| author | |
| committer | |
| log | 08a2311efd8b388cd431feb6000741f4a62da613 |
| tree | 528ba0056e9a54583c91b0787b9d5ffcdcc11de7 |
| parent | 1ed926c3216016d73e83377a295869d8579e2fde |
10 files changed, 400 insertions(+), 64 deletions(-)
README.md+19-3| ... | ... | @@ -44,7 +44,11 @@ make |
| 44 | 44 | |
| 45 | 45 | * variable declarations and assignment expressions |
| 46 | 46 | * Type checking |
| 47 | * loops | |
| 48 | * labels and goto | |
| 47 | 49 | * inline assembly and syscalls |
| 50 | * conditional compilation and ability to check target platform and architecture | |
| 51 | * main function with command line arguments | |
| 48 | 52 | * running code at compile time |
| 49 | 53 | * print! macro that takes var args |
| 50 | 54 | * panic! macro that prints a stack trace to stderr in debug mode and calls |
| ... | ... | @@ -104,14 +108,26 @@ Type : token(Symbol) | PointerType | token(Unreachable) |
| 104 | 108 | |
| 105 | 109 | PointerType : token(Star) token(Const) Type | token(Star) token(Mut) Type |
| 106 | 110 | |
| 107 | Block : token(LBrace) list(option(Expression), token(Semicolon)) token(RBrace) | |
| 111 | Block : token(LBrace) list(option(Statement), token(Semicolon)) token(RBrace) | |
| 108 | 112 | |
| 109 | Expression : BoolOrExpression | ReturnExpression | |
| 113 | Statement : NonBlockExpression token(Semicolon) | BlockExpression | |
| 114 | ||
| 115 | Expression : BlockExpression | NonBlockExpression | |
| 116 | ||
| 117 | NonBlockExpression : BoolOrExpression | ReturnExpression | |
| 118 | ||
| 119 | BlockExpression : IfExpression | Block | |
| 110 | 120 | |
| 111 | 121 | BoolOrExpression : BoolAndExpression token(BoolOr) BoolAndExpression | BoolAndExpression |
| 112 | 122 | |
| 113 | 123 | ReturnExpression : token(Return) option(Expression) |
| 114 | 124 | |
| 125 | IfExpression : token(If) Expression Block option(Else | ElseIf) | |
| 126 | ||
| 127 | ElseIf : token(Else) IfExpression | |
| 128 | ||
| 129 | Else : token(Else) Block | |
| 130 | ||
| 115 | 131 | BoolAndExpression : ComparisonExpression token(BoolAnd) ComparisonExpression | ComparisonExpression |
| 116 | 132 | |
| 117 | 133 | ComparisonExpression : BinaryOrExpression ComparisonOperator BinaryOrExpression | BinaryOrExpression |
| ... | ... | @@ -144,7 +160,7 @@ FnCallExpression : PrimaryExpression token(LParen) list(Expression, token(Comma) |
| 144 | 160 | |
| 145 | 161 | PrefixOp : token(Not) | token(Dash) | token(Tilde) |
| 146 | 162 | |
| 147 | PrimaryExpression : token(Number) | token(String) | token(Unreachable) | GroupedExpression | Block | token(Symbol) | |
| 163 | PrimaryExpression : token(Number) | token(String) | token(Unreachable) | GroupedExpression | token(Symbol) | |
| 148 | 164 | |
| 149 | 165 | GroupedExpression : token(LParen) Expression token(RParen) |
| 150 | 166 | ``` |
doc/vim/syntax/zig.vim+2-2| ... | ... | @@ -7,8 +7,8 @@ if exists("b:current_syntax") |
| 7 | 7 | finish |
| 8 | 8 | endif |
| 9 | 9 | |
| 10 | syn keyword zigKeyword fn return mut const extern unreachable export pub as use | |
| 11 | syn keyword zigType bool i8 u8 i16 u16 i32 u32 i64 u64 isize usize f32 f64 f128 void | |
| 10 | syn keyword zigKeyword fn return mut const extern unreachable export pub as use if else let void | |
| 11 | syn keyword zigType bool i8 u8 i16 u16 i32 u32 i64 u64 isize usize f32 f64 f128 | |
| 12 | 12 | |
| 13 | 13 | syn region zigCommentLine start="//" end="$" contains=zigTodo,@Spell |
| 14 | 14 | syn region zigCommentLineDoc start="//\%(//\@!\|!\)" end="$" contains=zigTodo,@Spell |
src/analyze.cpp+105-7| ... | ... | @@ -274,6 +274,7 @@ static void preview_function_declarations(CodeGen *g, ImportTableEntry *import, |
| 274 | 274 | case NodeTypeSymbol: |
| 275 | 275 | case NodeTypeCastExpr: |
| 276 | 276 | case NodeTypePrefixOpExpr: |
| 277 | case NodeTypeIfExpr: | |
| 277 | 278 | zig_unreachable(); |
| 278 | 279 | } |
| 279 | 280 | } |
| ... | ... | @@ -302,7 +303,9 @@ static void check_type_compatibility(CodeGen *g, AstNode *node, TypeTableEntry * |
| 302 | 303 | add_node_error(g, node, buf_sprintf("type mismatch. expected %s. got %s", buf_ptr(&expected_type->name), buf_ptr(&actual_type->name))); |
| 303 | 304 | } |
| 304 | 305 | |
| 305 | static TypeTableEntry * analyze_expression(CodeGen *g, ImportTableEntry *import, BlockContext *context, TypeTableEntry *expected_type, AstNode *node) { | |
| 306 | static TypeTableEntry * analyze_expression(CodeGen *g, ImportTableEntry *import, BlockContext *context, | |
| 307 | TypeTableEntry *expected_type, AstNode *node) | |
| 308 | { | |
| 306 | 309 | TypeTableEntry *return_type = nullptr; |
| 307 | 310 | switch (node->type) { |
| 308 | 311 | case NodeTypeBlock: |
| ... | ... | @@ -348,10 +351,64 @@ static TypeTableEntry * analyze_expression(CodeGen *g, ImportTableEntry *import, |
| 348 | 351 | |
| 349 | 352 | case NodeTypeBinOpExpr: |
| 350 | 353 | { |
| 351 | // TODO: think about expected types | |
| 352 | analyze_expression(g, import, context, expected_type, node->data.bin_op_expr.op1); | |
| 353 | analyze_expression(g, import, context, expected_type, node->data.bin_op_expr.op2); | |
| 354 | return_type = expected_type; | |
| 354 | switch (node->data.bin_op_expr.bin_op) { | |
| 355 | case BinOpTypeBoolOr: | |
| 356 | case BinOpTypeBoolAnd: | |
| 357 | analyze_expression(g, import, context, g->builtin_types.entry_bool, | |
| 358 | node->data.bin_op_expr.op1); | |
| 359 | analyze_expression(g, import, context, g->builtin_types.entry_bool, | |
| 360 | node->data.bin_op_expr.op2); | |
| 361 | return_type = g->builtin_types.entry_bool; | |
| 362 | break; | |
| 363 | case BinOpTypeCmpEq: | |
| 364 | case BinOpTypeCmpNotEq: | |
| 365 | case BinOpTypeCmpLessThan: | |
| 366 | case BinOpTypeCmpGreaterThan: | |
| 367 | case BinOpTypeCmpLessOrEq: | |
| 368 | case BinOpTypeCmpGreaterOrEq: | |
| 369 | // TODO think how should type checking for these work? | |
| 370 | analyze_expression(g, import, context, g->builtin_types.entry_i32, | |
| 371 | node->data.bin_op_expr.op1); | |
| 372 | analyze_expression(g, import, context, g->builtin_types.entry_i32, | |
| 373 | node->data.bin_op_expr.op2); | |
| 374 | return_type = g->builtin_types.entry_bool; | |
| 375 | break; | |
| 376 | case BinOpTypeBinOr: | |
| 377 | zig_panic("TODO bin or type"); | |
| 378 | break; | |
| 379 | case BinOpTypeBinXor: | |
| 380 | zig_panic("TODO bin xor type"); | |
| 381 | break; | |
| 382 | case BinOpTypeBinAnd: | |
| 383 | zig_panic("TODO bin and type"); | |
| 384 | break; | |
| 385 | case BinOpTypeBitShiftLeft: | |
| 386 | zig_panic("TODO bit shift left type"); | |
| 387 | break; | |
| 388 | case BinOpTypeBitShiftRight: | |
| 389 | zig_panic("TODO bit shift right type"); | |
| 390 | break; | |
| 391 | case BinOpTypeAdd: | |
| 392 | case BinOpTypeSub: | |
| 393 | // TODO think how should type checking for these work? | |
| 394 | analyze_expression(g, import, context, g->builtin_types.entry_i32, | |
| 395 | node->data.bin_op_expr.op1); | |
| 396 | analyze_expression(g, import, context, g->builtin_types.entry_i32, | |
| 397 | node->data.bin_op_expr.op2); | |
| 398 | return_type = g->builtin_types.entry_i32; | |
| 399 | break; | |
| 400 | case BinOpTypeMult: | |
| 401 | zig_panic("TODO mult type"); | |
| 402 | break; | |
| 403 | case BinOpTypeDiv: | |
| 404 | zig_panic("TODO div type"); | |
| 405 | break; | |
| 406 | case BinOpTypeMod: | |
| 407 | zig_panic("TODO modulus type"); | |
| 408 | break; | |
| 409 | case BinOpTypeInvalid: | |
| 410 | zig_unreachable(); | |
| 411 | } | |
| 355 | 412 | break; |
| 356 | 413 | } |
| 357 | 414 | |
| ... | ... | @@ -426,11 +483,46 @@ static TypeTableEntry * analyze_expression(CodeGen *g, ImportTableEntry *import, |
| 426 | 483 | |
| 427 | 484 | case NodeTypeSymbol: |
| 428 | 485 | // look up symbol in symbol table |
| 429 | zig_panic("TODO"); | |
| 486 | zig_panic("TODO analyze_expression symbol"); | |
| 430 | 487 | |
| 431 | 488 | case NodeTypeCastExpr: |
| 489 | zig_panic("TODO analyze_expression cast expr"); | |
| 490 | break; | |
| 491 | ||
| 432 | 492 | case NodeTypePrefixOpExpr: |
| 433 | zig_panic("TODO"); | |
| 493 | switch (node->data.prefix_op_expr.prefix_op) { | |
| 494 | case PrefixOpBoolNot: | |
| 495 | analyze_expression(g, import, context, g->builtin_types.entry_bool, | |
| 496 | node->data.prefix_op_expr.primary_expr); | |
| 497 | return_type = g->builtin_types.entry_bool; | |
| 498 | break; | |
| 499 | case PrefixOpBinNot: | |
| 500 | zig_panic("TODO type check bin not"); | |
| 501 | break; | |
| 502 | case PrefixOpNegation: | |
| 503 | zig_panic("TODO type check negation"); | |
| 504 | break; | |
| 505 | case PrefixOpInvalid: | |
| 506 | zig_unreachable(); | |
| 507 | } | |
| 508 | break; | |
| 509 | case NodeTypeIfExpr: | |
| 510 | { | |
| 511 | analyze_expression(g, import, context, g->builtin_types.entry_bool, node->data.if_expr.condition); | |
| 512 | ||
| 513 | TypeTableEntry *else_type; | |
| 514 | if (node->data.if_expr.else_node) { | |
| 515 | else_type = analyze_expression(g, import, context, expected_type, node->data.if_expr.else_node); | |
| 516 | } else { | |
| 517 | else_type = g->builtin_types.entry_void; | |
| 518 | } | |
| 519 | TypeTableEntry *then_type = analyze_expression(g, import, context, expected_type, | |
| 520 | node->data.if_expr.then_block); | |
| 521 | ||
| 522 | check_type_compatibility(g, node, expected_type, else_type); | |
| 523 | return_type = then_type; | |
| 524 | break; | |
| 525 | } | |
| 434 | 526 | case NodeTypeDirective: |
| 435 | 527 | case NodeTypeFnDecl: |
| 436 | 528 | case NodeTypeFnProto: |
| ... | ... | @@ -445,6 +537,11 @@ static TypeTableEntry * analyze_expression(CodeGen *g, ImportTableEntry *import, |
| 445 | 537 | } |
| 446 | 538 | assert(return_type); |
| 447 | 539 | check_type_compatibility(g, node, expected_type, return_type); |
| 540 | ||
| 541 | assert(!node->codegen_node); | |
| 542 | node->codegen_node = allocate<CodeGenNode>(1); | |
| 543 | node->codegen_node->data.expr_node.type_entry = return_type; | |
| 544 | ||
| 448 | 545 | return return_type; |
| 449 | 546 | } |
| 450 | 547 | |
| ... | ... | @@ -509,6 +606,7 @@ static void analyze_top_level_declaration(CodeGen *g, ImportTableEntry *import, |
| 509 | 606 | case NodeTypeSymbol: |
| 510 | 607 | case NodeTypeCastExpr: |
| 511 | 608 | case NodeTypePrefixOpExpr: |
| 609 | case NodeTypeIfExpr: | |
| 512 | 610 | zig_unreachable(); |
| 513 | 611 | } |
| 514 | 612 | } |
src/codegen.cpp+111-32| ... | ... | @@ -120,6 +120,10 @@ static LLVMValueRef get_variable_value(CodeGen *g, Buf *name) { |
| 120 | 120 | zig_unreachable(); |
| 121 | 121 | } |
| 122 | 122 | |
| 123 | static TypeTableEntry *get_expr_type(AstNode *node) { | |
| 124 | return node->codegen_node->data.expr_node.type_entry; | |
| 125 | } | |
| 126 | ||
| 123 | 127 | static LLVMValueRef gen_fn_call_expr(CodeGen *g, AstNode *node) { |
| 124 | 128 | assert(node->type == NodeTypeFnCallExpr); |
| 125 | 129 | |
| ... | ... | @@ -283,6 +287,7 @@ static LLVMValueRef gen_bool_and_expr(CodeGen *g, AstNode *node) { |
| 283 | 287 | |
| 284 | 288 | LLVMValueRef val1 = gen_expr(g, node->data.bin_op_expr.op1); |
| 285 | 289 | |
| 290 | LLVMBasicBlockRef orig_block = LLVMGetInsertBlock(g->builder); | |
| 286 | 291 | // block for when val1 == true |
| 287 | 292 | LLVMBasicBlockRef true_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "BoolAndTrue"); |
| 288 | 293 | // block for when val1 == false (don't even evaluate the second part) |
| ... | ... | @@ -297,13 +302,14 @@ static LLVMValueRef gen_bool_and_expr(CodeGen *g, AstNode *node) { |
| 297 | 302 | LLVMValueRef val2 = gen_expr(g, node->data.bin_op_expr.op2); |
| 298 | 303 | add_debug_source_node(g, node); |
| 299 | 304 | LLVMValueRef val2_i1 = LLVMBuildICmp(g->builder, LLVMIntEQ, val2, zero, ""); |
| 305 | LLVMBuildBr(g->builder, false_block); | |
| 300 | 306 | |
| 301 | 307 | LLVMPositionBuilderAtEnd(g->builder, false_block); |
| 302 | 308 | add_debug_source_node(g, node); |
| 303 | 309 | LLVMValueRef phi = LLVMBuildPhi(g->builder, LLVMInt1Type(), ""); |
| 304 | 310 | LLVMValueRef one_i1 = LLVMConstAllOnes(LLVMInt1Type()); |
| 305 | 311 | LLVMValueRef incoming_values[2] = {one_i1, val2_i1}; |
| 306 | LLVMBasicBlockRef incoming_blocks[2] = {LLVMGetInsertBlock(g->builder), true_block}; | |
| 312 | LLVMBasicBlockRef incoming_blocks[2] = {orig_block, true_block}; | |
| 307 | 313 | LLVMAddIncoming(phi, incoming_values, incoming_blocks, 2); |
| 308 | 314 | |
| 309 | 315 | return phi; |
| ... | ... | @@ -314,6 +320,8 @@ static LLVMValueRef gen_bool_or_expr(CodeGen *g, AstNode *expr_node) { |
| 314 | 320 | |
| 315 | 321 | LLVMValueRef val1 = gen_expr(g, expr_node->data.bin_op_expr.op1); |
| 316 | 322 | |
| 323 | LLVMBasicBlockRef orig_block = LLVMGetInsertBlock(g->builder); | |
| 324 | ||
| 317 | 325 | // block for when val1 == false |
| 318 | 326 | LLVMBasicBlockRef false_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "BoolOrFalse"); |
| 319 | 327 | // block for when val1 == true (don't even evaluate the second part) |
| ... | ... | @@ -328,13 +336,14 @@ static LLVMValueRef gen_bool_or_expr(CodeGen *g, AstNode *expr_node) { |
| 328 | 336 | LLVMValueRef val2 = gen_expr(g, expr_node->data.bin_op_expr.op2); |
| 329 | 337 | add_debug_source_node(g, expr_node); |
| 330 | 338 | LLVMValueRef val2_i1 = LLVMBuildICmp(g->builder, LLVMIntEQ, val2, zero, ""); |
| 339 | LLVMBuildBr(g->builder, true_block); | |
| 331 | 340 | |
| 332 | 341 | LLVMPositionBuilderAtEnd(g->builder, true_block); |
| 333 | 342 | add_debug_source_node(g, expr_node); |
| 334 | 343 | LLVMValueRef phi = LLVMBuildPhi(g->builder, LLVMInt1Type(), ""); |
| 335 | 344 | LLVMValueRef one_i1 = LLVMConstAllOnes(LLVMInt1Type()); |
| 336 | 345 | LLVMValueRef incoming_values[2] = {one_i1, val2_i1}; |
| 337 | LLVMBasicBlockRef incoming_blocks[2] = {LLVMGetInsertBlock(g->builder), false_block}; | |
| 346 | LLVMBasicBlockRef incoming_blocks[2] = {orig_block, false_block}; | |
| 338 | 347 | LLVMAddIncoming(phi, incoming_values, incoming_blocks, 2); |
| 339 | 348 | |
| 340 | 349 | return phi; |
| ... | ... | @@ -383,9 +392,91 @@ static LLVMValueRef gen_return_expr(CodeGen *g, AstNode *node) { |
| 383 | 392 | return LLVMBuildRetVoid(g->builder); |
| 384 | 393 | } |
| 385 | 394 | } |
| 386 | /* | |
| 387 | Expression : BoolOrExpression | ReturnExpression | |
| 388 | */ | |
| 395 | ||
| 396 | static LLVMValueRef gen_if_expr(CodeGen *g, AstNode *node) { | |
| 397 | assert(node->type == NodeTypeIfExpr); | |
| 398 | assert(node->data.if_expr.condition); | |
| 399 | assert(node->data.if_expr.then_block); | |
| 400 | ||
| 401 | LLVMValueRef cond_value = gen_expr(g, node->data.if_expr.condition); | |
| 402 | ||
| 403 | TypeTableEntry *then_type = get_expr_type(node->data.if_expr.then_block); | |
| 404 | bool use_expr_value = (then_type != g->builtin_types.entry_unreachable && | |
| 405 | then_type != g->builtin_types.entry_void); | |
| 406 | ||
| 407 | if (node->data.if_expr.else_node) { | |
| 408 | LLVMBasicBlockRef then_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "Then"); | |
| 409 | LLVMBasicBlockRef else_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "Else"); | |
| 410 | LLVMBasicBlockRef endif_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "EndIf"); | |
| 411 | ||
| 412 | LLVMBuildCondBr(g->builder, cond_value, then_block, else_block); | |
| 413 | ||
| 414 | LLVMPositionBuilderAtEnd(g->builder, then_block); | |
| 415 | LLVMValueRef then_expr_result = gen_expr(g, node->data.if_expr.then_block); | |
| 416 | LLVMBuildBr(g->builder, endif_block); | |
| 417 | ||
| 418 | LLVMPositionBuilderAtEnd(g->builder, else_block); | |
| 419 | LLVMValueRef else_expr_result = gen_expr(g, node->data.if_expr.else_node); | |
| 420 | LLVMBuildBr(g->builder, endif_block); | |
| 421 | ||
| 422 | LLVMPositionBuilderAtEnd(g->builder, endif_block); | |
| 423 | if (use_expr_value) { | |
| 424 | LLVMValueRef phi = LLVMBuildPhi(g->builder, LLVMTypeOf(then_expr_result), ""); | |
| 425 | LLVMValueRef incoming_values[2] = {then_expr_result, else_expr_result}; | |
| 426 | LLVMBasicBlockRef incoming_blocks[2] = {then_block, else_block}; | |
| 427 | LLVMAddIncoming(phi, incoming_values, incoming_blocks, 2); | |
| 428 | ||
| 429 | return phi; | |
| 430 | } | |
| 431 | ||
| 432 | return nullptr; | |
| 433 | } | |
| 434 | ||
| 435 | assert(!use_expr_value); | |
| 436 | ||
| 437 | LLVMBasicBlockRef then_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "Then"); | |
| 438 | LLVMBasicBlockRef endif_block = LLVMAppendBasicBlock(g->cur_fn->fn_value, "EndIf"); | |
| 439 | ||
| 440 | LLVMBuildCondBr(g->builder, cond_value, then_block, endif_block); | |
| 441 | ||
| 442 | LLVMPositionBuilderAtEnd(g->builder, then_block); | |
| 443 | gen_expr(g, node->data.if_expr.then_block); | |
| 444 | LLVMBuildBr(g->builder, endif_block); | |
| 445 | ||
| 446 | LLVMPositionBuilderAtEnd(g->builder, endif_block); | |
| 447 | return nullptr; | |
| 448 | } | |
| 449 | ||
| 450 | static LLVMValueRef gen_block(CodeGen *g, AstNode *block_node, TypeTableEntry *implicit_return_type) { | |
| 451 | assert(block_node->type == NodeTypeBlock); | |
| 452 | ||
| 453 | ImportTableEntry *import = g->cur_fn->import_entry; | |
| 454 | ||
| 455 | LLVMZigDILexicalBlock *di_block = LLVMZigCreateLexicalBlock(g->dbuilder, g->block_scopes.last(), | |
| 456 | import->di_file, block_node->line + 1, block_node->column + 1); | |
| 457 | g->block_scopes.append(LLVMZigLexicalBlockToScope(di_block)); | |
| 458 | ||
| 459 | add_debug_source_node(g, block_node); | |
| 460 | ||
| 461 | LLVMValueRef return_value; | |
| 462 | for (int i = 0; i < block_node->data.block.statements.length; i += 1) { | |
| 463 | AstNode *statement_node = block_node->data.block.statements.at(i); | |
| 464 | return_value = gen_expr(g, statement_node); | |
| 465 | } | |
| 466 | ||
| 467 | if (implicit_return_type) { | |
| 468 | if (implicit_return_type == g->builtin_types.entry_void) { | |
| 469 | LLVMBuildRetVoid(g->builder); | |
| 470 | } else if (implicit_return_type != g->builtin_types.entry_unreachable) { | |
| 471 | LLVMBuildRet(g->builder, return_value); | |
| 472 | } | |
| 473 | } | |
| 474 | ||
| 475 | g->block_scopes.pop(); | |
| 476 | ||
| 477 | return return_value; | |
| 478 | } | |
| 479 | ||
| 389 | 480 | static LLVMValueRef gen_expr(CodeGen *g, AstNode *node) { |
| 390 | 481 | switch (node->type) { |
| 391 | 482 | case NodeTypeBinOpExpr: |
| ... | ... | @@ -403,6 +494,8 @@ static LLVMValueRef gen_expr(CodeGen *g, AstNode *node) { |
| 403 | 494 | return LLVMBuildUnreachable(g->builder); |
| 404 | 495 | case NodeTypeVoid: |
| 405 | 496 | return nullptr; |
| 497 | case NodeTypeIfExpr: | |
| 498 | return gen_if_expr(g, node); | |
| 406 | 499 | case NodeTypeNumberLiteral: |
| 407 | 500 | { |
| 408 | 501 | Buf *number_str = &node->data.number; |
| ... | ... | @@ -427,6 +520,8 @@ static LLVMValueRef gen_expr(CodeGen *g, AstNode *node) { |
| 427 | 520 | Buf *name = &node->data.symbol; |
| 428 | 521 | return get_variable_value(g, name); |
| 429 | 522 | } |
| 523 | case NodeTypeBlock: | |
| 524 | return gen_block(g, node, nullptr); | |
| 430 | 525 | case NodeTypeRoot: |
| 431 | 526 | case NodeTypeRootExportDecl: |
| 432 | 527 | case NodeTypeFnProto: |
| ... | ... | @@ -434,7 +529,6 @@ static LLVMValueRef gen_expr(CodeGen *g, AstNode *node) { |
| 434 | 529 | case NodeTypeFnDecl: |
| 435 | 530 | case NodeTypeParamDecl: |
| 436 | 531 | case NodeTypeType: |
| 437 | case NodeTypeBlock: | |
| 438 | 532 | case NodeTypeExternBlock: |
| 439 | 533 | case NodeTypeDirective: |
| 440 | 534 | case NodeTypeUse: |
| ... | ... | @@ -443,30 +537,6 @@ static LLVMValueRef gen_expr(CodeGen *g, AstNode *node) { |
| 443 | 537 | zig_unreachable(); |
| 444 | 538 | } |
| 445 | 539 | |
| 446 | static void gen_block(CodeGen *g, ImportTableEntry *import, AstNode *block_node, TypeTableEntry *implicit_return_type) { | |
| 447 | assert(block_node->type == NodeTypeBlock); | |
| 448 | ||
| 449 | LLVMZigDILexicalBlock *di_block = LLVMZigCreateLexicalBlock(g->dbuilder, g->block_scopes.last(), | |
| 450 | import->di_file, block_node->line + 1, block_node->column + 1); | |
| 451 | g->block_scopes.append(LLVMZigLexicalBlockToScope(di_block)); | |
| 452 | ||
| 453 | add_debug_source_node(g, block_node); | |
| 454 | ||
| 455 | LLVMValueRef return_value; | |
| 456 | for (int i = 0; i < block_node->data.block.statements.length; i += 1) { | |
| 457 | AstNode *statement_node = block_node->data.block.statements.at(i); | |
| 458 | return_value = gen_expr(g, statement_node); | |
| 459 | } | |
| 460 | ||
| 461 | if (implicit_return_type == g->builtin_types.entry_void) { | |
| 462 | LLVMBuildRetVoid(g->builder); | |
| 463 | } else if (implicit_return_type != g->builtin_types.entry_unreachable) { | |
| 464 | LLVMBuildRet(g->builder, return_value); | |
| 465 | } | |
| 466 | ||
| 467 | g->block_scopes.pop(); | |
| 468 | } | |
| 469 | ||
| 470 | 540 | static LLVMZigDISubroutineType *create_di_function_type(CodeGen *g, AstNodeFnProto *fn_proto, |
| 471 | 541 | LLVMZigDIFile *di_file) |
| 472 | 542 | { |
| ... | ... | @@ -558,7 +628,7 @@ static void do_code_gen(CodeGen *g) { |
| 558 | 628 | LLVMGetParams(fn, codegen_fn_def->params); |
| 559 | 629 | |
| 560 | 630 | TypeTableEntry *implicit_return_type = codegen_fn_def->implicit_return_type; |
| 561 | gen_block(g, import, fn_def_node->data.fn_def.body, implicit_return_type); | |
| 631 | gen_block(g, fn_def_node->data.fn_def.body, implicit_return_type); | |
| 562 | 632 | |
| 563 | 633 | g->block_scopes.pop(); |
| 564 | 634 | } |
| ... | ... | @@ -585,6 +655,15 @@ static void define_primitive_types(CodeGen *g) { |
| 585 | 655 | buf_init_from_str(&entry->name, "(invalid)"); |
| 586 | 656 | g->builtin_types.entry_invalid = entry; |
| 587 | 657 | } |
| 658 | { | |
| 659 | TypeTableEntry *entry = allocate<TypeTableEntry>(1); | |
| 660 | entry->type_ref = LLVMInt1Type(); | |
| 661 | buf_init_from_str(&entry->name, "bool"); | |
| 662 | entry->di_type = LLVMZigCreateDebugBasicType(g->dbuilder, buf_ptr(&entry->name), 1, 8, | |
| 663 | LLVMZigEncoding_DW_ATE_unsigned()); | |
| 664 | g->type_table.put(&entry->name, entry); | |
| 665 | g->builtin_types.entry_bool = entry; | |
| 666 | } | |
| 588 | 667 | { |
| 589 | 668 | TypeTableEntry *entry = allocate<TypeTableEntry>(1); |
| 590 | 669 | entry->type_ref = LLVMInt8Type(); |
| ... | ... | @@ -803,7 +882,7 @@ static Buf *to_c_type(CodeGen *g, AstNode *type_node) { |
| 803 | 882 | g->c_stdint_used = true; |
| 804 | 883 | return buf_create_from_str("int32_t"); |
| 805 | 884 | } else { |
| 806 | zig_panic("TODO"); | |
| 885 | zig_panic("TODO to_c_type"); | |
| 807 | 886 | } |
| 808 | 887 | } |
| 809 | 888 |
src/parser.cpp+117-20| ... | ... | @@ -91,6 +91,8 @@ const char *node_type_str(NodeType node_type) { |
| 91 | 91 | return "Use"; |
| 92 | 92 | case NodeTypeVoid: |
| 93 | 93 | return "Void"; |
| 94 | case NodeTypeIfExpr: | |
| 95 | return "IfExpr"; | |
| 94 | 96 | } |
| 95 | 97 | zig_unreachable(); |
| 96 | 98 | } |
| ... | ... | @@ -236,7 +238,15 @@ void ast_print(AstNode *node, int indent) { |
| 236 | 238 | fprintf(stderr, "%s '%s'\n", node_type_str(node->type), buf_ptr(&node->data.use.path)); |
| 237 | 239 | break; |
| 238 | 240 | case NodeTypeVoid: |
| 239 | fprintf(stderr, "Void\n"); | |
| 241 | fprintf(stderr, "%s\n", node_type_str(node->type)); | |
| 242 | break; | |
| 243 | case NodeTypeIfExpr: | |
| 244 | fprintf(stderr, "%s\n", node_type_str(node->type)); | |
| 245 | if (node->data.if_expr.condition) | |
| 246 | ast_print(node->data.if_expr.condition, indent + 2); | |
| 247 | ast_print(node->data.if_expr.then_block, indent + 2); | |
| 248 | if (node->data.if_expr.else_node) | |
| 249 | ast_print(node->data.if_expr.else_node, indent + 2); | |
| 240 | 250 | break; |
| 241 | 251 | } |
| 242 | 252 | } |
| ... | ... | @@ -353,6 +363,7 @@ static void ast_invalid_token_error(ParseContext *pc, Token *token) { |
| 353 | 363 | |
| 354 | 364 | static AstNode *ast_parse_expression(ParseContext *pc, int *token_index, bool mandatory); |
| 355 | 365 | static AstNode *ast_parse_block(ParseContext *pc, int *token_index, bool mandatory); |
| 366 | static AstNode *ast_parse_if_expr(ParseContext *pc, int *token_index, bool mandatory); | |
| 356 | 367 | |
| 357 | 368 | |
| 358 | 369 | static void ast_expect_token(ParseContext *pc, Token *token, TokenId token_id) { |
| ... | ... | @@ -558,7 +569,7 @@ static AstNode *ast_parse_grouped_expr(ParseContext *pc, int *token_index, bool |
| 558 | 569 | } |
| 559 | 570 | |
| 560 | 571 | /* |
| 561 | PrimaryExpression : token(Number) | token(String) | token(Unreachable) | GroupedExpression | Block | token(Symbol) | |
| 572 | PrimaryExpression : token(Number) | token(String) | token(Unreachable) | GroupedExpression | token(Symbol) | |
| 562 | 573 | */ |
| 563 | 574 | static AstNode *ast_parse_primary_expr(ParseContext *pc, int *token_index, bool mandatory) { |
| 564 | 575 | Token *token = &pc->tokens->at(*token_index); |
| ... | ... | @@ -588,11 +599,6 @@ static AstNode *ast_parse_primary_expr(ParseContext *pc, int *token_index, bool |
| 588 | 599 | return node; |
| 589 | 600 | } |
| 590 | 601 | |
| 591 | AstNode *block_node = ast_parse_block(pc, token_index, false); | |
| 592 | if (block_node) { | |
| 593 | return block_node; | |
| 594 | } | |
| 595 | ||
| 596 | 602 | AstNode *grouped_expr_node = ast_parse_grouped_expr(pc, token_index, false); |
| 597 | 603 | if (grouped_expr_node) { |
| 598 | 604 | return grouped_expr_node; |
| ... | ... | @@ -975,6 +981,50 @@ static AstNode *ast_parse_bool_and_expr(ParseContext *pc, int *token_index, bool |
| 975 | 981 | return node; |
| 976 | 982 | } |
| 977 | 983 | |
| 984 | /* | |
| 985 | ElseIf : token(Else) IfExpression | |
| 986 | Else : token(Else) Block | |
| 987 | */ | |
| 988 | static AstNode *ast_parse_else_or_else_if(ParseContext *pc, int *token_index, bool mandatory) { | |
| 989 | Token *else_token = &pc->tokens->at(*token_index); | |
| 990 | ||
| 991 | if (else_token->id != TokenIdKeywordElse) { | |
| 992 | if (mandatory) { | |
| 993 | ast_invalid_token_error(pc, else_token); | |
| 994 | } else { | |
| 995 | return nullptr; | |
| 996 | } | |
| 997 | } | |
| 998 | *token_index += 1; | |
| 999 | ||
| 1000 | AstNode *if_expr = ast_parse_if_expr(pc, token_index, false); | |
| 1001 | if (if_expr) | |
| 1002 | return if_expr; | |
| 1003 | ||
| 1004 | return ast_parse_block(pc, token_index, true); | |
| 1005 | } | |
| 1006 | ||
| 1007 | /* | |
| 1008 | IfExpression : token(If) Expression Block option(Else | ElseIf) | |
| 1009 | */ | |
| 1010 | static AstNode *ast_parse_if_expr(ParseContext *pc, int *token_index, bool mandatory) { | |
| 1011 | Token *if_tok = &pc->tokens->at(*token_index); | |
| 1012 | if (if_tok->id != TokenIdKeywordIf) { | |
| 1013 | if (mandatory) { | |
| 1014 | ast_invalid_token_error(pc, if_tok); | |
| 1015 | } else { | |
| 1016 | return nullptr; | |
| 1017 | } | |
| 1018 | } | |
| 1019 | *token_index += 1; | |
| 1020 | ||
| 1021 | AstNode *node = ast_create_node(pc, NodeTypeIfExpr, if_tok); | |
| 1022 | node->data.if_expr.condition = ast_parse_expression(pc, token_index, true); | |
| 1023 | node->data.if_expr.then_block = ast_parse_block(pc, token_index, true); | |
| 1024 | node->data.if_expr.else_node = ast_parse_else_or_else_if(pc, token_index, false); | |
| 1025 | return node; | |
| 1026 | } | |
| 1027 | ||
| 978 | 1028 | /* |
| 979 | 1029 | ReturnExpression : token(Return) option(Expression) |
| 980 | 1030 | */ |
| ... | ... | @@ -1016,27 +1066,68 @@ static AstNode *ast_parse_bool_or_expr(ParseContext *pc, int *token_index, bool |
| 1016 | 1066 | } |
| 1017 | 1067 | |
| 1018 | 1068 | /* |
| 1019 | Expression : BoolOrExpression | ReturnExpression | |
| 1069 | BlockExpression : IfExpression | Block | |
| 1020 | 1070 | */ |
| 1021 | static AstNode *ast_parse_expression(ParseContext *pc, int *token_index, bool mandatory) { | |
| 1071 | static AstNode *ast_parse_block_expr(ParseContext *pc, int *token_index, bool mandatory) { | |
| 1022 | 1072 | Token *token = &pc->tokens->at(*token_index); |
| 1023 | 1073 | |
| 1024 | AstNode *return_expr = ast_parse_return_expr(pc, token_index, false); | |
| 1025 | if (return_expr) | |
| 1026 | return return_expr; | |
| 1074 | AstNode *if_expr = ast_parse_if_expr(pc, token_index, false); | |
| 1075 | if (if_expr) | |
| 1076 | return if_expr; | |
| 1077 | ||
| 1078 | AstNode *block = ast_parse_block(pc, token_index, false); | |
| 1079 | if (block) | |
| 1080 | return block; | |
| 1081 | ||
| 1082 | if (mandatory) | |
| 1083 | ast_invalid_token_error(pc, token); | |
| 1084 | ||
| 1085 | return nullptr; | |
| 1086 | } | |
| 1087 | ||
| 1088 | /* | |
| 1089 | NonBlockExpression : BoolOrExpression | ReturnExpression | |
| 1090 | */ | |
| 1091 | static AstNode *ast_parse_non_block_expr(ParseContext *pc, int *token_index, bool mandatory) { | |
| 1092 | Token *token = &pc->tokens->at(*token_index); | |
| 1027 | 1093 | |
| 1028 | 1094 | AstNode *bool_or_expr = ast_parse_bool_or_expr(pc, token_index, false); |
| 1029 | 1095 | if (bool_or_expr) |
| 1030 | 1096 | return bool_or_expr; |
| 1031 | 1097 | |
| 1032 | if (!mandatory) | |
| 1033 | return nullptr; | |
| 1098 | AstNode *return_expr = ast_parse_return_expr(pc, token_index, false); | |
| 1099 | if (return_expr) | |
| 1100 | return return_expr; | |
| 1034 | 1101 | |
| 1035 | ast_invalid_token_error(pc, token); | |
| 1102 | if (mandatory) | |
| 1103 | ast_invalid_token_error(pc, token); | |
| 1104 | ||
| 1105 | return nullptr; | |
| 1106 | } | |
| 1107 | ||
| 1108 | /* | |
| 1109 | Expression : BlockExpression | NonBlockExpression | |
| 1110 | */ | |
| 1111 | static AstNode *ast_parse_expression(ParseContext *pc, int *token_index, bool mandatory) { | |
| 1112 | Token *token = &pc->tokens->at(*token_index); | |
| 1113 | ||
| 1114 | AstNode *block_expr = ast_parse_block_expr(pc, token_index, false); | |
| 1115 | if (block_expr) | |
| 1116 | return block_expr; | |
| 1117 | ||
| 1118 | AstNode *non_block_expr = ast_parse_non_block_expr(pc, token_index, false); | |
| 1119 | if (non_block_expr) | |
| 1120 | return non_block_expr; | |
| 1121 | ||
| 1122 | if (mandatory) | |
| 1123 | ast_invalid_token_error(pc, token); | |
| 1124 | ||
| 1125 | return nullptr; | |
| 1036 | 1126 | } |
| 1037 | 1127 | |
| 1038 | 1128 | /* |
| 1039 | Block : token(LBrace) list(option(Expression), token(Semicolon)) token(RBrace) | |
| 1129 | Statement : NonBlockExpression token(Semicolon) | BlockExpression | |
| 1130 | Block : token(LBrace) list(option(Statement), token(Semicolon)) token(RBrace) | |
| 1040 | 1131 | */ |
| 1041 | 1132 | static AstNode *ast_parse_block(ParseContext *pc, int *token_index, bool mandatory) { |
| 1042 | 1133 | Token *last_token = &pc->tokens->at(*token_index); |
| ... | ... | @@ -1058,16 +1149,22 @@ static AstNode *ast_parse_block(ParseContext *pc, int *token_index, bool mandato |
| 1058 | 1149 | // {2;} -> {2;void} |
| 1059 | 1150 | // {;2} -> {void;2} |
| 1060 | 1151 | for (;;) { |
| 1061 | AstNode *expression_node = ast_parse_expression(pc, token_index, false); | |
| 1062 | if (!expression_node) { | |
| 1063 | expression_node = ast_create_node(pc, NodeTypeVoid, last_token); | |
| 1152 | AstNode *statement_node = ast_parse_block_expr(pc, token_index, false); | |
| 1153 | bool semicolon_expected = !statement_node; | |
| 1154 | if (!statement_node) { | |
| 1155 | statement_node = ast_parse_non_block_expr(pc, token_index, false); | |
| 1156 | if (!statement_node) { | |
| 1157 | statement_node = ast_create_node(pc, NodeTypeVoid, last_token); | |
| 1158 | } | |
| 1064 | 1159 | } |
| 1065 | node->data.block.statements.append(expression_node); | |
| 1160 | node->data.block.statements.append(statement_node); | |
| 1066 | 1161 | |
| 1067 | 1162 | last_token = &pc->tokens->at(*token_index); |
| 1068 | 1163 | if (last_token->id == TokenIdRBrace) { |
| 1069 | 1164 | *token_index += 1; |
| 1070 | 1165 | return node; |
| 1166 | } else if (!semicolon_expected) { | |
| 1167 | continue; | |
| 1071 | 1168 | } else if (last_token->id == TokenIdSemicolon) { |
| 1072 | 1169 | *token_index += 1; |
| 1073 | 1170 | } else { |
src/parser.hpp+8| ... | ... | @@ -39,6 +39,7 @@ enum NodeType { |
| 39 | 39 | NodeTypeFnCallExpr, |
| 40 | 40 | NodeTypeUse, |
| 41 | 41 | NodeTypeVoid, |
| 42 | NodeTypeIfExpr, | |
| 42 | 43 | }; |
| 43 | 44 | |
| 44 | 45 | struct AstNodeRoot { |
| ... | ... | @@ -167,6 +168,12 @@ struct AstNodeUse { |
| 167 | 168 | ZigList<AstNode *> *directives; |
| 168 | 169 | }; |
| 169 | 170 | |
| 171 | struct AstNodeIfExpr { | |
| 172 | AstNode *condition; | |
| 173 | AstNode *then_block; | |
| 174 | AstNode *else_node; // null, block node, or other if expr node | |
| 175 | }; | |
| 176 | ||
| 170 | 177 | struct AstNode { |
| 171 | 178 | enum NodeType type; |
| 172 | 179 | int line; |
| ... | ... | @@ -190,6 +197,7 @@ struct AstNode { |
| 190 | 197 | AstNodePrefixOpExpr prefix_op_expr; |
| 191 | 198 | AstNodeFnCallExpr fn_call_expr; |
| 192 | 199 | AstNodeUse use; |
| 200 | AstNodeIfExpr if_expr; | |
| 193 | 201 | Buf number; |
| 194 | 202 | Buf string; |
| 195 | 203 | Buf symbol; |
src/semantic_info.hpp+6| ... | ... | @@ -63,6 +63,7 @@ struct CodeGen { |
| 63 | 63 | HashMap<Buf *, ImportTableEntry *, buf_hash, buf_eql_buf> import_table; |
| 64 | 64 | |
| 65 | 65 | struct { |
| 66 | TypeTableEntry *entry_bool; | |
| 66 | 67 | TypeTableEntry *entry_u8; |
| 67 | 68 | TypeTableEntry *entry_i32; |
| 68 | 69 | TypeTableEntry *entry_string_literal; |
| ... | ... | @@ -111,10 +112,15 @@ struct FnDefNode { |
| 111 | 112 | LLVMValueRef *params; |
| 112 | 113 | }; |
| 113 | 114 | |
| 115 | struct ExprNode { | |
| 116 | TypeTableEntry *type_entry; | |
| 117 | }; | |
| 118 | ||
| 114 | 119 | struct CodeGenNode { |
| 115 | 120 | union { |
| 116 | 121 | TypeNode type_node; // for NodeTypeType |
| 117 | 122 | FnDefNode fn_def_node; // for NodeTypeFnDef |
| 123 | ExprNode expr_node; // for all the expression nodes | |
| 118 | 124 | } data; |
| 119 | 125 | }; |
| 120 | 126 |
src/tokenizer.cpp+6| ... | ... | @@ -183,6 +183,10 @@ static void end_token(Tokenize *t) { |
| 183 | 183 | t->cur_tok->id = TokenIdKeywordUse; |
| 184 | 184 | } else if (mem_eql_str(token_mem, token_len, "void")) { |
| 185 | 185 | t->cur_tok->id = TokenIdKeywordVoid; |
| 186 | } else if (mem_eql_str(token_mem, token_len, "if")) { | |
| 187 | t->cur_tok->id = TokenIdKeywordIf; | |
| 188 | } else if (mem_eql_str(token_mem, token_len, "else")) { | |
| 189 | t->cur_tok->id = TokenIdKeywordElse; | |
| 186 | 190 | } |
| 187 | 191 | |
| 188 | 192 | t->cur_tok = nullptr; |
| ... | ... | @@ -577,6 +581,8 @@ static const char * token_name(Token *token) { |
| 577 | 581 | case TokenIdKeywordAs: return "As"; |
| 578 | 582 | case TokenIdKeywordUse: return "Use"; |
| 579 | 583 | case TokenIdKeywordVoid: return "Void"; |
| 584 | case TokenIdKeywordIf: return "If"; | |
| 585 | case TokenIdKeywordElse: return "Else"; | |
| 580 | 586 | case TokenIdLParen: return "LParen"; |
| 581 | 587 | case TokenIdRParen: return "RParen"; |
| 582 | 588 | case TokenIdComma: return "Comma"; |
src/tokenizer.hpp+2| ... | ... | @@ -24,6 +24,8 @@ enum TokenId { |
| 24 | 24 | TokenIdKeywordAs, |
| 25 | 25 | TokenIdKeywordUse, |
| 26 | 26 | TokenIdKeywordVoid, |
| 27 | TokenIdKeywordIf, | |
| 28 | TokenIdKeywordElse, | |
| 27 | 29 | TokenIdLParen, |
| 28 | 30 | TokenIdRParen, |
| 29 | 31 | TokenIdComma, |
test/run_tests.cpp+24| ... | ... | @@ -189,6 +189,30 @@ static void add_compiling_test_cases(void) { |
| 189 | 189 | )SOURCE"); |
| 190 | 190 | } |
| 191 | 191 | |
| 192 | add_simple_case("if statements", R"SOURCE( | |
| 193 | #link("c") | |
| 194 | extern { | |
| 195 | fn puts(s: *const u8) -> i32; | |
| 196 | fn exit(code: i32) -> unreachable; | |
| 197 | } | |
| 198 | ||
| 199 | export fn _start() -> unreachable { | |
| 200 | if 1 != 0 { | |
| 201 | puts("1 is true"); | |
| 202 | } else { | |
| 203 | puts("1 is false"); | |
| 204 | } | |
| 205 | if 0 != 0 { | |
| 206 | puts("0 is true"); | |
| 207 | } else if 1 - 1 != 0 { | |
| 208 | puts("1 - 1 is true"); | |
| 209 | } | |
| 210 | if !(0 != 0) { | |
| 211 | puts("!0 is true"); | |
| 212 | } | |
| 213 | exit(0); | |
| 214 | } | |
| 215 | )SOURCE", "1 is true\n!0 is true\n"); | |
| 192 | 216 | } |
| 193 | 217 | |
| 194 | 218 | static void add_compile_failure_test_cases(void) { |