From f8ff51d46ab411a8b4244becf67c9748ef5550a7 Mon Sep 17 00:00:00 2001
From: Adam Dickmeiss <adam@indexdata.dk>
Date: Thu, 28 Nov 2013 14:27:14 +0100
Subject: [PATCH] danmarc iconv: fixes and more tests
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

0xa733 is converted to @å , 0xa732 is converted to @Å instead
of @a733, @a732 .
---
 src/iconv_decode_danmarc.c |   10 ++++------
 src/iconv_encode_danmarc.c |   28 ++++++++++++++++++++++------
 test/test_iconv.c          |   29 +++++++++++++++++++++++------
 3 files changed, 49 insertions(+), 18 deletions(-)

diff --git a/src/iconv_decode_danmarc.c b/src/iconv_decode_danmarc.c
index 4105137..ee87057 100644
--- a/src/iconv_decode_danmarc.c
+++ b/src/iconv_decode_danmarc.c
@@ -57,14 +57,12 @@ static unsigned long read_danmarc(yaz_iconv_t cd,
             *no_read = 2;
             break;
         case 0xe5: /* LATIN SMALL LETTER A WITH RING ABOVE */
-            x = 'a';
-            data->x_back = 'a';
-            *no_read = 1;
+            x = 0xa733;
+            *no_read = 2;
             break;
         case 0xc5: /* LATIN CAPITAL LETTER A WITH RING ABOVE */
-            x = 'A';
-            data->x_back = 'a';
-            *no_read = 1;
+            x = 0xa732;
+            *no_read = 2;
             break;
         default:
             if (inbytesleft < 5)
diff --git a/src/iconv_encode_danmarc.c b/src/iconv_encode_danmarc.c
index 1afcf7c..467008a 100644
--- a/src/iconv_encode_danmarc.c
+++ b/src/iconv_encode_danmarc.c
@@ -25,14 +25,14 @@ static size_t write_danmarc(yaz_iconv_t cd, yaz_iconv_encoder_t en,
 {
     unsigned char *outp = (unsigned char *) *outbuf;
 
-    if (x == '@')
+    if (x == '@' || x == '*' || x == 0xa4)
     {
         if (*outbytesleft < 2)
         {
             yaz_iconv_set_errno(cd, YAZ_ICONV_E2BIG);
             return (size_t)(-1);
         }
-        *outp++ = x;
+        *outp++ = '@';
         (*outbytesleft)--;
         *outp++ = x;
         (*outbytesleft)--;
@@ -48,15 +48,31 @@ static size_t write_danmarc(yaz_iconv_t cd, yaz_iconv_encoder_t en,
         (*outbytesleft)--;
     }
     else
-    {  /* full unicode, emit @XXXX */
+    {
         if (*outbytesleft < 6)
         {
             yaz_iconv_set_errno(cd, YAZ_ICONV_E2BIG);
             return (size_t)(-1);
         }
-        sprintf(*outbuf, "@%04lX", x);
-        outp += 5;
-        (*outbytesleft) -= 5;
+        switch (x)
+        {
+        case 0xa733:
+            *outp++ = '@';
+            *outp++ = 0xe5;
+            (*outbytesleft) -= 2;
+            break;
+        case 0xa732:
+            *outp++ = '@';
+            *outp++ = 0xc5;
+            (*outbytesleft) -= 2;
+            break;
+        default:
+            /* full unicode, emit @XXXX */
+            sprintf(*outbuf, "@%04lX", x);
+            outp += 5;
+            (*outbytesleft) -= 5;
+            break;
+        }
     }
     *outbuf = (char *) outp;
     return 0;
diff --git a/test/test_iconv.c b/test/test_iconv.c
index e341972..ab05cee 100644
--- a/test/test_iconv.c
+++ b/test/test_iconv.c
@@ -681,9 +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)
@@ -693,10 +693,17 @@ 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, "@\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);
 }
@@ -710,10 +717,20 @@ static void tst_utf8_to_danmarc(void)
         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);
 }
 
@@ -738,7 +755,7 @@ 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();
-- 
1.7.10.4