1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 2003-2006, 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 #include "intltest.h"
30
31
32 #define ASCII_SPACE 0x0020
33
34 class NamePrepTransform {
35
36 private :
37 Transliterator *mapping;
38 UnicodeSet unassigned;
39 UnicodeSet prohibited;
40 UnicodeSet labelSeparatorSet;
41 UResourceBundle *bundle;
42 NamePrepTransform(UParseError& parseError, UErrorCode& status);
43
44
45 public :
46
47 static NamePrepTransform* createInstance(UParseError& parseError, UErrorCode& status);
48
49 virtual ~NamePrepTransform();
50
51
52 inline UBool isProhibited(UChar32 ch);
53
54 /**
55 * ICU "poor man's RTTI", returns a UClassID for the actual class.
56 *
57 * @draft ICU 2.6
58 */
getDynamicClassID()59 inline UClassID getDynamicClassID() const { return getStaticClassID(); }
60
61 /**
62 * ICU "poor man's RTTI", returns a UClassID for this class.
63 *
64 * @draft ICU 2.6
65 */
getStaticClassID()66 static inline UClassID getStaticClassID() { return (UClassID)&fgClassID; }
67
68 /**
69 * Map every character in input stream with mapping character
70 * in the mapping table and populate the output stream.
71 * For any individual character the mapping table may specify
72 * that that a character be mapped to nothing, mapped to one
73 * other character or to a string of other characters.
74 *
75 * @param src Pointer to UChar buffer containing a single label
76 * @param srcLength Number of characters in the source label
77 * @param dest Pointer to the destination buffer to receive the output
78 * @param destCapacity The capacity of destination array
79 * @param allowUnassigned Unassigned values can be converted to ASCII for query operations
80 * If TRUE unassigned values are treated as normal Unicode code point.
81 * If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
82 * @param status ICU error code in/out parameter.
83 * Must fulfill U_SUCCESS before the function call.
84 * @return The number of UChars in the destination buffer
85 *
86 */
87 int32_t map(const UChar* src, int32_t srcLength,
88 UChar* dest, int32_t destCapacity,
89 UBool allowUnassigned,
90 UParseError* parseError,
91 UErrorCode& status );
92
93 /**
94 * Prepare the input stream with for use. This operation maps, normalizes(NFKC),
95 * checks for prohited and BiDi characters in the order defined by RFC 3454
96 *
97 * @param src Pointer to UChar buffer containing a single label
98 * @param srcLength Number of characters in the source label
99 * @param dest Pointer to the destination buffer to receive the output
100 * @param destCapacity The capacity of destination array
101 * @param allowUnassigned Unassigned values can be converted to ASCII for query operations
102 * If TRUE unassigned values are treated as normal Unicode code point.
103 * If FALSE the operation fails with U_UNASSIGNED_CODE_POINT error code.
104 * @param status ICU error code in/out parameter.
105 * Must fulfill U_SUCCESS before the function call.
106 * @return The number of UChars in the destination buffer
107 */
108 int32_t process(const UChar* src, int32_t srcLength,
109 UChar* dest, int32_t destCapacity,
110 UBool allowUnassigned,
111 UParseError* parseError,
112 UErrorCode& status );
113
114 /**
115 * Ascertain if the given code point is a label separator as specified by IDNA
116 *
117 * @return TRUE is the code point is a label separator
118 *
119 *
120 */
121 UBool isLabelSeparator(UChar32 ch, UErrorCode& status);
122
123
124 inline UBool isLDHChar(UChar32 ch);
125 private:
126 /**
127 * The address of this static class variable serves as this class's ID
128 * for ICU "poor man's RTTI".
129 */
130 static const char fgClassID;
131 };
132
isLDHChar(UChar32 ch)133 inline UBool NamePrepTransform::isLDHChar(UChar32 ch){
134 // high runner case
135 if(ch>0x007A){
136 return FALSE;
137 }
138 //[\\u002D \\u0030-\\u0039 \\u0041-\\u005A \\u0061-\\u007A]
139 if( (ch==0x002D) ||
140 (0x0030 <= ch && ch <= 0x0039) ||
141 (0x0041 <= ch && ch <= 0x005A) ||
142 (0x0061 <= ch && ch <= 0x007A)
143 ){
144 return TRUE;
145 }
146 return FALSE;
147 }
148
149 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
150 #else
151 class NamePrepTransform {
152 };
153 #endif /* #if !UCONFIG_NO_IDNA */
154
155 #endif
156
157 /*
158 * Hey, Emacs, please set the following:
159 *
160 * Local Variables:
161 * indent-tabs-mode: nil
162 * End:
163 *
164 */
165