icuSources/test/intltest/nptrans.h

   1 /*
   2  *******************************************************************************
   3  *
   4  *   Copyright (C) 2003, International Business Machines
   5  *   Corporation and others.  All Rights Reserved.
   6  *
   7  *******************************************************************************
   8  *   file name:  nptrans.h
   9  *   encoding:   US-ASCII
  10  *   tab size:   8 (not used)
  11  *   indentation:4
  12  *
  13  *   created on: 2003feb1
  14  *   created by: Ram Viswanadha
  15  */
  16
  17 #ifndef NPTRANS_H
  18 #define NPTRANS_H
  19
  20 #include "unicode/utypes.h"
  21
  22 #if !UCONFIG_NO_IDNA
  23 #if !UCONFIG_NO_TRANSLITERATION
  24
  25 #include "unicode/uniset.h"
  26 #include "unicode/ures.h"
  27 #include "unicode/translit.h"
  28
  29
  30
  31 #define ASCII_SPACE 0x0020
  32
  33 class NamePrepTransform {
  34
  35 private :
  36     Transliterator *mapping;
  37     UnicodeSet unassigned;
  38     UnicodeSet prohibited;
  39     UnicodeSet labelSeparatorSet;
  40     UResourceBundle *bundle;
  41     NamePrepTransform(UParseError& parseError, UErrorCode& status);
  42
  43
  44 public :
  45
  46     static NamePrepTransform* createInstance(UParseError& parseError, UErrorCode& status);
  47
  48     virtual ~NamePrepTransform();
  49
  50
  51     inline UBool isProhibited(UChar32 ch);
  52
  53     /**
  54      * ICU "poor man's RTTI", returns a UClassID for the actual class.
  55      *
  56      * @draft ICU 2.6
  57      */
  58     inline UClassID getDynamicClassID() const { return getStaticClassID(); }
  59
  60     /**
  61      * ICU "poor man's RTTI", returns a UClassID for this class.
  62      *
  63      * @draft ICU 2.6
  64      */
  65     static inline UClassID getStaticClassID() { return (UClassID)&fgClassID; }
  66
  67     /**
  68      * Map every character in input stream with mapping character
  69      * in the mapping table and populate the output stream.
  70      * For any individual character the mapping table may specify
  71      * that that a character be mapped to nothing, mapped to one
  72      * other character or to a string of other characters.
  73      *
  74      * @param src           Pointer to UChar buffer containing a single label
  75      * @param srcLength     Number of characters in the source label
  76      * @param dest          Pointer to the destination buffer to receive the output
  77      * @param destCapacity  The capacity of destination array
  78      * @param allowUnassigned   Unassigned values can be converted to ASCII for query operations
  79      *                          If TRUE unassigned values are treated as normal Unicode code point.
  80      *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
  81      * @param status        ICU error code in/out parameter.
  82      *                      Must fulfill U_SUCCESS before the function call.
  83      * @return The number of UChars in the destination buffer
  84      *
  85      */
  86     int32_t map(const UChar* src, int32_t srcLength,
  87                         UChar* dest, int32_t destCapacity,
  88                         UBool allowUnassigned,
  89                         UParseError* parseError,
  90                         UErrorCode& status );
  91
  92     /**
  93      * Prepare the input stream with for use. This operation maps, normalizes(NFKC),
  94      * checks for prohited and BiDi characters in the order defined by RFC 3454
  95      *
  96      * @param src           Pointer to UChar buffer containing a single label
  97      * @param srcLength     Number of characters in the source label
  98      * @param dest          Pointer to the destination buffer to receive the output
  99      * @param destCapacity  The capacity of destination array
 100      * @param allowUnassigned   Unassigned values can be converted to ASCII for query operations
 101      *                          If TRUE unassigned values are treated as normal Unicode code point.
 102      *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT error code.
 103      * @param status        ICU error code in/out parameter.
 104      *                      Must fulfill U_SUCCESS before the function call.
 105      * @return The number of UChars in the destination buffer
 106      */
 107     int32_t process(const UChar* src, int32_t srcLength,
 108                             UChar* dest, int32_t destCapacity,
 109                             UBool allowUnassigned,
 110                             UParseError* parseError,
 111                             UErrorCode& status );
 112
 113     /**
 114      * Ascertain if the given code point is a label separator as specified by IDNA
 115      *
 116      * @return TRUE is the code point is a label separator
 117      *
 118      *
 119      */
 120     UBool isLabelSeparator(UChar32 ch, UErrorCode& status);
 121
 122
 123     inline UBool isLDHChar(UChar32 ch);
 124 private:
 125     /**
 126      * The address of this static class variable serves as this class's ID
 127      * for ICU "poor man's RTTI".
 128      */
 129     static const char fgClassID;
 130 };
 131
 132 inline UBool NamePrepTransform::isLDHChar(UChar32 ch){
 133     // high runner case
 134     if(ch>0x007A){
 135         return FALSE;
 136     }
 137     //[\\u002D \\u0030-\\u0039 \\u0041-\\u005A \\u0061-\\u007A]
 138     if( (ch==0x002D) ||
 139         (0x0030 <= ch && ch <= 0x0039) ||
 140         (0x0041 <= ch && ch <= 0x005A) ||
 141         (0x0061 <= ch && ch <= 0x007A)
 142       ){
 143         return TRUE;
 144     }
 145     return FALSE;
 146 }
 147
 148 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
 149 #else
 150 class NamePrepTransform {
 151 };
 152 #endif /* #if !UCONFIG_NO_IDNA */
 153
 154 #endif
 155
 156 /*
 157  * Hey, Emacs, please set the following:
 158  *
 159  * Local Variables:
 160  * indent-tabs-mode: nil
 161  * End:
 162  *
 163  */