1 /******************************************************************** 2 * COPYRIGHT: 3 * Copyright (c) 1999-2006, 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 #if !UCONFIG_NO_LEGACY_CONVERSION 15 void addTestEuroRegression(TestNode** root) 16 { 17 addTest(root, &TestEuroRegression, "tsconv/eurocreg/TestEuroRegression"); 18 } 19 20 /* 21 * The table below lists codepages that are supposed to have roundtrip mappings for 22 * the U+20AC Euro sign. 23 * 24 * Changes made 2000nov28 and marked as such are due to the following: 25 * 26 * After updating all ibm-*.ucm files with precise fallback indicators (|0, |1, |3), 27 * some of these codepages failed the Euro regression test. 28 * This means that the actuall mappings changed when only the preciseness of fallback 29 * mappings should have changed. 30 * My (Markus) suspicion is that some files got Euro sign mappings added manually, 31 * changing their contents compared to the NLTC (IBM Toronto codepage database) definition. 32 * Such changes are highly undesirable because they effectively define new codepages. 33 * Codepage mapping files with "ibm-*.ucm" should always exactly match the files 34 * from the IBM codepage database. 35 * (If there are several mappings with the same number, then we choose the 36 * default mappings with Private-Use Area assignments.) 37 * 38 * Also, in the past, some aliases were set such that e.g. cp850 became an alias for ibm-858. 39 * This followed the practice of OS/2 that uses the old codepage number 850 for the new 40 * codepage 858, with the main difference being the additional Euro sign. 41 * However, we have documented that the "cp" prefix should be used for Microsoft-compatible 42 * codepages, and Microsoft Windows 2000's codepage 850 does not contain a Euro sign mapping. 43 * Therefore, cp850 must not support the Euro sign. 44 * In these cases, I have changed the codepage name here to point to a newer codepage with the 45 * Euro sign, using its new name. 46 * I could not find such "updates" for codepages 1362 and 1363 - we might want to supply them later. 47 */ 48 49 static const char convertersToCheck[][15] = { 50 "cp1250", 51 "cp1251", 52 "cp1252", 53 "cp1254", 54 "cp1255", 55 "cp1256", 56 "cp1257", 57 "cp1258", 58 "ibm1140", 59 "ibm1142", 60 "ibm1143", 61 "ibm1144", 62 "ibm1145", 63 "ibm1146", 64 "ibm1147", 65 "ibm1148", 66 "ibm1149", 67 "ibm1153", 68 "ibm1154", 69 "ibm1155", 70 "ibm1156", 71 "ibm1157", 72 "ibm1158", 73 /*"ibm-1159",*/ /* removed 2003Apr17 */ 74 "ibm12712", 75 "ibm16804", 76 "ibm-1160", 77 "ibm-1162", 78 "ibm-1164", 79 80 "ibm-858", /* was "cp850" changed 2000nov28 */ 81 /* duplicate "cp850" removed 2000nov28 */ 82 /*"ibm-9049",*/ /* was "cp857" changed 2002nov25 */ 83 "ibm-12712", /* was "cp424" changed 2000nov28 */ 84 "ibm-4899", /* was "cp803" changed 2000nov28 */ 85 "ibm-867", /* was "cp862" changed 2002nov25 */ 86 "cp1258", 87 "windows-950", 88 "cp1253", 89 /* "cp819", 90 "cp13488",*/ 91 "ibm-4971", 92 /*"ibm-9061",*/ /* was "cp869" changed 2002nov25 */ 93 /* "cp813",*/ 94 /*"ibm-9044",*/ /* was "cp852" changed 2002nov25 */ 95 /*"ibm-872",*/ /* was "cp855" changed 2002nov25 */ 96 /*"ibm-808",*/ /* was "cp866" changed 2002nov25 */ 97 /* "cp1131", 98 "cp1125",*/ 99 "ibm-902", /* was "cp922" changed 2003jan08 */ 100 "ibm-901", /* was "cp921" changed 2003jan09 */ 101 /*"ibm-17248",*/ /* was "cp864" changed 2002nov25 */ 102 /*"cp1008", 103 "cp1046",*/ 104 /* "cp9066", 105 "cp1129",*/ 106 "ibm-5123", /* was "cp1027" changed 2003jan08 */ 107 /* "cp300",*/ 108 /* "cp4930",*/ 109 "ibm-1364", 110 /* "cp1362" removed 2000nov28 */ 111 "cp1363", 112 /* "cp1114", removed 2002jul3 113 "cp947", removed 2002jul3 */ 114 "gb18030", 115 ""}; 116 117 UBool isEuroAware(UConverter*); 118 119 void TestEuroRegression() 120 { 121 int32_t i=0; 122 123 do 124 { 125 UErrorCode err = U_ZERO_ERROR; 126 UConverter* myConv = ucnv_open(convertersToCheck[i], &err); 127 if (U_FAILURE(err)&&convertersToCheck[i][0]) 128 log_data_err("%s \tMISSING [%s]\n", convertersToCheck[i], u_errorName(err)); 129 else 130 { 131 if (isEuroAware(myConv)) 132 log_verbose("%s \tsupports euro\n", convertersToCheck[i]); 133 else 134 log_err("%s \tDOES NOT support euro\n", convertersToCheck[i]); 135 ucnv_close(myConv); 136 } 137 } while (convertersToCheck[++i][0]); 138 } 139 140 UBool isEuroAware(UConverter* myConv) 141 { 142 static const UChar euroString[2] = { 0x20AC, 0x0000 }; 143 char target[20]; 144 UChar euroBack[2]; 145 int32_t targetSize, euroBackSize; 146 UErrorCode err = U_ZERO_ERROR; 147 /*const char* myName = ucnv_getName(myConv, &err);*/ 148 149 targetSize = ucnv_fromUChars(myConv, 150 target, 151 sizeof(target), 152 euroString, 153 -1, 154 &err); 155 if (U_FAILURE(err)) 156 { 157 log_err("Failure Occured in ucnv_fromUChars euro roundtrip test\n"); 158 return FALSE; 159 } 160 euroBackSize = ucnv_toUChars(myConv, 161 euroBack, 162 2, 163 target, 164 targetSize, 165 &err); 166 if (U_FAILURE(err)) 167 { 168 log_err("Failure Occured in ucnv_toUChars euro roundtrip test\n"); 169 return FALSE; 170 } 171 if (u_strcmp(euroString, euroBack)) 172 { 173 /* log_err("%s FAILED Euro rountrip\n", myName);*/ 174 return FALSE; 175 } 176 else 177 { 178 /* log_verbose("%s PASSED Euro rountrip\n", myName);*/ 179 return TRUE; 180 } 181 182 } 183 #else 184 void addTestEuroRegression(TestNode** root) 185 { 186 /* test nothing... */ 187 } 188 #endif 189
