Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 1 | //===- TGLexer.h - Lexer for TableGen Files ---------------------*- C++ -*-===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
Chris Lattner | 3060910 | 2007-12-29 20:37:13 +0000 | [diff] [blame] | 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This class represents the Lexer for tablegen files. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #ifndef TGLEXER_H |
| 15 | #define TGLEXER_H |
| 16 | |
Michael J. Spencer | 3cc52ea | 2010-11-29 18:47:54 +0000 | [diff] [blame^] | 17 | #include "llvm/Support/DataTypes.h" |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 18 | #include <vector> |
| 19 | #include <string> |
Chris Lattner | 8dcf751 | 2007-11-22 23:19:05 +0000 | [diff] [blame] | 20 | #include <cassert> |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 21 | |
| 22 | namespace llvm { |
| 23 | class MemoryBuffer; |
Chris Lattner | 8070ea3 | 2009-06-21 03:41:50 +0000 | [diff] [blame] | 24 | class SourceMgr; |
Chris Lattner | 1e3a8a4 | 2009-06-21 03:39:35 +0000 | [diff] [blame] | 25 | class SMLoc; |
Benjamin Kramer | d1e1703 | 2010-09-27 17:42:11 +0000 | [diff] [blame] | 26 | class Twine; |
| 27 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 28 | namespace tgtok { |
| 29 | enum TokKind { |
| 30 | // Markers |
| 31 | Eof, Error, |
| 32 | |
| 33 | // Tokens with no info. |
| 34 | minus, plus, // - + |
| 35 | l_square, r_square, // [ ] |
| 36 | l_brace, r_brace, // { } |
| 37 | l_paren, r_paren, // ( ) |
| 38 | less, greater, // < > |
| 39 | colon, semi, // ; : |
| 40 | comma, period, // , . |
| 41 | equal, question, // = ? |
| 42 | |
| 43 | // Keywords. |
| 44 | Bit, Bits, Class, Code, Dag, Def, Defm, Field, In, Int, Let, List, |
| 45 | MultiClass, String, |
| 46 | |
| 47 | // !keywords. |
Chris Lattner | c7252ce | 2010-10-06 00:19:21 +0000 | [diff] [blame] | 48 | XConcat, XSRA, XSRL, XSHL, XStrConcat, XCast, XSubst, |
David Greene | 6786d5e | 2010-01-05 19:11:42 +0000 | [diff] [blame] | 49 | XForEach, XCar, XCdr, XNull, XIf, XEq, |
David Greene | c7cafcd | 2009-04-22 20:18:10 +0000 | [diff] [blame] | 50 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 51 | // Integer value. |
| 52 | IntVal, |
| 53 | |
| 54 | // String valued tokens. |
| 55 | Id, StrVal, VarName, CodeFragment |
| 56 | }; |
| 57 | } |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 58 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 59 | /// TGLexer - TableGen Lexer class. |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 60 | class TGLexer { |
Chris Lattner | 8070ea3 | 2009-06-21 03:41:50 +0000 | [diff] [blame] | 61 | SourceMgr &SrcMgr; |
Chris Lattner | aa739d2 | 2009-03-13 07:05:43 +0000 | [diff] [blame] | 62 | |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 63 | const char *CurPtr; |
Chris Lattner | aa739d2 | 2009-03-13 07:05:43 +0000 | [diff] [blame] | 64 | const MemoryBuffer *CurBuf; |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 65 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 66 | // Information about the current token. |
| 67 | const char *TokStart; |
| 68 | tgtok::TokKind CurCode; |
| 69 | std::string CurStrVal; // This is valid for ID, STRVAL, VARNAME, CODEFRAGMENT |
Dan Gohman | 63f9720 | 2008-10-17 01:33:43 +0000 | [diff] [blame] | 70 | int64_t CurIntVal; // This is valid for INTVAL. |
Chris Lattner | aa739d2 | 2009-03-13 07:05:43 +0000 | [diff] [blame] | 71 | |
| 72 | /// CurBuffer - This is the current buffer index we're lexing from as managed |
| 73 | /// by the SourceMgr object. |
| 74 | int CurBuffer; |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 75 | |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 76 | public: |
Chris Lattner | 8070ea3 | 2009-06-21 03:41:50 +0000 | [diff] [blame] | 77 | TGLexer(SourceMgr &SrcMgr); |
Chris Lattner | aa739d2 | 2009-03-13 07:05:43 +0000 | [diff] [blame] | 78 | ~TGLexer() {} |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 79 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 80 | tgtok::TokKind Lex() { |
| 81 | return CurCode = LexToken(); |
| 82 | } |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 83 | |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 84 | tgtok::TokKind getCode() const { return CurCode; } |
| 85 | |
| 86 | const std::string &getCurStrVal() const { |
| 87 | assert((CurCode == tgtok::Id || CurCode == tgtok::StrVal || |
| 88 | CurCode == tgtok::VarName || CurCode == tgtok::CodeFragment) && |
| 89 | "This token doesn't have a string value"); |
| 90 | return CurStrVal; |
| 91 | } |
Dan Gohman | 63f9720 | 2008-10-17 01:33:43 +0000 | [diff] [blame] | 92 | int64_t getCurIntVal() const { |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 93 | assert(CurCode == tgtok::IntVal && "This token isn't an integer"); |
| 94 | return CurIntVal; |
| 95 | } |
| 96 | |
Chris Lattner | 1e3a8a4 | 2009-06-21 03:39:35 +0000 | [diff] [blame] | 97 | SMLoc getLoc() const; |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 98 | |
Benjamin Kramer | d1e1703 | 2010-09-27 17:42:11 +0000 | [diff] [blame] | 99 | void PrintError(const char *Loc, const Twine &Msg) const; |
| 100 | void PrintError(SMLoc Loc, const Twine &Msg) const; |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 101 | |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 102 | private: |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 103 | /// LexToken - Read the next token and return its code. |
| 104 | tgtok::TokKind LexToken(); |
| 105 | |
Benjamin Kramer | d1e1703 | 2010-09-27 17:42:11 +0000 | [diff] [blame] | 106 | tgtok::TokKind ReturnError(const char *Loc, const Twine &Msg); |
Chris Lattner | c8a9bbc | 2007-11-19 07:38:58 +0000 | [diff] [blame] | 107 | |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 108 | int getNextChar(); |
| 109 | void SkipBCPLComment(); |
| 110 | bool SkipCComment(); |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 111 | tgtok::TokKind LexIdentifier(); |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 112 | bool LexInclude(); |
Chris Lattner | f460165 | 2007-11-22 20:49:04 +0000 | [diff] [blame] | 113 | tgtok::TokKind LexString(); |
| 114 | tgtok::TokKind LexVarName(); |
| 115 | tgtok::TokKind LexNumber(); |
| 116 | tgtok::TokKind LexBracket(); |
| 117 | tgtok::TokKind LexExclaim(); |
Chris Lattner | a805874 | 2007-11-18 02:57:27 +0000 | [diff] [blame] | 118 | }; |
| 119 | |
| 120 | } // end namespace llvm |
| 121 | |
| 122 | #endif |