X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=test%2Ftsticonv.c;h=fd8f2c8de651ca9b62d974083a60c13468de1209;hp=c4e75d7132b956eb23e725e3e02abde55fb726a5;hb=8626b7019b5d3d9c9594f20025e97d06a2d590fc;hpb=43a7aff23f364fd3fac6e9746c53cda01be13738

diff --git a/test/tsticonv.c b/test/tsticonv.c
index c4e75d7..fd8f2c8 100644
--- a/test/tsticonv.c
+++ b/test/tsticonv.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2005, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: tsticonv.c,v 1.11 2005-08-22 20:34:23 adam Exp $
+ * $Id: tsticonv.c,v 1.17 2006-04-19 23:15:40 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -15,38 +15,43 @@
 #include <ctype.h>
 
 #include <yaz/yaz-util.h>
+#include <yaz/test.h>
 
 static int compare_buffers(char *msg, int no,
                            int expect_len, const char *expect_buf,
                            int got_len, const char *got_buf)
 {
-    int i;
     if (expect_len == got_len
         && !memcmp(expect_buf, got_buf, expect_len))
         return 1;
-    printf("tsticonv test=%s i=%d failed\n", msg, no);
-    printf("off got exp\n");
-    for (i = 0; i<got_len || i<expect_len; i++)
+    
+    if (0) /* use 1 see how the buffers differ (for debug purposes) */
     {
-        char got_char[10];
-        char expect_char[10];
-
-        if (i < got_len)
-            sprintf(got_char, "%02X", got_buf[i]);
-        else
-            sprintf(got_char, "?  ");
-
-        if (i < expect_len)
-            sprintf(expect_char, "%02X", expect_buf[i]);
-        else
-            sprintf(expect_char, "?  ");
-        
-        printf("%02d  %s  %s %c\n",
-               i, got_char, expect_char, got_buf[i] == expect_buf[i] ?
-               ' ' : '*');
-
+        int i;
+        printf("tsticonv test=%s i=%d failed\n", msg, no);
+        printf("off got exp\n");
+        for (i = 0; i<got_len || i<expect_len; i++)
+        {
+            char got_char[10];
+            char expect_char[10];
+            
+            if (i < got_len)
+                sprintf(got_char, "%02X", got_buf[i]);
+            else
+                sprintf(got_char, "?  ");
+            
+            if (i < expect_len)
+                sprintf(expect_char, "%02X", expect_buf[i]);
+            else
+                sprintf(expect_char, "?  ");
+            
+            printf("%02d  %s  %s %c\n",
+                   i, got_char, expect_char, got_buf[i] == expect_buf[i] ?
+                   ' ' : '*');
+            
+        }
     }
-    exit(1);
+    return 0;
 }
 
 /* some test strings in ISO-8859-1 format */
@@ -74,13 +79,12 @@ static void tst_marc8_to_iso_8859_1()
 {
     int i;
     yaz_iconv_t cd;
+    int ret;
 
     cd = yaz_iconv_open("ISO-8859-1", "MARC8");
+    YAZ_CHECK(cd);
     if (!cd)
-    {
-        printf("tsticonv 10 yaz_iconv_open failed\n");
-        exit(10);
-    }
+        return;
     for (i = 0; iso_8859_1_a[i]; i++)
     {
         size_t r;
@@ -91,16 +95,14 @@ static void tst_marc8_to_iso_8859_1()
         size_t outbytesleft = sizeof(outbuf0);
 
         r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        YAZ_CHECK(r != (size_t)(-1));
         if (r == (size_t) (-1))
-        {
-            int e = yaz_iconv_error(cd);
+            break;
 
-            printf ("tsticonv 11 i=%d e=%d\n", i, e);
-            exit(11);
-        }
-        compare_buffers("tsticonv 11", i,
-                        strlen(iso_8859_1_a[i]), iso_8859_1_a[i],
-                        outbuf - outbuf0, outbuf0);
+        ret = compare_buffers("tsticonv 11", i,
+                              strlen(iso_8859_1_a[i]), iso_8859_1_a[i],
+                              outbuf - outbuf0, outbuf0);
+        YAZ_CHECK(ret);
     }
     yaz_iconv_close(cd);
 }
@@ -159,19 +161,26 @@ static void tst_marc8_to_ucs4b()
         "\xe5\xe8\x41",
         12, "\x00\x00\x00\x41" "\x00\x00\x03\x04" "\x00\x00\x03\x08"
     }, 
+    { /* bug #416 */
+        "\xEB\x74\xEC\x73",
+        12, "\x00\x00\x00\x74" "\x00\x00\x03\x61" "\x00\x00\x00\x73"
+    },
+    { /* bug #416 */
+        "\xFA\x74\xFB\x73",
+        12, "\x00\x00\x00\x74" "\x00\x00\x03\x60" "\x00\x00\x00\x73"
+    },
     {
         0, 0, 0
     }
     };
     int i;
+    int ret;
     yaz_iconv_t cd;
 
     cd = yaz_iconv_open("UCS4", "MARC8");
+    YAZ_CHECK(cd);
     if (!cd)
-    {
-        printf ("tsticonv 20 yaz_iconv_open failed\n");
-        exit(20);
-    }
+        return;
     for (i = 0; ar[i].len; i++)
     {
         size_t r;
@@ -190,18 +199,17 @@ static void tst_marc8_to_ucs4b()
             if (r == (size_t) (-1))
             {
                 int e = yaz_iconv_error(cd);
+                YAZ_CHECK(e == YAZ_ICONV_E2BIG);
                 if (e != YAZ_ICONV_E2BIG)
-                {
-                    printf ("tsticonv 21 i=%d e=%d\n", i, e);
-                    exit(21);
-                }
+                    return;
             }
             else
                 break;
         }
-        compare_buffers("tsticonv 22", i,
-                        expect_len, ar[i].ucs4_b,
-                        outbuf - outbuf0, outbuf0);
+        ret = compare_buffers("tsticonv 22", i,
+                              expect_len, ar[i].ucs4_b,
+                              outbuf - outbuf0, outbuf0);
+        YAZ_CHECK(ret);
     }
     yaz_iconv_close(cd);
 }
@@ -220,14 +228,13 @@ static void tst_ucs4b_to_utf8()
     };
     
     int i;
+    int ret;
     yaz_iconv_t cd;
 
     cd = yaz_iconv_open("UTF8", "UCS4");
+    YAZ_CHECK(cd);
     if (!cd)
-    {
-        printf ("tsticonv 30 yaz_iconv_open failed\n");
-        exit(30);
-    }
+        return;
     for (i = 0; ucs4_c[i]; i++)
     {
         size_t r;
@@ -238,16 +245,13 @@ static void tst_ucs4b_to_utf8()
         size_t outbytesleft = sizeof(outbuf0);
 
         r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        YAZ_CHECK(r != (size_t) (-1));
         if (r == (size_t) (-1))
-        {
-            int e = yaz_iconv_error(cd);
-
-            printf ("tsticonv 31 i=%d e=%d\n", i, e);
-            exit(31);
-        }
-        compare_buffers("tsticonv 32", i,
-                        strlen(utf8_c[i]), utf8_c[i],
-                        outbuf - outbuf0, outbuf0);
+            return;
+        ret = compare_buffers("tsticonv 32", i,
+                              strlen(utf8_c[i]), utf8_c[i],
+                              outbuf - outbuf0, outbuf0);
+        YAZ_CHECK(ret);
     }
     yaz_iconv_close(cd);
 }
@@ -255,6 +259,7 @@ static void tst_ucs4b_to_utf8()
 static void dconvert(int mandatory, const char *tmpcode)
 {
     int i;
+    int ret;
     yaz_iconv_t cd;
     for (i = 0; iso_8859_1_a[i]; i++)
     {
@@ -267,52 +272,237 @@ static void dconvert(int mandatory, const char *tmpcode)
         size_t outbytesleft = sizeof(outbuf0);
 
         cd = yaz_iconv_open(tmpcode, "ISO-8859-1");
+        YAZ_CHECK(cd || !mandatory);
         if (!cd)
-        {
-            if (!mandatory)
-                return;
-            printf ("tsticonv code=%s i=%d 1\n", tmpcode, i);
-            exit(1);
-        }
+            return;
         r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
-        if (r == (size_t)(-1))
-        {
-            int e = yaz_iconv_error(cd);
-
-            printf ("tsticonv code=%s i=%d 2 e=%d\n", tmpcode, i, e);
-            exit(2);
-        }
+        YAZ_CHECK(r != (size_t) (-1));
         yaz_iconv_close(cd);
+        if (r == (size_t) (-1))
+            return;
         
         cd = yaz_iconv_open("ISO-8859-1", tmpcode);
+        YAZ_CHECK(cd || !mandatory);
         if (!cd)
-        {
-            if (!mandatory)
-                return;
-            printf ("tsticonv code=%s i=%d 3\n", tmpcode, i);
-            exit(3);
-        }
+            return;
         inbuf = outbuf0;
         inbytesleft = sizeof(outbuf0) - outbytesleft;
 
         outbuf = outbuf1;
         outbytesleft = sizeof(outbuf1);
         r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
-        if (r == (size_t)(-1)) {
-            int e = yaz_iconv_error(cd);
-
-            printf ("tsticonv code=%s i=%d 4 e=%d\n", tmpcode, i, e);
-            exit(4);
+        YAZ_CHECK(r != (size_t) (-1));
+        if (r != (size_t)(-1)) 
+        {
+            ret = compare_buffers("dconvert", i,
+                                  strlen(iso_8859_1_a[i]), iso_8859_1_a[i],
+                              sizeof(outbuf1) - outbytesleft, outbuf1);
+            YAZ_CHECK(ret);
         }
-        compare_buffers("dconvert", i,
-                        strlen(iso_8859_1_a[i]), iso_8859_1_a[i],
-                        sizeof(outbuf1) - outbytesleft, outbuf1);
         yaz_iconv_close(cd);
     }
 }
+
+int utf8_check(unsigned c)
+{
+    if (sizeof(c) >= 4)
+    {
+        size_t r;
+        char src[4];
+        char dst[4];
+        char utf8buf[6];
+        char *inbuf = src;
+        size_t inbytesleft = 4;
+        char *outbuf = utf8buf;
+        size_t outbytesleft = sizeof(utf8buf);
+        int i;
+        yaz_iconv_t cd = yaz_iconv_open("UTF-8", "UCS4LE");
+        if (!cd)
+            return 0;
+        for (i = 0; i<4; i++)
+            src[i] = c >> (i*8);
+        
+        r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        yaz_iconv_close(cd);
+
+        if (r == (size_t)(-1))
+            return 0;
+
+        cd = yaz_iconv_open("UCS4LE", "UTF-8");
+        if (!cd)
+            return 0;
+        inbytesleft = sizeof(utf8buf) - outbytesleft;
+        inbuf = utf8buf;
+
+        outbuf = dst;
+        outbytesleft = 4;
+
+        r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        if (r == (size_t)(-1))
+            return 0;
+
+        yaz_iconv_close(cd);
+
+        if (memcmp(src, dst, 4))
+            return 0;
+    }
+    return 1;
+}
         
+static int tst_convert(yaz_iconv_t cd, const char *buf, const char *cmpbuf)
+{
+    int ret = 0;
+    WRBUF b = wrbuf_alloc();
+    char outbuf[12];
+    size_t inbytesleft = strlen(buf);
+    const char *inp = buf;
+    while (inbytesleft)
+    {
+        size_t outbytesleft = sizeof(outbuf);
+        char *outp = outbuf;
+        size_t r = yaz_iconv(cd, (char**) &inp,  &inbytesleft,
+                             &outp, &outbytesleft);
+        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)))
+        ret = 1;
+    else
+        yaz_log(YLOG_LOG, "GOT (%.*s)", wrbuf_len(b), wrbuf_buf(b));
+    wrbuf_free(b, 1);
+    return ret;
+}
+
+static void tst_conversion_marc8_to_latin1()
+{
+    yaz_iconv_t cd = yaz_iconv_open("ISO-8859-1", "MARC8");
+
+    YAZ_CHECK(cd);
+    if (!cd)
+        return;
+
+    YAZ_CHECK(tst_convert(cd, "Cours de math", 
+                          "Cours de math"));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathâe", 
+                          "Cours de mathé"));
+    YAZ_CHECK(tst_convert(cd, "12345678âe", 
+                          "12345678é"));
+    YAZ_CHECK(tst_convert(cd, "123456789âe", 
+                          "123456789é"));
+    YAZ_CHECK(tst_convert(cd, "1234567890âe", 
+                          "1234567890é"));
+    YAZ_CHECK(tst_convert(cd, "12345678901âe", 
+                          "12345678901é"));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathâem", 
+                          "Cours de mathém"));
+    YAZ_CHECK(tst_convert(cd, "Cours de mathâematiques", 
+                          "Cours de mathématiques"));
+
+    yaz_iconv_close(cd);
+}
+
+static void tst_conversion_utf8_to_marc8()
+{
+    yaz_iconv_t cd = yaz_iconv_open("MARC8", "UTF-8");
+
+    YAZ_CHECK(cd);
+    if (!cd)
+        return;
+
+    YAZ_CHECK(tst_convert(cd, "Cours ", "Cours "));
+
+    /** Pure ASCII. 11 characters (sizeof(outbuf)-1) */
+    YAZ_CHECK(tst_convert(cd, "Cours de mat", "Cours de mat"));
+
+    /** Pure ASCII. 12 characters (sizeof(outbuf)) */
+    YAZ_CHECK(tst_convert(cd, "Cours de math", "Cours de math"));
+
+    /** Pure ASCII. 13 characters (sizeof(outbuf)) */
+    YAZ_CHECK(tst_convert(cd, "Cours de math.", "Cours de math."));
+
+    /** UPPERCASE SCANDINAVIAN O */
+    YAZ_CHECK(tst_convert(cd, "S\xc3\x98", "S\xa2"));
+
+    /** ARING */
+    YAZ_CHECK(tst_convert(cd, "A" "\xCC\x8A", "\xEA" "A"));
+
+    /** A MACRON + UMLAUT, DIAERESIS */
+    YAZ_CHECK(tst_convert(cd, "A" "\xCC\x84" "\xCC\x88",
+                          "\xE5\xE8\x41"));
+    
+    /* Ligature spanning two characters */
+    YAZ_CHECK(tst_convert(cd,
+                          "\x74" "\xCD\xA1" "\x73",  /* UTF-8 */
+                          "\xEB\x74\xEC\x73"));      /* MARC-8 */
+
+    /* Double title spanning two characters */
+    YAZ_CHECK(tst_convert(cd,
+                          "\x74" "\xCD\xA0" "\x73",  /* UTF-8 */
+                          "\xFA\x74\xFB\x73"));      /* MARC-8 */
+
+    /** Ideographic question mark (Unicode FF1F) */
+    YAZ_CHECK(tst_convert(cd,
+                          "\xEF\xBC\x9F" "o",        /* UTF-8 */
+                          "\033(1" "\x21\x2B\x3B" "\033(B" "o" ));
+
+    yaz_iconv_close(cd);
+}
+
+
+static void tst_conversion_latin1_to_marc8()
+{
+    yaz_iconv_t cd = yaz_iconv_open("MARC8", "ISO-8859-1");
+
+    YAZ_CHECK(cd);
+    if (!cd)
+        return;
+
+    YAZ_CHECK(tst_convert(cd, "Cours ", "Cours "));
+
+    /** Pure ASCII. 11 characters (sizeof(outbuf)-1) */
+    YAZ_CHECK(tst_convert(cd, "Cours de mat", "Cours de mat"));
+
+    /** Pure ASCII. 12 characters (sizeof(outbuf)) */
+    YAZ_CHECK(tst_convert(cd, "Cours de math", "Cours de math"));
+
+    /** Pure ASCII. 13 characters (sizeof(outbuf)) */
+    YAZ_CHECK(tst_convert(cd, "Cours de math.", "Cours de math."));
+
+    /** UPPERCASE SCANDINAVIAN O */
+    YAZ_CHECK(tst_convert(cd, "SØ", "S\xa2"));
+
+    yaz_iconv_close(cd);
+}
+
 int main (int argc, char **argv)
 {
+    YAZ_CHECK_INIT(argc, argv);
+
+    tst_conversion_marc8_to_latin1();
+
+    tst_conversion_utf8_to_marc8();
+
+    tst_conversion_latin1_to_marc8();
+
+    YAZ_CHECK(utf8_check(3));
+    YAZ_CHECK(utf8_check(127));
+    YAZ_CHECK(utf8_check(128));
+    YAZ_CHECK(utf8_check(255));
+    YAZ_CHECK(utf8_check(256));
+    YAZ_CHECK(utf8_check(900));
+    YAZ_CHECK(utf8_check(1000));
+    YAZ_CHECK(utf8_check(10000));
+    YAZ_CHECK(utf8_check(100000));
+    YAZ_CHECK(utf8_check(1000000));
+    YAZ_CHECK(utf8_check(10000000));
+    YAZ_CHECK(utf8_check(100000000));
+
     dconvert(1, "UTF-8");
     dconvert(1, "ISO-8859-1");
     dconvert(1, "UCS4");
@@ -321,7 +511,8 @@ int main (int argc, char **argv)
     tst_marc8_to_iso_8859_1();
     tst_marc8_to_ucs4b();
     tst_ucs4b_to_utf8();
-    exit(0);
+
+    YAZ_CHECK_TERM;
 }
 /*
  * Local variables:
@@ -330,4 +521,3 @@ int main (int argc, char **argv)
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab
  */
-