|
| 1 | +require_relative 'symbol_table' |
| 2 | +require_relative 'vm_writer' |
| 3 | + |
| 4 | +class CompilationEngine |
| 5 | + attr_reader :input, :vm_writer |
| 6 | + |
| 7 | + def initialize(input, output) |
| 8 | + @input = input |
| 9 | + @vm_writer = VMWriter.new(output) |
| 10 | + end |
| 11 | + |
| 12 | + def compile_class |
| 13 | + @symbols = SymbolTable.new |
| 14 | + |
| 15 | + # b.tag!(:class) do |
| 16 | + # Get the ball moving! |
| 17 | + input.advance |
| 18 | + |
| 19 | + consume(Tokenizer::KEYWORD, 'class') |
| 20 | + |
| 21 | + @class_name = current_token |
| 22 | + consume(Tokenizer::IDENTIFIER) |
| 23 | + |
| 24 | + consume_wrapped('{') do |
| 25 | + while %w[field static].include? current_token |
| 26 | + compile_class_var_dec |
| 27 | + end |
| 28 | + |
| 29 | + while %w[constructor function method].include? current_token |
| 30 | + compile_subroutine |
| 31 | + end |
| 32 | + end |
| 33 | + # end |
| 34 | + end |
| 35 | + |
| 36 | + def compile_class_var_dec |
| 37 | + # b.classVarDec do |
| 38 | + kind = current_token # field, static, etc. |
| 39 | + consume(Tokenizer::KEYWORD) |
| 40 | + |
| 41 | + type = current_token # int, char, etc. |
| 42 | + consume_type |
| 43 | + |
| 44 | + consume_separated(',') do |
| 45 | + name = current_token |
| 46 | + @symbols.define(name, type, kind) |
| 47 | + consume_identifier(name) |
| 48 | + end |
| 49 | + |
| 50 | + consume(Tokenizer::SYMBOL, ';') |
| 51 | + # end |
| 52 | + end |
| 53 | + |
| 54 | + def compile_subroutine |
| 55 | + @symbols.start_subroutine |
| 56 | + |
| 57 | + # b.subroutineDec do |
| 58 | + consume(Tokenizer::KEYWORD) |
| 59 | + |
| 60 | + try_consume(Tokenizer::KEYWORD, 'void') || consume_type |
| 61 | + |
| 62 | + method_name = full_method_name |
| 63 | + consume(Tokenizer::IDENTIFIER) |
| 64 | + |
| 65 | + n = 0 |
| 66 | + consume_wrapped('(') do |
| 67 | + n = compile_parameter_list |
| 68 | + end |
| 69 | + |
| 70 | + vm_writer.write_function(method_name, n) |
| 71 | + |
| 72 | + compile_subroutine_body |
| 73 | + # end |
| 74 | + end |
| 75 | + |
| 76 | + def compile_parameter_list |
| 77 | + # b.parameterList do |
| 78 | + return 0 if current_token == ')' |
| 79 | + |
| 80 | + n = 0 |
| 81 | + consume_separated(',') do |
| 82 | + kind = :arg |
| 83 | + |
| 84 | + type = current_token # int, char, etc. |
| 85 | + consume_type |
| 86 | + |
| 87 | + name = current_token |
| 88 | + @symbols.define(name, type, kind) |
| 89 | + consume_identifier(name) |
| 90 | + |
| 91 | + n += 1 |
| 92 | + end |
| 93 | + # end |
| 94 | + |
| 95 | + n |
| 96 | + end |
| 97 | + |
| 98 | + def compile_subroutine_body |
| 99 | + # b.subroutineBody do |
| 100 | + consume_wrapped('{') do |
| 101 | + while current_token == "var" |
| 102 | + compile_var_dec |
| 103 | + end |
| 104 | + |
| 105 | + compile_statements |
| 106 | + end |
| 107 | + # end |
| 108 | + end |
| 109 | + |
| 110 | + def compile_statements |
| 111 | + # b.statements do |
| 112 | + loop do |
| 113 | + case current_token |
| 114 | + when 'let' |
| 115 | + compile_let |
| 116 | + when 'do' |
| 117 | + compile_do |
| 118 | + when 'return' |
| 119 | + compile_return |
| 120 | + when 'if' |
| 121 | + compile_if |
| 122 | + when 'while' |
| 123 | + compile_while |
| 124 | + else |
| 125 | + break |
| 126 | + end |
| 127 | + end |
| 128 | + # end |
| 129 | + end |
| 130 | + |
| 131 | + def compile_while |
| 132 | + # b.whileStatement do |
| 133 | + consume(Tokenizer::KEYWORD, 'while') |
| 134 | + |
| 135 | + consume_wrapped('(') do |
| 136 | + compile_expression |
| 137 | + end |
| 138 | + |
| 139 | + consume_wrapped('{') do |
| 140 | + compile_statements |
| 141 | + end |
| 142 | + # end |
| 143 | + end |
| 144 | + |
| 145 | + def compile_var_dec |
| 146 | + # b.varDec do |
| 147 | + consume(Tokenizer::KEYWORD, 'var') |
| 148 | + |
| 149 | + kind = :var |
| 150 | + |
| 151 | + type = current_token |
| 152 | + consume_type |
| 153 | + |
| 154 | + consume_separated(',') do |
| 155 | + name = current_token |
| 156 | + @symbols.define(name, type, kind) |
| 157 | + consume_identifier(name) |
| 158 | + end |
| 159 | + |
| 160 | + consume(Tokenizer::SYMBOL, ';') |
| 161 | + # end |
| 162 | + end |
| 163 | + |
| 164 | + def compile_let |
| 165 | + # b.letStatement do |
| 166 | + consume(Tokenizer::KEYWORD, 'let') |
| 167 | + |
| 168 | + consume_identifier |
| 169 | + |
| 170 | + try_consume_wrapped('[') do |
| 171 | + compile_expression |
| 172 | + end |
| 173 | + |
| 174 | + consume(Tokenizer::SYMBOL, '=') |
| 175 | + |
| 176 | + compile_expression |
| 177 | + |
| 178 | + consume(Tokenizer::SYMBOL, ';') |
| 179 | + # end |
| 180 | + end |
| 181 | + |
| 182 | + def compile_do |
| 183 | + # b.doStatement do |
| 184 | + consume(Tokenizer::KEYWORD, 'do') |
| 185 | + |
| 186 | + consume_subroutine_call |
| 187 | + |
| 188 | + consume(Tokenizer::SYMBOL, ';') |
| 189 | + # end |
| 190 | + end |
| 191 | + |
| 192 | + def compile_return |
| 193 | + # b.returnStatement do |
| 194 | + consume(Tokenizer::KEYWORD, 'return') |
| 195 | + |
| 196 | + compile_expression unless current_token == ';' |
| 197 | + |
| 198 | + consume(Tokenizer::SYMBOL, ';') |
| 199 | + # end |
| 200 | + end |
| 201 | + |
| 202 | + def compile_if |
| 203 | + # b.ifStatement do |
| 204 | + consume(Tokenizer::KEYWORD, 'if') |
| 205 | + |
| 206 | + consume_wrapped('(') do |
| 207 | + compile_expression |
| 208 | + end |
| 209 | + |
| 210 | + consume_wrapped('{') do |
| 211 | + compile_statements |
| 212 | + end |
| 213 | + |
| 214 | + if try_consume(Tokenizer::KEYWORD, 'else') |
| 215 | + consume_wrapped('{') do |
| 216 | + compile_statements |
| 217 | + end |
| 218 | + end |
| 219 | + # end |
| 220 | + end |
| 221 | + |
| 222 | + def compile_expression_list |
| 223 | + # b.expressionList do |
| 224 | + return if current_token == ')' |
| 225 | + |
| 226 | + consume_separated(',') do |
| 227 | + compile_expression |
| 228 | + end |
| 229 | + # end |
| 230 | + end |
| 231 | + |
| 232 | + def compile_expression |
| 233 | + # b.expression do |
| 234 | + compile_term |
| 235 | + |
| 236 | + while %w[+ - * / & | < > =].include? current_token |
| 237 | + consume(Tokenizer::SYMBOL) |
| 238 | + compile_term |
| 239 | + end |
| 240 | + # end |
| 241 | + end |
| 242 | + |
| 243 | + def compile_term |
| 244 | + # b.term do |
| 245 | + return if try_consume(Tokenizer::INT_CONST) || |
| 246 | + try_consume(Tokenizer::STRING_CONST) || |
| 247 | + try_consume_wrapped('(') { compile_expression } |
| 248 | + |
| 249 | + case current_token |
| 250 | + when 'true', 'false', 'null', 'this' #keywordConst |
| 251 | + consume(Tokenizer::KEYWORD) |
| 252 | + when '-', '~' # unary op |
| 253 | + consume(Tokenizer::SYMBOL) |
| 254 | + compile_term |
| 255 | + else |
| 256 | + name = current_token |
| 257 | + input.advance |
| 258 | + |
| 259 | + case current_token |
| 260 | + when '[' |
| 261 | + # b.identifier( |
| 262 | + # name, |
| 263 | + # type: @symbols.type_of(name), |
| 264 | + # kind: @symbols.kind_of(name), |
| 265 | + # index: @symbols.index_of(name) |
| 266 | + # ) |
| 267 | + |
| 268 | + consume_wrapped('[') do |
| 269 | + compile_expression |
| 270 | + end |
| 271 | + when '.', '(' |
| 272 | + # b.identifier(name) |
| 273 | + consume_subroutine_call(skip_identifier: true) |
| 274 | + else |
| 275 | + # b.identifier( |
| 276 | + # name, |
| 277 | + # type: @symbols.type_of(name), |
| 278 | + # kind: @symbols.kind_of(name), |
| 279 | + # index: @symbols.index_of(name) |
| 280 | + # ) |
| 281 | + end |
| 282 | + end |
| 283 | + # end |
| 284 | + end |
| 285 | + |
| 286 | + private |
| 287 | + |
| 288 | + def consume(expected_type, expected_token = nil) |
| 289 | + unless try_consume(expected_type, expected_token) |
| 290 | + if expected_token |
| 291 | + raise "expected a #{expected_type} `#{expected_token}`, got #{current_type} `#{current_token}`" |
| 292 | + else |
| 293 | + raise "expected any #{expected_type}, got #{current_type} `#{current_token}`" |
| 294 | + end |
| 295 | + end |
| 296 | + end |
| 297 | + |
| 298 | + def try_consume(expected_type, expected_token = nil) |
| 299 | + return false unless current_type == expected_type && |
| 300 | + current_token == (expected_token || current_token) |
| 301 | + |
| 302 | + # b.tag!(current_type, current_token) |
| 303 | + |
| 304 | + input.advance if input.has_more_tokens? |
| 305 | + true |
| 306 | + end |
| 307 | + |
| 308 | + def consume_identifier(name = current_token) |
| 309 | + # b.identifier( |
| 310 | + # name, |
| 311 | + # type: @symbols.type_of(name), |
| 312 | + # kind: @symbols.kind_of(name), |
| 313 | + # index: @symbols.index_of(name) |
| 314 | + # ) |
| 315 | + |
| 316 | + input.advance if input.has_more_tokens? |
| 317 | + end |
| 318 | + |
| 319 | + def consume_wrapped(opening, &block) |
| 320 | + unless try_consume_wrapped(opening) { block.call } |
| 321 | + raise "expected wrapping `#{opening}`, got #{current_type}, `#{current_token}`" |
| 322 | + end |
| 323 | + end |
| 324 | + |
| 325 | + def try_consume_wrapped(opening) |
| 326 | + closing = case opening |
| 327 | + when '(' then ')' |
| 328 | + when '[' then ']' |
| 329 | + when '{' then '}' |
| 330 | + else raise "Unknown opening brace: `#{opening}`" |
| 331 | + end |
| 332 | + |
| 333 | + if try_consume(Tokenizer::SYMBOL, opening) |
| 334 | + yield |
| 335 | + consume(Tokenizer::SYMBOL, closing) |
| 336 | + true |
| 337 | + end |
| 338 | + end |
| 339 | + |
| 340 | + def b |
| 341 | + @builder |
| 342 | + end |
| 343 | + |
| 344 | + def current_type |
| 345 | + input.token_type |
| 346 | + end |
| 347 | + |
| 348 | + def current_token |
| 349 | + case current_type |
| 350 | + when Tokenizer::KEYWORD |
| 351 | + input.keyword |
| 352 | + when Tokenizer::SYMBOL |
| 353 | + input.symbol |
| 354 | + when Tokenizer::IDENTIFIER |
| 355 | + input.identifier |
| 356 | + when Tokenizer::INT_CONST |
| 357 | + input.int_val |
| 358 | + when Tokenizer::STRING_CONST |
| 359 | + input.string_val |
| 360 | + end |
| 361 | + end |
| 362 | + |
| 363 | + def consume_separated(sep) |
| 364 | + begin |
| 365 | + yield |
| 366 | + end while try_consume(Tokenizer::SYMBOL, sep) |
| 367 | + end |
| 368 | + |
| 369 | + def consume_type |
| 370 | + case current_type |
| 371 | + when Tokenizer::KEYWORD |
| 372 | + if %w[int char boolean].include? current_token |
| 373 | + consume(Tokenizer::KEYWORD) |
| 374 | + end |
| 375 | + else |
| 376 | + consume(Tokenizer::IDENTIFIER) |
| 377 | + end |
| 378 | + end |
| 379 | + |
| 380 | + def consume_subroutine_call(skip_identifier: false) |
| 381 | + consume(Tokenizer::IDENTIFIER) unless skip_identifier |
| 382 | + |
| 383 | + if try_consume(Tokenizer::SYMBOL, '.') |
| 384 | + consume(Tokenizer::IDENTIFIER) |
| 385 | + end |
| 386 | + |
| 387 | + consume_wrapped('(') do |
| 388 | + compile_expression_list |
| 389 | + end |
| 390 | + end |
| 391 | + |
| 392 | + def full_method_name |
| 393 | + "#{@class_name}.#{current_token}" |
| 394 | + end |
| 395 | +end |
0 commit comments