=================================================================== RCS file: /cvs/mandoc/mdoc_html.c,v retrieving revision 1.308 retrieving revision 1.323 diff -u -p -r1.308 -r1.323 --- mandoc/mdoc_html.c 2018/06/25 14:06:07 1.308 +++ mandoc/mdoc_html.c 2019/01/07 07:26:29 1.323 @@ -1,4 +1,4 @@ -/* $Id: mdoc_html.c,v 1.308 2018/06/25 14:06:07 schwarze Exp $ */ +/* $Id: mdoc_html.c,v 1.323 2019/01/07 07:26:29 schwarze Exp $ */ /* * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons * Copyright (c) 2014,2015,2016,2017,2018 Ingo Schwarze @@ -34,8 +34,6 @@ #include "html.h" #include "main.h" -#define INDENT 5 - #define MDOC_ARGS const struct roff_meta *meta, \ struct roff_node *n, \ struct html *h @@ -44,7 +42,7 @@ #define MIN(a,b) ((/*CONSTCOND*/(a)<(b))?(a):(b)) #endif -struct htmlmdoc { +struct mdoc_html_act { int (*pre)(MDOC_ARGS); void (*post)(MDOC_ARGS); }; @@ -64,6 +62,7 @@ static int mdoc_root_pre(const struct roff_meta *, static void mdoc__x_post(MDOC_ARGS); static int mdoc__x_pre(MDOC_ARGS); +static int mdoc_abort_pre(MDOC_ARGS); static int mdoc_ad_pre(MDOC_ARGS); static int mdoc_an_pre(MDOC_ARGS); static int mdoc_ap_pre(MDOC_ARGS); @@ -121,7 +120,7 @@ static int mdoc_vt_pre(MDOC_ARGS); static int mdoc_xr_pre(MDOC_ARGS); static int mdoc_xx_pre(MDOC_ARGS); -static const struct htmlmdoc __mdocs[MDOC_MAX - MDOC_Dd] = { +static const struct mdoc_html_act mdoc_html_acts[MDOC_MAX - MDOC_Dd] = { {NULL, NULL}, /* Dd */ {NULL, NULL}, /* Dt */ {NULL, NULL}, /* Os */ @@ -156,7 +155,7 @@ static const struct htmlmdoc __mdocs[MDOC_MAX - MDOC_D {mdoc_nd_pre, NULL}, /* Nd */ {mdoc_nm_pre, NULL}, /* Nm */ {mdoc_quote_pre, mdoc_quote_post}, /* Op */ - {mdoc_ft_pre, NULL}, /* Ot */ + {mdoc_abort_pre, NULL}, /* Ot */ {mdoc_pa_pre, NULL}, /* Pa */ {mdoc_ex_pre, NULL}, /* Rv */ {mdoc_st_pre, NULL}, /* St */ @@ -229,7 +228,7 @@ static const struct htmlmdoc __mdocs[MDOC_MAX - MDOC_D {mdoc_em_pre, NULL}, /* Fr */ {NULL, NULL}, /* Ud */ {mdoc_lb_pre, NULL}, /* Lb */ - {mdoc_pp_pre, NULL}, /* Lp */ + {mdoc_abort_pre, NULL}, /* Lp */ {mdoc_lk_pre, NULL}, /* Lk */ {mdoc_mt_pre, NULL}, /* Mt */ {mdoc_quote_pre, mdoc_quote_post}, /* Brq */ @@ -243,7 +242,6 @@ static const struct htmlmdoc __mdocs[MDOC_MAX - MDOC_D {mdoc__x_pre, mdoc__x_post}, /* %U */ {NULL, NULL}, /* Ta */ }; -static const struct htmlmdoc *const mdocs = __mdocs - MDOC_Dd; /* @@ -270,22 +268,21 @@ 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 -html_mdoc(void *arg, const struct roff_man *mdoc) +html_mdoc(void *arg, const struct roff_meta *mdoc) { struct html *h; struct roff_node *n; @@ -300,16 +297,16 @@ html_mdoc(void *arg, const struct roff_man *mdoc) if (n->type == ROFFT_COMMENT) print_gen_comment(h, n); t = print_otag(h, TAG_HEAD, ""); - print_mdoc_head(&mdoc->meta, h); + print_mdoc_head(mdoc, h); print_tagq(h, t); print_otag(h, TAG_BODY, ""); } - mdoc_root_pre(&mdoc->meta, h); + mdoc_root_pre(mdoc, h); t = print_otag(h, TAG_DIV, "c", "manual-text"); - print_mdoc_nodelist(&mdoc->meta, n, h); + print_mdoc_nodelist(mdoc, n, h); print_tagq(h, t); - mdoc_root_post(&mdoc->meta, h); + mdoc_root_post(mdoc, h); print_tagq(h, NULL); } @@ -348,16 +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... */ @@ -368,14 +369,15 @@ print_mdoc_node(MDOC_ARGS) * (i.e., within a
) don't print the newline.
 		 */
 		if (*n->string == ' ' && n->flags & NODE_LINE &&
-		    (h->flags & (HTML_LITERAL | HTML_NONEWLINE)) == 0)
+		    (h->flags & HTML_NONEWLINE) == 0 &&
+		    (n->flags & NODE_NOFILL) == 0)
 			print_otag(h, TAG_BR, "");
 		if (NODE_DELIMC & n->flags)
 			h->flags |= HTML_NOSPACE;
 		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;
@@ -400,13 +402,14 @@ 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 (mdocs[n->tok].pre != NULL &&
+		if (mdoc_html_acts[n->tok - MDOC_Dd].pre != NULL &&
 		    (n->end == ENDBODY_NOT || n->child != NULL))
-			child = (*mdocs[n->tok].pre)(meta, n, h);
+			child = (*mdoc_html_acts[n->tok - MDOC_Dd].pre)(meta,
+			    n, h);
 		break;
 	}
 
@@ -415,24 +418,30 @@ 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 ||
-		    mdocs[n->tok].post == NULL ||
+		if (mdoc_html_acts[n->tok - MDOC_Dd].post == NULL ||
 		    n->flags & NODE_ENDED)
 			break;
-		(*mdocs[n->tok].post)(meta, n, h);
+		(*mdoc_html_acts[n->tok - MDOC_Dd].post)(meta, n, h);
 		if (n->end != ENDBODY_NOT)
 			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
@@ -509,9 +518,59 @@ cond_id(const struct roff_node *n)
 static int
 mdoc_sh_pre(MDOC_ARGS)
 {
-	char	*id;
+	struct roff_node	*sn, *subn;
+	struct tag		*t, *tsec, *tsub;
+	char			*id;
+	int			 sc;
 
 	switch (n->type) {
+	case ROFFT_BLOCK:
+		html_close_paragraph(h);
+		if ((h->oflags & HTML_TOC) == 0 ||
+		    h->flags & HTML_TOCDONE ||
+		    n->sec <= SEC_SYNOPSIS)
+			break;
+		h->flags |= HTML_TOCDONE;
+		sc = 0;
+		for (sn = n->next; sn != NULL; sn = sn->next)
+			if (sn->sec == SEC_CUSTOM)
+				if (++sc == 2)
+					break;
+		if (sc < 2)
+			break;
+		t = print_otag(h, TAG_H1, "c", "Sh");
+		print_text(h, "TABLE OF CONTENTS");
+		print_tagq(h, t);
+		t = print_otag(h, TAG_UL, "c", "Bl-compact");
+		for (sn = n; sn != NULL; sn = sn->next) {
+			tsec = print_otag(h, TAG_LI, "");
+			id = html_make_id(sn->head, 0);
+			tsub = print_otag(h, TAG_A, "hR", id);
+			free(id);
+			print_mdoc_nodelist(meta, sn->head->child, h);
+			print_tagq(h, tsub);
+			tsub = NULL;
+			for (subn = sn->body->child; subn != NULL;
+			    subn = subn->next) {
+				if (subn->tok != MDOC_Ss)
+					continue;
+				id = html_make_id(subn->head, 0);
+				if (id == NULL)
+					continue;
+				if (tsub == NULL)
+					print_otag(h, TAG_UL,
+					    "c", "Bl-compact");
+				tsub = print_otag(h, TAG_LI, "");
+				print_otag(h, TAG_A, "hR", id);
+				free(id);
+				print_mdoc_nodelist(meta,
+				    subn->head->child, h);
+				print_tagq(h, tsub);
+			}
+			print_tagq(h, tsec);
+		}
+		print_tagq(h, t);
+		break;
 	case ROFFT_HEAD:
 		id = html_make_id(n, 1);
 		print_otag(h, TAG_H1, "cTi", "Sh", id);
@@ -533,8 +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);
@@ -576,9 +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");
@@ -589,6 +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 */
@@ -599,8 +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, "");
@@ -613,7 +692,7 @@ mdoc_xr_pre(MDOC_ARGS)
 	if (NULL == n->child)
 		return 0;
 
-	if (h->base_man)
+	if (h->base_man1)
 		print_otag(h, TAG_A, "cThM", "Xr",
 		    n->child->string, n->child->next == NULL ?
 		    NULL : n->child->next->string);
@@ -719,8 +798,7 @@ mdoc_it_pre(MDOC_ARGS)
 			break;
 		case ROFFT_BODY:
 			if (n->child == NULL) {
-				print_otag(h, TAG_DD, "ss?",
-				    "width", "auto");
+				print_otag(h, TAG_DD, "s", "width", "auto");
 				print_text(h, "\\ ");
 			} else
 				print_otag(h, TAG_DD, "");
@@ -749,17 +827,20 @@ mdoc_it_pre(MDOC_ARGS)
 static int
 mdoc_bl_pre(MDOC_ARGS)
 {
-	char		 cattr[28];
+	char		 cattr[32];
 	struct mdoc_bl	*bl;
 	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;
@@ -843,14 +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;
 }
 
@@ -868,80 +955,45 @@ mdoc_sx_pre(MDOC_ARGS)
 static int
 mdoc_bd_pre(MDOC_ARGS)
 {
-	int			 comp, sv;
+	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. */
+	/* Handle preceding whitespace. */
 
-	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;
-
-	print_otag(h, TAG_PRE, "c", "Li");
-
-	/* This can be recursive: save & set our literal state. */
-
-	sv = h->flags & HTML_LITERAL;
-	h->flags |= HTML_LITERAL;
-
-	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_Lp:
-		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;
 	}
+	(void)strlcpy(buf, "Bd", sizeof(buf));
+	if (comp == 0)
+		(void)strlcat(buf, " Pp", sizeof(buf));
 
-	if (0 == sv)
-		h->flags &= ~HTML_LITERAL;
+	/* Handle the -offset argument. */
 
-	return 0;
+	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
@@ -1175,7 +1227,7 @@ mdoc_fn_pre(MDOC_ARGS)
 
 	for (n = n->child->next; n; n = n->next) {
 		if (NODE_SYNPRETTY & n->flags)
-			t = print_otag(h, TAG_VAR, "cTss?", "Fa",
+			t = print_otag(h, TAG_VAR, "cTs", "Fa",
 			    "white-space", "nowrap");
 		else
 			t = print_otag(h, TAG_VAR, "cT", "Fa");
@@ -1225,8 +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;
 }
 
@@ -1401,10 +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";
@@ -1450,13 +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;
 }
 
@@ -1633,9 +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 ||
@@ -1762,4 +1837,10 @@ mdoc_eo_post(MDOC_ARGS)
 		h->flags |= HTML_NOSPACE;
 	else if ( ! tail)
 		h->flags &= ~HTML_NOSPACE;
+}
+
+static int
+mdoc_abort_pre(MDOC_ARGS)
+{
+	abort();
 }