Fix encoding of 0x00a4 (unicode to danmarc) YAZ-713

[yaz-moved-to-github.git] / test / test_iconv.c

diff --git a/test/test_iconv.c b/test/test_iconv.c
index 27596b2..76d3965 100644 (file)
--- a/test/test_iconv.c
+++ b/test/test_iconv.c
@@ -156,16 +156,6 @@ static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf)
     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");
@@ -288,6 +278,15 @@ static void tst_ucs4b_to_utf8(void)
 
 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;
@@ -682,9 +681,9 @@ static void tst_utf8_codes(void)
     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)
@@ -694,14 +693,49 @@ static void tst_danmarc_to_latin1(void)
 
     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)
 {
@@ -722,7 +756,8 @@ int main (int argc, char **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();