Copy tagged strings before marking hyphens as breakable.

For example, this makes ":tCo-processes" work in ksh(1).

17 files changed, 48 insertions, 29 deletions

diff --git a/regress/usr.bin/mandoc/mdoc/Em/tag.in b/regress/usr.bin/mandoc/mdoc/Em/tag.in
index 70e6eb447ec..eb6ac085160 100644
--- a/regress/usr.bin/mandoc/mdoc/Em/tag.in
+++ b/regress/usr.bin/mandoc/mdoc/Em/tag.in
@@ -1,5 +1,5 @@
-.\" $OpenBSD: tag.in,v 1.1 2020/03/13 00:31:05 schwarze Exp $
-.Dd $Mdocdate: March 13 2020 $
+.\" $OpenBSD: tag.in,v 1.2 2020/04/02 14:55:29 schwarze Exp $
+.Dd $Mdocdate: April 2 2020 $
 .Dt EM-TAG 1
 .Os
 .Sh NAME
@@ -11,7 +11,7 @@ BEGINTEST
 .It Em one | two
 text
 .It Xo
-.Em three
+.Em three-with-hyphens
 .Xc
 text
 .El
diff --git a/regress/usr.bin/mandoc/mdoc/Em/tag.out_ascii b/regress/usr.bin/mandoc/mdoc/Em/tag.out_ascii
index 74f8212bf71..283c1120b82 100644
--- a/regress/usr.bin/mandoc/mdoc/Em/tag.out_ascii
+++ b/regress/usr.bin/mandoc/mdoc/Em/tag.out_ascii
@@ -9,9 +9,10 @@ DDEESSCCRRIIPPTTIIOONN
      _o_n_e | _t_w_o
              text
 
-     _t_h_r_e_e   text
+     _t_h_r_e_e_-_w_i_t_h_-_h_y_p_h_e_n_s
+             text
      _f_o_u_r _o_n_e _f_i_v_e
 
      ENDTEST
 
-OpenBSD                         March 13, 2020                         OpenBSD
+OpenBSD                          April 2, 2020                         OpenBSD
diff --git a/regress/usr.bin/mandoc/mdoc/Em/tag.out_html b/regress/usr.bin/mandoc/mdoc/Em/tag.out_html
index 4046673f833..225343eebd0 100644
--- a/regress/usr.bin/mandoc/mdoc/Em/tag.out_html
+++ b/regress/usr.bin/mandoc/mdoc/Em/tag.out_html
@@ -2,7 +2,7 @@
   <dt><a class="permalink" href="#one"><i class="Em" id="one">one</i></a> |
     <a class="permalink" href="#two"><i class="Em" id="two">two</i></a></dt>
   <dd>text</dd>
-  <dt><a class="permalink" href="#three"><i class="Em" id="three">three</i></a></dt>
+  <dt><a class="permalink" href="#three-with-hyphens"><i class="Em" id="three-with-hyphens">three-with-hyphens</i></a></dt>
   <dd>text</dd>
 </dl>
 <a class="permalink" href="#four"><i class="Em" id="four">four</i></a>
diff --git a/regress/usr.bin/mandoc/mdoc/Em/tag.out_markdown b/regress/usr.bin/mandoc/mdoc/Em/tag.out_markdown
index 4c10715982f..c881c2d9cb4 100644
--- a/regress/usr.bin/mandoc/mdoc/Em/tag.out_markdown
+++ b/regress/usr.bin/mandoc/mdoc/Em/tag.out_markdown
@@ -12,7 +12,7 @@ BEGINTEST
 
 > text
 
-*three*
+*three-with-hyphens*
 
 > text
 
@@ -22,4 +22,4 @@ BEGINTEST
 
 ENDTEST
 
-OpenBSD - March 13, 2020
+OpenBSD - April 2, 2020
diff --git a/regress/usr.bin/mandoc/mdoc/Em/tag.out_tag b/regress/usr.bin/mandoc/mdoc/Em/tag.out_tag
index dd032f4f1f0..965d3e47b78 100644
--- a/regress/usr.bin/mandoc/mdoc/Em/tag.out_tag
+++ b/regress/usr.bin/mandoc/mdoc/Em/tag.out_tag
@@ -2,6 +2,6 @@ NAME 3
 DESCRIPTION 6
 one 9
 two 9
-three 12
-four 13
-explicit 13
+three-with-hyphens 12
+four 14
+explicit 14
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.in b/regress/usr.bin/mandoc/mdoc/Sh/tag.in
index da8a884a200..c44abbbfd2b 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.in
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.in
@@ -1,5 +1,5 @@
-.\" $OpenBSD: tag.in,v 1.2 2020/04/01 20:10:18 schwarze Exp $
-.Dd $Mdocdate: April 1 2020 $
+.\" $OpenBSD: tag.in,v 1.3 2020/04/02 14:55:29 schwarze Exp $
+.Dd $Mdocdate: April 2 2020 $
 .Dt SH-TAG 1
 .Os
 .Sh NAME
@@ -19,6 +19,8 @@ Text introducing examples.
 .Tg example
 .Ss Subsection
 Example text.
+.Ss Sub-section
+More example text.
 .Sh "\&	 WEIRD SECTION	 "
 Text in weird section.
 .Sh \ \&
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_ascii b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_ascii
index 309cb5597bf..8e6cce9d1f6 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_ascii
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_ascii
@@ -20,6 +20,9 @@ EEXXAAMMPPLLEESS
    SSuubbsseeccttiioonn
      Example text.
 
+   SSuubb--sseeccttiioonn
+     More example text.
+
       WWEEIIRRDD SSEECCTTIIOONN
      Text in weird section.
 
@@ -28,4 +31,4 @@ EEXXAAMMPPLLEESS
 
      ENDTEST
 
-OpenBSD                          April 1, 2020                         OpenBSD
+OpenBSD                          April 2, 2020                         OpenBSD
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_html b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_html
index 49c45fdb18e..d99c6d4da69 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_html
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_html
@@ -12,6 +12,10 @@
 <h2 class="Ss" id="example"><a class="permalink" href="#example">Subsection</a></h2>
 <p class="Pp">Example text.</p>
 </section>
+<section class="Ss">
+<h2 class="Ss" id="Sub-section"><a class="permalink" href="#Sub-section">Sub-section</a></h2>
+<p class="Pp">More example text.</p>
+</section>
 </section>
 <section class="Sh">
 <h1 class="Sh" id="WEIRD_SECTION"><a class="permalink" href="#WEIRD_SECTION">	
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_lint b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_lint
index cfb1dfad13e..63f9d2e4884 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_lint
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_lint
@@ -1,3 +1,3 @@
 mandoc: tag.in:14:2: WARNING: duplicate section title: Sh DESCRIPTION
-mandoc: tag.in:22:7: WARNING: tab in filled text
-mandoc: tag.in:22:22: WARNING: tab in filled text
+mandoc: tag.in:24:7: WARNING: tab in filled text
+mandoc: tag.in:24:22: WARNING: tab in filled text
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_markdown b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_markdown
index 75a1e58e437..4b87b144e38 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_markdown
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_markdown
@@ -26,6 +26,10 @@ Text introducing examples.
 
 Example text.
 
+## Sub-section
+
+More example text.
+
 # 	 WEIRD SECTION	 
 
 Text in weird section.
@@ -36,4 +40,4 @@ Text in section with empty header.
 
 ENDTEST
 
-OpenBSD - April 1, 2020
+OpenBSD - April 2, 2020
diff --git a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_tag b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_tag
index 04b11882b7d..b4adea63444 100644
--- a/regress/usr.bin/mandoc/mdoc/Sh/tag.out_tag
+++ b/regress/usr.bin/mandoc/mdoc/Sh/tag.out_tag
@@ -4,4 +4,5 @@ Subsection 9
 DESCRIPTION 14
 examples 17
 example 20
-WEIRD_SECTION 23
+Sub-section 23
+WEIRD_SECTION 26
diff --git a/regress/usr.bin/mandoc/mdoc/Tg/warn.in b/regress/usr.bin/mandoc/mdoc/Tg/warn.in
index e36f4506150..d4dd3296f3b 100644
--- a/regress/usr.bin/mandoc/mdoc/Tg/warn.in
+++ b/regress/usr.bin/mandoc/mdoc/Tg/warn.in
@@ -1,5 +1,5 @@
-.\" $OpenBSD: warn.in,v 1.1 2020/03/13 00:31:06 schwarze Exp $
-.Dd $Mdocdate: March 13 2020 $
+.\" $OpenBSD: warn.in,v 1.2 2020/04/02 14:55:29 schwarze Exp $
+.Dd $Mdocdate: April 2 2020 $
 .Dt TG-WARN 1
 .Os
 .Sh NAME
@@ -8,7 +8,7 @@
 .Sh DESCRIPTION
 BEGINTEST
 .Pp
-.Tg start
+.Tg start-tag
 initial
 text
 .Tg
diff --git a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_ascii b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_ascii
index 8dee9aed406..cb199b1f4e7 100644
--- a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_ascii
+++ b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_ascii
@@ -16,4 +16,4 @@ EEXXAAMMPPLLEESS
 
      ENDTEST
 
-OpenBSD                         March 13, 2020                         OpenBSD
+OpenBSD                          April 2, 2020                         OpenBSD
diff --git a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_html b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_html
index b4d3cf74c57..fdc333b6577 100644
--- a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_html
+++ b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_html
@@ -1,4 +1,4 @@
-<p class="Pp"><mark id="start"></mark>initial text
+<p class="Pp"><mark id="start-tag"></mark>initial text
     <a class="permalink" href="#macro"><code class="Ic" id="macro">macro</code></a>
     too many badstart badend whitespace <mark id="sub"></mark></p>
 <section class="Ss">
diff --git a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_markdown b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_markdown
index 872044244a7..ccc13bcec44 100644
--- a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_markdown
+++ b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_markdown
@@ -27,4 +27,4 @@ text
 
 ENDTEST
 
-OpenBSD - March 13, 2020
+OpenBSD - April 2, 2020
diff --git a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_tag b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_tag
index 6426c3b57e3..9009af0ea4f 100644
--- a/regress/usr.bin/mandoc/mdoc/Tg/warn.out_tag
+++ b/regress/usr.bin/mandoc/mdoc/Tg/warn.out_tag
@@ -1,6 +1,6 @@
 NAME 3
 DESCRIPTION 6
-start 9
+start-tag 9
 macro 9
 sub 9
 double 11
diff --git a/usr.bin/mandoc/mdoc_validate.c b/usr.bin/mandoc/mdoc_validate.c
index 689cab24a95..4b22e617e22 100644
--- a/usr.bin/mandoc/mdoc_validate.c
+++ b/usr.bin/mandoc/mdoc_validate.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: mdoc_validate.c,v 1.296 2020/04/01 20:10:17 schwarze Exp $ */
+/* $OpenBSD: mdoc_validate.c,v 1.297 2020/04/02 14:55:29 schwarze Exp $ */
 /*
  * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
  * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
@@ -2149,10 +2149,11 @@ post_rs(POST_ARGS)
 static void
 post_hyph(POST_ARGS)
 {
-	struct roff_node	*nch;
+	struct roff_node	*n, *nch;
 	char			*cp;
 
-	for (nch = mdoc->last->child; nch != NULL; nch = nch->next) {
+	n = mdoc->last;
+	for (nch = n->child; nch != NULL; nch = nch->next) {
 		if (nch->type != ROFFT_TEXT)
 			continue;
 		cp = nch->string;
@@ -2161,8 +2162,11 @@ post_hyph(POST_ARGS)
 		while (*(++cp) != '\0')
 			if (*cp == '-' &&
 			    isalpha((unsigned char)cp[-1]) &&
-			    isalpha((unsigned char)cp[1]))
+			    isalpha((unsigned char)cp[1])) {
+				if (n->string == NULL && n->flags & NODE_ID)
+					n->string = mandoc_strdup(nch->string);
 				*cp = ASCII_HYPH;
+			}
 	}
 }