icuSources/tools/gencolusb/extract_unsafe_backwards.cpp

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