MimIR 0.1
MimIR is my Intermediate Representation
Loading...
Searching...
No Matches
lexer.cpp
Go to the documentation of this file.
1#include "mim/ast/lexer.h"
2
3#include "mim/ast/ast.h"
4
5using namespace std::literals;
6
7namespace mim::ast {
8
9namespace utf8 = fe::utf8;
10using Tag = Tok::Tag;
11
12Lexer::Lexer(AST& ast, std::istream& istream, const fs::path* path /*= nullptr*/, std::ostream* md /*= nullptr*/)
13 : Super(istream, path)
14 , ast_(ast)
15 , md_(md) {
16#define CODE(t, str) keywords_[ast.sym(str)] = Tag::t;
17 MIM_KEY(CODE)
18#undef CODE
19
20#define CODE(str, t) \
21 if (Tag::t != Tag::Nil) keywords_[ast.sym(str)] = Tag::t;
22 MIM_SUBST(CODE)
23#undef CODE
24
25 if (start_md())
26 emit_md(true);
27 else
28 md_fence();
29}
30
31Tok Lexer::lex() {
32 while (true) {
33 start();
34
35 if (accept(utf8::EoF)) return tok(Tag::EoF);
36 if (accept(utf8::isspace)) continue;
37 if (accept(utf8::Null)) {
38 ast().error(loc_, "invalid UTF-8 character");
39 continue;
40 }
41
42 // clang-format off
43 // delimiters
44 if (accept( '(')) return tok(Tag::D_paren_l);
45 if (accept( ')')) return tok(Tag::D_paren_r);
46 if (accept( '[')) return tok(Tag::D_brckt_l);
47 if (accept( ']')) return tok(Tag::D_brckt_r);
48 if (accept( '{')) return tok(Tag::D_brace_l);
49 if (accept( '}')) return tok(Tag::D_brace_r);
50 if (accept(U'⦃')) return tok(Tag::D_curly_l);
51 if (accept(U'⦄')) return tok(Tag::D_curly_r);
52 if (accept(U'«')) return tok(Tag::D_quote_l);
53 if (accept(U'»')) return tok(Tag::D_quote_r);
54 if (accept(U'⟪')) return tok(Tag::D_quote_l);
55 if (accept(U'⟫')) return tok(Tag::D_quote_r);
56 if (accept(U'‹')) return tok(Tag::D_angle_l);
57 if (accept(U'›')) return tok(Tag::D_angle_r);
58 if (accept(U'⟨')) return tok(Tag::D_angle_l);
59 if (accept(U'⟩')) return tok(Tag::D_angle_r);
60 if (accept( '<')) {
61 if (accept( '<')) return tok(Tag::D_quote_l);
62 return tok(Tag::D_angle_l);
63 }
64 if (accept( '>')) {
65 if (accept( '>')) return tok(Tag::D_quote_r);
66 return tok(Tag::D_angle_r);
67 }
68 // further tokens
69 if (accept(U'→')) return tok(Tag::T_arrow);
70 if (accept( '@')) return tok(Tag::T_at);
71 if (accept( '=')) return tok(Tag::T_assign);
72 if (accept(U'⊥')) return tok(Tag::T_bot);
73 if (accept(U'⊤')) return tok(Tag::T_top);
74 if (accept(U'□')) return tok(Tag::T_box);
75 if (accept( ',')) return tok(Tag::T_comma);
76 if (accept( '$')) return tok(Tag::T_dollar);
77 if (accept( '#')) return tok(Tag::T_extract);
78 if (accept(U'λ')) return tok(Tag::T_lm);
79 if (accept( ';')) return tok(Tag::T_semicolon);
80 if (accept(U'★')) return tok(Tag::T_star);
81 if (accept( '*')) return tok(Tag::T_star);
82 if (accept( ':')) return tok(Tag::T_colon);
83 // clang-format on
84
85 if (accept('%')) {
86 if (lex_id()) return {loc_, Tag::M_anx, sym()};
87 ast().error(loc_, "invalid axiom name '{}'", str_);
88 continue;
89 }
90
91 if (accept('.')) {
92 if (accept(utf8::isdigit)) {
93 parse_digits();
94 parse_exp();
95 return {loc_, f64(std::strtod(str_.c_str(), nullptr))};
96 }
97
98 return tok(Tag::T_dot);
99 }
100
101 if (accept('\'')) {
102 auto c = lex_char();
103 if (accept('\'')) return {loc_, c};
104 ast().error(loc_, "invalid character literal {}", str_);
105 continue;
106 }
107
108 if (accept<Append::Off>('\"')) {
109 while (lex_char() != '"') {}
110 str_.pop_back(); // remove final '"'
111 return {loc_, Tag::L_str, sym()};
112 }
113
114 if (lex_id()) {
115 if (auto i = keywords_.find(sym()); i != keywords_.end()) return tok(i->second);
116 return {loc_, Tag::M_id, sym()};
117 }
118
119 if (utf8::isdigit(ahead()) || utf8::any('+', '-')(ahead())) {
120 if (auto lit = parse_lit()) return *lit;
121 continue;
122 }
123
124 if (start_md()) {
125 emit_md();
126 continue;
127 }
128
129 // comments
130 if (accept('/')) {
131 if (accept('*')) {
132 eat_comments();
133 continue;
134 }
135 if (accept('/')) {
136 while (ahead() != utf8::EoF && ahead() != '\n') next();
137 continue;
138 }
139
140 ast().error({loc_.path, peek_}, "invalid input char '/'; maybe you wanted to start a comment?");
141 continue;
142 }
143
144 ast().error({loc_.path, peek_}, "invalid input char '{}'", utf8::Char32(ahead()));
145 next();
146 }
147}
148
149bool Lexer::lex_id() {
150 if (accept([](char32_t c) { return c == '_' || utf8::isalpha(c); })) {
151 while (accept([](char32_t c) { return c == '_' || c == '.' || utf8::isalnum(c); })) {}
152 return true;
153 }
154 return false;
155}
156
157// clang-format off
158std::optional<Tok> Lexer::parse_lit() {
159 int base = 10;
160 std::optional<bool> sign;
161
162 if (accept<Append::Off>('+')) {
163 sign = false;
164 } else if (accept<Append::Off>('-')) {
165 if (accept('>')) return tok(Tag::T_arrow);
166 sign = true;
167 }
168
169 // prefix starting with '0'
170 if (accept<Append::Off>('0')) {
171 if (accept<Append::Off>('b')) base = 2;
172 else if (accept<Append::Off>('B')) base = 2;
173 else if (accept<Append::Off>('o')) base = 8;
174 else if (accept<Append::Off>('O')) base = 8;
175 else if (accept<Append::Off>('x')) base = 16;
176 else if (accept<Append::Off>('X')) base = 16;
177 }
178
179 parse_digits(base);
180
181 if (accept(utf8::any('i', 'I'))) {
182 if (sign) str_.insert(0, "-"sv);
183 auto val = std::strtoull(str_.c_str(), nullptr, base);
184 str_.clear();
185 parse_digits();
186 auto width = std::strtoull(str_.c_str(), nullptr, 10);
187 return Tok{loc_, ast().world().lit_int(width, val)};
188 }
189
190 if (!sign && base == 10) {
191 if (utf8::isrange(ahead(), U'₀', U'₉')) {
192 auto i = std::strtoull(str_.c_str(), nullptr, 10);
193 std::string mod;
194 while (utf8::isrange(ahead(), U'₀', U'₉')) mod += next() - U'₀' + '0';
195 auto m = std::strtoull(mod.c_str(), nullptr, 10);
196 return Tok{loc_, ast().world().lit_idx_mod(m, i)};
197 } else if (accept<Append::Off>('_')) {
198 auto i = std::strtoull(str_.c_str(), nullptr, 10);
199 str_.clear();
200 if (accept(utf8::isdigit)) {
201 parse_digits(10);
202 auto m = std::strtoull(str_.c_str(), nullptr, 10);
203 return Tok{loc_, ast().world().lit_idx_mod(m, i)};
204 } else {
205 ast().error(loc_, "stray underscore in Idx literal; size is missing");
206 auto i = std::strtoull(str_.c_str(), nullptr, 10);
207 return Tok{loc_, u64(i)};
208 }
209 }
210 }
211
212 bool is_float = false;
213 if (base == 10 || base == 16) {
214 // parse fractional part
215 if (accept('.')) {
216 is_float = true;
217 parse_digits(base);
218 }
219
220 bool has_exp = parse_exp(base);
221 if (base == 16 && is_float && !has_exp) ast().error(loc_, "hexadecimal floating constants require an exponent");
222 is_float |= has_exp;
223 }
224
225 if (sign && str_.empty()) {
226 ast().error(loc_, "stray '{}'", *sign ? "-" : "+");
227 return {};
228 }
229
230 if (is_float && base == 16) str_.insert(0, "0x"sv);
231 if (sign && *sign) str_.insert(0, "-"sv);
232
233 if (is_float) return Tok{loc_, f64(std::strtod (str_.c_str(), nullptr ))};
234 if (sign) return Tok{loc_, u64(std::strtoll (str_.c_str(), nullptr, base))};
235 else return Tok{loc_, u64(std::strtoull(str_.c_str(), nullptr, base))};
236}
237
238void Lexer::parse_digits(int base /*= 10*/) {
239 switch (base) {
240 // clang-format off
241 case 2: while (accept(utf8::isbdigit)) {} break;
242 case 8: while (accept(utf8::isodigit)) {} break;
243 case 10: while (accept(utf8::isdigit)) {} break;
244 case 16: while (accept(utf8::isxdigit)) {} break;
245 // clang-format on
246 default: fe::unreachable();
247 }
248}
249
250bool Lexer::parse_exp(int base /*= 10*/) {
251 if (accept(base == 10 ? utf8::any('e', 'E') : utf8::any('p', 'P'))) {
252 accept(utf8::any('+', '-'));
253 if (!utf8::isdigit(ahead())) ast().error(loc_, "exponent has no digits");
254 parse_digits();
255 return true;
256 }
257 return false;
258}
259// clang-format on
260
261char8_t Lexer::lex_char() {
262 if (accept<Append::Off>('\\')) {
263 // clang-format off
264 if (false) {}
265 else if (accept<Append::Off>('\'')) str_ += '\'';
266 else if (accept<Append::Off>('\\')) str_ += '\\';
267 else if (accept<Append::Off>( '"')) str_ += '\"';
268 else if (accept<Append::Off>( '0')) str_ += '\0';
269 else if (accept<Append::Off>( 'a')) str_ += '\a';
270 else if (accept<Append::Off>( 'b')) str_ += '\b';
271 else if (accept<Append::Off>( 'f')) str_ += '\f';
272 else if (accept<Append::Off>( 'n')) str_ += '\n';
273 else if (accept<Append::Off>( 'r')) str_ += '\r';
274 else if (accept<Append::Off>( 't')) str_ += '\t';
275 else if (accept<Append::Off>( 'v')) str_ += '\v';
276 else ast().error(loc_.anew_finis(), "invalid escape character '\\{}'", (char)ahead());
277 // clang-format on
278 return str_.back();
279 }
280 auto c = next();
281 str_ += c;
282 if (utf8::isascii(c)) return c;
283 ast().error(loc_, "invalid character '{}'", (char)c);
284 return '\0';
285}
286
287void Lexer::eat_comments() {
288 while (true) {
289 while (ahead() != utf8::EoF && ahead() != '*') next();
290 if (accept(utf8::EoF)) {
291 ast().error(loc_, "non-terminated multiline comment");
292 return;
293 }
294 next();
295 if (accept('/')) break;
296 }
297}
298
299void Lexer::emit_md(bool start_of_file) {
300 if (!start_of_file) md_fence();
301
302 do {
303 out_ = false;
304 for (int i = 0; i < 3; ++i) next();
305 accept(' ');
306 out_ = true;
307
308 while (ahead() != utf8::EoF && ahead() != '\n') next();
309 accept('\n');
310 } while (start_md());
311
312 if (ahead() == utf8::EoF)
313 out_ = false;
314 else
315 md_fence();
316}
317
318Sym Lexer::sym() { return ast().sym(str_); }
319
320} // namespace mim::ast
mim::World::lit_idx_mod
const Lit * lit_idx_mod(nat_t mod, u64 val)
Constructs a Lit of type Idx of size mod.
Definition world.h:422
mim::World::lit_int
const Lit * lit_int(nat_t width, u64 val)
Constructs a Lit of type Idx of size $2^width$.
Definition world.h:409
mim::ast::AST::world
World & world()
Definition ast.h:61
mim::ast::AST::error
Error & error()
Definition ast.h:63
mim::ast::AST::sym
Sym sym(const char *s)
Definition ast.h:69
mim::ast::Lexer::Lexer
Lexer(AST &, std::istream &, const fs::path *path=nullptr, std::ostream *md=nullptr)
Creates a lexer to read *.mim files (see Lexical Structure).
Definition lexer.cpp:12
mim::ast::Lexer::ast
AST & ast()
Definition lexer.h:22
mim::ast
Definition ast.h:14
mim::f64
double f64
Definition types.h:41
mim::u64
uint64_t u64
Definition types.h:34
MIM_SUBST
#define MIM_SUBST(m)
Definition tok.h:99
CODE
#define CODE(t, str)
Definition tok.h:54
MIM_KEY
#define MIM_KEY(m)
Definition tok.h:13