/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2012 Index Data
+ * Copyright (C) 1995-2013 Index Data
* See the file LICENSE for details.
*/
#if HAVE_CONFIG_H
return tst_convert_x(cd, buf, cmpbuf, 0);
}
-/* some test strings in ISO-8859-1 format */
-static const char *iso_8859_1_a[] = {
- "ax" ,
- "\xd8",
- "eneb\346r",
- "\xe5" "\xd8",
- "\xe5" "\xd8" "b",
- "\xe5" "\xe5",
- 0 };
-
static void tst_marc8_to_ucs4b(void)
{
yaz_iconv_t cd = yaz_iconv_open("UCS4", "MARC8");
static void dconvert(int mandatory, const char *tmpcode)
{
+ /* some test strings in ISO-8859-1 format */
+ static const char *iso_8859_1_a[] = {
+ "ax" ,
+ "\xd8",
+ "eneb\346r",
+ "\xe5" "\xd8",
+ "\xe5" "\xd8" "b",
+ "\xe5" "\xe5",
+ 0 };
int i;
int ret;
yaz_iconv_t cd;
/** UPPERCASE SCANDINAVIAN O */
YAZ_CHECK(tst_convert(cd, "S\xc3\x98", "S\xa2"));
- /** ARING */
+ /** ARING (NFD) */
YAZ_CHECK(tst_convert(cd, "A" "\xCC\x8A", "\xEA" "A"));
+ /** ARING (NFC) */
+ YAZ_CHECK(tst_convert(cd, "\xC3\x85", "\xEA" "A"));
+
/** A MACRON + UMLAUT, DIAERESIS */
YAZ_CHECK(tst_convert(cd, "A" "\xCC\x84" "\xCC\x88",
"\xE5\xE8\x41"));
YAZ_CHECK(utf8_check(100000000));
}
-static void tst_danmarc_to_latin1(void)
+static void tst_danmarc_to_utf8(void)
{
- yaz_iconv_t cd = yaz_iconv_open("iso-8859-1", "danmarc");
+ yaz_iconv_t cd = yaz_iconv_open("utf-8", "danmarc");
YAZ_CHECK(cd);
if (!cd)
YAZ_CHECK(tst_convert(cd, "a@@b", "a@b"));
YAZ_CHECK(tst_convert(cd, "a@@@@b", "a@@b"));
- YAZ_CHECK(tst_convert(cd, "@000ab", "\nb"));
- YAZ_CHECK(tst_convert(cd, "@\xe5", "aa"));
- YAZ_CHECK(tst_convert(cd, "@\xc5.", "Aa."));
+ YAZ_CHECK(tst_convert(cd, "@*", "*"));
+ YAZ_CHECK(tst_convert(cd, "@@", "@"));
+ YAZ_CHECK(tst_convert(cd, "@\xa4", "\xC2\xA4"));
+ YAZ_CHECK(tst_convert(cd, "\xa4", "\xC2\xA4"));
+ YAZ_CHECK(tst_convert(cd, "@\xe5", "\xEA\x9C\xB3"));
+ YAZ_CHECK(tst_convert(cd, "@\xc5.", "\xEA\x9C\xB2" "."));
+
+ YAZ_CHECK(tst_convert(cd, "@a733", "\xEA\x9C\xB3"));
+ YAZ_CHECK(tst_convert(cd, "@a732.", "\xEA\x9C\xB2" "."));
+
+ YAZ_CHECK(tst_convert(cd, "a@03BBb", "a\xce\xbb" "b")); /* lambda */
+
+ yaz_iconv_close(cd);
+}
+
+static void tst_utf8_to_danmarc(void)
+{
+ yaz_iconv_t cd = yaz_iconv_open("danmarc", "utf-8");
+
+ YAZ_CHECK(cd);
+ if (!cd)
+ return;
+
+ YAZ_CHECK(tst_convert(cd, "ax", "ax"));
+
+ YAZ_CHECK(tst_convert(cd, "a@b", "a@@b"));
+ YAZ_CHECK(tst_convert(cd, "a@@b", "a@@@@b"));
+
+ YAZ_CHECK(tst_convert(cd, "*", "@*"));
+ YAZ_CHECK(tst_convert(cd, "@", "@@"));
+ YAZ_CHECK(tst_convert(cd, "\xC2\xA4", "\xa4"));
+
+ YAZ_CHECK(tst_convert(cd, "a\xc3\xa5" "b", "a\xe5" "b")); /* aring */
+ YAZ_CHECK(tst_convert(cd, "a\xce\xbb" "b", "a@03BBb")); /* lambda */
+
+ YAZ_CHECK(tst_convert(cd, "\xEA\x9C\xB2" ".", "@\xc5."));
+ YAZ_CHECK(tst_convert(cd, "\xEA\x9C\xB3", "@\xe5"));
yaz_iconv_close(cd);
}
+
int main (int argc, char **argv)
{
YAZ_CHECK_INIT(argc, argv);
tst_utf8_to_marc8("marc8lossy");
tst_utf8_to_marc8("marc8lossless");
- tst_danmarc_to_latin1();
+ tst_danmarc_to_utf8();
+ tst_utf8_to_danmarc();
tst_latin1_to_marc8();