icuSources/tools/gencolusb/extract_unsafe_backwards.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /**
   4  * Copyright (c) 1999-2016, International Business Machines Corporation and
   5  * others. All Rights Reserved.
   6  *
   7  * Generator for source/i18n/collunsafe.h
   8  * see Makefile
   9  */
  10
  11 #include <stdio.h>
  12 #include "unicode/uversion.h"
  13 #include "unicode/uniset.h"
  14 #include "collationroot.h"
  15 #include "collationtailoring.h"
  16
  17 /**
  18  * Define the type of generator to use. Choose one.
  19  */
  20 #define SERIALIZE 1   //< Default: use UnicodeSet.serialize() and a new internal c'tor
  21 #define RANGES 0      //< Enumerate ranges (works, not as fast. No support in collationdatareader.cpp)
  22 #define PATTERN 0     //< Generate a UnicodeSet pattern (depends on #11891 AND probably slower. No support in collationdatareader.cpp)
  23
  24 int main(int argc, const char *argv[]) {
  25     UErrorCode errorCode = U_ZERO_ERROR;
  26
  27     // Get the unsafeBackwardsSet
  28     const CollationCacheEntry *rootEntry = CollationRoot::getRootCacheEntry(errorCode);
  29     if(U_FAILURE(errorCode)) {
  30       fprintf(stderr, "Err: %s getting root cache entry\n", u_errorName(errorCode));
  31       return 1;
  32     }
  33     const UVersionInfo &version = rootEntry->tailoring->version;
  34     const UnicodeSet *unsafeBackwardSet = rootEntry->tailoring->unsafeBackwardSet;
  35     char verString[20];
  36     u_versionToString(version, verString);
  37     fprintf(stderr, "Generating data for ICU %s, Collation %s\n", U_ICU_VERSION, verString);
  38     int32_t rangeCount = unsafeBackwardSet->getRangeCount();
  39
  40 #if SERIALIZE
  41     fprintf(stderr, ".. serializing\n");
  42     // UnicodeSet serialization
  43
  44     UErrorCode preflightCode = U_ZERO_ERROR;
  45     // preflight
  46     int32_t serializedCount = unsafeBackwardSet->serialize(NULL,0,preflightCode);
  47     if(U_FAILURE(preflightCode) && preflightCode != U_BUFFER_OVERFLOW_ERROR) {
  48       fprintf(stderr, "Err: %s preflighting unicode set\n", u_errorName(preflightCode));
  49       return 1;
  50     }
  51     uint16_t *serializedData = new uint16_t[serializedCount];
  52     // serialize
  53     unsafeBackwardSet->serialize(serializedData, serializedCount, errorCode);
  54     if(U_FAILURE(errorCode)) {
  55       delete [] serializedData;
  56       fprintf(stderr, "Err: %s serializing unicodeset\n", u_errorName(errorCode));
  57       return 1;
  58     }
  59 #endif
  60
  61 #if PATTERN
  62     fprintf(stderr,".. pattern. (Note: collationdatareader.cpp does not support this form also see #11891)\n");
  63     // attempt to use pattern
  64
  65     UnicodeString pattern;
  66     UnicodeSet set(*unsafeBackwardSet);
  67     set.compact();
  68     set.toPattern(pattern, FALSE);
  69
  70     if(U_SUCCESS(errorCode)) {
  71       // This fails (bug# ?) - which is why this method was abandoned.
  72
  73       // UnicodeSet usA(pattern, errorCode);
  74       // fprintf(stderr, "\n%s:%d: err creating set A %s\n", __FILE__, __LINE__, u_errorName(errorCode));
  75       // return 1;
  76     }
  77
  78
  79     const UChar *buf = pattern.getBuffer();
  80     int32_t needed = pattern.length();
  81
  82     // print
  83     {
  84       char buf2[2048];
  85       int32_t len2 = pattern.extract(0, pattern.length(), buf2, "utf-8");
  86       buf2[len2]=0;
  87       fprintf(stderr,"===\n%s\n===\n", buf2);
  88     }
  89
  90     const UnicodeString unsafeBackwardPattern(FALSE, buf, needed);
  91   if(U_SUCCESS(errorCode)) {
  92     //UnicodeSet us(unsafeBackwardPattern, errorCode);
  93     //    fprintf(stderr, "\n%s:%d: err creating set %s\n", __FILE__, __LINE__, u_errorName(errorCode));
  94   } else {
  95     fprintf(stderr, "Uset OK - \n");
  96   }
  97 #endif
  98
  99
 100   // Generate the output file.
 101
 102   printf("// collunsafe.h\n");
 103   printf("// %s\n", U_COPYRIGHT_STRING);
 104   printf("\n");
 105   printf("// To be included by collationdatareader.cpp, and generated by gencolusb.\n");
 106   printf("// Machine generated, do not edit.\n");
 107   printf("\n");
 108   printf("#ifndef COLLUNSAFE_H\n"
 109          "#define COLLUNSAFE_H\n"
 110          "\n"
 111          "#include \"unicode/utypes.h\"\n"
 112          "\n"
 113          "#define COLLUNSAFE_ICU_VERSION \"" U_ICU_VERSION "\"\n");
 114   printf("#define COLLUNSAFE_COLL_VERSION \"%s\"\n", verString);
 115
 116
 117
 118 #if PATTERN
 119   printf("#define COLLUNSAFE_PATTERN 1\n");
 120   printf("static const int32_t collunsafe_len = %d;\n", needed);
 121   printf("static const UChar collunsafe_pattern[collunsafe_len] = {\n");
 122   for(int i=0;i<needed;i++) {
 123     if( (i>0) && (i%8 == 0) ) {
 124       printf(" // %d\n", i);
 125     }
 126     printf("0x%04X", buf[i]); // TODO check
 127     if(i != (needed-1)) {
 128       printf(", ");
 129     }
 130     }
 131   printf(" //%d\n};\n", (needed-1));
 132 #endif
 133
 134 #if RANGE
 135     fprintf(stderr, "COLLUNSAFE_RANGE - no code support in collationdatareader.cpp for this\n");
 136     printf("#define COLLUNSAFE_RANGE 1\n");
 137     printf("static const int32_t unsafe_rangeCount = %d;\n", rangeCount);
 138     printf("static const UChar32 unsafe_ranges[%d] = { \n", rangeCount*2);
 139     for(int32_t i=0;i<rangeCount;i++) {
 140       printf(" 0x%04X, 0x%04X, // %d\n",
 141              unsafeBackwardSet->getRangeStart(i),
 142              unsafeBackwardSet->getRangeEnd(i),
 143              i);
 144     }
 145     printf("};\n");
 146 #endif
 147
 148 #if SERIALIZE
 149     printf("#define COLLUNSAFE_SERIALIZE 1\n");
 150     printf("static const int32_t unsafe_serializedCount = %d;\n", serializedCount);
 151     printf("static const uint16_t unsafe_serializedData[%d] = { \n", serializedCount);
 152     for(int32_t i=0;i<serializedCount;i++) {
 153       if( (i>0) && (i%8 == 0) ) {
 154         printf(" // %d\n", i);
 155       }
 156       printf("0x%04X", serializedData[i]); // TODO check
 157       if(i != (serializedCount-1)) {
 158         printf(", ");
 159       }
 160     }
 161     printf("};\n");
 162 #endif
 163
 164     printf("#endif\n");
 165     fflush(stderr);
 166     fflush(stdout);
 167     return(U_SUCCESS(errorCode)?0:1);
 168 }