current/SOURCES/groff-1.20.1-nroff-convert-encoding.patch

There are manpages written in many different encodings. This patch makes nroff
recognize its input encoding and then pass it to groff as the "-K" argument
(which will call preconv).

The patch also adds a "2>/dev/null" which is used to prevent error messages when
you open a big manpage and quickly close it.

diff -Nru groff-1.20.1/src/roff/nroff/nroff.sh i18n/src/roff/nroff/nroff.sh
--- groff-1.20.1/src/roff/nroff/nroff.sh        2009-01-09 12:25:52.000000000 -0200
+++ i18n/src/roff/nroff/nroff.sh        2009-11-19 10:35:14.000000000 -0200
@@ -125,6 +125,33 @@
     T=-T$Tloc ;;
 esac
 
+# copy the man page to a temp file
+TMPFILE=$(mktemp /tmp/man.XXXXXX)
+trap "rm -f $TMPFILE" 0 1 2 3 15
+cat ${1+"$@"} >| ${TMPFILE}
+
+# test the charset encoding of the man page
+if iconv -f utf-8 -t utf-8 -o /dev/null ${TMPFILE} 2>/dev/null
+then
+  charset_in=utf-8
+else
+  # non utf-8 encoding
+  case "${LANGUAGE-${LC_CTYPE-${LANG}}}" in
+    cs*|hr*|hu*|pl*|ro*|sk*|sl*) charset_in=iso-8859-2 ;;
+    el*) charset_in=iso-8859-7 ;;
+    tr*) charset_in=iso-8859-9 ;;
+    bg*) charset_in=cp1251 ;;
+    ja*) charset_in=euc-jp ;;
+    ko*) charset_in=euc-kr ;;
+    ru*) charset_in=koi8-r ;;
+    uk*) charset_in=koi8-u ;;
+    zh_TW*|zh_HK*) charset_in=big5 ;;
+    zh_CN*|zh*) charset_in=gb2312 ;;
+    *) charset_in=iso-8859-1 ;;
+  esac
+fi
+
+
 # Set up the `GROFF_BIN_PATH' variable
 # to be exported in the current `GROFF_RUNTIME' environment.
 
@@ -133,6 +160,6 @@
 
 # Load nroff-style character definitions too.
 
-PATH="$GROFF_RUNTIME$PATH" groff -mtty-char $T $opts ${1+"$@"}
+PATH="$GROFF_RUNTIME$PATH" cat $TMPFILE | groff -K $charset_in -mtty-char $T $opts 2>/dev/null
 
 # eof
1	There are manpages written in many different encodings. This patch makes nroff
2	recognize its input encoding and then pass it to groff as the "-K" argument
3	(which will call preconv).
4
5	The patch also adds a "2>/dev/null" which is used to prevent error messages when
6	you open a big manpage and quickly close it.
7
8	diff -Nru groff-1.20.1/src/roff/nroff/nroff.sh i18n/src/roff/nroff/nroff.sh
9	--- groff-1.20.1/src/roff/nroff/nroff.sh 2009-01-09 12:25:52.000000000 -0200
10	+++ i18n/src/roff/nroff/nroff.sh 2009-11-19 10:35:14.000000000 -0200
11	@@ -125,6 +125,33 @@
12	T=-T$Tloc ;;
13	esac
14
15	+# copy the man page to a temp file
16	+TMPFILE=$(mktemp /tmp/man.XXXXXX)
17	+trap "rm -f $TMPFILE" 0 1 2 3 15
18	+cat ${1+"$@"} >\| ${TMPFILE}
19	+
20	+# test the charset encoding of the man page
21	+if iconv -f utf-8 -t utf-8 -o /dev/null ${TMPFILE} 2>/dev/null
22	+then
23	+ charset_in=utf-8
24	+else
25	+ # non utf-8 encoding
26	+ case "${LANGUAGE-${LC_CTYPE-${LANG}}}" in
27	+ cs\|hr\|hu\|pl\|ro\|sk\|sl*) charset_in=iso-8859-2 ;;
28	+ el*) charset_in=iso-8859-7 ;;
29	+ tr*) charset_in=iso-8859-9 ;;
30	+ bg*) charset_in=cp1251 ;;
31	+ ja*) charset_in=euc-jp ;;
32	+ ko*) charset_in=euc-kr ;;
33	+ ru*) charset_in=koi8-r ;;
34	+ uk*) charset_in=koi8-u ;;
35	+ zh_TW\|zh_HK) charset_in=big5 ;;
36	+ zh_CN\|zh) charset_in=gb2312 ;;
37	+ *) charset_in=iso-8859-1 ;;
38	+ esac
39	+fi
40	+
41	+
42	# Set up the `GROFF_BIN_PATH' variable
43	# to be exported in the current `GROFF_RUNTIME' environment.
44
45	@@ -133,6 +160,6 @@
46
47	# Load nroff-style character definitions too.
48
49	-PATH="$GROFF_RUNTIME$PATH" groff -mtty-char $T $opts ${1+"$@"}
50	+PATH="$GROFF_RUNTIME$PATH" cat $TMPFILE \| groff -K $charset_in -mtty-char $T $opts 2>/dev/null
51
52	# eof