===================================================================
RCS file: /cvs/docbook2mdoc/parse.c,v
retrieving revision 1.31
retrieving revision 1.35
diff -u -p -r1.31 -r1.35
--- docbook2mdoc/parse.c	2019/04/10 14:34:08	1.31
+++ docbook2mdoc/parse.c	2019/04/12 06:46:45	1.35
@@ -1,4 +1,4 @@
-/* $Id: parse.c,v 1.31 2019/04/10 14:34:08 schwarze Exp $ */
+/* $Id: parse.c,v 1.35 2019/04/12 06:46:45 schwarze Exp $ */
 /*
  * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2019 Ingo Schwarze <schwarze@openbsd.org>
@@ -322,47 +322,88 @@ warn_msg(struct parse *p, const char *fmt, ...)
  * Otherwise, create a new one as a child of the current node.
  */
 static void
-xml_char(struct parse *p, const char *word, int sz)
+xml_text(struct parse *p, const char *word, int sz)
 {
-	struct pnode	*n;
-	size_t		 newsz;
+	struct pnode	*n, *np;
+	size_t		 oldsz, newsz;
+	int		 i;
 
+	assert(sz > 0);
 	if (p->del > 0)
 		return;
 
-	if (p->cur == NULL) {
-		error_msg(p, "discarding text before document: %.*s", sz, word);
+	if ((n = p->cur) == NULL) {
+		error_msg(p, "discarding text before document: %.*s",
+		    sz, word);
 		return;
 	}
 
-	if (p->cur->node != NODE_TEXT) {
-		if ((n = calloc(1, sizeof(*n))) == NULL)
+	/* Append to the current text node, if one is open. */
+
+	if (n->node == NODE_TEXT) {
+		oldsz = strlen(n->b);
+		newsz = oldsz + sz;
+		if (oldsz && (p->flags & PFLAG_SPC))
+			newsz++;
+		if ((n->b = realloc(n->b, newsz + 1)) == NULL)
 			fatal(p);
-		n->node = NODE_TEXT;
-		n->spc = (p->flags & PFLAG_SPC) != 0;
-		n->parent = p->cur;
-		TAILQ_INIT(&n->childq);
-		TAILQ_INIT(&n->attrq);
-		TAILQ_INSERT_TAIL(&p->cur->childq, n, child);
-		p->cur = n;
+		if (oldsz && (p->flags & PFLAG_SPC))
+			n->b[oldsz++] = ' ';
+		memcpy(n->b + oldsz, word, sz);
+		n->b[newsz] = '\0';
+		p->flags &= ~PFLAG_SPC;
+		return;
 	}
 
-	if (p->tree->flags & TREE_CLOSED &&
-	    p->cur->parent == p->tree->root)
+	if (p->tree->flags & TREE_CLOSED && n == p->tree->root)
 		warn_msg(p, "text after end of document: %.*s", sz, word);
 
-	/* Append to the current text node. */
+	/* Create a new text node. */
 
-	assert(sz >= 0);
-	newsz = p->cur->bsz + (p->cur->bsz && (p->flags & PFLAG_SPC)) + sz;
-	if ((p->cur->b = realloc(p->cur->b, newsz + 1)) == NULL)
+	if ((n = pnode_alloc(p->cur)) == NULL)
 		fatal(p);
-	if (p->cur->bsz && (p->flags & PFLAG_SPC))
-		p->cur->b[p->cur->bsz++] = ' ';
-	memcpy(p->cur->b + p->cur->bsz, word, sz);
-	p->cur->b[p->cur->bsz = newsz] = '\0';
-	p->cur->real = p->cur->b;
+	n->node = NODE_TEXT;
+	n->spc = (p->flags & PFLAG_SPC) != 0;
 	p->flags &= ~PFLAG_SPC;
+
+	/*
+	 * If this node follows a non-text node without intervening
+	 * whitespace, keep the text in it as short as possible,
+	 * and do not keep it open.
+	 */
+
+	if (n->spc == 0 &&
+	    (np = TAILQ_PREV(n, pnodeq, child)) != NULL &&
+	    np->node != NODE_TEXT && np->node != NODE_ESCAPE) {
+		i = 0;
+		while (i < sz && !isspace((unsigned char)word[i]))
+			i++;
+		if ((n->b = strndup(word, i)) == NULL)
+			fatal(p);
+		if (i == sz)
+			return;
+		while (i < sz && isspace((unsigned char)word[i]))
+			i++;
+		if (i == sz) {
+			p->flags |= PFLAG_SPC;
+			return;
+		}
+
+		/* Put any remaining text into a second node. */
+
+		if ((n = pnode_alloc(p->cur)) == NULL)
+			fatal(p);
+		n->node = NODE_TEXT;
+		n->spc = 1;
+		word += i;
+		sz -= i;
+	}
+	if ((n->b = strndup(word, sz)) == NULL)
+		fatal(p);
+
+	/* The new node remains open for later pnode_closetext(). */
+
+	p->cur = n;
 }
 
 /*
@@ -372,14 +413,15 @@ static void
 pnode_closetext(struct parse *p)
 {
 	struct pnode	*n;
+	char		*cp;
 
 	if ((n = p->cur) == NULL || n->node != NODE_TEXT)
 		return;
 	p->cur = n->parent;
-	while (n->bsz > 0 && isspace((unsigned char)n->b[n->bsz - 1])) {
-		n->b[--n->bsz] = '\0';
+	for (cp = strchr(n->b, '\0');
+	    cp > n->b && isspace((unsigned char)cp[-1]);
+	    *--cp = '\0')
 		p->flags |= PFLAG_SPC;
-	}
 }
 
 static void
@@ -438,16 +480,11 @@ xml_entity(struct parse *p, const char *name)
 	}
 
 	/* Create, append, and close out an entity node. */
-	if ((n = calloc(1, sizeof(*n))) == NULL ||
-	    (n->b = n->real = strdup(entity->roff)) == NULL)
+	if ((n = pnode_alloc(p->cur)) == NULL ||
+	    (n->b = strdup(entity->roff)) == NULL)
 		fatal(p);
 	n->node = NODE_ESCAPE;
-	n->bsz = strlen(n->b);
 	n->spc = (p->flags & PFLAG_SPC) != 0;
-	n->parent = p->cur;
-	TAILQ_INIT(&n->childq);
-	TAILQ_INIT(&n->attrq);
-	TAILQ_INSERT_TAIL(&p->cur->childq, n, child);
 	p->flags &= ~PFLAG_SPC;
 }
 
@@ -503,7 +540,7 @@ xml_elem_start(struct parse *p, const char *name)
 	if (p->tree->flags & TREE_CLOSED && p->cur->parent == NULL)
 		warn_msg(p, "element after end of document: <%s>", name);
 
-	if ((n = calloc(1, sizeof(*n))) == NULL)
+	if ((n = pnode_alloc(p->cur)) == NULL)
 		fatal(p);
 
 	/*
@@ -555,13 +592,6 @@ xml_elem_start(struct parse *p, const char *name)
 		n->spc = (p->flags & PFLAG_SPC) != 0;
 		break;
 	}
-	n->parent = p->cur;
-	TAILQ_INIT(&n->childq);
-	TAILQ_INIT(&n->attrq);
-
-	if (p->cur != NULL)
-		TAILQ_INSERT_TAIL(&p->cur->childq, n, child);
-
 	p->cur = n;
 	if (n->node == NODE_DOCTYPE) {
 		if (p->doctype == NULL)
@@ -680,6 +710,7 @@ xml_elem_end(struct parse *p, const char *name)
 		p->flags &= ~PFLAG_SPC;
 		break;
 	case NODE_DOCTYPE:
+	case NODE_SBR:
 		p->flags &= ~PFLAG_EEND;
 		/* FALLTHROUGH */
 	default:
@@ -970,9 +1001,11 @@ parse_string(struct parse *p, char *b, size_t rlen,
 
 		} else {
 			advance(p, b, rlen, &pend,
-			    p->ncur == NODE_DOCTYPE ? "<&]" : "<&",
+			    p->ncur == NODE_DOCTYPE ? "<&]\n" : "<&\n",
 			    refill);
-			xml_char(p, b + poff, pend - poff);
+			xml_text(p, b + poff, pend - poff);
+			if (b[pend] == '\n')
+				pnode_closetext(p);
 		}
 	}
 	return poff;