Represent mdoc(7) .Pp (and .sp, and some SYNOPSIS and .Rs features)

by the <p> HTML element and use the html_fillmode() mechanism
for .Bd -unfilled, just like it was done for man(7) earlier, finally
getting rid both of the horrible <div class="Pp"></div> hack and
of the worst HTML syntax violations caused by nested displays.

Care is needed because in some situations, paragraphs have to remain
open across several subsequent macros, whereas in other situations,
they must get closed together with a block containing them.

Some implementation details include:
* Always close paragraphs before emitting HTML flow content.
* Let html_close_paragraph() also close <pre> for extra safety.
* Drop the old, now unused function print_paragraph().
* Minor adjustments in the top-level man(7) node formatter for symmetry.
* Bugfix: .Ss heads suspend no-fill mode, even though .Ss doesn't end it.
* Bugfix: give up on .Op semantic markup for now, see the comment.

1 files changed, 132 insertions, 94 deletions

diff --git a/mdoc_html.c b/mdoc_html.c
index e4523f95..2978f0c8 100644
--- a/mdoc_html.c
+++ b/mdoc_html.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_html.c,v 1.322 2018/12/31 10:35:56 schwarze Exp $ */
+/*	$Id: mdoc_html.c,v 1.323 2019/01/07 07:26:29 schwarze Exp $ */
 /*
  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2014,2015,2016,2017,2018 Ingo Schwarze <schwarze@openbsd.org>
@@ -268,18 +268,17 @@ synopsis_pre(struct html *h, const struct roff_node *n)
 	case MDOC_Fo:
 	case MDOC_In:
 	case MDOC_Vt:
-		print_paragraph(h);
 		break;
 	case MDOC_Ft:
-		if (MDOC_Fn != n->tok && MDOC_Fo != n->tok) {
-			print_paragraph(h);
+		if (n->tok != MDOC_Fn && n->tok != MDOC_Fo)
 			break;
-		}
 		/* FALLTHROUGH */
 	default:
 		print_otag(h, TAG_BR, "");
-		break;
+		return;
 	}
+	html_close_paragraph(h);
+	print_otag(h, TAG_P, "c", "Pp");
 }
 
 void
@@ -346,16 +345,20 @@ print_mdoc_nodelist(MDOC_ARGS)
 static void
 print_mdoc_node(MDOC_ARGS)
 {
-	int		 child;
 	struct tag	*t;
+	int		 child;
 
 	if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
 		return;
 
+	html_fillmode(h, n->flags & NODE_NOFILL ? ROFF_nf : ROFF_fi);
+
 	child = 1;
 	t = h->tag;
-	n->flags &= ~NODE_ENDED;
+	if (t->tag == TAG_P || t->tag == TAG_PRE)
+		t = t->next;
 
+	n->flags &= ~NODE_ENDED;
 	switch (n->type) {
 	case ROFFT_TEXT:
 		/* No tables in this mode... */
@@ -374,7 +377,7 @@ print_mdoc_node(MDOC_ARGS)
 		print_text(h, n->string);
 		if (NODE_DELIMO & n->flags)
 			h->flags |= HTML_NOSPACE;
-		return;
+		break;
 	case ROFFT_EQN:
 		print_eqn(h, n->eqn);
 		break;
@@ -399,8 +402,8 @@ print_mdoc_node(MDOC_ARGS)
 		assert(h->tblt == NULL);
 		if (n->tok < ROFF_MAX) {
 			roff_html_pre(h, n);
-			child = 0;
-			break;
+			print_stagq(h, t);
+			return;
 		}
 		assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
 		if (mdoc_html_acts[n->tok - MDOC_Dd].pre != NULL &&
@@ -415,17 +418,17 @@ print_mdoc_node(MDOC_ARGS)
 		h->flags |= HTML_PREKEEP;
 	}
 
-	if (child && n->child)
+	if (child && n->child != NULL)
 		print_mdoc_nodelist(meta, n->child, h);
 
 	print_stagq(h, t);
 
 	switch (n->type) {
+	case ROFFT_TEXT:
 	case ROFFT_EQN:
 		break;
 	default:
-		if (n->tok < ROFF_MAX ||
-		    mdoc_html_acts[n->tok - MDOC_Dd].post == NULL ||
+		if (mdoc_html_acts[n->tok - MDOC_Dd].post == NULL ||
 		    n->flags & NODE_ENDED)
 			break;
 		(*mdoc_html_acts[n->tok - MDOC_Dd].post)(meta, n, h);
@@ -433,6 +436,12 @@ print_mdoc_node(MDOC_ARGS)
 			n->body->flags |= NODE_ENDED;
 		break;
 	}
+
+	if (n->flags & NODE_NOFILL &&
+	    (n->next == NULL || n->next->flags & NODE_LINE)) {
+		h->col++;
+		print_endline(h);
+	}
 }
 
 static void
@@ -516,6 +525,7 @@ mdoc_sh_pre(MDOC_ARGS)
 
 	switch (n->type) {
 	case ROFFT_BLOCK:
+		html_close_paragraph(h);
 		if ((h->oflags & HTML_TOC) == 0 ||
 		    h->flags & HTML_TOCDONE ||
 		    n->sec <= SEC_SYNOPSIS)
@@ -582,8 +592,17 @@ mdoc_ss_pre(MDOC_ARGS)
 {
 	char	*id;
 
-	if (n->type != ROFFT_HEAD)
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
+		return 1;
+	case ROFFT_HEAD:
+		break;
+	case ROFFT_BODY:
 		return 1;
+	default:
+		abort();
+	}
 
 	id = html_make_id(n, 1);
 	print_otag(h, TAG_H2, "cTi", "Ss", id);
@@ -625,9 +644,17 @@ mdoc_cm_pre(MDOC_ARGS)
 static int
 mdoc_nd_pre(MDOC_ARGS)
 {
-	if (n->type != ROFFT_BODY)
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
 		return 1;
-
+	case ROFFT_HEAD:
+		return 0;
+	case ROFFT_BODY:
+		break;
+	default:
+		abort();
+	}
 	print_text(h, "\\(em");
 	/* Cannot use TAG_SPAN because it may contain blocks. */
 	print_otag(h, TAG_DIV, "cT", "Nd");
@@ -638,6 +665,8 @@ static int
 mdoc_nm_pre(MDOC_ARGS)
 {
 	switch (n->type) {
+	case ROFFT_BLOCK:
+		break;
 	case ROFFT_HEAD:
 		print_otag(h, TAG_TD, "");
 		/* FALLTHROUGH */
@@ -648,8 +677,9 @@ mdoc_nm_pre(MDOC_ARGS)
 		print_otag(h, TAG_TD, "");
 		return 1;
 	default:
-		break;
+		abort();
 	}
+	html_close_paragraph(h);
 	synopsis_pre(h, n);
 	print_otag(h, TAG_TABLE, "c", "Nm");
 	print_otag(h, TAG_TR, "");
@@ -802,12 +832,15 @@ mdoc_bl_pre(MDOC_ARGS)
 	enum htmltag	 elemtype;
 
 	switch (n->type) {
-	case ROFFT_BODY:
-		return 1;
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
+		break;
 	case ROFFT_HEAD:
 		return 0;
+	case ROFFT_BODY:
+		return 1;
 	default:
-		break;
+		abort();
 	}
 
 	bl = &n->norm->Bl;
@@ -891,14 +924,20 @@ mdoc_em_pre(MDOC_ARGS)
 static int
 mdoc_d1_pre(MDOC_ARGS)
 {
-	if (n->type != ROFFT_BLOCK)
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
+		break;
+	case ROFFT_HEAD:
+		return 0;
+	case ROFFT_BODY:
 		return 1;
-
+	default:
+		abort();
+	}
 	print_otag(h, TAG_DIV, "c", "Bd Bd-indent");
-
 	if (n->tok == MDOC_Dl)
 		print_otag(h, TAG_CODE, "c", "Li");
-
 	return 1;
 }
 
@@ -916,69 +955,45 @@ mdoc_sx_pre(MDOC_ARGS)
 static int
 mdoc_bd_pre(MDOC_ARGS)
 {
-	int			 comp;
+	char			 buf[16];
 	struct roff_node	*nn;
+	int			 comp;
 
-	if (n->type == ROFFT_HEAD)
-		return 0;
-
-	if (n->type == ROFFT_BLOCK) {
-		comp = n->norm->Bd.comp;
-		for (nn = n; nn && ! comp; nn = nn->parent) {
-			if (nn->type != ROFFT_BLOCK)
-				continue;
-			if (MDOC_Ss == nn->tok || MDOC_Sh == nn->tok)
-				comp = 1;
-			if (nn->prev)
-				break;
-		}
-		if ( ! comp)
-			print_paragraph(h);
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
 		return 1;
+	case ROFFT_HEAD:
+		return 0;
+	case ROFFT_BODY:
+		break;
+	default:
+		abort();
 	}
 
-	/* Handle the -offset argument. */
-
-	if (n->norm->Bd.offs == NULL ||
-	    ! strcmp(n->norm->Bd.offs, "left"))
-		print_otag(h, TAG_DIV, "c", "Bd");
-	else
-		print_otag(h, TAG_DIV, "c", "Bd Bd-indent");
-
-	if (n->norm->Bd.type != DISP_unfilled &&
-	    n->norm->Bd.type != DISP_literal)
-		return 1;
+	/* Handle preceding whitespace. */
 
-	print_otag(h, TAG_PRE, "c", "Li");
-	for (nn = n->child; nn; nn = nn->next) {
-		print_mdoc_node(meta, nn, h);
-		/*
-		 * If the printed node flushes its own line, then we
-		 * needn't do it here as well.  This is hacky, but the
-		 * notion of selective eoln whitespace is pretty dumb
-		 * anyway, so don't sweat it.
-		 */
-		switch (nn->tok) {
-		case ROFF_br:
-		case ROFF_sp:
-		case MDOC_Sm:
-		case MDOC_Bl:
-		case MDOC_D1:
-		case MDOC_Dl:
-		case MDOC_Pp:
+	comp = n->norm->Bd.comp;
+	for (nn = n; nn != NULL && comp == 0; nn = nn->parent) {
+		if (nn->type != ROFFT_BLOCK)
 			continue;
-		default:
+		if (nn->tok == MDOC_Sh || nn->tok == MDOC_Ss)
+			comp = 1;
+		if (nn->prev != NULL)
 			break;
-		}
-		if (h->flags & HTML_NONEWLINE ||
-		    (nn->next && ! (nn->next->flags & NODE_LINE)))
-			continue;
-		else if (nn->next)
-			print_text(h, "\n");
-
-		h->flags |= HTML_NOSPACE;
 	}
-	return 0;
+	(void)strlcpy(buf, "Bd", sizeof(buf));
+	if (comp == 0)
+		(void)strlcat(buf, " Pp", sizeof(buf));
+
+	/* Handle the -offset argument. */
+
+	if (n->norm->Bd.offs != NULL &&
+	    strcmp(n->norm->Bd.offs, "left") != 0)
+		(void)strlcat(buf, " Bd-indent", sizeof(buf));
+
+	print_otag(h, TAG_DIV, "c", buf);
+	return 1;
 }
 
 static int
@@ -1262,8 +1277,10 @@ mdoc_skip_pre(MDOC_ARGS)
 static int
 mdoc_pp_pre(MDOC_ARGS)
 {
-
-	print_paragraph(h);
+	if ((n->flags & NODE_NOFILL) == 0) {
+		html_close_paragraph(h);
+		print_otag(h, TAG_P, "c", "Pp");
+	}
 	return 0;
 }
 
@@ -1438,10 +1455,17 @@ mdoc_bf_pre(MDOC_ARGS)
 {
 	const char	*cattr;
 
-	if (n->type == ROFFT_HEAD)
-		return 0;
-	else if (n->type != ROFFT_BODY)
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
 		return 1;
+	case ROFFT_HEAD:
+		return 0;
+	case ROFFT_BODY:
+		break;
+	default:
+		abort();
+	}
 
 	if (FONT_Em == n->norm->Bf.font)
 		cattr = "Bf Em";
@@ -1487,13 +1511,21 @@ mdoc_pf_post(MDOC_ARGS)
 static int
 mdoc_rs_pre(MDOC_ARGS)
 {
-	if (n->type != ROFFT_BLOCK)
-		return 1;
-
-	if (n->prev && SEC_SEE_ALSO == n->sec)
-		print_paragraph(h);
-
-	print_otag(h, TAG_CITE, "cT", "Rs");
+	switch (n->type) {
+	case ROFFT_BLOCK:
+		if (n->sec == SEC_SEE_ALSO)
+			html_close_paragraph(h);
+		break;
+	case ROFFT_HEAD:
+		return 0;
+	case ROFFT_BODY:
+		if (n->sec == SEC_SEE_ALSO)
+			print_otag(h, TAG_P, "c", "Pp");
+		print_otag(h, TAG_CITE, "cT", "Rs");
+		break;
+	default:
+		abort();
+	}
 	return 1;
 }
 
@@ -1670,9 +1702,15 @@ mdoc_quote_pre(MDOC_ARGS)
 	case MDOC_Oo:
 	case MDOC_Op:
 		print_text(h, "\\(lB");
-		h->flags |= HTML_NOSPACE;
-		/* Cannot use TAG_SPAN because it may contain blocks. */
-		print_otag(h, TAG_IDIV, "c", "Op");
+		/*
+		 * Give up on semantic markup for now.
+		 * We cannot use TAG_SPAN because .Oo may contain blocks.
+		 * We cannot use TAG_IDIV because we might be in a
+		 * phrasing context (like .Dl or .Pp); we cannot
+		 * close out a .Pp at this point either because
+		 * that would break the line.
+		 */
+		/* XXX print_otag(h, TAG_???, "c", "Op"); */
 		break;
 	case MDOC_En:
 		if (NULL == n->norm->Es ||