Home | History | Annotate | Download | only in toolutil
      1 //  2016 and later: Unicode, Inc. and others.
      2 // License & terms of use: http://www.unicode.org/copyright.html
      3 /*
      4 *******************************************************************************
      5 * Copyright (C) 2013-2015, International Business Machines
      6 * Corporation and others.  All Rights Reserved.
      7 *******************************************************************************
      8 * collationinfo.cpp
      9 *
     10 * created on: 2013aug05
     11 * created by: Markus W. Scherer
     12 */
     13 
     14 #include <stdio.h>
     15 #include <string.h>
     16 
     17 #include "unicode/utypes.h"
     18 
     19 #if !UCONFIG_NO_COLLATION
     20 
     21 #include "collationdata.h"
     22 #include "collationdatareader.h"
     23 #include "collationinfo.h"
     24 #include "uassert.h"
     25 #include "uvectr32.h"
     26 
     27 U_NAMESPACE_BEGIN
     28 
     29 void
     30 CollationInfo::printSizes(int32_t sizeWithHeader, const int32_t indexes[]) {
     31     int32_t totalSize = indexes[CollationDataReader::IX_TOTAL_SIZE];
     32     if(sizeWithHeader > totalSize) {
     33         printf("  header size:                  %6ld\n", (long)(sizeWithHeader - totalSize));
     34     }
     35 
     36     int32_t length = indexes[CollationDataReader::IX_INDEXES_LENGTH];
     37     printf("  indexes:          %6ld *4 = %6ld\n", (long)length, (long)length * 4);
     38 
     39     length = getDataLength(indexes, CollationDataReader::IX_REORDER_CODES_OFFSET);
     40     if(length != 0) {
     41         printf("  reorder codes:    %6ld *4 = %6ld\n", (long)length / 4, (long)length);
     42     }
     43 
     44     length = getDataLength(indexes, CollationDataReader::IX_REORDER_TABLE_OFFSET);
     45     if(length != 0) {
     46         U_ASSERT(length >= 256);
     47         printf("  reorder table:                %6ld\n", (long)length);
     48     }
     49 
     50     length = getDataLength(indexes, CollationDataReader::IX_TRIE_OFFSET);
     51     if(length != 0) {
     52         printf("  trie size:                    %6ld\n", (long)length);
     53     }
     54 
     55     length = getDataLength(indexes, CollationDataReader::IX_RESERVED8_OFFSET);
     56     if(length != 0) {
     57         printf("  reserved (offset 8):          %6ld\n", (long)length);
     58     }
     59 
     60     length = getDataLength(indexes, CollationDataReader::IX_CES_OFFSET);
     61     if(length != 0) {
     62         printf("  CEs:              %6ld *8 = %6ld\n", (long)length / 8, (long)length);
     63     }
     64 
     65     length = getDataLength(indexes, CollationDataReader::IX_RESERVED10_OFFSET);
     66     if(length != 0) {
     67         printf("  reserved (offset 10):         %6ld\n", (long)length);
     68     }
     69 
     70     length = getDataLength(indexes, CollationDataReader::IX_CE32S_OFFSET);
     71     if(length != 0) {
     72         printf("  CE32s:            %6ld *4 = %6ld\n", (long)length / 4, (long)length);
     73     }
     74 
     75     length = getDataLength(indexes, CollationDataReader::IX_ROOT_ELEMENTS_OFFSET);
     76     if(length != 0) {
     77         printf("  rootElements:     %6ld *4 = %6ld\n", (long)length / 4, (long)length);
     78     }
     79 
     80     length = getDataLength(indexes, CollationDataReader::IX_CONTEXTS_OFFSET);
     81     if(length != 0) {
     82         printf("  contexts:         %6ld *2 = %6ld\n", (long)length / 2, (long)length);
     83     }
     84 
     85     length = getDataLength(indexes, CollationDataReader::IX_UNSAFE_BWD_OFFSET);
     86     if(length != 0) {
     87         printf("  unsafeBwdSet:     %6ld *2 = %6ld\n", (long)length / 2, (long)length);
     88     }
     89 
     90     length = getDataLength(indexes, CollationDataReader::IX_FAST_LATIN_TABLE_OFFSET);
     91     if(length != 0) {
     92         printf("  fastLatin table:  %6ld *2 = %6ld\n", (long)length / 2, (long)length);
     93     }
     94 
     95     length = getDataLength(indexes, CollationDataReader::IX_SCRIPTS_OFFSET);
     96     if(length != 0) {
     97         printf("  scripts data:     %6ld *2 = %6ld\n", (long)length / 2, (long)length);
     98     }
     99 
    100     length = getDataLength(indexes, CollationDataReader::IX_COMPRESSIBLE_BYTES_OFFSET);
    101     if(length != 0) {
    102         U_ASSERT(length >= 256);
    103         printf("  compressibleBytes:            %6ld\n", (long)length);
    104     }
    105 
    106     length = getDataLength(indexes, CollationDataReader::IX_RESERVED18_OFFSET);
    107     if(length != 0) {
    108         printf("  reserved (offset 18):         %6ld\n", (long)length);
    109     }
    110 
    111     printf(" collator binary total size:    %6ld\n", (long)sizeWithHeader);
    112 }
    113 
    114 int32_t
    115 CollationInfo::getDataLength(const int32_t indexes[], int32_t startIndex) {
    116     return indexes[startIndex + 1] - indexes[startIndex];
    117 }
    118 
    119 void
    120 CollationInfo::printReorderRanges(const CollationData &data, const int32_t *codes, int32_t length) {
    121     UErrorCode errorCode = U_ZERO_ERROR;
    122     UVector32 ranges(errorCode);
    123     data.makeReorderRanges(codes, length, ranges, errorCode);
    124     if(U_FAILURE(errorCode)) {
    125         printf("  error building reorder ranges: %s\n", u_errorName(errorCode));
    126         return;
    127     }
    128 
    129     int32_t start = 0;
    130     for(int32_t i = 0; i < ranges.size(); ++i) {
    131         int32_t pair = ranges.elementAti(i);
    132         int32_t limit = (pair >> 16) & 0xffff;
    133         int16_t offset = (int16_t)pair;
    134         if(offset == 0) {
    135             // [inclusive-start, exclusive-limit[
    136             printf("          [%04x, %04x[\n", start, limit);
    137         } else if(offset > 0) {
    138             printf("  reorder [%04x, %04x[ by offset  %02x to [%04x, %04x[\n",
    139                     start, limit, offset,
    140                     start + (offset << 8), limit + (offset << 8));
    141         } else /* offset < 0 */ {
    142             printf("  reorder [%04x, %04x[ by offset -%02x to [%04x, %04x[\n",
    143                     start, limit, -offset,
    144                     start + (offset << 8), limit + (offset << 8));
    145         }
    146         start = limit;
    147     }
    148 }
    149 
    150 U_NAMESPACE_END
    151 
    152 #endif  // !UCONFIG_NO_COLLATION
    153