proper handling of quoted tab-separated column lists

by kristaps@

3 files changed, 64 insertions, 55 deletions

diff --git a/usr.bin/mandoc/libmdoc.h b/usr.bin/mandoc/libmdoc.h
index 3b59e440d82..6ac17daf5d7 100644
--- a/usr.bin/mandoc/libmdoc.h
+++ b/usr.bin/mandoc/libmdoc.h
@@ -1,4 +1,4 @@
-/*	$Id: libmdoc.h,v 1.31 2010/05/15 12:30:59 schwarze Exp $ */
+/*	$Id: libmdoc.h,v 1.32 2010/05/15 13:12:55 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -32,6 +32,8 @@ struct	mdoc {
 #define	MDOC_LITERAL	 (1 << 1) /* in a literal scope */
 #define	MDOC_PBODY	 (1 << 2) /* in the document body */
 #define	MDOC_NEWLINE	 (1 << 3) /* first macro/text in a line */
+#define	MDOC_PHRASELIT	 (1 << 4) /* literal within a partila phrase */
+#define	MDOC_PPHRASE	 (1 << 5) /* within a partial phrase */
 	int		  pflags;
 	enum mdoc_next	  next;
 	struct mdoc_node *last;
@@ -193,9 +195,9 @@ enum margserr	  mdoc_args(struct mdoc *, int,
 			int *, char *, enum mdoct, char **);
 enum margserr	  mdoc_zargs(struct mdoc *, int, 
 			int *, char *, int, char **);
-#define	ARGS_DELIM	(1 << 1)	/* See args(). */
-#define	ARGS_TABSEP	(1 << 2)	/* See args(). */
-#define	ARGS_NOWARN	(1 << 3)	/* See args(). */
+#define	ARGS_DELIM	(1 << 1)
+#define	ARGS_TABSEP	(1 << 2)
+#define	ARGS_NOWARN	(1 << 3)
 
 int		  mdoc_macroend(struct mdoc *);
 
diff --git a/usr.bin/mandoc/mdoc_argv.c b/usr.bin/mandoc/mdoc_argv.c
index 6e284601a2f..d4b1cc4bb6c 100644
--- a/usr.bin/mandoc/mdoc_argv.c
+++ b/usr.bin/mandoc/mdoc_argv.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_argv.c,v 1.27 2010/05/15 12:30:59 schwarze Exp $ */
+/*	$Id: mdoc_argv.c,v 1.28 2010/05/15 13:12:55 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -394,8 +394,21 @@ args(struct mdoc *m, int line, int *pos,
 	assert(*pos);
 	assert(' ' != buf[*pos]);
 
-	if ('\0' == buf[*pos])
+	if ('\0' == buf[*pos]) {
+		if (MDOC_PPHRASE & m->flags)
+			return(ARGS_EOLN);
+		/*
+		 * If we're not in a partial phrase and the flag for
+		 * being a phrase literal is still set, the punctuation
+		 * is unterminated.
+		 */
+		if (MDOC_PHRASELIT & m->flags)
+			if ( ! mdoc_pwarn(m, line, *pos, EQUOTTERM))
+				return(ARGS_ERROR);
+
+		m->flags &= ~MDOC_PHRASELIT;
 		return(ARGS_EOLN);
+	}
 
 	/* 
 	 * If the first character is a closing delimiter and we're to
@@ -441,16 +454,18 @@ args(struct mdoc *m, int line, int *pos,
 	if (ARGS_TABSEP & fl) {
 		/* Scan ahead to tab (can't be escaped). */
 		p = strchr(*v, '\t');
+		pp = NULL;
 
 		/* Scan ahead to unescaped `Ta'. */
-		for (pp = *v; ; pp++) {
-			if (NULL == (pp = strstr(pp, "Ta")))
-				break;
-			if (pp > *v && ' ' != *(pp - 1))
-				continue;
-			if (' ' == *(pp + 2) || 0 == *(pp + 2))
-				break;
-		}
+		if ( ! (MDOC_PHRASELIT & m->flags)) 
+			for (pp = *v; ; pp++) {
+				if (NULL == (pp = strstr(pp, "Ta")))
+					break;
+				if (pp > *v && ' ' != *(pp - 1))
+					continue;
+				if (' ' == *(pp + 2) || 0 == *(pp + 2))
+					break;
+			}
 
 		/* By default, assume a phrase. */
 		rc = ARGS_PHRASE;
@@ -503,8 +518,12 @@ args(struct mdoc *m, int line, int *pos,
 	 * Whitespace is NOT involved in literal termination.
 	 */
 
-	if ('\"' == buf[*pos]) {
-		*v = &buf[++(*pos)];
+	if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
+		if ( ! (MDOC_PHRASELIT & m->flags))
+			*v = &buf[++(*pos)];
+
+		if (MDOC_PPHRASE & m->flags)
+			m->flags |= MDOC_PHRASELIT;
 
 		for ( ; buf[*pos]; (*pos)++) {
 			if ('\"' != buf[*pos])
@@ -514,17 +533,18 @@ args(struct mdoc *m, int line, int *pos,
 			(*pos)++;
 		}
 
-		if (0 == buf[*pos]) {
-			if (ARGS_NOWARN & fl)
+		if ('\0' == buf[*pos]) {
+			if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
 				return(ARGS_QWORD);
 			if ( ! mdoc_pwarn(m, line, *pos, EQUOTTERM))
 				return(ARGS_ERROR);
 			return(ARGS_QWORD);
 		}
 
-		buf[(*pos)++] = 0;
+		m->flags &= ~MDOC_PHRASELIT;
+		buf[(*pos)++] = '\0';
 
-		if (0 == buf[*pos])
+		if ('\0' == buf[*pos])
 			return(ARGS_QWORD);
 
 		while (' ' == buf[*pos])
@@ -546,15 +566,15 @@ args(struct mdoc *m, int line, int *pos,
 		if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
 			break;
 
-	if (0 == buf[*pos])
+	if ('\0' == buf[*pos])
 		return(ARGS_WORD);
 
-	buf[(*pos)++] = 0;
+	buf[(*pos)++] = '\0';
 
 	while (' ' == buf[*pos])
 		(*pos)++;
 
-	if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
+	if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
 		if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
 			return(ARGS_ERROR);
 
diff --git a/usr.bin/mandoc/mdoc_macro.c b/usr.bin/mandoc/mdoc_macro.c
index a87b7827372..4f24b9adb77 100644
--- a/usr.bin/mandoc/mdoc_macro.c
+++ b/usr.bin/mandoc/mdoc_macro.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_macro.c,v 1.40 2010/05/15 12:30:59 schwarze Exp $ */
+/*	$Id: mdoc_macro.c,v 1.41 2010/05/15 13:12:55 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -45,7 +45,7 @@ static	int	  	append_delims(struct mdoc *,
 static	enum mdoct	lookup(enum mdoct, const char *);
 static	enum mdoct	lookup_raw(const char *);
 static	int	  	phrase(struct mdoc *, int, int, 
-				char *, enum margserr, int);
+				char *, enum margserr);
 static	enum mdoct 	rew_alt(enum mdoct);
 static	int	  	rew_dobreak(enum mdoct, 
 				const struct mdoc_node *);
@@ -883,7 +883,7 @@ in_line(MACRO_PROT_ARGS)
 static int
 blk_full(MACRO_PROT_ARGS)
 {
-	int		  la, pcnt;
+	int		  la;
 	struct mdoc_arg	 *arg;
 	struct mdoc_node *head; /* save of head macro */
 	struct mdoc_node *body; /* save of body macro */
@@ -952,7 +952,7 @@ blk_full(MACRO_PROT_ARGS)
 
 	ac = ARGS_ERROR;
 
-	for (pcnt = 0; ; ) {
+	for ( ; ; ) {
 		la = *pos;
 		lac = ac;
 		ac = mdoc_args(m, line, pos, buf, tok, &p);
@@ -962,12 +962,18 @@ blk_full(MACRO_PROT_ARGS)
 		if (ARGS_EOLN == ac)
 			break;
 
+		if (ARGS_PEND == ac) {
+			if (ARGS_PPHRASE == lac)
+				ac = ARGS_PPHRASE;
+			else
+				ac = ARGS_PHRASE;
+		}
+
 		/* Don't emit leading punct. for phrases. */
 
 		if (NULL == head && 
 				ARGS_PHRASE != ac &&
 				ARGS_PPHRASE != ac &&
-				ARGS_PEND != ac &&
 				ARGS_QWORD != ac &&
 				DELIM_OPEN == mdoc_isdelim(p)) {
 			if ( ! mdoc_word_alloc(m, line, la, p))
@@ -979,29 +985,18 @@ blk_full(MACRO_PROT_ARGS)
 
 		if (NULL == head || 
 				ARGS_PHRASE == ac || 
-				ARGS_PEND == ac || 
 				ARGS_PPHRASE == ac) {
 			if ( ! mdoc_head_alloc(m, line, ppos, tok))
 				return(0);
 			head = m->last;
 		}
 
-		if (ARGS_PHRASE == ac || 
-				ARGS_PEND == ac ||
-				ARGS_PPHRASE == ac) {
-			/*
-			 * Special treatment for the last phrase.  A
-			 * prior ARGS_PHRASE gets is handled as a
-			 * regular ARGS_PHRASE, but a prior ARGS_PPHRASE
-			 * has special handling.
-			 */
-			if (ARGS_PEND == ac && ARGS_ERROR == lac)
-				ac = ARGS_PHRASE;
-			else if (ARGS_PEND == ac && ARGS_PHRASE == lac)
-				ac = ARGS_PHRASE;
-
-			if ( ! phrase(m, line, la, buf, ac, pcnt++))
+		if (ARGS_PHRASE == ac || ARGS_PPHRASE == ac) {
+			if (ARGS_PPHRASE == ac)
+				m->flags |= MDOC_PPHRASE;
+			if ( ! phrase(m, line, la, buf, ac))
 				return(0);
+			m->flags &= ~MDOC_PPHRASE;
 			if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
 				return(0);
 			continue;
@@ -1543,25 +1538,18 @@ obsolete(MACRO_PROT_ARGS)
  * macro is encountered.
  */
 static int
-phrase(struct mdoc *m, int line, int ppos, char *buf, 
-		enum margserr ac, int count)
+phrase(struct mdoc *m, int line, int ppos, char *buf, enum margserr ac)
 {
 	int		 la, pos;
 	enum margserr	 aac;
 	enum mdoct	 ntok;
 	char		*p;
 
-	assert(ARGS_PHRASE == ac || 
-			ARGS_PEND == ac ||
-			ARGS_PPHRASE == ac);
-
-	if (count && ARGS_PPHRASE == ac)
-		return(mdoc_word_alloc(m, line, ppos, &buf[ppos]));
+	assert(ARGS_PHRASE == ac || ARGS_PPHRASE == ac);
 
 	for (pos = ppos; ; ) {
 		la = pos;
 
-		/* Note: no calling context! */
 		aac = mdoc_zargs(m, line, &pos, buf, 0, &p);
 
 		if (ARGS_ERROR == aac)
@@ -1569,8 +1557,7 @@ phrase(struct mdoc *m, int line, int ppos, char *buf,
 		if (ARGS_EOLN == aac)
 			break;
 
-		ntok = ARGS_QWORD == aac || ARGS_PEND == ac ? 
-			MDOC_MAX : lookup_raw(p);
+		ntok = ARGS_QWORD == aac ? MDOC_MAX : lookup_raw(p);
 
 		if (MDOC_MAX == ntok) {
 			if ( ! mdoc_word_alloc(m, line, la, p))