1ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru/* 2ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ******************************************************************************* 3ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho * Copyright (C) 2005-2009, International Business Machines 5ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * Corporation and others. All Rights Reserved. 6ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 7ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ******************************************************************************* 8ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 9ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * created on: 2005jun15 10ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * created by: Raymond Yang 11ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru */ 12ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 13ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#if !UCONFIG_NO_IDNA 14ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 15ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include <stdio.h> 16ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include <stdlib.h> 17ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include <string.h> 18ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include "unicode/utypes.h" 19ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include "unicode/ucnv.h" 20ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include "unicode/ustring.h" 21ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include "unicode/uidna.h" 22ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 23ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#include "idnaconf.h" 24ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 25ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_TAG[] = {0x3D, 0x3D, 0x3D, 0x3D, 0x3D, 0}; // ===== 26ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_NAMEZONE[] = {0x6E, 0x61, 0x6D, 0x65, 0x7A, 0x6F, 0x6E, 0x65, 0}; // namezone 27ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_NAMEBASE[] = {0x6E, 0x61, 0x6D, 0x65, 0x62, 0x61, 0x73, 0x65, 0}; // namebase 28ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_NAMEUTF8[] = {0x6E, 0x61, 0x6D, 0x65, 0x75, 0x74, 0x66, 0x38, 0}; // nameutf8 29ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 30ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_TYPE[] = {0x74, 0x79, 0x70, 0x65, 0}; // type 31ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_TOASCII[] = {0x74, 0x6F, 0x61, 0x73, 0x63, 0x69, 0x69, 0}; // toascii 32ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_TOUNICODE[] = {0x74, 0x6F, 0x75, 0x6E, 0x69, 0x63, 0x6F, 0x64, 0x65, 0}; // tounicode 33ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 34ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_PASSFAIL[] = {0x70, 0x61, 0x73, 0x73, 0x66, 0x61, 0x69, 0x6C, 0}; // passfail 35ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_PASS[] = {0x70, 0x61, 0x73, 0x73, 0}; // pass 36ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_FAIL[] = {0x66, 0x61, 0x69, 0x6C, 0}; // fail 37ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 38ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_DESC[] = {0x64, 0x65, 0x73, 0x63, 0}; // desc 39ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Querustatic const UChar C_USESTD3ASCIIRULES[] = {0x55, 0x73, 0x65, 0x53, 0x54, 0x44, 40ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 0x33, 0x41, 0x53, 0x43, 0x49, 0x49, 0x52, 0x75, 0x6C, 0x65, 0x73, 0}; // UseSTD3ASCIIRules 41ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 42ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste QueruIdnaConfTest::IdnaConfTest(){ 43ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru base = NULL; 44ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru len = 0; 45ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru curOffset = 0; 46ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 47ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru type = option = passfail = -1; 48ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namebase.setToBogus(); 49ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namezone.setToBogus(); 50ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 51ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste QueruIdnaConfTest::~IdnaConfTest(){ 52ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru delete [] base; 53ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 54ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 55ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#if !UCONFIG_NO_IDNA 56ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru/* this function is modified from RBBITest::ReadAndConvertFile() 57ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 58ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru */ 59ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste QueruUBool IdnaConfTest::ReadAndConvertFile(){ 60ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 61ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru char * source = NULL; 62ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru size_t source_len; 63ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 64ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // read the test data file to memory 65ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru FILE* f = NULL; 66ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UErrorCode status = U_ZERO_ERROR; 67ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 68ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru const char *path = IntlTest::getSourceTestData(status); 69ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (U_FAILURE(status)) { 70ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("%s", u_errorName(status)); 71ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return FALSE; 72ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 73ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 74ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru const char* name = "idna_conf.txt"; // test data file 75ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru int t = strlen(path) + strlen(name) + 1; 76ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru char* absolute_name = new char[t]; 77ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru strcpy(absolute_name, path); 78ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru strcat(absolute_name, name); 79ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru f = fopen(absolute_name, "rb"); 80ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru delete [] absolute_name; 81ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 82ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (f == NULL){ 8385bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho dataerrln("fopen error on %s", name); 84ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return FALSE; 85ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 86ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 87ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru fseek( f, 0, SEEK_END); 88ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ((source_len = ftell(f)) <= 0){ 89ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("Error reading test data file."); 90ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru fclose(f); 91ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return FALSE; 92ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 93ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 94ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru source = new char[source_len]; 95ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru fseek(f, 0, SEEK_SET); 96ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (fread(source, 1, source_len, f) != source_len) { 97ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("Error reading test data file."); 98ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru delete [] source; 99ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru fclose(f); 100ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return FALSE; 101ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 102ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru fclose(f); 103ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 104ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // convert the UTF-8 encoded stream to UTF-16 stream 105ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UConverter* conv = ucnv_open("utf-8", &status); 106ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru int dest_len = ucnv_toUChars(conv, 107ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru NULL, // dest, 108ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 0, // destCapacity, 109ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru source, 110ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru source_len, 111ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru &status); 112ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (status == U_BUFFER_OVERFLOW_ERROR) { 113ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // Buffer Overflow is expected from the preflight operation. 114ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru status = U_ZERO_ERROR; 115ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar * dest = NULL; 116ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru dest = new UChar[ dest_len + 1]; 117ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ucnv_toUChars(conv, dest, dest_len + 1, source, source_len, &status); 118ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // Do not know the "if possible" behavior of ucnv_toUChars() 119ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // Do it by ourself. 120ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru dest[dest_len] = 0; 121ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru len = dest_len; 122ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru base = dest; 123ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru delete [] source; 124ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ucnv_close(conv); 125ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return TRUE; // The buffer will owned by caller. 126ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 127ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("UConverter error: %s", u_errorName(status)); 128ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru delete [] source; 129ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ucnv_close(conv); 130ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return FALSE; 131ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 132ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 133ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruint IdnaConfTest::isNewlineMark(){ 134ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru static const UChar LF = 0x0a; 135ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru static const UChar CR = 0x0d; 136ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar c = base[curOffset]; 137ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // CR LF 138ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ( c == CR && curOffset + 1 < len && base[curOffset + 1] == LF){ 139ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return 2; 140ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 141ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 142ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // CR or LF 143ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ( c == CR || c == LF) { 144ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return 1; 145ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 146ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 147ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return 0; 148ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 149ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 150ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru/* Read a logical line. 151ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 152ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * All lines ending in a backslash (\) and immediately followed by a newline 153ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * character are joined with the next line in the source file forming logical 154ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * lines from the physical lines. 155ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 156ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru */ 157ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste QueruUBool IdnaConfTest::ReadOneLine(UnicodeString& buf){ 158ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ( !(curOffset < len) ) return FALSE; // stream end 159ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 160ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru static const UChar BACKSLASH = 0x5c; 161ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru buf.remove(); 162ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru int t = 0; 163ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru while (curOffset < len){ 164ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ((t = isNewlineMark())) { // end of line 165ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru curOffset += t; 166ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru break; 167ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 168ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar c = base[curOffset]; 169ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (c == BACKSLASH && curOffset < len -1){ // escaped new line mark 170ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ((t = isNewlineMark())){ 171ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru curOffset += 1 + t; // BACKSLAH and NewlineMark 172ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru continue; 173ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 174ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru }; 175ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru buf.append(c); 176ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru curOffset++; 177ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 178ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return TRUE; 179ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 180ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 181ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru// 182ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru//=============================================================== 183ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru// 184ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 185ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru/* Explain <xxxxx> tag to a native value 186ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * 187ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * Since <xxxxx> is always larger than the native value, 188ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * the operation will replace the tag directly in the buffer, 189ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru * and, of course, will shift tail elements. 190ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru */ 191ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruvoid IdnaConfTest::ExplainCodePointTag(UnicodeString& buf){ 192ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru buf.append((UChar)0); // add a terminal NULL 193ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar* bufBase = buf.getBuffer(buf.length()); 194ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar* p = bufBase; 195ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru while (*p != 0){ 196ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if ( *p != 0x3C){ // < 197ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru *bufBase++ = *p++; 198ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 199ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru p++; // skip < 200ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar32 cp = 0; 201ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru for ( ;*p != 0x3E; p++){ // > 202ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (0x30 <= *p && *p <= 0x39){ // 0-9 203ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru cp = (cp * 16) + (*p - 0x30); 204ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (0x61 <= *p && *p <= 0x66){ // a-f 205ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru cp = (cp * 16) + (*p - 0x61) + 10; 206ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (0x41 <= *p && *p <= 0x46) {// A-F 207ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru cp = (cp * 16) + (*p - 0x41) + 10; 208ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 209ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // no else. hope everything is good. 210ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 211ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru p++; // skip > 212ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (U_IS_BMP(cp)){ 213ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru *bufBase++ = cp; 214ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 215ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru *bufBase++ = U16_LEAD(cp); 216ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru *bufBase++ = U16_TRAIL(cp); 217ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 218ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 219ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 220ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru *bufBase = 0; // close our buffer 221ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru buf.releaseBuffer(); 222ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 223ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 224ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruvoid IdnaConfTest::Call(){ 225ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (type == -1 || option == -1 || passfail == -1 || namebase.isBogus() || namezone.isBogus()){ 226ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("Incomplete record"); 227ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 228ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UErrorCode status = U_ZERO_ERROR; 229ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UChar result[200] = {0,}; // simple life 230ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru const UChar *p = namebase.getTerminatedBuffer(); 231ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru const int p_len = namebase.length(); 232ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 233ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (type == 0 && option == 0){ 234ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru uidna_IDNToASCII(p, p_len, result, 200, UIDNA_USE_STD3_RULES, NULL, &status); 235ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (type == 0 && option == 1){ 236ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru uidna_IDNToASCII(p, p_len, result, 200, UIDNA_ALLOW_UNASSIGNED, NULL, &status); 237ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (type == 1 && option == 0){ 238ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru uidna_IDNToUnicode(p, p_len, result, 200, UIDNA_USE_STD3_RULES, NULL, &status); 239ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (type == 1 && option == 1){ 240ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru uidna_IDNToUnicode(p, p_len, result, 200, UIDNA_ALLOW_UNASSIGNED, NULL, &status); 241ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 242ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (passfail == 0){ 243ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (U_FAILURE(status)){ 24485bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho id.append(" should pass, but failed. - "); 24585bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho id.append(u_errorName(status)); 24685bf2e2fbc60a9f938064abc8127d61da7d19882Claire Ho errcheckln(status, id); 247ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else{ 248ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (namezone.compare(result, -1) == 0){ 249ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // expected 250ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru logln(UnicodeString("namebase: ") + prettify(namebase) + UnicodeString(" result: ") + prettify(result)); 251ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 252ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru id.append(" no error, but result is not as expected."); 253ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln(id); 254ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 255ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 256ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (passfail == 1){ 257ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (U_FAILURE(status)){ 258ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // expected 259ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // TODO: Uncomment this when U_IDNA_ZERO_LENGTH_LABEL_ERROR is added to u_errorName 260ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru //logln("Got the expected error: " + UnicodeString(u_errorName(status))); 261ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else{ 262ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (namebase.compare(result, -1) == 0){ 263ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // garbage in -> garbage out 264ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru logln(UnicodeString("ICU will not recognize malformed ACE-Prefixes or incorrect ACE-Prefixes. ") + UnicodeString("namebase: ") + prettify(namebase) + UnicodeString(" result: ") + prettify(result)); 265ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 266ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru id.append(" should fail, but not failed. "); 267ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru id.append(u_errorName(status)); 268ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln(id); 269ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 270ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 271ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 272ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 273ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru type = option = passfail = -1; 274ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namebase.setToBogus(); 275ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namezone.setToBogus(); 276ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru id.remove(); 277ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru return; 278ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 279ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 280ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruvoid IdnaConfTest::Test(void){ 281ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (!ReadAndConvertFile())return; 282ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 283ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UnicodeString s; 284ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UnicodeString key; 285ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru UnicodeString value; 286ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 287ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // skip everything before the first "=====" and "=====" itself 288ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru do { 289ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (!ReadOneLine(s)) { 290ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru errln("End of file prematurely found"); 291ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru break; 292ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 293ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 294ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru while (s.compare(C_TAG, -1) != 0); //"=====" 295ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 296ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru while(ReadOneLine(s)){ 297ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru s.trim(); 298ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru key.remove(); 299ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru value.remove(); 300ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (s.compare(C_TAG, -1) == 0){ //"=====" 301ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru Call(); 302ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 303ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // explain key:value 304ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru int p = s.indexOf((UChar)0x3A); // : 305ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru key.setTo(s,0,p).trim(); 306ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru value.setTo(s,p+1).trim(); 307ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (key.compare(C_TYPE, -1) == 0){ 308ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (value.compare(C_TOASCII, -1) == 0) { 309ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru type = 0; 310ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (value.compare(C_TOUNICODE, -1) == 0){ 311ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru type = 1; 312ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 313ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (key.compare(C_PASSFAIL, -1) == 0){ 314ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (value.compare(C_PASS, -1) == 0){ 315ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru passfail = 0; 316ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (value.compare(C_FAIL, -1) == 0){ 317ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru passfail = 1; 318ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 319ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (key.compare(C_DESC, -1) == 0){ 320ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru if (value.indexOf(C_USESTD3ASCIIRULES, u_strlen(C_USESTD3ASCIIRULES), 0) == -1){ 321ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru option = 1; // not found 322ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else { 323ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru option = 0; 324ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 325ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru id.setTo(value, 0, value.indexOf((UChar)0x20)); // space 326ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (key.compare(C_NAMEZONE, -1) == 0){ 327ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ExplainCodePointTag(value); 328ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namezone.setTo(value); 329ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } else if (key.compare(C_NAMEBASE, -1) == 0){ 330ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru ExplainCodePointTag(value); 331ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru namebase.setTo(value); 332ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 333ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // just skip other lines 334ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 335ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 336ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 337ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru Call(); // for last record 338ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 339ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#else 340ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruvoid IdnaConfTest::Test(void) 341ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru{ 342ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru // test nothing... 343ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 344ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#endif 345ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 346ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queruvoid IdnaConfTest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par*/){ 347ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru switch (index) { 348ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru TESTCASE(0,Test); 349ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru default: name = ""; break; 350ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru } 351ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru} 352ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru 353ac04d0bbe12b3ef54518635711412f178cb4d16Jean-Baptiste Queru#endif 354