1// SPDX-License-Identifier: MIT OR MPL-2.0 OR LGPL-2.1-or-later OR GPL-2.0-or-later
2// Copyright 2011, SIL International, All rights reserved.
3
4#pragma once
5
6#include <cstdlib>
7#include "inc/Main.h"
8
9namespace graphite2 {
10
11typedef uint32 uchar_t;
12
13template <int N>
14struct _utf_codec
15{
16 typedef uchar_t codeunit_t;
17
18 static void put(codeunit_t * cp, const uchar_t , int8 & len) throw();
19 static uchar_t get(const codeunit_t * cp, int8 & len) throw();
20 static bool validate(const codeunit_t * s, const codeunit_t * const e) throw();
21};
22
23
24template <>
25struct _utf_codec<32>
26{
27private:
28 static const uchar_t limit = 0x110000;
29public:
30 typedef uint32 codeunit_t;
31
32 inline
33 static void put(codeunit_t * cp, const uchar_t usv, int8 & l) throw()
34 {
35 *cp = usv; l = 1;
36 }
37
38 inline
39 static uchar_t get(const codeunit_t * cp, int8 & l) throw()
40 {
41 if (cp[0] < limit) { l = 1; return cp[0]; }
42 else { l = -1; return 0xFFFD; }
43 }
44
45 inline
46 static bool validate(const codeunit_t * s, const codeunit_t * const e) throw()
47 {
48 return s <= e;
49 }
50};
51
52
53template <>
54struct _utf_codec<16>
55{
56private:
57 static const int32 lead_offset = 0xD800 - (0x10000 >> 10);
58 static const int32 surrogate_offset = 0x10000 - (0xD800 << 10) - 0xDC00;
59public:
60 typedef uint16 codeunit_t;
61
62 inline
63 static void put(codeunit_t * cp, const uchar_t usv, int8 & l) throw()
64 {
65 if (usv < 0x10000) { l = 1; cp[0] = codeunit_t(usv); }
66 else
67 {
68 cp[0] = codeunit_t(lead_offset + (usv >> 10));
69 cp[1] = codeunit_t(0xDC00 + (usv & 0x3FF));
70 l = 2;
71 }
72 }
73
74 inline
75 static uchar_t get(const codeunit_t * cp, int8 & l) throw()
76 {
77 const uint32 uh = cp[0];
78 l = 1;
79
80 if (uh < 0xD800|| uh > 0xDFFF) { return uh; }
81 if (uh > 0xDBFF) { l = -1; return 0xFFFD; }
82 const uint32 ul = cp[1];
83 if (ul < 0xDC00 || ul > 0xDFFF) { l = -1; return 0xFFFD; }
84 ++l;
85 return (uh<<10) + ul + surrogate_offset;
86 }
87
88 inline
89 static bool validate(const codeunit_t * s, const codeunit_t * const e) throw()
90 {
91 const ptrdiff_t n = e-s;
92 if (n <= 0) return n == 0;
93 const uint32 u = *(e-1); // Get the last codepoint
94 return (u < 0xD800 || u > 0xDBFF);
95 }
96};
97
98
99template <>
100struct _utf_codec<8>
101{
102private:
103 static const int8 sz_lut[16];
104 static const byte mask_lut[5];
105 static const uchar_t limit = 0x110000;
106
107public:
108 typedef uint8 codeunit_t;
109
110 inline
111 static void put(codeunit_t * cp, const uchar_t usv, int8 & l) throw()
112 {
113 if (usv < 0x80) {l = 1; cp[0] = usv; return; }
114 if (usv < 0x0800) {l = 2; cp[0] = 0xC0 + (usv >> 6); cp[1] = 0x80 + (usv & 0x3F); return; }
115 if (usv < 0x10000) {l = 3; cp[0] = 0xE0 + (usv >> 12); cp[1] = 0x80 + ((usv >> 6) & 0x3F); cp[2] = 0x80 + (usv & 0x3F); return; }
116 else {l = 4; cp[0] = 0xF0 + (usv >> 18); cp[1] = 0x80 + ((usv >> 12) & 0x3F); cp[2] = 0x80 + ((usv >> 6) & 0x3F); cp[3] = 0x80 + (usv & 0x3F); return; }
117 }
118
119 inline
120 static uchar_t get(const codeunit_t * cp, int8 & l) throw()
121 {
122 const int8 seq_sz = sz_lut[*cp >> 4];
123 uchar_t u = *cp & mask_lut[seq_sz];
124 l = 1;
125 bool toolong = false;
126
127 switch(seq_sz) {
128 case 4: u <<= 6; u |= *++cp & 0x3F; if (*cp >> 6 != 2) break; ++l; toolong = (u < 0x10); GR_FALLTHROUGH;
129 // no break
130 case 3: u <<= 6; u |= *++cp & 0x3F; if (*cp >> 6 != 2) break; ++l; toolong |= (u < 0x20); GR_FALLTHROUGH;
131 // no break
132 case 2: u <<= 6; u |= *++cp & 0x3F; if (*cp >> 6 != 2) break; ++l; toolong |= (u < 0x80); GR_FALLTHROUGH;
133 // no break
134 case 1: break;
135 case 0: l = -1; return 0xFFFD;
136 }
137
138 if (l != seq_sz || toolong || u >= limit)
139 {
140 l = -l;
141 return 0xFFFD;
142 }
143 return u;
144 }
145
146 inline
147 static bool validate(const codeunit_t * s, const codeunit_t * const e) throw()
148 {
149 const ptrdiff_t n = e-s;
150 if (n <= 0) return n == 0;
151 s += (n-1);
152 if (*s < 0x80) return true;
153 if (*s >= 0xC0) return false;
154 if (n == 1) return true;
155 if (*--s < 0x80) return true;
156 if (*s >= 0xE0) return false;
157 if (n == 2 || *s >= 0xC0) return true;
158 if (*--s < 0x80) return true;
159 if (*s >= 0xF0) return false;
160 return true;
161 }
162
163};
164
165
166template <typename C>
167class _utf_iterator
168{
169 typedef _utf_codec<sizeof(C)*8> codec;
170
171 C * cp;
172 mutable int8 sl;
173
174public:
175 typedef C codeunit_type;
176 typedef uchar_t value_type;
177 typedef uchar_t * pointer;
178
179 class reference
180 {
181 const _utf_iterator & _i;
182
183 reference(const _utf_iterator & i): _i(i) {}
184 public:
185 operator value_type () const throw () { return codec::get(_i.cp, _i.sl); }
186 reference & operator = (const value_type usv) throw() { codec::put(_i.cp, usv, _i.sl); return *this; }
187
188 friend class _utf_iterator;
189 };
190
191
192 _utf_iterator(const void * us=0) : cp(reinterpret_cast<C *>(const_cast<void *>(us))), sl(1) { }
193
194 _utf_iterator & operator ++ () { cp += abs(sl); return *this; }
195 _utf_iterator operator ++ (int) { _utf_iterator tmp(*this); operator++(); return tmp; }
196
197 bool operator == (const _utf_iterator & rhs) const throw() { return cp >= rhs.cp; }
198 bool operator != (const _utf_iterator & rhs) const throw() { return !operator==(rhs); }
199
200 reference operator * () const throw() { return *this; }
201 pointer operator ->() const throw() { return &operator *(); }
202
203 operator codeunit_type * () const throw() { return cp; }
204
205 bool error() const throw() { return sl < 1; }
206 bool validate(const _utf_iterator & e) { return codec::validate(cp, e.cp); }
207};
208
209template <typename C>
210struct utf
211{
212 typedef typename _utf_codec<sizeof(C)*8>::codeunit_t codeunit_t;
213
214 typedef _utf_iterator<C> iterator;
215 typedef _utf_iterator<const C> const_iterator;
216
217 inline
218 static bool validate(codeunit_t * s, codeunit_t * e) throw() {
219 return _utf_codec<sizeof(C)*8>::validate(s,e);
220 }
221};
222
223
224typedef utf<uint32> utf32;
225typedef utf<uint16> utf16;
226typedef utf<uint8> utf8;
227
228} // namespace graphite2
229