CDC Badge OS
Firmware for the CDC Badge v1.0 hardware security key
Loading...
Searching...
No Matches
Cp437.cpp
Go to the documentation of this file.
1
5
6#include "cdc_core/Cp437.h"
7
8namespace cdc::core::cp437 {
9
10namespace {
11
14constexpr uint16_t kHigh[128] = {
15 0x00C7, 0x00FC, 0x00E9, 0x00E2, 0x00E4, 0x00E0, 0x00E5, 0x00E7, // 80
16 0x00EA, 0x00EB, 0x00E8, 0x00EF, 0x00EE, 0x00EC, 0x00C4, 0x00C5, // 88
17 0x00C9, 0x00E6, 0x00C6, 0x00F4, 0x00F6, 0x00F2, 0x00FB, 0x00F9, // 90
18 0x00FF, 0x00D6, 0x00DC, 0x00A2, 0x00A3, 0x00A5, 0x20A7, 0x0192, // 98
19 0x00E1, 0x00ED, 0x00F3, 0x00FA, 0x00F1, 0x00D1, 0x00AA, 0x00BA, // A0
20 0x00BF, 0x2310, 0x00AC, 0x00BD, 0x00BC, 0x00A1, 0x00AB, 0x00BB, // A8
21 0x2591, 0x2592, 0x2593, 0x2502, 0x2524, 0x2561, 0x2562, 0x2556, // B0
22 0x2555, 0x2563, 0x2551, 0x2557, 0x255D, 0x255C, 0x255B, 0x2510, // B8
23 0x2514, 0x2534, 0x252C, 0x251C, 0x2500, 0x253C, 0x255E, 0x255F, // C0
24 0x255A, 0x2554, 0x2569, 0x2566, 0x2560, 0x2550, 0x256C, 0x2567, // C8
25 0x2568, 0x2564, 0x2565, 0x2559, 0x2558, 0x2552, 0x2553, 0x256B, // D0
26 0x256A, 0x2518, 0x250C, 0x2588, 0x2584, 0x258C, 0x2590, 0x2580, // D8
27 0x03B1, 0x00DF, 0x0393, 0x03C0, 0x03A3, 0x03C3, 0x00B5, 0x03C4, // E0
28 0x03A6, 0x0398, 0x03A9, 0x03B4, 0x221E, 0x03C6, 0x03B5, 0x2229, // E8
29 0x2261, 0x00B1, 0x2265, 0x2264, 0x2320, 0x2321, 0x00F7, 0x2248, // F0
30 0x00B0, 0x2219, 0x00B7, 0x221A, 0x207F, 0x00B2, 0x25A0, 0x00A0, // F8
31};
32
33} // namespace
34
35uint32_t toUnicode(uint8_t b)
36{
37 return (b < 0x80) ? b : kHigh[b - 0x80];
38}
39
40uint8_t fromUnicode(uint32_t cp)
41{
42 if (cp < 0x80) return static_cast<uint8_t>(cp);
43 // Glyphs CP437 only carries outside the 0x80..0xFF block, mapped to their
44 // closest renderable byte (the built-in font draws these positions).
45 switch (cp) {
46 case 0x00A6: return 0x7C; // broken bar -> '|'
47 case 0x00E3: return 0x83; // a-tilde -> a-circ (closest glyph)
48 // CP437 control-area pictographs (0x01-0x1F) addressed by Unicode
49 // codepoint. 0x0A and 0x0D are intentionally absent: the text renderer
50 // consumes those bytes as newline / carriage-return.
51 case 0x263A: return 0x01; // white smiling face
52 case 0x263B: return 0x02; // black smiling face
53 case 0x2665: return 0x03; // heart
54 case 0x2666: return 0x04; // diamond
55 case 0x2663: return 0x05; // club
56 case 0x2660: return 0x06; // spade
57 case 0x2022: return 0x07; // bullet
58 case 0x25D8: return 0x08; // inverse bullet
59 case 0x25CB: return 0x09; // circle
60 case 0x2642: return 0x0B; // male sign
61 case 0x2640: return 0x0C; // female sign
62 case 0x266B: return 0x0E; // beamed eighth notes
63 case 0x263C: return 0x0F; // sun with rays
64 case 0x25BA: return 0x10; // right-pointing pointer
65 case 0x25C4: return 0x11; // left-pointing pointer
66 case 0x2195: return 0x12; // up-down arrow
67 case 0x203C: return 0x13; // double exclamation mark
68 case 0x00B6: return 0x14; // pilcrow / paragraph
69 case 0x00A7: return 0x15; // section sign
70 case 0x25AC: return 0x16; // black rectangle / bar
71 case 0x21A8: return 0x17; // up-down arrow with base
72 case 0x2191: return 0x18; // up arrow
73 case 0x2193: return 0x19; // down arrow
74 case 0x2192: return 0x1A; // right arrow
75 case 0x2190: return 0x1B; // left arrow
76 case 0x221F: return 0x1C; // right angle
77 case 0x2194: return 0x1D; // left-right arrow
78 case 0x25B2: return 0x1E; // up-pointing triangle
79 case 0x25BC: return 0x1F; // down-pointing triangle
80 default: break;
81 }
82 for (int i = 0; i < 128; ++i) {
83 if (kHigh[i] == cp) return static_cast<uint8_t>(0x80 + i);
84 }
85 return 0;
86}
87
88std::string fromUtf8(const char* s)
89{
90 std::string out;
91 if (!s) return out;
92 const uint8_t* r = reinterpret_cast<const uint8_t*>(s);
93 while (*r) {
94 uint8_t c = *r;
95 uint32_t cp = 0;
96 uint8_t cont = 0;
97 if ((c & 0x80) == 0) { out.push_back(static_cast<char>(c)); ++r; continue; }
98 else if ((c & 0xE0) == 0xC0) { cp = c & 0x1F; cont = 1; }
99 else if ((c & 0xF0) == 0xE0) { cp = c & 0x0F; cont = 2; }
100 else if ((c & 0xF8) == 0xF0) { cp = c & 0x07; cont = 3; }
101 else { ++r; continue; }
102 ++r;
103 bool ok = true;
104 for (uint8_t i = 0; i < cont; ++i) {
105 if ((*r & 0xC0) != 0x80) { ok = false; break; }
106 cp = (cp << 6) | (*r & 0x3F);
107 ++r;
108 }
109 if (!ok) continue;
110 uint8_t mapped = fromUnicode(cp);
111 if (mapped) out.push_back(static_cast<char>(mapped));
112 }
113 return out;
114}
115
116std::string toUtf8(const char* s)
117{
118 std::string out;
119 if (!s) return out;
120 for (const uint8_t* r = reinterpret_cast<const uint8_t*>(s); *r; ++r) {
121 uint32_t cp = toUnicode(*r);
122 if (cp < 0x80) {
123 out.push_back(static_cast<char>(cp));
124 } else if (cp < 0x800) {
125 out.push_back(static_cast<char>(0xC0 | (cp >> 6)));
126 out.push_back(static_cast<char>(0x80 | (cp & 0x3F)));
127 } else {
128 out.push_back(static_cast<char>(0xE0 | (cp >> 12)));
129 out.push_back(static_cast<char>(0x80 | ((cp >> 6) & 0x3F)));
130 out.push_back(static_cast<char>(0x80 | (cp & 0x3F)));
131 }
132 }
133 return out;
134}
135
136} // namespace cdc::core::cp437
Cp437.h
Canonical CP437 <-> Unicode/UTF-8 codec.
cdc::core::cp437::fromUtf8
std::string fromUtf8(const char *s)
Convert a UTF-8 string to CP437 bytes (unmapped chars dropped).
Definition Cp437.cpp:88
cdc::core::cp437::toUtf8
std::string toUtf8(const char *s)
Convert CP437 bytes to a UTF-8 string.
Definition Cp437.cpp:116
cdc::core::cp437::fromUnicode
uint8_t fromUnicode(uint32_t cp)
Map a Unicode codepoint to its CP437 byte, or 0 if it has none.
Definition Cp437.cpp:40
cdc::core::cp437::toUnicode
uint32_t toUnicode(uint8_t b)
Map a CP437 byte to its Unicode codepoint. 0x00-0x7F is ASCII.
Definition Cp437.cpp:35