annotate src/dil/Unicode.d @ 806:bcb74c9b895c

Moved out files in the trunk folder to the root.
author Aziz K?ksal <aziz.koeksal@gmail.com>
date Sun, 09 Mar 2008 00:12:19 +0100
parents trunk/src/dil/Unicode.d@c1d5cfd7aa44
children
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
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
57 /// Decodes a character from str at index.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
58 /// Params:
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
59 /// index = set to one past the ASCII char or one past the last trail byte
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
60 /// of the valid UTF-8 sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
61 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
62 in { assert(str.length && index < str.length); }
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
63 out { assert(index <= str.length); }
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
64 body
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
65 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
66 char* p = str.ptr + index;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
67 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
68 dchar c = decode(p, end);
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
69 if (c != ERROR_CHAR)
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
70 index = p - str.ptr;
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
71 return c;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
72 }
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
73
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
74 /// Decodes a character starting at ref_p.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
75 /// Params:
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
76 /// ref_p = set to one past the ASCII char or one past the last trail byte
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
77 /// of the valid UTF-8 sequence.
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
78 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
79 in { assert(ref_p && ref_p < end); }
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
80 out(c) { assert(ref_p <= end && (isValidChar(c) || c == ERROR_CHAR)); }
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
81 body
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
82 {
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
83 char* p = ref_p;
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
84 dchar c = *p;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
85
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
86 if (c < 0x80)
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
87 return ref_p++, c;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
88
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
89 p++; // Move to second byte.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
90 if (!(p < end))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
91 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
92
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
93 // Error if second byte is not a trail byte.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
94 if (!isTrailByte(*p))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
95 return ERROR_CHAR;
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 // Check for overlong sequences.
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
98 switch (c)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
99 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
100 case 0xE0, // 11100000 100xxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
101 0xF0, // 11110000 1000xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
102 0xF8, // 11111000 10000xxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
103 0xFC: // 11111100 100000xx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
104 if ((*p & c) == 0x80)
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 default:
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
107 if ((c & 0xFE) == 0xC0) // 1100000x
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
108 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
109 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
110
722
ceaac6a24258 Added isUnicodeAlpha() for DDocParser and MacroParser.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 629
diff changeset
111 const char[] checkNextByte = "if (!(++p < end && isTrailByte(*p)))"
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
112 " return ERROR_CHAR;";
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
113 const char[] appendSixBits = "c = (c << 6) | *p & 0b0011_1111;";
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 // Decode
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
116 if ((c & 0b1110_0000) == 0b1100_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
117 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
118 // 110xxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
119 c &= 0b0001_1111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
120 mixin(appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
121 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
122 else if ((c & 0b1111_0000) == 0b1110_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
123 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
124 // 1110xxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
125 c &= 0b0000_1111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
126 mixin(appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
127 checkNextByte ~ appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
128 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
129 else if ((c & 0b1111_1000) == 0b1111_0000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
130 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
131 // 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
132 c &= 0b0000_0111;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
133 mixin(appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
134 checkNextByte ~ appendSixBits ~
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
135 checkNextByte ~ appendSixBits);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
136 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
137 else
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
138 // 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
139 // 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
140 // 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
141 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
142
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
143 assert(isTrailByte(*p));
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
144
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
145 if (!isValidChar(c))
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
146 return ERROR_CHAR;
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
147 ref_p = p+1;
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
148 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
149 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
150
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
151 /// Encodes c 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
152 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
153 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
154 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
155
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
156 char[6] b = void;
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
157 if (c < 0x80)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
158 str ~= c;
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
159 else if (c < 0x800)
518
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
160 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
161 b[0] = 0xC0 | (c >> 6);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
162 b[1] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
163 str ~= b[0..2];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
164 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
165 else if (c < 0x10000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
166 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
167 b[0] = 0xE0 | (c >> 12);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
168 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
169 b[2] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
170 str ~= b[0..3];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
171 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
172 else if (c < 0x200000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
173 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
174 b[0] = 0xF0 | (c >> 18);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
175 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
176 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
177 b[3] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
178 str ~= b[0..4];
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 /+ // 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
181 else if (c < 0x4000000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
182 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
183 b[0] = 0xF8 | (c >> 24);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
184 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
185 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
186 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
187 b[4] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
188 str ~= b[0..5];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
189 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
190 else if (c < 0x80000000)
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
191 {
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
192 b[0] = 0xFC | (c >> 30);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
193 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
194 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
195 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
196 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
197 b[5] = 0x80 | (c & 0x3F);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
198 str ~= b[0..6];
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
199 }
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 else
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
202 assert(0);
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
203 }
8f86bb9ef715 Added module dil.Converter and dil.FileBOM.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 510
diff changeset
204
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
205 /// Encodes c and appends it to str.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
206 void encode(ref wchar[] str, dchar c)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
207 in { assert(isValidChar(c)); }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
208 body
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
209 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
210 if (c < 0x10000)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
211 str ~= cast(wchar)c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
212 else
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
213 { // Encode with surrogate pair.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
214 wchar[2] pair = void;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
215 c -= 0x10000; // c'
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
216 // higher10bits(c') | 0b1101_10xx_xxxx_xxxx
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
217 pair[0] = (c >> 10) | 0xD800;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
218 // lower10bits(c') | 0b1101_11yy_yyyy_yyyy
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
219 pair[1] = (c & 0x3FF) | 0xDC00;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
220 str ~= pair;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
221 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
222 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
223
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
224 /// Decodes a character from a UTF-16 sequence.
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
225 /// Params:
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
226 /// str = the UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
227 /// index = where to start from.
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
228 /// Returns: ERROR_CHAR in case of an error in the sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
229 dchar decode(wchar[] str, ref size_t index)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
230 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
231 assert(str.length && index < str.length);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
232 dchar c = str[index];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
233 if (0xD800 > c || c > 0xDFFF)
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 ++index;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
236 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
237 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
238 if (c <= 0xDBFF && index+1 != str.length)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
239 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
240 wchar c2 = str[index+1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
241 if (0xDC00 <= c2 && c2 <= 0xDFFF)
764
4579e8505d5e Fixed unittests and removed dil.File.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 747
diff changeset
242 { // Decode surrogate pair.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
243 // (c - 0xD800) << 10 + 0x10000 ->
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
244 // (c - 0xD800 + 0x40) << 10 ->
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
245 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
246 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
247 index += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
248 return c;
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 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
251 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
252 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
253
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
254 /// Decodes a character from a UTF-16 sequence.
769
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
255 /// Params:
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
256 /// p = start of the UTF-16 sequence.
5e3ef1b2011c Added and improved documentation.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 764
diff changeset
257 /// end = one past the end of the sequence.
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
258 /// Returns: ERROR_CHAR in case of an error in the sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
259 dchar decode(ref wchar* p, wchar* end)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
260 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
261 assert(p && p < end);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
262 dchar c = *p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
263 if (0xD800 > c || c > 0xDFFF)
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 ++p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
266 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
267 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
268 if (c <= 0xDBFF && p+1 != end)
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 wchar c2 = p[1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
271 if (0xDC00 <= c2 && c2 <= 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
272 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
273 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
274 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
275 p += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
276 return c;
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 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
280 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
281
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
282 /// Decodes a character from a zero-terminated UTF-16 string.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
283 /// Params:
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
284 /// p = start of the UTF-16 sequence.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
285 /// Returns: ERROR_CHAR in case of an error in the sequence.
510
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
286 dchar decode(ref wchar* p)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
287 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
288 assert(p);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
289 dchar c = *p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
290 if (0xD800 > c || c > 0xDFFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
291 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
292 ++p;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
293 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
294 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
295 if (c <= 0xDBFF)
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
296 {
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
297 wchar c2 = p[1];
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
298 if (0xDC00 <= c2 && c2 <= 0xDFFF)
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 c = (c - 0xD7C0) << 10;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
301 c |= (c2 & 0x3FF);
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
302 p += 2;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
303 return c;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
304 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
305 }
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
306 return ERROR_CHAR;
dd3ce87b3569 Added module dil.Unicode.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents:
diff changeset
307 }
789
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
308
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
309 /// Converts a UTF-8 string to a UTF-16 string.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
310 wchar[] toUTF16(char[] str)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
311 {
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
312 wchar[] result;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
313 size_t idx;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
314 while (idx < str.length)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
315 {
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
316 auto c = decode(str, idx);
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
317 if (c == ERROR_CHAR)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
318 { // Skip trail bytes.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
319 while (++idx < str.length && isTrailByte(str[idx]))
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
320 {}
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
321 c = REPLACEMENT_CHAR;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
322 }
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
323 encode(result, c);
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
324 }
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
325 return result;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
326 }
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
327
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
328 /// Converts a UTF-8 string to a UTF-32 string.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
329 dchar[] toUTF32(char[] str)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
330 {
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
331 dchar[] result;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
332 size_t idx;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
333 while (idx < str.length)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
334 {
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
335 auto c = decode(str, idx);
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
336 if (c == ERROR_CHAR)
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
337 { // Skip trail bytes.
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
338 while (++idx < str.length && isTrailByte(str[idx]))
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
339 {}
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
340 c = REPLACEMENT_CHAR;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
341 }
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
342 result ~= c;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
343 }
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
344 return result;
c1d5cfd7aa44 Implemented string literal conversion.
Aziz K?ksal <aziz.koeksal@gmail.com>
parents: 769
diff changeset
345 }