Lines Matching +full:- +full:c
1 //===- MILexer.cpp - Machine instructions lexer implementation ------------===//
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
7 //===----------------------------------------------------------------------===//
11 //===----------------------------------------------------------------------===//
44 char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
48 StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
50 StringRef upto(Cursor C) const {
51 assert(C.Ptr >= Ptr && C.Ptr <= End);
52 return StringRef(Ptr, C.Ptr - Ptr);
63 this->Kind = Kind;
64 this->Range = Range;
80 this->IntVal = std::move(IntVal);
85 static Cursor skipWhitespace(Cursor C) {
86 while (isblank(C.peek()))
87 C.advance();
88 return C;
91 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
94 static Cursor skipComment(Cursor C) {
95 if (C.peek() != ';')
96 return C;
97 while (!isNewlineChar(C.peek()) && !C.isEOF())
98 C.advance();
99 return C;
104 static Cursor skipMachineOperandComment(Cursor C) {
105 if (C.peek() != '/' || C.peek(1) != '*')
106 return C;
108 while (C.peek() != '*' || C.peek(1) != '/')
109 C.advance();
111 C.advance();
112 C.advance();
113 return C;
117 /// expression: [-a-zA-Z$._0-9]
118 static bool isIdentifierChar(char C) {
119 return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
120 C == '$';
128 Cursor C = Cursor(Value.substr(1, Value.size() - 2));
131 Str.reserve(C.remaining().size());
132 while (!C.isEOF()) {
133 char Char = C.peek();
135 if (C.peek(1) == '\\') {
138 C.advance(2);
141 if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
142 Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
143 C.advance(3);
148 C.advance();
154 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
155 assert(C.peek() == '"');
156 for (C.advance(); C.peek() != '"'; C.advance()) {
157 if (C.isEOF() || isNewlineChar(C.peek())) {
159 C.location(),
164 C.advance();
165 return C;
168 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
170 auto Range = C;
171 C.advance(PrefixLength);
172 if (C.peek() == '"') {
173 if (Cursor R = lexStringConstant(C, ErrorCallback)) {
183 while (isIdentifierChar(C.peek()))
184 C.advance();
185 Token.reset(Type, Range.upto(C))
186 .setStringValue(Range.upto(C).drop_front(PrefixLength));
187 return C;
194 .Case("implicit-def", MIToken::kw_implicit_define)
200 .Case("early-clobber", MIToken::kw_early_clobber)
201 .Case("debug-use", MIToken::kw_debug_use)
203 .Case("tied-def", MIToken::kw_tied_def)
204 .Case("frame-setup", MIToken::kw_frame_setup)
205 .Case("frame-destroy", MIToken::kw_frame_destroy)
222 .Case("debug-location", MIToken::kw_debug_location)
223 .Case("debug-instr-number", MIToken::kw_debug_instr_number)
224 .Case("dbg-instr-ref", MIToken::kw_dbg_instr_ref)
246 .Case("target-index", MIToken::kw_target_index)
254 .Case("target-flags", MIToken::kw_target_flags)
256 .Case("non-temporal", MIToken::kw_non_temporal)
264 .Case("jump-table", MIToken::kw_jump_table)
265 .Case("constant-pool", MIToken::kw_constant_pool)
266 .Case("call-entry", MIToken::kw_call_entry)
269 .Case("landing-pad", MIToken::kw_landing_pad)
270 .Case("inlineasm-br-indirect-target",
272 .Case("ehfunclet-entry", MIToken::kw_ehfunclet_entry)
278 .Case("pre-instr-symbol", MIToken::kw_pre_instr_symbol)
279 .Case("post-instr-symbol", MIToken::kw_post_instr_symbol)
280 .Case("heap-alloc-marker", MIToken::kw_heap_alloc_marker)
282 .Case("cfi-type", MIToken::kw_cfi_type)
285 .Case("unknown-size", MIToken::kw_unknown_size)
286 .Case("unknown-address", MIToken::kw_unknown_address)
288 .Case("ir-block-address-taken", MIToken::kw_ir_block_address_taken)
289 .Case("machine-block-address-taken",
291 .Case("call-frame-size", MIToken::kw_call_frame_size)
296 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
297 if (!isalpha(C.peek()) && C.peek() != '_')
299 auto Range = C;
300 while (isIdentifierChar(C.peek()))
301 C.advance();
302 auto Identifier = Range.upto(C);
305 return C;
308 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
310 bool IsReference = C.remaining().starts_with("%bb.");
311 if (!IsReference && !C.remaining().starts_with("bb."))
313 auto Range = C;
315 C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
316 if (!isdigit(C.peek())) {
317 Token.reset(MIToken::Error, C.remaining());
318 ErrorCallback(C.location(), "expected a number after '%bb.'");
319 return C;
321 auto NumberRange = C;
322 while (isdigit(C.peek()))
323 C.advance();
324 StringRef Number = NumberRange.upto(C);
329 if (C.peek() == '.') {
330 C.advance(); // Skip '.'
332 while (isIdentifierChar(C.peek()))
333 C.advance();
337 Range.upto(C))
339 .setStringValue(Range.upto(C).drop_front(StringOffset));
340 return C;
343 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
345 if (!C.remaining().starts_with(Rule) || !isdigit(C.peek(Rule.size())))
347 auto Range = C;
348 C.advance(Rule.size());
349 auto NumberRange = C;
350 while (isdigit(C.peek()))
351 C.advance();
352 Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
353 return C;
356 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
358 if (!C.remaining().starts_with(Rule) || !isdigit(C.peek(Rule.size())))
360 auto Range = C;
361 C.advance(Rule.size());
362 auto NumberRange = C;
363 while (isdigit(C.peek()))
364 C.advance();
365 StringRef Number = NumberRange.upto(C);
367 if (C.peek() == '.') {
368 C.advance();
370 while (isIdentifierChar(C.peek()))
371 C.advance();
373 Token.reset(Kind, Range.upto(C))
375 .setStringValue(Range.upto(C).drop_front(StringOffset));
376 return C;
379 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
380 return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
383 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
384 return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
387 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
388 return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
391 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
392 return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
395 static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token,
398 if (!C.remaining().starts_with(Rule))
400 return lexName(C, Token, MIToken::SubRegisterIndex, Rule.size(),
404 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
406 const StringRef Rule = "%ir-block.";
407 if (!C.remaining().starts_with(Rule))
409 if (isdigit(C.peek(Rule.size())))
410 return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
411 return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
414 static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
417 if (!C.remaining().starts_with(Rule))
419 if (isdigit(C.peek(Rule.size())))
420 return maybeLexIndex(C, Token, Rule, MIToken::IRValue);
421 return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
424 static Cursor maybeLexStringConstant(Cursor C, MIToken &Token,
426 if (C.peek() != '"')
428 return lexName(C, Token, MIToken::StringConstant, /*PrefixLength=*/0,
432 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
433 auto Range = C;
434 C.advance(); // Skip '%'
435 auto NumberRange = C;
436 while (isdigit(C.peek()))
437 C.advance();
438 Token.reset(MIToken::VirtualRegister, Range.upto(C))
439 .setIntegerValue(APSInt(NumberRange.upto(C)));
440 return C;
444 static bool isRegisterChar(char C) {
445 return isIdentifierChar(C) && C != '.';
448 static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) {
449 Cursor Range = C;
450 C.advance(); // Skip '%'
451 while (isRegisterChar(C.peek()))
452 C.advance();
453 Token.reset(MIToken::NamedVirtualRegister, Range.upto(C))
454 .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
455 return C;
458 static Cursor maybeLexRegister(Cursor C, MIToken &Token,
460 if (C.peek() != '%' && C.peek() != '$')
463 if (C.peek() == '%') {
464 if (isdigit(C.peek(1)))
465 return lexVirtualRegister(C, Token);
467 if (isRegisterChar(C.peek(1)))
468 return lexNamedVirtualRegister(C, Token);
473 assert(C.peek() == '$');
474 auto Range = C;
475 C.advance(); // Skip '$'
476 while (isRegisterChar(C.peek()))
477 C.advance();
478 Token.reset(MIToken::NamedRegister, Range.upto(C))
479 .setStringValue(Range.upto(C).drop_front(1)); // Drop the '$'
480 return C;
483 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
485 if (C.peek() != '@')
487 if (!isdigit(C.peek(1)))
488 return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
490 auto Range = C;
491 C.advance(1); // Skip the '@'
492 auto NumberRange = C;
493 while (isdigit(C.peek()))
494 C.advance();
495 Token.reset(MIToken::GlobalValue, Range.upto(C))
496 .setIntegerValue(APSInt(NumberRange.upto(C)));
497 return C;
500 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
502 if (C.peek() != '&')
504 return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1,
508 static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token,
511 if (!C.remaining().starts_with(Rule))
513 auto Start = C;
514 C.advance(Rule.size());
517 if (C.peek() != '"') {
518 while (isIdentifierChar(C.peek()))
519 C.advance();
520 StringRef String = Start.upto(C).drop_front(Rule.size());
521 if (C.peek() != '>') {
522 ErrorCallback(C.location(),
527 C.advance();
529 Token.reset(MIToken::MCSymbol, Start.upto(C)).setStringValue(String);
530 return C;
534 Cursor R = lexStringConstant(C, ErrorCallback);
536 ErrorCallback(C.location(),
555 static bool isValidHexFloatingPointPrefix(char C) {
556 return C == 'H' || C == 'K' || C == 'L' || C == 'M' || C == 'R';
559 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
560 C.advance();
561 // Skip over [0-9]*([eE][-+]?[0-9]+)?
562 while (isdigit(C.peek()))
563 C.advance();
564 if ((C.peek() == 'e' || C.peek() == 'E') &&
565 (isdigit(C.peek(1)) ||
566 ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) {
567 C.advance(2);
568 while (isdigit(C.peek()))
569 C.advance();
571 Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
572 return C;
575 static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) {
576 if (C.peek() != '0' || (C.peek(1) != 'x' && C.peek(1) != 'X'))
578 Cursor Range = C;
579 C.advance(2);
581 if (isValidHexFloatingPointPrefix(C.peek())) {
582 C.advance();
585 while (isxdigit(C.peek()))
586 C.advance();
587 StringRef StrVal = Range.upto(C);
591 Token.reset(MIToken::HexLiteral, Range.upto(C));
592 else // It must be 3, which means that there was a floating-point prefix.
593 Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
594 return C;
597 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
598 if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
600 auto Range = C;
601 C.advance();
602 while (isdigit(C.peek()))
603 C.advance();
604 if (C.peek() == '.')
605 return lexFloatingPointLiteral(Range, C, Token);
606 StringRef StrVal = Range.upto(C);
608 return C;
622 static Cursor maybeLexExclaim(Cursor C, MIToken &Token,
624 if (C.peek() != '!')
626 auto Range = C;
627 C.advance(1);
628 if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
629 Token.reset(MIToken::exclaim, Range.upto(C));
630 return C;
632 while (isIdentifierChar(C.peek()))
633 C.advance();
634 StringRef StrVal = Range.upto(C);
639 return C;
642 static MIToken::TokenKind symbolToken(char C) {
643 switch (C) {
662 case '-':
673 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
676 if (C.peek() == ':' && C.peek(1) == ':') {
680 Kind = symbolToken(C.peek());
683 auto Range = C;
684 C.advance(Length);
685 Token.reset(Kind, Range.upto(C));
686 return C;
689 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
690 if (!isNewlineChar(C.peek()))
692 auto Range = C;
693 C.advance();
694 Token.reset(MIToken::Newline, Range.upto(C));
695 return C;
698 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
700 if (C.peek() != '`')
702 auto Range = C;
703 C.advance();
704 auto StrRange = C;
705 while (C.peek() != '`') {
706 if (C.isEOF() || isNewlineChar(C.peek())) {
708 C.location(),
711 return C;
713 C.advance();
715 StringRef Value = StrRange.upto(C);
716 C.advance();
717 Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value);
718 return C;
723 auto C = skipComment(skipWhitespace(Cursor(Source)));
724 if (C.isEOF()) {
725 Token.reset(MIToken::Eof, C.remaining());
726 return C.remaining();
729 C = skipWhitespace(skipMachineOperandComment(C));
731 if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
733 if (Cursor R = maybeLexIdentifier(C, Token))
735 if (Cursor R = maybeLexJumpTableIndex(C, Token))
737 if (Cursor R = maybeLexStackObject(C, Token))
739 if (Cursor R = maybeLexFixedStackObject(C, Token))
741 if (Cursor R = maybeLexConstantPoolItem(C, Token))
743 if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback))
745 if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
747 if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
749 if (Cursor R = maybeLexRegister(C, Token, ErrorCallback))
751 if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
753 if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
755 if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback))
757 if (Cursor R = maybeLexHexadecimalLiteral(C, Token))
759 if (Cursor R = maybeLexNumericalLiteral(C, Token))
761 if (Cursor R = maybeLexExclaim(C, Token, ErrorCallback))
763 if (Cursor R = maybeLexSymbol(C, Token))
765 if (Cursor R = maybeLexNewline(C, Token))
767 if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
769 if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback))
772 Token.reset(MIToken::Error, C.remaining());
773 ErrorCallback(C.location(),
774 Twine("unexpected character '") + Twine(C.peek()) + "'");
775 return C.remaining();