icuSources/test/cintltst/eurocreg.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1999-2003, International Business Machines Corporation and
   4  * others. All Rights Reserved.
   5  ********************************************************************/
   6 #include "unicode/utypes.h"
   7 #include "unicode/ustring.h"
   8 #include "unicode/ctest.h"
   9 #include "unicode/ucnv.h"
  10
  11 void TestEuroRegression(void);
  12 void addTestEuroRegression(TestNode** root);
  13
  14 void addTestEuroRegression(TestNode** root)
  15 {
  16     addTest(root, &TestEuroRegression, "tsconv/eurocreg/TestEuroRegression");
  17 }
  18
  19 /*
  20  * The table below lists codepages that are supposed to have roundtrip mappings for
  21  * the U+20AC Euro sign.
  22  *
  23  * Changes made 2000nov28 and marked as such are due to the following:
  24  *
  25  * After updating all ibm-*.ucm files with precise fallback indicators (|0, |1, |3),
  26  * some of these codepages failed the Euro regression test.
  27  * This means that the actuall mappings changed when only the preciseness of fallback
  28  * mappings should have changed.
  29  * My (Markus) suspicion is that some files got Euro sign mappings added manually,
  30  * changing their contents compared to the NLTC (IBM Toronto codepage database) definition.
  31  * Such changes are highly undesirable because they effectively define new codepages.
  32  * Codepage mapping files with "ibm-*.ucm" should always exactly match the files
  33  * from the IBM codepage database.
  34  * (If there are several mappings with the same number, then we choose the
  35  * default mappings with Private-Use Area assignments.)
  36  *
  37  * Also, in the past, some aliases were set such that e.g. cp850 became an alias for ibm-858.
  38  * This followed the practice of OS/2 that uses the old codepage number 850 for the new
  39  * codepage 858, with the main difference being the additional Euro sign.
  40  * However, we have documented that the "cp" prefix should be used for Microsoft-compatible
  41  * codepages, and Microsoft Windows 2000's codepage 850 does not contain a Euro sign mapping.
  42  * Therefore, cp850 must not support the Euro sign.
  43  * In these cases, I have changed the codepage name here to point to a newer codepage with the
  44  * Euro sign, using its new name.
  45  * I could not find such "updates" for codepages 1362 and 1363 - we might want to supply them later.
  46  */
  47
  48 char convertersToCheck[][15] = {
  49   "cp1250",
  50   "cp1251",
  51   "cp1252",
  52   "cp1254",
  53   "cp1255",
  54   "cp1256",
  55   "cp1257",
  56   "cp1258",
  57   "ibm1140",
  58   "ibm1142",
  59   "ibm1143",
  60   "ibm1144",
  61   "ibm1145",
  62   "ibm1146",
  63   "ibm1147",
  64   "ibm1148",
  65   "ibm1149",
  66   "ibm1153",
  67   "ibm1154",
  68   "ibm1155",
  69   "ibm1156",
  70   "ibm1157",
  71   "ibm1158",
  72   /*"ibm-1159",*/ /* removed 2003Apr17 */
  73   "ibm12712",
  74   "ibm16804",
  75   "ibm-1160",
  76   "ibm-1162",
  77   "ibm-1164",
  78
  79   "ibm-858", /* was "cp850" changed 2000nov28 */
  80   /* duplicate "cp850" removed 2000nov28 */
  81   /*"ibm-9049",*/ /* was "cp857" changed 2002nov25 */
  82   "ibm-12712", /* was "cp424" changed 2000nov28 */
  83   "ibm-4899", /* was "cp803" changed 2000nov28 */
  84   "ibm-867", /* was "cp862" changed 2002nov25 */
  85   "cp1258",
  86   "windows-950",
  87   "cp1253",
  88   /*  "cp819",
  89       "cp13488",*/
  90   "cpibm4971",
  91   /*"ibm-9061",*/ /* was "cp869" changed 2002nov25 */
  92   /* "cp813",*/
  93   /*"ibm-9044",*/ /* was "cp852" changed 2002nov25 */
  94   /*"ibm-872",*/ /* was "cp855" changed 2002nov25 */
  95   /*"ibm-808",*/ /* was "cp866" changed 2002nov25 */
  96 /*  "cp1131",
  97   "cp1125",*/
  98   "ibm-902", /* was "cp922" changed 2003jan08 */
  99   "ibm-901", /* was "cp921" changed 2003jan09 */
 100   /*"ibm-17248",*/ /* was "cp864" changed 2002nov25 */
 101   /*"cp1008",
 102   "cp1046",*/
 103   /*  "cp9066",
 104       "cp1129",*/
 105   "ibm-5123", /* was "cp1027" changed 2003jan08 */
 106   /* "cp300",*/
 107   /* "cp4930",*/
 108   "cp1364",
 109   /* "cp1362" removed 2000nov28 */
 110   "cp1363",
 111   /* "cp1114", removed 2002jul3
 112      "cp947", removed 2002jul3 */
 113   "gb18030",
 114   ""};
 115
 116 UBool isEuroAware(UConverter*);
 117
 118 void TestEuroRegression()
 119 {
 120     int32_t i=0;
 121
 122     do
 123     {
 124         UErrorCode err = U_ZERO_ERROR;
 125         UConverter* myConv =  ucnv_open(convertersToCheck[i], &err);
 126         if (U_FAILURE(err)&&convertersToCheck[i][0])
 127             log_data_err("%s  \tMISSING [%s]\n", convertersToCheck[i], u_errorName(err));
 128         else
 129         {
 130             if (isEuroAware(myConv))
 131                 log_verbose("%s  \tsupports euro\n", convertersToCheck[i]);
 132             else
 133                 log_err("%s  \tDOES NOT support euro\n", convertersToCheck[i]);
 134             ucnv_close(myConv);
 135         }
 136     } while (convertersToCheck[++i][0]);
 137 }
 138
 139 UBool isEuroAware(UConverter* myConv)
 140 {
 141     static const UChar euroString[2] = { 0x20AC, 0x0000 };
 142     char target[20];
 143     UChar euroBack[2];
 144     int32_t targetSize, euroBackSize;
 145     UErrorCode err = U_ZERO_ERROR;
 146     /*const char* myName =   ucnv_getName(myConv, &err);*/
 147
 148     targetSize = ucnv_fromUChars(myConv,
 149             target,
 150             sizeof(target),
 151             euroString,
 152             -1,
 153             &err);
 154     if (U_FAILURE(err))
 155     {
 156       log_err("Failure Occured in ucnv_fromUChars euro roundtrip test\n");
 157       return FALSE;
 158     }
 159     euroBackSize = ucnv_toUChars(myConv,
 160             euroBack,
 161             2,
 162             target,
 163             targetSize,
 164             &err);
 165     if (U_FAILURE(err))
 166     {
 167         log_err("Failure Occured in ucnv_toUChars euro roundtrip test\n");
 168         return FALSE;
 169     }
 170     if (u_strcmp(euroString, euroBack))
 171     {
 172         /*      log_err("%s FAILED Euro rountrip\n", myName);*/
 173         return FALSE;
 174     }
 175     else
 176     {
 177         /*      log_verbose("%s PASSED Euro rountrip\n", myName);*/
 178         return TRUE;
 179     }
 180
 181 }