===================================================================
RCS file: /cvs/mandoc/mdoc_html.c,v
retrieving revision 1.248
retrieving revision 1.255
diff -u -p -r1.248 -r1.255
--- mandoc/mdoc_html.c	2017/01/17 01:47:51	1.248
+++ mandoc/mdoc_html.c	2017/01/19 16:59:30	1.255
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_html.c,v 1.248 2017/01/17 01:47:51 schwarze Exp $ */
+/*	$Id: mdoc_html.c,v 1.255 2017/01/19 16:59:30 schwarze Exp $ */
 /*
  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2014, 2015, 2016, 2017 Ingo Schwarze <schwarze@openbsd.org>
@@ -48,6 +48,7 @@ struct	htmlmdoc {
 	void		(*post)(MDOC_ARGS);
 };
 
+static	char		 *make_id(const struct roff_node *);
 static	void		  print_mdoc_head(MDOC_ARGS);
 static	void		  print_mdoc_node(MDOC_ARGS);
 static	void		  print_mdoc_nodelist(MDOC_ARGS);
@@ -299,42 +300,47 @@ void
 html_mdoc(void *arg, const struct roff_man *mdoc)
 {
 	struct html	*h;
-	struct tag	*t, *tt;
+	struct tag	*t;
 
 	h = (struct html *)arg;
 
-	if ( ! (HTML_FRAGMENT & h->oflags)) {
+	if ((h->oflags & HTML_FRAGMENT) == 0) {
 		print_gen_decls(h);
-		t = print_otag(h, TAG_HTML, "");
-		tt = print_otag(h, TAG_HEAD, "");
+		print_otag(h, TAG_HTML, "");
+		t = print_otag(h, TAG_HEAD, "");
 		print_mdoc_head(&mdoc->meta, mdoc->first->child, h);
-		print_tagq(h, tt);
+		print_tagq(h, t);
 		print_otag(h, TAG_BODY, "");
-		print_otag(h, TAG_DIV, "c", "mandoc");
-	} else
-		t = print_otag(h, TAG_DIV, "c", "mandoc");
+	}
 
 	mdoc_root_pre(&mdoc->meta, mdoc->first->child, h);
+	t = print_otag(h, TAG_DIV, "c", "manual-text");
 	print_mdoc_nodelist(&mdoc->meta, mdoc->first->child, h);
-	mdoc_root_post(&mdoc->meta, mdoc->first->child, h);
 	print_tagq(h, t);
-	putchar('\n');
+	mdoc_root_post(&mdoc->meta, mdoc->first->child, h);
+	print_tagq(h, NULL);
 }
 
 static void
 print_mdoc_head(MDOC_ARGS)
 {
+	char	*cp;
 
 	print_gen_head(h);
-	bufinit(h);
-	bufcat(h, meta->title);
-	if (meta->msec)
-		bufcat_fmt(h, "(%s)", meta->msec);
-	if (meta->arch)
-		bufcat_fmt(h, " (%s)", meta->arch);
 
+	if (meta->arch != NULL && meta->msec != NULL)
+		mandoc_asprintf(&cp, "%s(%s) (%s)", meta->title,
+		    meta->msec, meta->arch);
+	else if (meta->msec != NULL)
+		mandoc_asprintf(&cp, "%s(%s)", meta->title, meta->msec);
+	else if (meta->arch != NULL)
+		mandoc_asprintf(&cp, "%s (%s)", meta->title, meta->arch);
+	else
+		cp = mandoc_strdup(meta->title);
+
 	print_otag(h, TAG_TITLE, "");
-	print_text(h, h->buf);
+	print_text(h, cp);
+	free(cp);
 }
 
 static void
@@ -379,8 +385,6 @@ print_mdoc_node(MDOC_ARGS)
 			h->flags |= HTML_NOSPACE;
 		return;
 	case ROFFT_EQN:
-		if (n->flags & NODE_LINE)
-			putchar('\n');
 		print_eqn(h, n->eqn);
 		break;
 	case ROFFT_TBL:
@@ -489,12 +493,35 @@ mdoc_root_pre(MDOC_ARGS)
 	return 1;
 }
 
+static char *
+make_id(const struct roff_node *n)
+{
+	const struct roff_node	*nch;
+	char			*buf, *cp;
+
+	for (nch = n->child; nch != NULL; nch = nch->next)
+		if (nch->type != ROFFT_TEXT)
+			return NULL;
+
+	buf = NULL;
+	deroff(&buf, n);
+
+	/* http://www.w3.org/TR/html5/dom.html#the-id-attribute */
+
+	for (cp = buf; *cp != '\0'; cp++)
+		if (*cp == ' ')
+			*cp = '_';
+
+	return buf;
+}
+
 static int
 mdoc_sh_pre(MDOC_ARGS)
 {
+	char	*id;
+
 	switch (n->type) {
 	case ROFFT_BLOCK:
-		print_otag(h, TAG_DIV, "c", "section");
 		return 1;
 	case ROFFT_BODY:
 		if (n->sec == SEC_AUTHORS)
@@ -504,44 +531,29 @@ mdoc_sh_pre(MDOC_ARGS)
 		break;
 	}
 
-	bufinit(h);
+	if ((id = make_id(n)) != NULL) {
+		print_otag(h, TAG_H1, "ci", "Sh", id);
+		free(id);
+	} else
+		print_otag(h, TAG_H1, "c", "Sh");
 
-	for (n = n->child; n != NULL && n->type == ROFFT_TEXT; ) {
-		bufcat_id(h, n->string);
-		if (NULL != (n = n->next))
-			bufcat_id(h, " ");
-	}
-
-	if (NULL == n)
-		print_otag(h, TAG_H1, "i", h->buf);
-	else
-		print_otag(h, TAG_H1, "");
-
 	return 1;
 }
 
 static int
 mdoc_ss_pre(MDOC_ARGS)
 {
-	if (n->type == ROFFT_BLOCK) {
-		print_otag(h, TAG_DIV, "c", "subsection");
+	char	*id;
+
+	if (n->type != ROFFT_HEAD)
 		return 1;
-	} else if (n->type == ROFFT_BODY)
-		return 1;
 
-	bufinit(h);
+	if ((id = make_id(n)) != NULL) {
+		print_otag(h, TAG_H2, "ci", "Ss", id);
+		free(id);
+	} else
+		print_otag(h, TAG_H2, "c", "Ss");
 
-	for (n = n->child; n != NULL && n->type == ROFFT_TEXT; ) {
-		bufcat_id(h, n->string);
-		if (NULL != (n = n->next))
-			bufcat_id(h, " ");
-	}
-
-	if (NULL == n)
-		print_otag(h, TAG_H2, "i", h->buf);
-	else
-		print_otag(h, TAG_H2, "");
-
 	return 1;
 }
 
@@ -623,13 +635,12 @@ mdoc_xr_pre(MDOC_ARGS)
 	if (NULL == n->child)
 		return 0;
 
-	if (h->base_man) {
-		buffmt_man(h, n->child->string,
-		    n->child->next ?
-		    n->child->next->string : NULL);
-		print_otag(h, TAG_A, "ch", "link-man", h->buf);
-	} else
-		print_otag(h, TAG_A, "c", "link-man");
+	if (h->base_man)
+		print_otag(h, TAG_A, "chM", "Xr",
+		    n->child->string, n->child->next == NULL ?
+		    NULL : n->child->next->string);
+	else
+		print_otag(h, TAG_A, "c", "Xr");
 
 	n = n->child;
 	print_text(h, n->string);
@@ -835,32 +846,25 @@ mdoc_d1_pre(MDOC_ARGS)
 	if (n->type != ROFFT_BLOCK)
 		return 1;
 
-	print_otag(h, TAG_BLOCKQUOTE, "svtvb", 0, 0);
+	print_otag(h, TAG_DIV, "c", "D1");
 
-	/* BLOCKQUOTE needs a block body. */
+	if (n->tok == MDOC_Dl)
+		print_otag(h, TAG_CODE, "c", "Li");
 
-	print_otag(h, TAG_DIV, "c", "display");
-
-	if (MDOC_Dl == n->tok)
-		print_otag(h, TAG_CODE, "c", "lit");
-
 	return 1;
 }
 
 static int
 mdoc_sx_pre(MDOC_ARGS)
 {
-	bufinit(h);
-	bufcat(h, "#");
+	char	*id;
 
-	for (n = n->child; n; ) {
-		bufcat_id(h, n->string);
-		if (NULL != (n = n->next))
-			bufcat_id(h, " ");
-	}
+	if ((id = make_id(n)) != NULL) {
+		print_otag(h, TAG_A, "chR", "Sx", id);
+		free(id);
+	} else
+		print_otag(h, TAG_A, "c", "Sx");
 
-	print_otag(h, TAG_I, "c", "link-sec");
-	print_otag(h, TAG_A, "ch", "link-sec", h->buf);
 	return 1;
 }
 
@@ -901,15 +905,15 @@ mdoc_bd_pre(MDOC_ARGS)
 		offs = -1;
 
 	if (offs == -1)
-		print_otag(h, TAG_DIV, "cswl", "display", n->norm->Bd.offs);
+		print_otag(h, TAG_DIV, "cswl", "Bd", n->norm->Bd.offs);
 	else
-		print_otag(h, TAG_DIV, "cshl", "display", offs);
+		print_otag(h, TAG_DIV, "cshl", "Bd", offs);
 
 	if (n->norm->Bd.type != DISP_unfilled &&
 	    n->norm->Bd.type != DISP_literal)
 		return 1;
 
-	print_otag(h, TAG_PRE, "c", "lit");
+	print_otag(h, TAG_PRE, "c", "Li");
 
 	/* This can be recursive: save & set our literal state. */
 
@@ -1051,9 +1055,8 @@ mdoc_fa_pre(MDOC_ARGS)
 static int
 mdoc_fd_pre(MDOC_ARGS)
 {
-	char		 buf[BUFSIZ];
-	size_t		 sz;
 	struct tag	*t;
+	char		*buf, *cp;
 
 	synopsis_pre(h, n);
 
@@ -1073,25 +1076,16 @@ mdoc_fd_pre(MDOC_ARGS)
 	if (NULL != (n = n->next)) {
 		assert(n->type == ROFFT_TEXT);
 
-		/*
-		 * XXX This is broken and not easy to fix.
-		 * When using -Oincludes, truncation may occur.
-		 * Dynamic allocation wouldn't help because
-		 * passing long strings to buffmt_includes()
-		 * does not work either.
-		 */
-
-		strlcpy(buf, '<' == *n->string || '"' == *n->string ?
-		    n->string + 1 : n->string, BUFSIZ);
-
-		sz = strlen(buf);
-		if (sz && ('>' == buf[sz - 1] || '"' == buf[sz - 1]))
-			buf[sz - 1] = '\0';
-
 		if (h->base_includes) {
-			buffmt_includes(h, buf);
-			t = print_otag(h, TAG_A, "ch", "link-includes",
-			    h->buf);
+			cp = n->string;
+			if (*cp == '<' || *cp == '"')
+				cp++;
+			buf = mandoc_strdup(cp);
+			cp = strchr(buf, '\0') - 1;
+			if (cp >= buf && (*cp == '>' || *cp == '"'))
+				*cp = '\0';
+			t = print_otag(h, TAG_A, "chI", "link-includes", buf);
+			free(buf);
 		} else
 			t = print_otag(h, TAG_A, "c", "link-includes");
 
@@ -1279,16 +1273,16 @@ static int
 mdoc_mt_pre(MDOC_ARGS)
 {
 	struct tag	*t;
+	char		*cp;
 
 	for (n = n->child; n; n = n->next) {
 		assert(n->type == ROFFT_TEXT);
 
-		bufinit(h);
-		bufcat(h, "mailto:");
-		bufcat(h, n->string);
-		t = print_otag(h, TAG_A, "ch", "link-mail", h->buf);
+		mandoc_asprintf(&cp, "mailto:%s", n->string);
+		t = print_otag(h, TAG_A, "ch", "link-mail", cp);
 		print_text(h, n->string);
 		print_tagq(h, t);
+		free(cp);
 	}
 
 	return 0;
@@ -1355,11 +1349,10 @@ mdoc_in_pre(MDOC_ARGS)
 	if (NULL != (n = n->child)) {
 		assert(n->type == ROFFT_TEXT);
 
-		if (h->base_includes) {
-			buffmt_includes(h, n->string);
-			t = print_otag(h, TAG_A, "ch", "link-includes",
-			    h->buf);
-		} else
+		if (h->base_includes)
+			t = print_otag(h, TAG_A, "chI", "link-includes",
+			    n->string);
+		else
 			t = print_otag(h, TAG_A, "c", "link-includes");
 		print_text(h, n->string);
 		print_tagq(h, t);
@@ -1417,7 +1410,7 @@ mdoc_bf_pre(MDOC_ARGS)
 	else if (FONT_Sy == n->norm->Bf.font)
 		cattr = "symb";
 	else if (FONT_Li == n->norm->Bf.font)
-		cattr = "lit";
+		cattr = "Li";
 	else
 		cattr = "none";
 
@@ -1476,7 +1469,7 @@ mdoc_no_pre(MDOC_ARGS)
 static int
 mdoc_li_pre(MDOC_ARGS)
 {
-	print_otag(h, TAG_CODE, "c", "lit");
+	print_otag(h, TAG_CODE, "c", "Li");
 	return 1;
 }
 
@@ -1659,7 +1652,7 @@ mdoc_quote_pre(MDOC_ARGS)
 	case MDOC_Ql:
 		print_text(h, "\\(oq");
 		h->flags |= HTML_NOSPACE;
-		print_otag(h, TAG_CODE, "c", "lit");
+		print_otag(h, TAG_CODE, "c", "Li");
 		break;
 	case MDOC_So:
 	case MDOC_Sq: