Surprisingly, every escape sequence can also be used as an argument

delimiter for an outer escape sequence, in which case the delimiting
escape sequence retains its syntax but usually ignores its argument
and loses its inherent effect.  Add rudimentary support for this
syntax quirk in order to improve parsing compatibility with groff.

10 files changed, 281 insertions, 4 deletions

diff --git a/regress/char/C/Makefile b/regress/char/C/Makefile
new file mode 100644
index 00000000..3d938fe7
--- /dev/null
+++ b/regress/char/C/Makefile
@@ -0,0 +1,6 @@
+# $OpenBSD: Makefile,v 1.1 2022/06/08 13:08:00 schwarze Exp $
+
+REGRESS_TARGETS	 = delim
+LINT_TARGETS	 = delim
+
+.include <bsd.regress.mk>
diff --git a/regress/char/C/delim.in b/regress/char/C/delim.in
new file mode 100644
index 00000000..e048e0a8
--- /dev/null
+++ b/regress/char/C/delim.in
@@ -0,0 +1,51 @@
+.\" $OpenBSD: delim.in,v 1.1 2022/06/08 13:08:00 schwarze Exp $
+.Dd $Mdocdate: June 8 2022 $
+.Dt C-DELIM 1
+.Os
+.Sh NAME
+.Nm C-delim
+.Nd argument delimiters for the C escape sequence
+.Sh DESCRIPTION
+empty: \C''
+.br
+single byte: \C'_'\C'-'
+.br
+two bytes: \C'hy'
+.br
+multiple bytes: \C'integral'
+.Ss Argument delimiters
+unsupported \er: \C\rat\ru
+.br
+ignored \e&: \C\&at\&u
+.br
+useless \e.: \C\.at.u
+.br
+invalid \eG: \C\GatGu
+.br
+special \e-: \C\-at\-u
+.br
+break \ep: \C\pat\pu
+.br
+nospace \ec: \C\cat\cu
+.\".br
+.\"XXX skipchar \ec: \C\zat\zu
+.br
+.ds mystr xatxu
+string expansion: \C\*[mystr]
+.br
+.nr myreg 1341
+register expansion: \C\n[myreg]u
+.br
+ignored \eON: \C\O1at\O2u
+.br
+special character: \C\(hyat\(hyu
+.br
+ignored \eZ\(aqstr\(aq: \C\Z'foo'at\Z'bar'u
+.br
+horizontal motion: \C\h'1'at\h'3'u
+.br
+horizontal line: \C\l'4'at\l'2'u
+.br
+overstrike: \C\o'ab'at\o'cd'u
+.br
+unterminated: \C'at
diff --git a/regress/char/C/delim.out_ascii b/regress/char/C/delim.out_ascii
new file mode 100644
index 00000000..7276d7c2
--- /dev/null
+++ b/regress/char/C/delim.out_ascii
@@ -0,0 +1,30 @@
+C-DELIM(1)                  General Commands Manual                 C-DELIM(1)
+
+NNAAMMEE
+     CC--ddeelliimm - argument delimiters for the C escape sequence
+
+DDEESSCCRRIIPPTTIIOONN
+     empty:
+     single byte: -
+     two bytes: -
+     multiple bytes: <integral>
+
+   AArrgguummeenntt ddeelliimmiitteerrss
+     unsupported \r: @u
+     ignored \&: @u
+     useless \.: @u
+     invalid \G: @u
+     special \-: @u
+     break \p: @u
+     nospace \c: @u
+     string expansion: @u
+     register expansion: 3/4u
+     ignored \ON: @u
+     special character: @u
+     ignored \Z'str': @u
+     horizontal motion: @u
+     horizontal line: @u
+     overstrike: @u
+     unterminated:
+
+OpenBSD                          June 8, 2022                          OpenBSD
diff --git a/regress/char/C/delim.out_lint b/regress/char/C/delim.out_lint
new file mode 100644
index 00000000..2dc23d44
--- /dev/null
+++ b/regress/char/C/delim.out_lint
@@ -0,0 +1,6 @@
+mandoc: delim.in:9:8: ERROR: invalid special character: \C''
+mandoc: delim.in:11:14: ERROR: invalid special character: \C'_'
+mandoc: delim.in:17:20: UNSUPP: unsupported escape sequence: \r
+mandoc: delim.in:17:24: UNSUPP: unsupported escape sequence: \r
+mandoc: delim.in:23:16: WARNING: undefined escape, printing literally: \G
+mandoc: delim.in:51:15: ERROR: incomplete escape sequence: \C'at
diff --git a/regress/char/Makefile b/regress/char/Makefile
index dbd922e0..b3cacfda 100644
--- a/regress/char/Makefile
+++ b/regress/char/Makefile
@@ -1,6 +1,6 @@
-# $OpenBSD: Makefile,v 1.9 2014/06/20 18:27:51 schwarze Exp $
+# $OpenBSD: Makefile,v 1.10 2022/06/08 13:08:00 schwarze Exp $
 
-SUBDIR = accent bar hyphen space unicode N
+SUBDIR = accent bar hyphen space unicode C N
 
 .include "../Makefile.sub"
 .include <bsd.subdir.mk>
diff --git a/regress/char/N/Makefile b/regress/char/N/Makefile
index 3630b154..30b19147 100644
--- a/regress/char/N/Makefile
+++ b/regress/char/N/Makefile
@@ -1,5 +1,6 @@
-# $OpenBSD: Makefile,v 1.2 2011/11/17 16:28:45 schwarze Exp $
+# $OpenBSD: Makefile,v 1.3 2022/06/08 13:08:00 schwarze Exp $
 
-REGRESS_TARGETS=basic
+REGRESS_TARGETS	 = basic delim
+LINT_TARGETS	 = basic delim
 
 .include <bsd.regress.mk>
diff --git a/regress/char/N/basic.out_lint b/regress/char/N/basic.out_lint
new file mode 100644
index 00000000..2dd8bb90
--- /dev/null
+++ b/regress/char/N/basic.out_lint
@@ -0,0 +1 @@
+mandoc: basic.in:23:14: ERROR: invalid escape argument delimiter: \N6
diff --git a/regress/char/N/delim.in b/regress/char/N/delim.in
new file mode 100644
index 00000000..f9ac0099
--- /dev/null
+++ b/regress/char/N/delim.in
@@ -0,0 +1,101 @@
+.\" $OpenBSD: delim.in,v 1.1 2022/06/08 13:08:00 schwarze Exp $
+.Dd $Mdocdate: June 8 2022 $
+.Dt N-DELIM 1
+.Os
+.Sh NAME
+.Nm N-delim
+.Nd argument delimiters for the N escape sequence
+.Sh DESCRIPTION
+unsupported \e!: \N\!42\!u
+.br
+unsupported \er: \N\r42u
+.br
+ignored \e%: \N\%42\%u
+.br
+ignored \e&: \N\&42\&u
+.br
+ignored \e): \N\)42\)u
+.br
+ignored \e,: \N\,42\,u
+.br
+ignored \e/: \N\/42\/u
+.br
+ignored \e^: \N\^42\^u
+.br
+ignored \ea: \N\a42\au
+.br
+ignored \ed: \N\d42\du
+.br
+ignored \et: \N\t42\tu
+.br
+ignored \eu: \N\u42\uu
+.br
+ignored \e{: \N\{42\{u
+.br
+ignored \e|: \N\|42\|u
+.br
+ignored \e}: \N\}42\}u
+.br
+useless \e.: \N\.42.u
+.\".br
+.\"XXX useless \e\e: \N\\42\\u
+.br
+invalid \eG: \N\G42Gu
+.br
+special \e\ : \N\ 42\ u
+.br
+special \e\(aq: \N\'42\'u
+.br
+special \e-: \N\-42\-u
+.br
+special \e0: \N\042\0u
+.br
+special \e:: \N\:42\:u
+.br
+special \e_: \N\_42\_u
+.br
+special \e\(ga: \N\`42\`u
+.br
+special \ee: \N\e42\eu
+.br
+special \e~: \N\~42\~u
+.br
+break \ep: \N\p42\pu
+.br
+nospace \ec: \N\c42\cu
+.\".br
+.\"XXX skipchar \ez: \N\z42\zu
+.br
+.ds mystr x42xu
+string expansion: \N\*[mystr]
+.br
+.nr myreg 23
+register expansion: \N\n[myreg]u
+.br
+ignored \eO: \N\O142\O2u
+.br
+ignored \eY: \N\Y[mystr]42\Y[mystr]u
+.br
+ignored \ek: \N\k[myreg]42\k[myreg]u
+.br
+special character: \N\(hy42\(hyu
+.br
+ignored \eD: \N\D't 1'42\D't 2'u
+.br
+ignored \eL: \N\L'2'42u
+.br
+ignored \eX: \N\X'foo'42\X'bar'u
+.br
+ignored \eZ: \N\Z'foo'42\Z''u
+.br
+ignored \eb: \N\b'2'42\b''u
+.br
+ignored \ev: \N\v'2'42\v'0'u
+.br
+ignored \ex: \N\x'2'42\v'0'u
+.br
+horizontal motion: \N\h'1'42\h'3'u
+.br
+horizontal line: \N\l'4'42\l'2'u
+.br
+overstrike: \N\o'ab'42\o'cd'u
diff --git a/regress/char/N/delim.out_ascii b/regress/char/N/delim.out_ascii
new file mode 100644
index 00000000..4e7cb1c2
--- /dev/null
+++ b/regress/char/N/delim.out_ascii
@@ -0,0 +1,52 @@
+N-DELIM(1)                  General Commands Manual                 N-DELIM(1)
+
+NNAAMMEE
+     NN--ddeelliimm - argument delimiters for the N escape sequence
+
+DDEESSCCRRIIPPTTIIOONN
+     unsupported \!: *u
+     unsupported \r: 42u
+     ignored \%: *u
+     ignored \&: *u
+     ignored \): *u
+     ignored \,: 42u
+     ignored \/: *u
+     ignored \^: 42u
+     ignored \a: 42u
+     ignored \d: 42u
+     ignored \t: 42u
+     ignored \u: 42u
+     ignored \{: *u
+     ignored \|: 42u
+     ignored \}: *u
+     useless \.: 42.u
+     invalid \G: *u
+     special \ : 42 u
+     special \': *u
+     special \-: *u
+     special \0: 42 u
+     special \:: *u
+     special \_: *u
+     special \`: *u
+     special \e: *u
+     special \~: 42 u
+     break \p: *u
+     nospace \c: *u
+     string expansion: *u
+     register expansion: 3u
+     ignored \O: 42u
+     ignored \Y: 42u
+     ignored \k: *u
+     special character: *u
+     ignored \D: 42u
+     ignored \L: 42u
+     ignored \X: 42u
+     ignored \Z: 42u
+     ignored \b: 42u
+     ignored \v: 42u
+     ignored \x: 42u
+     horizontal motion: 42   u
+     horizontal line: 42__u
+     overstrike: 42cdu
+
+OpenBSD                          June 8, 2022                          OpenBSD
diff --git a/regress/char/N/delim.out_lint b/regress/char/N/delim.out_lint
new file mode 100644
index 00000000..698c58c3
--- /dev/null
+++ b/regress/char/N/delim.out_lint
@@ -0,0 +1,29 @@
+mandoc: delim.in:9:20: UNSUPP: unsupported escape sequence: \!
+mandoc: delim.in:9:24: UNSUPP: unsupported escape sequence: \!
+mandoc: delim.in:11:20: UNSUPP: unsupported escape sequence: \r
+mandoc: delim.in:11:18: ERROR: invalid escape argument delimiter: \N\r
+mandoc: delim.in:19:14: ERROR: invalid escape argument delimiter: \N\,
+mandoc: delim.in:23:14: ERROR: invalid escape argument delimiter: \N\^
+mandoc: delim.in:25:14: ERROR: invalid escape argument delimiter: \N\a
+mandoc: delim.in:27:14: ERROR: invalid escape argument delimiter: \N\d
+mandoc: delim.in:29:14: ERROR: invalid escape argument delimiter: \N\t
+mandoc: delim.in:31:14: ERROR: invalid escape argument delimiter: \N\u
+mandoc: delim.in:35:14: ERROR: invalid escape argument delimiter: \N\|
+mandoc: delim.in:39:14: ERROR: invalid escape argument delimiter: \N\.
+mandoc: delim.in:43:16: WARNING: undefined escape, printing literally: \G
+mandoc: delim.in:45:15: ERROR: invalid escape argument delimiter: \N\ 
+mandoc: delim.in:51:14: ERROR: invalid escape argument delimiter: \N\0
+mandoc: delim.in:61:14: ERROR: invalid escape argument delimiter: \N\~
+mandoc: delim.in:73:21: ERROR: invalid escape argument delimiter: \N2
+mandoc: delim.in:75:14: ERROR: invalid escape argument delimiter: \N\O1
+mandoc: delim.in:77:14: ERROR: invalid escape argument delimiter: \N\Y[mystr]
+mandoc: delim.in:83:14: ERROR: invalid escape argument delimiter: \N\D't 1'
+mandoc: delim.in:85:14: ERROR: invalid escape argument delimiter: \N\L'2'
+mandoc: delim.in:87:14: ERROR: invalid escape argument delimiter: \N\X'foo'
+mandoc: delim.in:89:14: ERROR: invalid escape argument delimiter: \N\Z'foo'
+mandoc: delim.in:91:14: ERROR: invalid escape argument delimiter: \N\b'2'
+mandoc: delim.in:93:14: ERROR: invalid escape argument delimiter: \N\v'2'
+mandoc: delim.in:95:14: ERROR: invalid escape argument delimiter: \N\x'2'
+mandoc: delim.in:97:20: ERROR: invalid escape argument delimiter: \N\h'1'
+mandoc: delim.in:99:18: ERROR: invalid escape argument delimiter: \N\l'4'
+mandoc: delim.in:101:13: ERROR: invalid escape argument delimiter: \N\o'ab'