===================================================================
RCS file: /cvs/docbook2mdoc/parse.c,v
retrieving revision 1.15
retrieving revision 1.19
diff -u -p -r1.15 -r1.19
--- docbook2mdoc/parse.c	2019/04/06 13:45:58	1.15
+++ docbook2mdoc/parse.c	2019/04/07 17:39:26	1.19
@@ -1,4 +1,4 @@
-/* $Id: parse.c,v 1.15 2019/04/06 13:45:58 schwarze Exp $ */
+/* $Id: parse.c,v 1.19 2019/04/07 17:39:26 schwarze Exp $ */
 /*
  * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2019 Ingo Schwarze <schwarze@openbsd.org>
@@ -52,6 +52,7 @@ struct	parse {
 	int		 nline;  /* Line number of next token. */
 	int		 ncol;   /* Column number of next token. */
 	int		 del;    /* Levels of nested nodes being deleted. */
+	int		 spc;	 /* Whitespace before the next element. */
 	int		 attr;   /* The most recent attribute is valid. */
 	int		 warn;
 };
@@ -189,7 +190,7 @@ static	const struct element elements[] = {
 	{ "title",		NODE_TITLE },
 	{ "trademark",		NODE_IGNORE },
 	{ "type",		NODE_TYPE },
-	{ "ulink",		NODE_ULINK },
+	{ "ulink",		NODE_LINK },
 	{ "userinput",		NODE_LITERAL },
 	{ "variablelist",	NODE_VARIABLELIST },
 	{ "varlistentry",	NODE_VARLISTENTRY },
@@ -294,6 +295,7 @@ static void
 xml_char(struct parse *ps, const char *p, int sz)
 {
 	struct pnode	*dat;
+	size_t		 newsz;
 
 	if (ps->del > 0)
 		return;
@@ -309,6 +311,7 @@ xml_char(struct parse *ps, const char *p, int sz)
 			exit(1);
 		}
 		dat->node = NODE_TEXT;
+		dat->spc = ps->spc;
 		dat->parent = ps->cur;
 		TAILQ_INIT(&dat->childq);
 		TAILQ_INIT(&dat->attrq);
@@ -323,24 +326,35 @@ xml_char(struct parse *ps, const char *p, int sz)
 	/* Append to the current text node. */
 
 	assert(sz >= 0);
-	ps->cur->b = realloc(ps->cur->b, ps->cur->bsz + sz + 1);
+	newsz = ps->cur->bsz + (ps->cur->bsz && ps->spc) + sz;
+	ps->cur->b = realloc(ps->cur->b, newsz + 1);
 	if (ps->cur->b == NULL) {
 		perror(NULL);
 		exit(1);
 	}
+	if (ps->cur->bsz && ps->spc)
+		ps->cur->b[ps->cur->bsz++] = ' ';
 	memcpy(ps->cur->b + ps->cur->bsz, p, sz);
-	ps->cur->bsz += sz;
-	ps->cur->b[ps->cur->bsz] = '\0';
+	ps->cur->b[ps->cur->bsz = newsz] = '\0';
 	ps->cur->real = ps->cur->b;
+	ps->spc = 0;
 }
 
+/*
+ * Close out the text node and strip trailing whitespace, if one is open.
+ */
 static void
-pnode_trim(struct pnode *pn)
+pnode_closetext(struct parse *p)
 {
-	assert(pn->node == NODE_TEXT);
-	for (; pn->bsz > 0; pn->b[--pn->bsz] = '\0')
-		if (isspace((unsigned char)pn->b[pn->bsz - 1]) == 0)
-			break;
+	struct pnode	*n;
+
+	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';
+		p->spc = 1;
+	}
 }
 
 static void
@@ -357,11 +371,7 @@ xml_entity(struct parse *p, const char *name)
 		return;
 	}
 
-	/* Close out the text node, if there is one. */
-	if (p->cur->node == NODE_TEXT) {
-		pnode_trim(p->cur);
-		p->cur = p->cur->parent;
-	}
+	pnode_closetext(p);
 
 	if (p->tree->flags & TREE_CLOSED && p->cur == p->tree->root)
 		warn_msg(p, "entity after end of document: &%s;", name);
@@ -383,10 +393,12 @@ xml_entity(struct parse *p, const char *name)
 	}
 	dat->node = NODE_ESCAPE;
 	dat->bsz = strlen(dat->b);
+	dat->spc = p->spc;
 	dat->parent = p->cur;
 	TAILQ_INIT(&dat->childq);
 	TAILQ_INIT(&dat->attrq);
 	TAILQ_INSERT_TAIL(&p->cur->childq, dat, child);
+	p->spc = 0;
 }
 
 /*
@@ -410,11 +422,7 @@ xml_elem_start(struct parse *ps, const char *name)
 		return;
 	}
 
-	/* Close out the text node, if there is one. */
-	if (ps->cur != NULL && ps->cur->node == NODE_TEXT) {
-		pnode_trim(ps->cur);
-		ps->cur = ps->cur->parent;
-	}
+	pnode_closetext(ps);
 
 	for (elem = elements; elem->name != NULL; elem++)
 		if (strcmp(elem->name, name) == 0)
@@ -448,7 +456,50 @@ xml_elem_start(struct parse *ps, const char *name)
 		perror(NULL);
 		exit(1);
 	}
-	dat->node = elem->node;
+
+	/*
+	 * Nodes that begin a new macro or request line or start by
+	 * printing text always want whitespace before themselves.
+	 */
+
+	switch (dat->node = elem->node) {
+	case NODE_AUTHORGROUP:
+	case NODE_BOOKINFO:
+	case NODE_CAUTION:
+	case NODE_EDITOR:
+	case NODE_ENTRY:
+	case NODE_FUNCDEF:
+	case NODE_FUNCPROTOTYPE:
+	case NODE_INFORMALEQUATION:
+	case NODE_INLINEEQUATION:
+	case NODE_ITEMIZEDLIST:
+	case NODE_LEGALNOTICE:
+	case NODE_LITERALLAYOUT:
+	case NODE_NOTE:
+	case NODE_ORDEREDLIST:
+	case NODE_PARA:
+	case NODE_PREFACE:
+	case NODE_PROGRAMLISTING:
+	case NODE_REFMETA:
+	case NODE_REFNAMEDIV:
+	case NODE_REFSYNOPSISDIV:
+	case NODE_ROW:
+	case NODE_SBR:
+	case NODE_SCREEN:
+	case NODE_SECTION:
+	case NODE_SYNOPSIS:
+	case NODE_TGROUP:
+	case NODE_TIP:
+	case NODE_TITLE:
+	case NODE_VARIABLELIST:
+	case NODE_VARLISTENTRY:
+	case NODE_WARNING:
+		dat->spc = 1;
+		break;
+	default:
+		dat->spc = ps->spc;
+		break;
+	}
 	dat->parent = ps->cur;
 	TAILQ_INIT(&dat->childq);
 	TAILQ_INIT(&dat->attrq);
@@ -467,7 +518,7 @@ xml_attrkey(struct parse *ps, const char *name)
 	struct pattr	*attr;
 	enum attrkey	 key;
 
-	if (ps->del > 0 || *name == '\0')
+	if (ps->del > 0 || ps->ncur == NODE_IGNORE || *name == '\0')
 		return;
 	if ((key = attrkey_parse(name)) == ATTRKEY__MAX) {
 		ps->attr = 0;
@@ -489,7 +540,7 @@ xml_attrval(struct parse *ps, const char *name)
 {
 	struct pattr	*attr;
 
-	if (ps->del > 0 || ps->attr == 0)
+	if (ps->del > 0 || ps->ncur == NODE_IGNORE || ps->attr == 0)
 		return;
 	if ((attr = TAILQ_LAST(&ps->cur->attrq, pattrq)) == NULL)
 		return;
@@ -519,11 +570,8 @@ xml_elem_end(struct parse *ps, const char *name)
 		return;
 	}
 
-	/* Close out the text node, if there is one. */
-	if (ps->del == 0 && ps->cur != NULL && ps->cur->node == NODE_TEXT) {
-		pnode_trim(ps->cur);
-		ps->cur = ps->cur->parent;
-	}
+	if (ps->del == 0)
+		pnode_closetext(ps);
 
 	if (name != NULL) {
 		for (elem = elements; elem->name != NULL; elem++)
@@ -558,6 +606,7 @@ xml_elem_end(struct parse *ps, const char *name)
 			ps->tree->flags |= TREE_CLOSED;
 		else
 			ps->cur = ps->cur->parent;
+		ps->spc = 0;
 		break;
 	}
 	assert(ps->del == 0);
@@ -651,6 +700,7 @@ parse_string(struct parse *p, char *b, size_t rlen,
 	size_t		 pend;  /* Offset of the end of the current word. */
 	int		 elem_end;
 
+	p->spc = 0;
 	pend = 0;
 	for (;;) {
 
@@ -663,6 +713,7 @@ parse_string(struct parse *p, char *b, size_t rlen,
 		if ((poff = pend) == rlen)
 			break;
 		if (isspace((unsigned char)b[pend])) {
+			p->spc = 1;
 			increment(p, b, &pend, refill);
 			continue;
 		}
@@ -836,10 +887,7 @@ parse_file(struct parse *p, int fd, const char *fname)
 		perror(fname);
 		p->tree->flags |= TREE_FAIL;
 	}
-	if (p->cur != NULL && p->cur->node == NODE_TEXT) {
-		pnode_trim(p->cur);
-		p->cur = p->cur->parent;
-	}
+	pnode_closetext(p);
 	if ((p->tree->flags & TREE_CLOSED) == 0)
 		warn_msg(p, "document not closed");
 	return p->tree;