1fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius/* 2fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius******************************************************************************* 31b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert* Copyright (C) 2013-2015, International Business Machines 4fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* Corporation and others. All Rights Reserved. 5fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius******************************************************************************* 6fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* collationinfo.cpp 7fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* 8fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* created on: 2013aug05 9fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* created by: Markus W. Scherer 10fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius*/ 11fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 12fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include <stdio.h> 13fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include <string.h> 14fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 15fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "unicode/utypes.h" 16fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 17fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#if !UCONFIG_NO_COLLATION 18fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 191b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert#include "collationdata.h" 20fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "collationdatareader.h" 21fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "collationinfo.h" 22fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "uassert.h" 231b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert#include "uvectr32.h" 24fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 25fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusU_NAMESPACE_BEGIN 26fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 27fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusvoid 28fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusCollationInfo::printSizes(int32_t sizeWithHeader, const int32_t indexes[]) { 29fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t totalSize = indexes[CollationDataReader::IX_TOTAL_SIZE]; 30fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(sizeWithHeader > totalSize) { 31fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" header size: %6ld\n", (long)(sizeWithHeader - totalSize)); 32fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 33fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 34fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t length = indexes[CollationDataReader::IX_INDEXES_LENGTH]; 35fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" indexes: %6ld *4 = %6ld\n", (long)length, (long)length * 4); 36fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 37fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_REORDER_CODES_OFFSET); 38fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 39fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" reorder codes: %6ld *4 = %6ld\n", (long)length / 4, (long)length); 40fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 41fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 42fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_REORDER_TABLE_OFFSET); 43fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 44fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius U_ASSERT(length >= 256); 45fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" reorder table: %6ld\n", (long)length); 46fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 47fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 48fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_TRIE_OFFSET); 49fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 50fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" trie size: %6ld\n", (long)length); 51fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 52fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 53fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_RESERVED8_OFFSET); 54fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 55fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" reserved (offset 8): %6ld\n", (long)length); 56fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 57fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 58fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_CES_OFFSET); 59fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 60fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" CEs: %6ld *8 = %6ld\n", (long)length / 8, (long)length); 61fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 62fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 63fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_RESERVED10_OFFSET); 64fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 65fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" reserved (offset 10): %6ld\n", (long)length); 66fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 67fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 68fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_CE32S_OFFSET); 69fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 70fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" CE32s: %6ld *4 = %6ld\n", (long)length / 4, (long)length); 71fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 72fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 73fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_ROOT_ELEMENTS_OFFSET); 74fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 75fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" rootElements: %6ld *4 = %6ld\n", (long)length / 4, (long)length); 76fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 77fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 78fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_CONTEXTS_OFFSET); 79fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 80fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" contexts: %6ld *2 = %6ld\n", (long)length / 2, (long)length); 81fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 82fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 83fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_UNSAFE_BWD_OFFSET); 84fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 85fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" unsafeBwdSet: %6ld *2 = %6ld\n", (long)length / 2, (long)length); 86fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 87fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 88fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_FAST_LATIN_TABLE_OFFSET); 89fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 90fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" fastLatin table: %6ld *2 = %6ld\n", (long)length / 2, (long)length); 91fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 92fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 93fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_SCRIPTS_OFFSET); 94fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 95fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" scripts data: %6ld *2 = %6ld\n", (long)length / 2, (long)length); 96fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 97fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 98fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_COMPRESSIBLE_BYTES_OFFSET); 99fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 100fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius U_ASSERT(length >= 256); 101fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" compressibleBytes: %6ld\n", (long)length); 102fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 103fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 104fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius length = getDataLength(indexes, CollationDataReader::IX_RESERVED18_OFFSET); 105fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if(length != 0) { 106fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" reserved (offset 18): %6ld\n", (long)length); 107fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 108fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 109fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius printf(" collator binary total size: %6ld\n", (long)sizeWithHeader); 110fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius} 111fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 112fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusint32_t 113fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusCollationInfo::getDataLength(const int32_t indexes[], int32_t startIndex) { 114fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return indexes[startIndex + 1] - indexes[startIndex]; 115fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius} 116fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 1171b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubertvoid 1181b7d32f919554dda9c193b32188251337bc756f1Fredrik RoubertCollationInfo::printReorderRanges(const CollationData &data, const int32_t *codes, int32_t length) { 1191b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert UErrorCode errorCode = U_ZERO_ERROR; 1201b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert UVector32 ranges(errorCode); 1211b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert data.makeReorderRanges(codes, length, ranges, errorCode); 1221b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert if(U_FAILURE(errorCode)) { 1231b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert printf(" error building reorder ranges: %s\n", u_errorName(errorCode)); 1241b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert return; 1251b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert } 1261b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert 1271b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int32_t start = 0; 1281b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert for(int32_t i = 0; i < ranges.size(); ++i) { 1291b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int32_t pair = ranges.elementAti(i); 1301b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int32_t limit = (pair >> 16) & 0xffff; 1311b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert int16_t offset = (int16_t)pair; 1321b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert if(offset == 0) { 1331b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert // [inclusive-start, exclusive-limit[ 1341b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert printf(" [%04x, %04x[\n", start, limit); 1351b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert } else if(offset > 0) { 1361b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert printf(" reorder [%04x, %04x[ by offset %02x to [%04x, %04x[\n", 1371b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert start, limit, offset, 1381b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert start + (offset << 8), limit + (offset << 8)); 1391b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert } else /* offset < 0 */ { 1401b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert printf(" reorder [%04x, %04x[ by offset -%02x to [%04x, %04x[\n", 1411b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert start, limit, -offset, 1421b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert start + (offset << 8), limit + (offset << 8)); 1431b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert } 1441b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert start = limit; 1451b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert } 1461b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert} 1471b7d32f919554dda9c193b32188251337bc756f1Fredrik Roubert 148fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusU_NAMESPACE_END 149fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 150fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#endif // !UCONFIG_NO_COLLATION 151