annotate trunk/src/dil/Unicode.d @ 769:5e3ef1b2011c

Added and improved documentation.
author Aziz K?ksal <aziz.koeksal@gmail.com>
date Sun, 17 Feb 2008 02:21:55 +0100
parents 4579e8505d5e
children c1d5cfd7aa44
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
1 /++
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
2 Author: Aziz Köksal
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
3 License: GPL3
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
4 +/
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
5 module dil.Unicode;
629
d050e211402b Moved files in src/std/ to src/util/.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 518
diff changeset
6 public import util.uni : isUniAlpha;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
7
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
8 /// U+FFFD = �. Used to replace invalid Unicode characters.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
9 const dchar REPLACEMENT_CHAR = '\uFFFD';
739
49fe21aa387c Added sanitizeText() to dil.Converter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 737
diff changeset
10 const char[3] REPLACEMENT_STR = \uFFFD; /// Ditto
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
11 /// Invalid character, returned on errors.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
12 const dchar ERROR_CHAR = 0xD800;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
13
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
14 /// Returns: true if this character is not a surrogate
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
15 /// code point and not higher than 0x10FFFF.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
16 bool isValidChar(dchar d)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
17 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
18 return d < 0xD800 || d > 0xDFFF && d <= 0x10FFFF;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
19 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
20
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
21 /// There are a total of 66 noncharacters.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
22 /// Returns: true if this is one of them.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
23 /// See_also: Chapter 16.7 Noncharacters in Unicode 5.0
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
24 bool isNoncharacter(dchar d)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
25 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
26 return 0xFDD0 <= d && d <= 0xFDEF || // 32
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
27 d <= 0x10FFFF && (d & 0xFFFF) >= 0xFFFE; // 34
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
28 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
29
747
00f872d949ea Added method scanCommentText() to DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 741
diff changeset
30 /// Returns: true if this is a trail byte of a UTF-8 sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
31 bool isTrailByte(ubyte b)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
32 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
33 return (b & 0xC0) == 0x80; // 10xx_xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
34 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
35
747
00f872d949ea Added method scanCommentText() to DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 741
diff changeset
36 /// Returns: true if this is a lead byte of a UTF-8 sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
37 bool isLeadByte(ubyte b)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
38 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
39 return (b & 0xC0) == 0xC0; // 11xx_xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
40 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
41
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
42 /// Advances ref_p only if this is a valid Unicode alpha character.
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
43 bool isUnicodeAlpha(ref char* ref_p, char* end)
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
44 in { assert(ref_p && ref_p < end); }
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
45 body
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
46 {
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
47 if (*ref_p < 0x80)
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
48 return false;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
49 auto p = ref_p;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
50 auto c = decode(p, end);
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
51 if (!isUniAlpha(c))
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
52 return false;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
53 ref_p = p;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
54 return true;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
55 }
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
56
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
57 /// index is set one past the last trail byte of the valid UTF-8 sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
58 dchar decode(char[] str, ref size_t index)
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
59 in { assert(str.length && index < str.length); }
737
f88b5285b86b Implemented DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 722
diff changeset
60 out(c) { assert(isValidChar(c) || c == ERROR_CHAR); }
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
61 body
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
62 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
63 char* p = str.ptr + index;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
64 char* end = str.ptr + str.length;
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
65 dchar c = decode(p, end);
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
66 if (c != ERROR_CHAR)
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
67 index = p - str.ptr + 1;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
68 return c;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
69 }
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
70
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
71 /// ref_p is set to the last trail byte of the valid UTF-8 sequence.
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
72 dchar decode(ref char* ref_p, char* end)
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
73 in { assert(ref_p && ref_p < end); }
737
f88b5285b86b Implemented DDocEmitter.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 722
diff changeset
74 out(c) { assert(isValidChar(c) || c == ERROR_CHAR); }
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
75 body
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
76 {
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
77 char* p = ref_p;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
78 dchar c = *p;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
79
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
80 if (c < 0x80)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
81 {
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
82 ref_p++;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
83 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
84 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
85
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
86 p++; // Move to second byte.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
87 if (!(p < end))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
88 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
89
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
90 // Error if second byte is not a trail byte.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
91 if (!isTrailByte(*p))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
92 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
93
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
94 // Check for overlong sequences.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
95 switch (c)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
96 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
97 case 0xE0, // 11100000 100xxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
98 0xF0, // 11110000 1000xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
99 0xF8, // 11111000 10000xxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
100 0xFC: // 11111100 100000xx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
101 if ((*p & c) == 0x80)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
102 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
103 default:
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
104 if ((c & 0xFE) == 0xC0) // 1100000x
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
105 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
106 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
107
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
108 const char[] checkNextByte = "if (!(++p < end && isTrailByte(*p)))"
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
109 " return ERROR_CHAR;";
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
110 const char[] appendSixBits = "c = (c << 6) | *p & 0b0011_1111;";
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
111
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
112 // Decode
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
113 if ((c & 0b1110_0000) == 0b1100_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
114 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
115 // 110xxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
116 c &= 0b0001_1111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
117 mixin(appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
118 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
119 else if ((c & 0b1111_0000) == 0b1110_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
120 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
121 // 1110xxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
122 c &= 0b0000_1111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
123 mixin(appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
124 checkNextByte ~ appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
125 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
126 else if ((c & 0b1111_1000) == 0b1111_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
127 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
128 // 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
129 c &= 0b0000_0111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
130 mixin(appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
131 checkNextByte ~ appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
132 checkNextByte ~ appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
133 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
134 else
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
135 // 5 and 6 byte UTF-8 sequences are not allowed yet.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
136 // 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
137 // 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
138 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
139
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
140 assert(isTrailByte(*p));
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
141
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
142 if (!isValidChar(c))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
143 return ERROR_CHAR;
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
144 ref_p = p;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
145 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
146 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
147
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
148 /// Encodes a character and appends it to str.
518
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
149 void encode(ref char[] str, dchar c)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
150 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
151 assert(isValidChar(c), "check if character is valid before calling encode().");
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
152
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
153 char[6] b = void;
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
154 if (c < 0x80)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
155 str ~= c;
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
156 else if (c < 0x800)
518
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
157 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
158 b[0] = 0xC0 | (c >> 6);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
159 b[1] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
160 str ~= b[0..2];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
161 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
162 else if (c < 0x10000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
163 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
164 b[0] = 0xE0 | (c >> 12);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
165 b[1] = 0x80 | ((c >> 6) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
166 b[2] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
167 str ~= b[0..3];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
168 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
169 else if (c < 0x200000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
170 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
171 b[0] = 0xF0 | (c >> 18);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
172 b[1] = 0x80 | ((c >> 12) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
173 b[2] = 0x80 | ((c >> 6) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
174 b[3] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
175 str ~= b[0..4];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
176 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
177 /+ // There are no 5 and 6 byte UTF-8 sequences yet.
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
178 else if (c < 0x4000000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
179 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
180 b[0] = 0xF8 | (c >> 24);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
181 b[1] = 0x80 | ((c >> 18) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
182 b[2] = 0x80 | ((c >> 12) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
183 b[3] = 0x80 | ((c >> 6) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
184 b[4] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
185 str ~= b[0..5];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
186 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
187 else if (c < 0x80000000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
188 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
189 b[0] = 0xFC | (c >> 30);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
190 b[1] = 0x80 | ((c >> 24) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
191 b[2] = 0x80 | ((c >> 18) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
192 b[3] = 0x80 | ((c >> 12) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
193 b[4] = 0x80 | ((c >> 6) & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
194 b[5] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
195 str ~= b[0..6];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
196 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
197 +/
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
198 else
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
199 assert(0);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
200 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
201
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
202 /// Encodes a character and appends it to str.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
203 void encode(ref wchar[] str, dchar c)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
204 in { assert(isValidChar(c)); }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
205 body
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
206 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
207 if (c < 0x10000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
208 str ~= cast(wchar)c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
209 else
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
210 { // Encode with surrogate pair.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
211 wchar[2] pair = void;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
212 c -= 0x10000; // c'
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
213 // higher10bits(c') | 0b1101_10xx_xxxx_xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
214 pair[0] = (c >> 10) | 0xD800;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
215 // lower10bits(c') | 0b1101_11yy_yyyy_yyyy
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
216 pair[1] = (c & 0x3FF) | 0xDC00;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
217 str ~= pair;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
218 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
219 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
220
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
221 /// Returns a decoded character from a UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
222 /// Returns: ERROR_CHAR in case of an error in the sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
223 /// Params:
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
224 /// str = the UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
225 /// index = where to start from.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
226 dchar decode(wchar[] str, ref size_t index)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
227 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
228 assert(str.length && index < str.length);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
229 dchar c = str[index];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
230 if (0xD800 > c || c > 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
231 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
232 ++index;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
233 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
234 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
235 if (c <= 0xDBFF && index+1 != str.length)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
236 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
237 wchar c2 = str[index+1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
238 if (0xDC00 <= c2 && c2 <= 0xDFFF)
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
239 { // Decode surrogate pair.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
240 // (c - 0xD800) << 10 + 0x10000 ->
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
241 // (c - 0xD800 + 0x40) << 10 ->
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
242 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
243 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
244 index += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
245 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
246 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
247 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
248 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
249 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
250
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
251 /// Returns a decoded character from a UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
252 /// Returns: ERROR_CHAR in case of an error in the sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
253 /// Params:
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
254 /// p = start of the UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
255 /// end = one past the end of the sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
256 dchar decode(ref wchar* p, wchar* end)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
257 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
258 assert(p && p < end);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
259 dchar c = *p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
260 if (0xD800 > c || c > 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
261 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
262 ++p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
263 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
264 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
265 if (c <= 0xDBFF && p+1 != end)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
266 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
267 wchar c2 = p[1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
268 if (0xDC00 <= c2 && c2 <= 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
269 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
270 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
271 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
272 p += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
273 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
274 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
275 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
276 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
277 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
278
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
279 /// Decode a character from a zero-terminated string.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
280 dchar decode(ref wchar* p)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
281 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
282 assert(p);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
283 dchar c = *p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
284 if (0xD800 > c || c > 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
285 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
286 ++p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
287 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
288 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
289 if (c <= 0xDBFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
290 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
291 wchar c2 = p[1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
292 if (0xDC00 <= c2 && c2 <= 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
293 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
294 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
295 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
296 p += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
297 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
298 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
299 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
300 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
301 }