Test handling of combining and non-combining accents

for the ASCII and UTF-8 output modes.
Related to mandoc/chars.in rev. 1.21.

9 files changed, 174 insertions, 2 deletions

diff --git a/regress/usr.bin/mandoc/char/Makefile b/regress/usr.bin/mandoc/char/Makefile
index 114208867f6..59d722105bc 100644
--- a/regress/usr.bin/mandoc/char/Makefile
+++ b/regress/usr.bin/mandoc/char/Makefile
@@ -1,6 +1,6 @@
-# $OpenBSD: Makefile,v 1.7 2013/11/10 19:26:31 schwarze Exp $
+# $OpenBSD: Makefile,v 1.8 2014/03/08 18:00:59 schwarze Exp $
 
-SUBDIR = bar hyphen space unicode N
+SUBDIR = accent bar hyphen space unicode N
 
 ascii groff groff-clean obj-clean tman: _SUBDIRUSE
 
diff --git a/regress/usr.bin/mandoc/char/accent/Makefile b/regress/usr.bin/mandoc/char/accent/Makefile
new file mode 100644
index 00000000000..4bc149a7b1b
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/Makefile
@@ -0,0 +1,7 @@
+# $OpenBSD: Makefile,v 1.1 2014/03/08 18:00:59 schwarze Exp $
+
+REGRESS_TARGETS = nocombine utf8only combine
+SKIP_ASCII = utf8only combine
+UTF8_TARGETS = nocombine utf8only combine
+
+.include <bsd.regress.mk>
diff --git a/regress/usr.bin/mandoc/char/accent/combine.in b/regress/usr.bin/mandoc/char/accent/combine.in
new file mode 100644
index 00000000000..124d82d511d
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/combine.in
@@ -0,0 +1,23 @@
+.TH CHAR-ACCENT-COMBINE 1 "February 28, 2014" OpenBSD
+.SH NAME
+\fBchar-accent-combine\fR - combining accents
+.SH DESCRIPTION
+char + combine char U: e\U'0301'
+.br
+char + combine char C: e\C'u0301'
+.br
+char + combine char named: e\[u0301]
+.\" XXX not implemented in mandoc
+.\" .br
+.\" combined chars named numeric: \[u0065_0301]
+.\" .br
+.\" combined chars named: \[e aa]
+.br
+combined char pre N: \['e]
+.br
+combined char pre 2: \('e
+.\" XXX Plan 9 roff only
+.\" .br
+.\" combined char post N: \[e']
+.\" .br
+.\" combined char post 2: \(e'
diff --git a/regress/usr.bin/mandoc/char/accent/combine.out_utf8 b/regress/usr.bin/mandoc/char/accent/combine.out_utf8
new file mode 100644
index 00000000000..92ee28bd9b1
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/combine.out_utf8
@@ -0,0 +1,17 @@
+CHAR-ACCENT-COMBINE(1)     OpenBSD Reference Manual     CHAR-ACCENT-COMBINE(1)
+
+
+
+NNAAMMEE
+       cchhaarr--aacccceenntt--ccoommbbiinnee - combining accents
+
+DDEESSCCRRIIPPTTIIOONN
+       char + combine char U: eU'0301'
+       char + combine char C: é
+       char + combine char named: é
+       combined char pre N: é
+       combined char pre 2: é
+
+
+
+OpenBSD                        February 28, 2014        CHAR-ACCENT-COMBINE(1)
diff --git a/regress/usr.bin/mandoc/char/accent/nocombine.in b/regress/usr.bin/mandoc/char/accent/nocombine.in
new file mode 100644
index 00000000000..39428f76b51
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/nocombine.in
@@ -0,0 +1,45 @@
+.TH CHAR-ACCENT-NOCOMBINE 1 "March 8, 2014" OpenBSD
+.SH NAME
+\fBchar-accent-nocombine\fR - non-combining accents
+.SH DESCRIPTION
+bare acute accent: e'e
+.br
+escaped acute accent: e\'e
+.br
+acute accent sequence: e\(aae
+.br
+bare grave accent: e`e
+.br
+escaped grave accent: e\`e
+.br
+acute grave sequence: e\(gae
+.br
+hungarian umlaut: e\(a"e
+.br
+.\" XXX This is ridiculous.
+.\" XXX groff prints the macron as an underscore in the previous line.
+.\" macron: e\(a-e
+.br
+.\" XXX groff doesn't have a dot in ASCII mode, only in UTF-8 mode.
+.\" dotted: e\(a.e
+.br
+circumflex: e\(a^e
+.br
+.\" XXX groff uses a backspace for this one in ASCII mode.
+.\" breve: e\(abe
+.br
+cedilla: e\(ace
+.br
+dieresis: e\(ade
+.br
+caron: e\(ahe
+.br
+ring: e\(aoe
+.br
+tilde: e\(a~e
+.br
+ogonek: e\(hoe
+.br
+text hat: e\(hae
+.br
+text tilde: e\(tie
diff --git a/regress/usr.bin/mandoc/char/accent/nocombine.out_ascii b/regress/usr.bin/mandoc/char/accent/nocombine.out_ascii
new file mode 100644
index 00000000000..764110ddfa6
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/nocombine.out_ascii
@@ -0,0 +1,28 @@
+CHAR-ACCENT-NOCOMBINE(1)   OpenBSD Reference Manual   CHAR-ACCENT-NOCOMBINE(1)
+
+
+
+NNAAMMEE
+       cchhaarr--aacccceenntt--nnooccoommbbiinnee - non-combining accents
+
+DDEESSCCRRIIPPTTIIOONN
+       bare acute accent: e'e
+       escaped acute accent: e'e
+       acute accent sequence: e'e
+       bare grave accent: e`e
+       escaped grave accent: e`e
+       acute grave sequence: e`e
+       hungarian umlaut: e"e
+       circumflex: e^e
+       cedilla: e,e
+       dieresis: e"e
+       caron: eve
+       ring: eoe
+       tilde: e~e
+       ogonek: e,e
+       text hat: e^e
+       text tilde: e~e
+
+
+
+OpenBSD                          March 8, 2014        CHAR-ACCENT-NOCOMBINE(1)
diff --git a/regress/usr.bin/mandoc/char/accent/nocombine.out_utf8 b/regress/usr.bin/mandoc/char/accent/nocombine.out_utf8
new file mode 100644
index 00000000000..33389badfc7
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/nocombine.out_utf8
@@ -0,0 +1,28 @@
+CHAR-ACCENT-NOCOMBINE(1)   OpenBSD Reference Manual   CHAR-ACCENT-NOCOMBINE(1)
+
+
+
+NNAAMMEE
+       cchhaarr--aacccceenntt--nnooccoommbbiinnee - non-combining accents
+
+DDEESSCCRRIIPPTTIIOONN
+       bare acute accent: e'e
+       escaped acute accent: e´e
+       acute accent sequence: e´e
+       bare grave accent: e`e
+       escaped grave accent: e`e
+       acute grave sequence: e`e
+       hungarian umlaut: e˝e
+       circumflex: e^e
+       cedilla: e¸e
+       dieresis: e¨e
+       caron: eˇe
+       ring: e˚e
+       tilde: e~e
+       ogonek: e˛e
+       text hat: e^e
+       text tilde: e~e
+
+
+
+OpenBSD                          March 8, 2014        CHAR-ACCENT-NOCOMBINE(1)
diff --git a/regress/usr.bin/mandoc/char/accent/utf8only.in b/regress/usr.bin/mandoc/char/accent/utf8only.in
new file mode 100644
index 00000000000..8c7d2ca0650
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/utf8only.in
@@ -0,0 +1,9 @@
+.TH CHAR-ACCENT-UTF8ONLY 1 "March 8, 2014" OpenBSD
+.SH NAME
+\fBchar-accent-utf8only\fR - accents available in UTF-8 only
+.SH DESCRIPTION
+macron: e\(a-e
+.br
+dotted: e\(a.e
+.br
+breve: e\(abe
diff --git a/regress/usr.bin/mandoc/char/accent/utf8only.out_utf8 b/regress/usr.bin/mandoc/char/accent/utf8only.out_utf8
new file mode 100644
index 00000000000..afca2e69d83
--- /dev/null
+++ b/regress/usr.bin/mandoc/char/accent/utf8only.out_utf8
@@ -0,0 +1,15 @@
+CHAR-ACCENT-UTF8ONLY(1)    OpenBSD Reference Manual    CHAR-ACCENT-UTF8ONLY(1)
+
+
+
+NNAAMMEE
+       cchhaarr--aacccceenntt--uuttff88oonnllyy - accents available in UTF-8 only
+
+DDEESSCCRRIIPPTTIIOONN
+       macron: e¯e
+       dotted: e˙e
+       breve: e˘e
+
+
+
+OpenBSD                          March 8, 2014         CHAR-ACCENT-UTF8ONLY(1)