FWIW here is a patch that
a) tries to detect if user is running in utf-8 locale by a heuristic
similar that that used in the autoconf test
b) allows setting the unicode locale at runtime (changes the define to
a variable)
The autoconf script is not modified, removing the define should be
trivial for anybody who knows autoconf, though.
Thanks
Michal
diff -ur vilistextum-2.6.9/debian/changelog vilistextum-2.6.9.mod/debian/changelog
--- vilistextum-2.6.9/debian/changelog 2010-02-04 23:41:53.000000000 +0100
+++ vilistextum-2.6.9.mod/debian/changelog 2010-02-04 23:48:05.000000000 +0100
@@ -1,3 +1,9 @@
+vilistextum (2.6.9-1.m) unstable; urgency=low
+
+ * add multibyte support
+
+ -- hramrach <hramr...@heretic.chello.upc.cz> Thu, 04 Feb 2010 23:47:41 +0100
+
vilistextum (2.6.9-1) unstable; urgency=low
* Initial release (Closes: #516972).
Only in vilistextum-2.6.9.mod/debian: files
diff -ur vilistextum-2.6.9/debian/rules vilistextum-2.6.9.mod/debian/rules
--- vilistextum-2.6.9/debian/rules 2010-02-04 23:41:53.000000000 +0100
+++ vilistextum-2.6.9.mod/debian/rules 2010-02-04 23:50:27.000000000 +0100
@@ -4,3 +4,4 @@
include /usr/share/cdbs/1/class/autotools.mk
DEB_INSTAL_DOCS_ALL = CHANGES
+DEB_CONFIGURE_EXTRA_FLAGS = --enable-multibyte
Only in vilistextum-2.6.9.mod/debian: stamp-autotools-files
Only in vilistextum-2.6.9.mod/debian: stamp-makefile-build
Only in vilistextum-2.6.9.mod/debian: vilistextum
Only in vilistextum-2.6.9.mod/debian: vilistextum.debhelper.log
Only in vilistextum-2.6.9.mod/debian: vilistextum.substvars
Only in vilistextum-2.6.9.mod/doc: Makefile
Only in vilistextum-2.6.9.mod: Makefile
diff -ur vilistextum-2.6.9/src/charset.c vilistextum-2.6.9.mod/src/charset.c
--- vilistextum-2.6.9/src/charset.c 2006-01-22 20:30:26.000000000 +0100
+++ vilistextum-2.6.9.mod/src/charset.c 2010-02-04 22:07:49.000000000 +0100
@@ -42,9 +42,9 @@
void init_multibyte()
{
char *ret;
- ret = setlocale(LC_CTYPE, INTERNAL_LOCALE);
+ ret = setlocale(LC_CTYPE, internal_locale);
if (ret==NULL) {
- fprintf(stderr, "setlocale failed with: %s\n\n", INTERNAL_LOCALE);
+ fprintf(stderr, "setlocale failed with: %s\n\n", internal_locale);
exit(1);
}
/*else { fprintf(stderr, "%s\n", ret); } */
Only in vilistextum-2.6.9.mod/src: charset.o
diff -ur vilistextum-2.6.9/src/fileio.c vilistextum-2.6.9.mod/src/fileio.c
--- vilistextum-2.6.9/src/fileio.c 2006-04-22 18:42:26.000000000 +0200
+++ vilistextum-2.6.9.mod/src/fileio.c 2010-02-04 22:07:14.000000000 +0100
@@ -100,9 +100,9 @@
output[strlen(output)] = '\0';
- ret = setlocale(LC_CTYPE, INTERNAL_LOCALE);
+ ret = setlocale(LC_CTYPE, internal_locale);
if (ret==NULL) {
- fprintf(stderr, "setlocale failed with: %s\n\n", INTERNAL_LOCALE);
+ fprintf(stderr, "setlocale failed with: %s\n\n", internal_locale);
exit(1);
}
mbstowcs(converted_string, output, strlen(output));
Only in vilistextum-2.6.9.mod/src: fileio.o
Only in vilistextum-2.6.9.mod/src: html.o
Only in vilistextum-2.6.9.mod/src: html_tag.o
Only in vilistextum-2.6.9.mod/src: latin1.o
Only in vilistextum-2.6.9.mod/src: lists.o
diff -ur vilistextum-2.6.9/src/main.c vilistextum-2.6.9.mod/src/main.c
--- vilistextum-2.6.9/src/main.c 2006-05-01 07:37:43.000000000 +0200
+++ vilistextum-2.6.9.mod/src/main.c 2010-02-05 00:30:17.000000000 +0100
@@ -104,12 +104,35 @@
#include <stdio.h>
#include <stdlib.h>
+#ifdef MULTIBYTE
+#include <locale.h>
+#endif
#include "html.h"
#include "fileio.h"
#include "charset.h"
#include "debug.h"
+#ifdef MULTIBYTE
+
+static int suffix(const char * str, const char * suffix)
+{
+ if ( strlen(str) < strlen(suffix) ) return 0;
+ if ( ! strcmp(suffix, str + ( strlen(str) - strlen(suffix) ) ) ) return 1;
+ return 0;
+}
+
+static int utf_8_locale(const char * locale)
+{
+ if(!locale) return 0;
+ return suffix(locale,".utf8") || suffix(locale, ".UTF-8");
+}
+
+char * internal_locale = "";
+
+#endif
+
+
/* commandline options */
int palm = 0, /* if true, enable pda specific restrictions */
convert_tags = 0, /* if true, convert some tags (eg B,I) to ASCII characters */
@@ -151,9 +174,11 @@
" -s, --shrink-lines [NUMBER] \n"
" -t, --no-title \n"
" -w, --width NUMBER \n"
+" -8, --no-utf-8 \n"
#ifdef MULTIBYTE
"\n"
" -u, --output-utf-8 \n"
+" -L, --utf-8-locale LOCALE \n"
" -x, --translit \n"
" -y, --charset CHARSET \n"
#endif
@@ -202,6 +227,8 @@
{"charset", 1, 0, 'y'},
{"translit", 0, 0, 'x'},
+ {"no-utf-8", 0, 0, '8'},
+ {"utf-8-locale", 1, 0, 'L'},
{"output-utf-8", 0, 0, 'u'},
{0, 0, 0, 0}
@@ -216,11 +243,17 @@
}
printf("\n");
#endif
-
+
+#ifdef MULTIBYTE
+#define OPTSTRING "pmsi:ce:hkltrnavy:xuL:8w:"
+#else
+#define OPTSTRING "pmsi:ce:hkltrnav8w:"
+#endif
+
#ifdef HAVE_GETOPT_H
- c = getopt_long (argc, argv, "pmsi:ce:hkltrnavy:xuw:", long_options, &option_index);
+ c = getopt_long (argc, argv, OPTSTRING, long_options, &option_index);
#else
- c = getopt(argc, argv, "pmsi:ce:hkltrnavy:xuw:");
+ c = getopt(argc, argv, OPTSTRING);
#endif
#ifdef DEBUG
fprintf(stderr, "c: %d c: %c \n\n", c,c);
@@ -241,12 +274,20 @@
printf("Vilistextum %s (%s)\n", VERSION, RELEASEDATE);
#endif
exit(0);
+ break;
#ifdef MULTIBYTE
case 'y': set_iconv_charset(argument); break;
+ case 'L': internal_locale = argument;
+ option_output_utf8 = 1;
+ break;
case 'x': transliteration = 1; break;
case 'u': option_output_utf8 = 1; break;
#endif
- break;
+ case '8':
+#ifdef MULTIBYTE
+ option_output_utf8 = 0;
+#endif
+ break;
case 'p':
/* settings for PDAs */
hr_breite = 30;
@@ -305,7 +346,10 @@
int main(int argc, char *argv[])
{
+
#ifdef MULTIBYTE
+ internal_locale = setlocale(LC_CTYPE, "");
+ option_output_utf8 = utf_8_locale(internal_locale);
init_multibyte();
use_default_charset();
#endif
Only in vilistextum-2.6.9.mod/src: main.c~
diff -ur vilistextum-2.6.9/src/main.h vilistextum-2.6.9.mod/src/main.h
--- vilistextum-2.6.9/src/main.h 2005-07-23 18:47:44.000000000 +0200
+++ vilistextum-2.6.9.mod/src/main.h 2010-02-04 23:24:19.000000000 +0100
@@ -19,6 +19,8 @@
int option_no_alt;
int option_output_utf8;
+char * internal_locale;
+
CHAR *default_image;
#endif
Only in vilistextum-2.6.9.mod/src: main.o
Only in vilistextum-2.6.9.mod/src: Makefile
Only in vilistextum-2.6.9.mod/src: microsoft.o
Only in vilistextum-2.6.9.mod/src: text.o
Only in vilistextum-2.6.9.mod/src: unicode_entities.o
Only in vilistextum-2.6.9.mod/src: util.o
Only in vilistextum-2.6.9.mod/src: vilistextum
Only in vilistextum-2.6.9.mod: stamp-h1
diff -ur vilistextum-2.6.9/tests/check_entities vilistextum-2.6.9.mod/tests/check_entities
--- vilistextum-2.6.9/tests/check_entities 2006-04-25 13:42:45.000000000 +0200
+++ vilistextum-2.6.9.mod/tests/check_entities 2010-02-05 00:18:25.000000000 +0100
@@ -3,7 +3,7 @@
echo ""
echo "TEST: $0"
rm -f entities.output.test
-../src/vilistextum 2>/dev/null -m entities.html entities.output.test
+../src/vilistextum 2>/dev/null -m8 entities.html entities.output.test
diff >/dev/null entities.output entities.output.test
diff -ur vilistextum-2.6.9/tests/check_latin2 vilistextum-2.6.9.mod/tests/check_latin2
--- vilistextum-2.6.9/tests/check_latin2 2006-04-25 13:42:51.000000000 +0200
+++ vilistextum-2.6.9.mod/tests/check_latin2 2010-02-05 00:17:51.000000000 +0100
@@ -7,7 +7,7 @@
../src/vilistextum -v | grep >/dev/null multibyte || exit 0
rm -f latin2.output.test
-../src/vilistextum latin2.html latin2.output.test
+../src/vilistextum -8 latin2.html latin2.output.test
diff >/dev/null latin2.output latin2.output.test
Only in vilistextum-2.6.9.mod/tests: entities.output.test
Only in vilistextum-2.6.9.mod/tests: latin2.output.test
Only in vilistextum-2.6.9.mod/tests: Makefile
Only in vilistextum-2.6.9.mod/tests: tags.output.test
Only in vilistextum-2.6.9.mod/tests: utf-8-sampler.output.test
