Fixed yaz_iconv to return YAZ_ICONV_EINVAL if an incomplete MARC-8

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

diff --git a/test/tsticonv.c b/test/tsticonv.c
index 55cd9d1..43455ad 100644 (file)
--- a/test/tsticonv.c
+++ b/test/tsticonv.c
@@ -1,8 +1,8 @@
 /*
- * Copyright (C) 1995-2006, Index Data ApS
+ * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: tsticonv.c,v 1.19 2006-05-03 13:01:04 adam Exp $
+ * $Id: tsticonv.c,v 1.25 2007-03-09 08:39:38 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -90,19 +90,20 @@ static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf)
     char outbuf[12];
     size_t inbytesleft = strlen(buf);
     const char *inp = buf;
-    while (inbytesleft)
+    int rounds = 0;
+    for (rounds = 0; inbytesleft && rounds < sizeof(outbuf); rounds++)
     {
         size_t outbytesleft = sizeof(outbuf);
         char *outp = outbuf;
         size_t r = yaz_iconv(cd, (char**) &inp,  &inbytesleft,
                              &outp, &outbytesleft);
+        wrbuf_write(b, outbuf, outp - outbuf);
         if (r == (size_t) (-1))
         {
             int e = yaz_iconv_error(cd);
             if (e != YAZ_ICONV_E2BIG)
                 break;
         }
-        wrbuf_write(b, outbuf, outp - outbuf);
     }
     if (wrbuf_len(b) == strlen(cmpbuf) 
         && !memcmp(cmpbuf, wrbuf_buf(b), wrbuf_len(b)))
@@ -124,7 +125,7 @@ static const char *iso_8859_1_a[] = {
     "\xe5" "\xe5",
     0 };
 
-static void tst_marc8_to_ucs4b()
+static void tst_marc8_to_ucs4b(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("UCS4", "MARC8");
     YAZ_CHECK(cd);
@@ -222,7 +223,7 @@ static void tst_marc8_to_ucs4b()
     yaz_iconv_close(cd);
 }
 
-static void tst_ucs4b_to_utf8()
+static void tst_ucs4b_to_utf8(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("UTF8", "UCS4");
     YAZ_CHECK(cd);
@@ -337,7 +338,7 @@ int utf8_check(unsigned c)
     return 1;
 }
         
-static void tst_marc8_to_utf8()
+static void tst_marc8_to_utf8(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("UTF-8", "MARC8");
 
@@ -350,10 +351,15 @@ static void tst_marc8_to_utf8()
     /* COMBINING ACUTE ACCENT */
     YAZ_CHECK(tst_convert(cd, "Cours de mathâe", 
                           "Cours de mathe\xcc\x81"));
+
+
+    YAZ_CHECK(tst_convert(cd, "a\xea\x1e", "a\x1e\xcc\x8a"));
+
+    YAZ_CHECK(tst_convert(cd, "a\xea", "a"));
     yaz_iconv_close(cd);
 }
 
-static void tst_marc8s_to_utf8()
+static void tst_marc8s_to_utf8(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("UTF-8", "MARC8s");
 
@@ -371,7 +377,7 @@ static void tst_marc8s_to_utf8()
 }
 
 
-static void tst_marc8_to_latin1()
+static void tst_marc8_to_latin1(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("ISO-8859-1", "MARC8");
 
@@ -413,7 +419,7 @@ static void tst_marc8_to_latin1()
     yaz_iconv_close(cd);
 }
 
-static void tst_utf8_to_marc8()
+static void tst_utf8_to_marc8(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("MARC8", "UTF-8");
 
@@ -455,13 +461,20 @@ static void tst_utf8_to_marc8()
     /** Ideographic question mark (Unicode FF1F) */
     YAZ_CHECK(tst_convert(cd,
                           "\xEF\xBC\x9F" "o",        /* UTF-8 */
-                          "\033(1" "\x21\x2B\x3B" "\033(B" "o" ));
+                          "\033$1" "\x21\x2B\x3B" "\033(B" "o" ));
+
 
+    /** Superscript 0 . bug #642 */
+    YAZ_CHECK(tst_convert(cd,
+                          "(\xe2\x81\xb0)",        /* UTF-8 */
+                          "(\033p0\x1bs)"));
+    
+ 
     yaz_iconv_close(cd);
 }
 
 
-static void tst_latin1_to_marc8()
+static void tst_latin1_to_marc8(void)
 {
     yaz_iconv_t cd = yaz_iconv_open("MARC8", "ISO-8859-1");
 
@@ -483,17 +496,22 @@ static void tst_latin1_to_marc8()
     /** D8: UPPERCASE SCANDINAVIAN O */
     YAZ_CHECK(tst_convert(cd, "S\xd8", "S\xa2"));
 
-#if 0
-    /* does not pass yet */
     /** E9: LATIN SMALL LETTER E WITH ACUTE */
-    YAZ_CHECK(tst_convert(cd,
-                          "Cours de math\xe9",
-                          "Cours de mathâe"));
-#endif         
+    YAZ_CHECK(tst_convert(cd, "Cours de math\xe9", "Cours de mathâe"));
+    YAZ_CHECK(tst_convert(cd, "Cours de math", "Cours de math"
+                  ));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathé", "Cours de mathâe" ));
+    YAZ_CHECK(tst_convert(cd, "12345678é","12345678âe"));
+    YAZ_CHECK(tst_convert(cd, "123456789é", "123456789âe"));
+    YAZ_CHECK(tst_convert(cd, "1234567890é","1234567890âe"));
+    YAZ_CHECK(tst_convert(cd, "12345678901é", "12345678901âe"));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathém", "Cours de mathâem"));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathématiques",
+                          "Cours de mathâematiques"));
     yaz_iconv_close(cd);
 }
 
-static void tst_utf8_codes()
+static void tst_utf8_codes(void)
 {
     YAZ_CHECK(utf8_check(3));
     YAZ_CHECK(utf8_check(127));