OXYGENHUB
◆ main openmsx / src / input / UnicodeKeymap.cc
6.0 KiB · C++ · last changed chore(): initial commit 1 month ago source blame history raw
1 
#include "UnicodeKeymap.hh"
2
3 
#include "File.hh"
4 
#include "FileContext.hh"
5 
#include "FileException.hh"
6 
#include "MSXException.hh"
7
8 
#include "narrow.hh"
9 
#include "one_of.hh"
10 
#include "ranges.hh"
11 
#include "stl.hh"
12
13 
#include <algorithm>
14 
#include <bit>
15 
#include <optional>
16
17 
namespace openmsx {
18
19 
using namespace std::literals;
20
21 
/** Parses the given string as a hexadecimal integer.
22 
  */
23 
[[nodiscard]] static constexpr std::optional<unsigned> parseHex(std::string_view str)
24 
{
25 
	if (str.empty()) {
26 
		return {};
27 
	}
28 
	unsigned value = 0;
29 
	for (const char c : str) {
30 
		value *= 16;
31 
		if ('0' <= c && c <= '9') {
32 
			value += c - '0';
33 
		} else if ('A' <= c && c <= 'F') {
34 
			value += c - 'A' + 10;
35 
		} else if ('a' <= c && c <= 'f') {
36 
			value += c - 'a' + 10;
37 
		} else {
38 
			return {};
39 
		}
40 
	}
41 
	return value;
42 
}
43
44 
/** Returns true iff the given character is a separator.
45 
  * Separators are: comma, whitespace and hash mark.
46 
  * Newline (\n) is not considered a separator.
47 
  */
48 
[[nodiscard]] static constexpr bool isSep(char c)
49 
{
50 
	return c == one_of(',',             // comma
51 
	                   ' ', '\t', '\r', // whitespace
52 
	                   '#');            // comment
53 
}
54
55 
/** Removes separator characters at the start of the given string reference.
56 
  * Characters between a hash mark and the following newline are also skipped.
57 
  */
58 
static constexpr void skipSep(std::string_view& str)
59 
{
60 
	while (!str.empty()) {
61 
		const char c = str.front();
62 
		if (!isSep(c)) break;
63 
		if (c == '#') {
64 
			// Skip till end of line.
65 
			while (!str.empty() && str.front() != '\n') str.remove_prefix(1);
66 
			break;
67 
		}
68 
		str.remove_prefix(1);
69 
	}
70 
}
71
72 
/** Returns the next token in the given string.
73 
  * The token and any separators preceding it are removed from the string.
74 
  */
75 
[[nodiscard]] static constexpr std::string_view nextToken(std::string_view& str)
76 
{
77 
	skipSep(str);
78 
	const auto* tokenBegin = str.data();
79 
	while (!str.empty() && str.front() != '\n' && !isSep(str.front())) {
80 
		// Pop non-separator character.
81 
		str.remove_prefix(1);
82 
	}
83 
	return {tokenBegin, size_t(str.data() - tokenBegin)};
84 
}
85
86
87 
UnicodeKeymap::UnicodeKeymap(std::string_view keyboardType)
88 
{
89 
	auto filename = systemFileContext().resolve(
90 
		tmpStrCat("unicodemaps/unicodemap.", keyboardType));
91 
	try {
92 
		auto buf = File(filename).mmap<const char>();
93 
		parseUnicodeKeyMapFile(std::string_view(buf.data(), buf.size())); // TODO c++23
94 
		// TODO in the future we'll require the presence of
95 
		//      "MSX-Video-Characterset" in the keyboard information
96 
		//      file, then we don't need this fallback.
97 
		if (!msxChars.has_value()) {
98 
			msxChars.emplace("MSXVID.TXT");
99 
		}
100 
	} catch (FileException&) {
101 
		throw MSXException("Couldn't load unicode keymap file: ", filename);
102 
	}
103 
}
104
105 
UnicodeKeymap::KeyInfo UnicodeKeymap::get(unsigned unicode) const
106 
{
107 
	auto m = binary_find(mapData, unicode, {}, &Entry::unicode);
108 
	return m ? m->keyInfo : KeyInfo();
109 
}
110
111 
UnicodeKeymap::KeyInfo UnicodeKeymap::getDeadKey(unsigned n) const
112 
{
113 
	assert(n < NUM_DEAD_KEYS);
114 
	return deadKeys[n];
115 
}
116
117 
void UnicodeKeymap::parseUnicodeKeyMapFile(std::string_view data)
118 
{
119 
	std::ranges::fill(relevantMods, 0);
120
121 
	while (!data.empty()) {
122 
		if (data.front() == '\n') {
123 
			// Next line.
124 
			data.remove_prefix(1);
125 
		}
126
127 
		std::string_view token = nextToken(data);
128 
		if (token.empty()) {
129 
			// Skip empty line.
130 
			continue;
131 
		}
132
133 
		if (token == "MSX-Video-Characterset:") {
134 
			auto vidFileName = nextToken(data);
135 
			if (vidFileName.empty()) {
136 
				throw MSXException("Missing filename for MSX-Video-Characterset");
137 
			}
138 
			msxChars.emplace(vidFileName);
139 
			continue;
140 
		}
141
142 
		// Parse first token: a unicode value or the keyword DEADKEY.
143 
		unsigned unicode = 0;
144 
		unsigned deadKeyIndex = 0;
145 
		bool isDeadKey = token.starts_with("DEADKEY");
146 
		if (isDeadKey) {
147 
			token.remove_prefix(strlen("DEADKEY"));
148 
			if (token.empty()) {
149 
				// The normal keywords are
150 
				//    DEADKEY1  DEADKEY2  DEADKEY3
151 
				// but for backwards compatibility also still recognize
152 
				//    DEADKEY
153 
			} else {
154 
				auto d = parseHex(token);
155 
				if (!d || *d > NUM_DEAD_KEYS) {
156 
					throw MSXException(
157 
						"Wrong deadkey number in keymap file. "
158 
						"It must be 1..", NUM_DEAD_KEYS);
159 
				}
160 
				deadKeyIndex = *d - 1; // Make index 0 based instead of 1 based
161 
			}
162 
		} else {
163 
			auto u = parseHex(token);
164 
			if (!u || *u > 0x1FBFF) {
165 
				throw MSXException("Wrong unicode value in keymap file");
166 
			}
167 
			unicode = *u;
168 
		}
169
170 
		// Parse second token. It must be <ROW><COL>
171 
		token = nextToken(data);
172 
		if (token == "--") {
173 
			// Skip -- for now, it means the character cannot be typed.
174 
			continue;
175 
		}
176 
		auto rowcol = parseHex(token);
177 
		if (!rowcol || *rowcol >= 0x100) {
178 
			throw MSXException(
179 
				(token.empty() ? "Missing"sv : "Wrong"sv),
180 
				" <ROW><COL> value in keymap file");
181 
		}
182 
		if ((*rowcol >> 4) >= KeyMatrixPosition::NUM_ROWS) {
183 
			throw MSXException("Too high row value in keymap file");
184 
		}
185 
		if ((*rowcol & 0x0F) >= KeyMatrixPosition::NUM_COLS) {
186 
			throw MSXException("Too high column value in keymap file");
187 
		}
188 
		auto pos = KeyMatrixPosition(narrow_cast<uint8_t>(*rowcol));
189
190 
		// Parse remaining tokens. It is an optional list of modifier keywords.
191 
		uint8_t modMask = 0;
192 
		while (true) {
193 
			token = nextToken(data);
194 
			if (token.empty()) {
195 
				break;
196 
			} else if (token == "SHIFT") {
197 
				modMask |= KeyInfo::SHIFT_MASK;
198 
			} else if (token == "CTRL") {
199 
				modMask |= KeyInfo::CTRL_MASK;
200 
			} else if (token == "GRAPH") {
201 
				modMask |= KeyInfo::GRAPH_MASK;
202 
			} else if (token == "CAPSLOCK") {
203 
				modMask |= KeyInfo::CAPS_MASK;
204 
			} else if (token == "CODE") {
205 
				modMask |= KeyInfo::CODE_MASK;
206 
			} else {
207 
				throw MSXException(
208 
					"Invalid modifier \"", token, "\" in keymap file");
209 
			}
210 
		}
211
212 
		if (isDeadKey) {
213 
			if (modMask != 0) {
214 
				throw MSXException(
215 
					"DEADKEY entry in keymap file cannot have modifiers");
216 
			}
217 
			deadKeys[deadKeyIndex] = KeyInfo(pos, 0);
218 
		} else {
219 
			mapData.emplace_back(Entry{.unicode = unicode, .keyInfo = KeyInfo(pos, modMask)});
220 
			// Note: getRowCol() uses 3 bits for column, rowcol uses 4.
221 
			relevantMods[pos.getRowCol()] |= modMask;
222 
		}
223 
	}
224
225 
	std::ranges::sort(mapData, {}, &Entry::unicode);
226 
}
227
228 
} // namespace openmsx