Moved MDOC_NEXT_SIBLING into mdoc_word (libmdoc).

De-chunking of word terms in libman.

3 files changed, 67 insertions, 28 deletions

diff --git a/man.c b/man.c
index 993d9f4c..59cfba98 100644
--- a/man.c
+++ b/man.c
@@ -65,6 +65,8 @@ static	int		 man_ptext(struct man *, int, char *);
 static	int		 man_pmacro(struct man *, int, char *);
 static	void		 man_free1(struct man *);
 static	int		 man_alloc1(struct man *);
+static	int		 pstring(struct man *, int, int, 
+				const char *, size_t);
 
 
 const struct man_node *
@@ -318,23 +320,43 @@ man_block_alloc(struct man *m, int line, int pos, int tok)
 }
 
 
-int
-man_word_alloc(struct man *m, int line, int pos, const char *word)
+static int
+pstring(struct man *m, int line, int pos, 
+		const char *p, size_t len)
 {
-	struct man_node	*p;
+	struct man_node	*n;
+	size_t		 sv;
 
-	p = man_node_alloc(line, pos, MAN_TEXT, -1);
-	if (NULL == p)
+	n = man_node_alloc(line, pos, MAN_TEXT, -1);
+	if (NULL == n)
 		return(0);
-	if (NULL == (p->string = strdup(word)))
+
+	n->string = malloc(len + 1);
+	if (NULL == n->string) {
+		free(n);
 		return(0);
-	if ( ! man_node_append(m, p))
+	}
+
+	sv = strlcpy(n->string, p, len + 1);
+
+	/* Prohibit truncation. */
+	assert(sv < len + 1);
+
+	if ( ! man_node_append(m, n))
 		return(0);
 	m->next = MAN_NEXT_SIBLING;
 	return(1);
 }
 
 
+int
+man_word_alloc(struct man *m, int line, int pos, const char *word)
+{
+
+	return(pstring(m, line, pos, word, strlen(word)));
+}
+
+
 void
 man_node_free(struct man_node *p)
 {
@@ -364,14 +386,42 @@ man_node_freelist(struct man_node *p)
 static int
 man_ptext(struct man *m, int line, char *buf)
 {
+	int		 i, j;
+
+	/* First de-chunk and allocate words. */
+
+	for (i = 0; ' ' == buf[i]; i++)
+		/* Skip leading whitespace. */ ;
+	if (0 == buf[i]) {
+		if ( ! pstring(m, line, 0, &buf[i], 0))
+			return(0);
+		goto descope;
+	}
 
-	/* First allocate word. */
+	for (j = i; buf[i]; i++) {
+		if (' ' != buf[i])
+			continue;
 
-	/* FIXME: dechunk words! */
+		/* Escaped whitespace. */
+		if (i && ' ' == buf[i] && '\\' == buf[i - 1])
+			continue;
+
+		buf[i++] = 0;
+		if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
+			return(0);
 
-	if ( ! man_word_alloc(m, line, 0, buf))
+		for ( ; ' ' == buf[i]; i++)
+			/* Skip trailing whitespace. */ ;
+
+		j = i;
+		if (0 == buf[i])
+			break;
+	}
+
+	if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 		return(0);
-	m->next = MAN_NEXT_SIBLING;
+
+descope:
 
 	/*
 	 * Co-ordinate what happens with having a next-line scope open:
diff --git a/mdoc.c b/mdoc.c
index f9ae1ede..b09dfa30 100644
--- a/mdoc.c
+++ b/mdoc.c
@@ -543,7 +543,10 @@ pstring(struct mdoc *m, int line, int pos, const char *p, size_t len)
 	/* Prohibit truncation. */
 	assert(sv < len + 1);
 
-	return(node_append(m, n));
+	if ( ! node_append(m, n))
+		return(0);
+	m->next = MDOC_NEXT_SIBLING;
+	return(1);
 }
 
 
@@ -600,12 +603,8 @@ parsetext(struct mdoc *m, int line, char *buf)
 	 * back-end, as it should be preserved as a single term.
 	 */
 
-	if (MDOC_LITERAL & m->flags) {
-		if ( ! mdoc_word_alloc(m, line, 0, buf))
-			return(0);
-		m->next = MDOC_NEXT_SIBLING;
-		return(1);
-	}
+	if (MDOC_LITERAL & m->flags)
+		return(mdoc_word_alloc(m, line, 0, buf));
 
 	/* Disallow blank/white-space lines in non-literal mode. */
 
@@ -630,7 +629,6 @@ parsetext(struct mdoc *m, int line, char *buf)
 		buf[i++] = 0;
 		if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 			return(0);
-		m->next = MDOC_NEXT_SIBLING;
 
 		for ( ; ' ' == buf[i]; i++)
 			/* Skip trailing whitespace. */ ;
diff --git a/mdoc_macro.c b/mdoc_macro.c
index 379e8c7d..022a86a5 100644
--- a/mdoc_macro.c
+++ b/mdoc_macro.c
@@ -645,7 +645,6 @@ append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
 		assert(mdoc_isdelim(p));
 		if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	return(1);
@@ -722,7 +721,6 @@ blk_exp_close(MACRO_PROT_ARGS)
 
 		if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	if ( ! flushed && ! rew_expblock(mdoc, tok, line, ppos))
@@ -852,7 +850,6 @@ in_line(MACRO_PROT_ARGS)
 			cnt++;
 		if ( ! mdoc_word_alloc(mdoc, line, la, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	if (0 == lastpunct && ! rew_elem(mdoc, tok))
@@ -996,7 +993,6 @@ blk_full(MACRO_PROT_ARGS)
 		if (MDOC_MAX == (c = lookup(mdoc, tok, p))) {
 			if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 				return(0);
-			mdoc->next = MDOC_NEXT_SIBLING;
 			continue;
 		} 
 
@@ -1063,7 +1059,6 @@ blk_part_imp(MACRO_PROT_ARGS)
 		if (MDOC_MAX == (c = lookup(mdoc, tok, p))) {
 			if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 				return(0);
-			mdoc->next = MDOC_NEXT_SIBLING;
 			continue;
 		} 
 
@@ -1191,7 +1186,6 @@ blk_part_exp(MACRO_PROT_ARGS)
 	
 		if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	if ( ! flushed) {
@@ -1298,7 +1292,6 @@ in_line_argn(MACRO_PROT_ARGS)
 	
 		if ( ! mdoc_word_alloc(mdoc, line, lastarg, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	if ( ! flushed && ! rew_elem(mdoc, tok))
@@ -1366,7 +1359,6 @@ in_line_eoln(MACRO_PROT_ARGS)
 
 		if ( ! mdoc_word_alloc(mdoc, line, la, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	return(rew_elem(mdoc, tok));
@@ -1414,7 +1406,6 @@ phrase(struct mdoc *mdoc, int line, int ppos, char *buf)
 
 		if ( ! mdoc_word_alloc(mdoc, line, la, p))
 			return(0);
-		mdoc->next = MDOC_NEXT_SIBLING;
 	}
 
 	return(1);