ! (MDOC_IGN_SCOPE & mdoc->pflags))
-		return(mdoc_perr(mdoc, line, pos, 
+		return(mdoc_verr(mdoc, line, pos, 
 				"%s scope breaks %s scope of %s", 
 				tt, t, n));
-	return(mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX,
+	return(mdoc_vwarn(mdoc, line, pos, 
 				"%s scope breaks %s scope of %s", 
 				tt, t, n));
 }
@@ -314,24 +247,31 @@ mdoc_macroend(struct mdoc *mdoc)
 			continue;
 		if ( ! (MDOC_EXPLICIT & mdoc_macros[n->tok].flags))
 			continue;
-		return(nerr(mdoc, n, EOPEN));
+		return(mdoc_nerr(mdoc, n, EOPEN));
 	}
 
 	return(rew_last(mdoc, mdoc->first));
 }
 
 static int
-lookup(struct mdoc *mdoc, int line, int pos, int from, const char *p)
+lookup(struct mdoc *mdoc, int from, const char *p)
 {
+
+	if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
+		return(MDOC_MAX);
+	return(lookup_raw(mdoc, p));
+}
+
+
+static int
+lookup_raw(struct mdoc *mdoc, const char *p)
+{
 	int		 res;
 
-	res = mdoc_hash_find(mdoc->htab, p);
-	if (MDOC_PARSED & mdoc_macros[from].flags)
+	if (MDOC_MAX == (res = mdoc_hash_find(mdoc->htab, p)))
+		return(MDOC_MAX);
+	if (MDOC_CALLABLE & mdoc_macros[res].flags)
 		return(res);
-	if (MDOC_MAX == res)
-		return(res);
-	if ( ! pwarn(mdoc, line, pos, WMACPARM))
-		return(-1);
 	return(MDOC_MAX);
 }
 
@@ -456,6 +396,8 @@ rew_dohalt(int tok, enum mdoc_type type, const struct 
 		if (type == p->type && tok == p->tok)
 			return(REWIND_REWIND);
 		break;
+	case (MDOC_Nd):
+		/* FALLTHROUGH */
 	case (MDOC_Ss):
 		assert(MDOC_TAIL != type);
 		if (type == p->type && tok == p->tok)
@@ -561,9 +503,13 @@ rew_dobreak(int tok, const struct mdoc_node *p)
 	switch (tok) {
 	case (MDOC_It):
 		return(MDOC_It == p->tok);
+	case (MDOC_Nd):
+		return(MDOC_Nd == p->tok);
 	case (MDOC_Ss):
 		return(MDOC_Ss == p->tok);
 	case (MDOC_Sh):
+		if (MDOC_Nd == p->tok)
+			return(1);
 		if (MDOC_Ss == p->tok)
 			return(1);
 		return(MDOC_Sh == p->tok);
@@ -639,7 +585,7 @@ rew_expblock(struct mdoc *mdoc, int tok, int line, int
 	for (n = mdoc->last; n; n = n->parent) {
 		c = rew_dohalt(tok, MDOC_BLOCK, n);
 		if (REWIND_HALT == c)
-			return(perr(mdoc, line, ppos, ENOCTX));
+			return(mdoc_perr(mdoc, line, ppos, ENOCTX));
 		if (REWIND_REWIND == c)
 			break;
 		else if (rew_dobreak(tok, n))
@@ -724,13 +670,13 @@ blk_exp_close(MACRO_PROT_ARGS)
 	}
 
 	if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
-		if (0 == buf[*pos]) {
-			if ( ! rew_subblock(MDOC_BODY, mdoc, 
-						tok, line, ppos))
+		if (buf[*pos]) 
+			if ( ! mdoc_pwarn(mdoc, line, ppos, ENOLINE))
 				return(0);
-			return(rew_expblock(mdoc, tok, line, ppos));
-		}
-		return(perr(mdoc, line, ppos, ENOPARMS));
+
+		if ( ! rew_subblock(MDOC_BODY, mdoc, tok, line, ppos))
+			return(0);
+		return(rew_expblock(mdoc, tok, line, ppos));
 	}
 
 	if ( ! rew_subblock(MDOC_BODY, mdoc, tok, line, ppos))
@@ -743,7 +689,7 @@ blk_exp_close(MACRO_PROT_ARGS)
 		mdoc->next = MDOC_NEXT_CHILD;
 	}
 
-	for (lastarg = ppos, flushed = j = 0; ; j++) {
+	for (flushed = j = 0; ; j++) {
 		lastarg = *pos;
 
 		if (j == maxargs && ! flushed) {
@@ -761,7 +707,7 @@ blk_exp_close(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (-1 == (c = lookup(mdoc, line, lastarg, tok, p)))
+		if (-1 == (c = lookup(mdoc, tok, p)))
 			return(0);
 		else if (MDOC_MAX != c) {
 			if ( ! flushed) {
@@ -804,17 +750,17 @@ in_line(MACRO_PROT_ARGS)
 	 * usually because of reserved words) to squeak by.
 	 */
 	switch (tok) {
-	case (MDOC_Lp):
+	case (MDOC_An):
 		/* FALLTHROUGH */
-	case (MDOC_Pp):
+	case (MDOC_Ar):
 		/* FALLTHROUGH */
-	case (MDOC_Nm):
-		/* FALLTHROUGH */
 	case (MDOC_Fl):
 		/* FALLTHROUGH */
 	case (MDOC_Lk):
 		/* FALLTHROUGH */
-	case (MDOC_Ar):
+	case (MDOC_Nm):
+		/* FALLTHROUGH */
+	case (MDOC_Pa):
 		nc = 1;
 		break;
 	default:
@@ -822,7 +768,7 @@ in_line(MACRO_PROT_ARGS)
 		break;
 	}
 
-	for (la = ppos, arg = NULL;; ) {
+	for (arg = NULL;; ) {
 		la = *pos;
 		c = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
 
@@ -852,8 +798,7 @@ in_line(MACRO_PROT_ARGS)
 
 		/* Quoted words shouldn't be looked-up. */
 
-		c = ARGS_QWORD == w ? MDOC_MAX :
-			lookup(mdoc, line, la, tok, p);
+		c = ARGS_QWORD == w ? MDOC_MAX : lookup(mdoc, tok, p);
 
 		/* 
 		 * In this case, we've located a submacro and must
@@ -873,7 +818,7 @@ in_line(MACRO_PROT_ARGS)
 					return(0);
 			} else if ( ! nc && 0 == cnt) {
 				mdoc_argv_free(arg);
-				if ( ! pwarn(mdoc, line, ppos, WIGNE))
+				if ( ! mdoc_pwarn(mdoc, line, ppos, EIGNE))
 					return(0);
 			}
 			c = mdoc_macro(mdoc, c, line, la, pos, buf);
@@ -929,7 +874,7 @@ in_line(MACRO_PROT_ARGS)
 			return(0);
 	} else if ( ! nc && 0 == cnt) {
 		mdoc_argv_free(arg);
-		if ( ! pwarn(mdoc, line, ppos, WIGNE))
+		if ( ! mdoc_pwarn(mdoc, line, ppos, EIGNE))
 			return(0);
 	}
 
@@ -945,10 +890,26 @@ in_line(MACRO_PROT_ARGS)
 static int
 blk_full(MACRO_PROT_ARGS)
 {
-	int		  c, lastarg, reopen;
+	int		  c, lastarg, reopen, dohead;
 	struct mdoc_arg	 *arg;
 	char		 *p;
 
+	/* 
+	 * Whether to process a block-head section.  If this is
+	 * non-zero, then a head will be opened for all line arguments.
+	 * If not, then the head will always be empty and only a body
+	 * will be opened, which will stay open at the eoln.
+	 */
+
+	switch (tok) {
+	case (MDOC_Nd):
+		dohead = 0;
+		break;
+	default:
+		dohead = 1;
+		break;
+	}
+
 	if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
 		if ( ! rew_subblock(MDOC_BODY, mdoc, 
 					tok, line, ppos))
@@ -993,6 +954,16 @@ blk_full(MACRO_PROT_ARGS)
 
 	if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
 		return(0);
+
+	/* Immediately close out head and enter body, if applicable. */
+
+	if (0 == dohead) {
+		if ( ! rew_subblock(MDOC_HEAD, mdoc, tok, line, ppos))
+			return(0);
+		if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
+			return(0);
+	} 
+
 	mdoc->next = MDOC_NEXT_CHILD;
 
 	for (reopen = 0;; ) {
@@ -1004,6 +975,7 @@ blk_full(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 		if (ARGS_PHRASE == c) {
+			assert(dohead);
 			if (reopen && ! mdoc_head_alloc
 					(mdoc, line, ppos, tok))
 				return(0);
@@ -1023,7 +995,7 @@ blk_full(MACRO_PROT_ARGS)
 			continue;
 		}
 
-		if (-1 == (c = lookup(mdoc, line, lastarg, tok, p)))
+		if (-1 == (c = lookup(mdoc, tok, p)))
 			return(0);
 
 		if (MDOC_MAX == c) {
@@ -1040,9 +1012,13 @@ blk_full(MACRO_PROT_ARGS)
 	
 	if (1 == ppos && ! append_delims(mdoc, line, pos, buf))
 		return(0);
+
+	/* If the body's already open, then just return. */
+	if (0 == dohead) 
+		return(1);
+
 	if ( ! rew_subblock(MDOC_HEAD, mdoc, tok, line, ppos))
 		return(0);
-
 	if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
 		return(0);
 	mdoc->next = MDOC_NEXT_CHILD;
@@ -1077,7 +1053,7 @@ blk_part_imp(MACRO_PROT_ARGS)
 
 	/* XXX - no known argument macros. */
 
-	for (lastarg = ppos;; ) {
+	for (;;) {
 		lastarg = *pos;
 		c = mdoc_args(mdoc, line, pos, buf, tok, &p);
 		assert(ARGS_PHRASE != c);
@@ -1089,7 +1065,7 @@ blk_part_imp(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (-1 == (c = lookup(mdoc, line, lastarg, tok, p)))
+		if (-1 == (c = lookup(mdoc, tok, p)))
 			return(0);
 		else if (MDOC_MAX == c) {
 			if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
@@ -1113,7 +1089,7 @@ blk_part_imp(MACRO_PROT_ARGS)
 		if (body == n)
 			break;
 
-	if (NULL == n && ! pwarn(mdoc, body->line, body->pos, WIMPBRK))
+	if (NULL == n && ! mdoc_nwarn(mdoc, body, EIMPBRK))
 			return(0);
 
 	if (n && ! rew_last(mdoc, body))
@@ -1138,7 +1114,6 @@ blk_part_exp(MACRO_PROT_ARGS)
 	int		  lastarg, flushed, j, c, maxargs;
 	char		 *p;
 
-	lastarg = ppos;
 	flushed = 0;
 
 	/*
@@ -1194,7 +1169,7 @@ blk_part_exp(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (-1 == (c = lookup(mdoc, line, lastarg, tok, p)))
+		if (-1 == (c = lookup(mdoc, tok, p)))
 			return(0);
 		else if (MDOC_MAX != c) {
 			if ( ! flushed) {
@@ -1274,7 +1249,7 @@ in_line_argn(MACRO_PROT_ARGS)
 		break;
 	}
 
-	for (lastarg = ppos, arg = NULL;; ) {
+	for (arg = NULL;; ) {
 		lastarg = *pos;
 		c = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
 
@@ -1314,7 +1289,7 @@ in_line_argn(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == c)
 			break;
 
-		if (-1 == (c = lookup(mdoc, line, lastarg, tok, p)))
+		if (-1 == (c = lookup(mdoc, tok, p)))
 			return(0);
 		else if (MDOC_MAX != c) {
 			if ( ! flushed && ! rew_elem(mdoc, tok))
@@ -1392,8 +1367,7 @@ in_line_eoln(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == w)
 			break;
 
-		c = ARGS_QWORD == w ? MDOC_MAX :
-			lookup(mdoc, line, la, tok, p);
+		c = ARGS_QWORD == w ? MDOC_MAX : lookup(mdoc, tok, p);
 
 		if (MDOC_MAX != c && -1 != c) {
 			if ( ! rew_elem(mdoc, tok))
@@ -1416,82 +1390,48 @@ static int
 obsolete(MACRO_PROT_ARGS)
 {
 
-	return(pwarn(mdoc, line, ppos, WOBS));
+	return(mdoc_pwarn(mdoc, line, ppos, EOBS));
 }
 
 
+/*
+ * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
+ * They're unusual because they're basically free-form text until a
+ * macro is encountered.
+ */
 static int
 phrase(struct mdoc *mdoc, int line, int ppos, char *buf)
 {
-	int		 i, la, c, quoted;
+	int		  c, w, la, pos;
+	char		 *p;
 
-	/*
-	 * Parse over words in a phrase.  We have to handle this
-	 * specially because we assume no calling context -- in normal
-	 * circumstances, we switch argument parsing based on whether
-	 * the parent macro accepts quotes, tabs, etc.  Here, anything
-	 * goes.
-	 */
+	for (pos = ppos; ; ) {
+		la = pos;
 
-	for (i = ppos; buf[i]; ) {
-		assert(' ' != buf[i]);
-		la = i;
-		quoted = 0;
+		/* Note: no calling context! */
+		w = mdoc_zargs(mdoc, line, &pos, buf, &p);
 
-		/* 
-		 * Read to next token.  If quoted (check not escaped),
-		 * scan ahead to next unescaped quote.  If not quoted or
-		 * escape-quoted, then scan ahead to next space.
-		 */
+		if (ARGS_ERROR == w)
+			return(0);
+		if (ARGS_EOLN == w)
+			break;
 
-		if ((i && '\"' == buf[i] && '\\' != buf[i - 1]) || 
-				(0 == i && '\"' == buf[i])) {
-			for (la = ++i; buf[i]; i++) 
-				if ('\"' != buf[i])
-					continue;
-				else if ('\\' != buf[i - 1])
-					break;
-			if (0 == buf[i]) 
-				return(perr(mdoc, line, la, EQUOT));
-			quoted = 1;
-		} else
-			for ( ; buf[i]; i++)
-				if (i && ' ' == buf[i]) {
-					if ('\\' != buf[i - 1])
-						break;
-				} else if (' ' == buf[i])
-					break;
+		c = ARGS_QWORD == w ? MDOC_MAX :
+			mdoc_hash_find(mdoc->htab, p);
 
-		/* If not end-of-line, terminate argument. */
-
-		if (buf[i])
-			buf[i++] = 0;
-
-		/* Read to next argument. */
-
-		for ( ; buf[i] && ' ' == buf[i]; i++)
-			/* Spin. */ ;
-
-		/* 
-		 * If we're a non-quoted string, try to look up the
-		 * value as a macro and execute it, if found.
-		 */
-
-		c = quoted ? MDOC_MAX :
-			mdoc_hash_find(mdoc->htab, &buf[la]);
-
-		if (MDOC_MAX != c) {
-			if ( ! mdoc_macro(mdoc, c, line, la, &i, buf))
+		if (MDOC_MAX != c && -1 != c) {
+			if ( ! mdoc_macro(mdoc, c, line, la, &pos, buf))
 				return(0);
-			return(append_delims(mdoc, line, &i, buf));
-		}
+			return(append_delims(mdoc, line, &pos, buf));
+		} else if (-1 == c)
+			return(0);
 
-		/* A regular word or quoted string. */
-
-		if ( ! mdoc_word_alloc(mdoc, line, la, &buf[la]))
+		if ( ! mdoc_word_alloc(mdoc, line, la, p))
 			return(0);
 		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	return(1);
 }
+
+