blob: 0246ab6b900b2d7aaf9f3be17805b77db6b3da15 [file] [log] [blame]
Chris Lattnera8058742007-11-18 02:57:27 +00001//===- TGLexer.h - Lexer for TableGen Files ---------------------*- C++ -*-===//
2//
3// The LLVM Compiler Infrastructure
4//
Chris Lattner30609102007-12-29 20:37:13 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Chris Lattnera8058742007-11-18 02:57:27 +00007//
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. Spencer3cc52ea2010-11-29 18:47:54 +000017#include "llvm/Support/DataTypes.h"
Chris Lattnera8058742007-11-18 02:57:27 +000018#include <string>
Joerg Sonnenbergerdd137902011-06-01 13:10:15 +000019#include <vector>
Chris Lattner8dcf7512007-11-22 23:19:05 +000020#include <cassert>
Chris Lattnera8058742007-11-18 02:57:27 +000021
22namespace llvm {
23class MemoryBuffer;
Chris Lattner8070ea32009-06-21 03:41:50 +000024class SourceMgr;
Chris Lattner1e3a8a42009-06-21 03:39:35 +000025class SMLoc;
Benjamin Kramerd1e17032010-09-27 17:42:11 +000026class Twine;
27
Chris Lattnerf4601652007-11-22 20:49:04 +000028namespace 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, // < >
Francois Pichet606957f2011-03-14 02:30:32 +000039 colon, semi, // : ;
Chris Lattnerf4601652007-11-22 20:49:04 +000040 comma, period, // , .
41 equal, question, // = ?
David Greened3d1cad2011-10-19 13:04:43 +000042 paste, // #
43
Chris Lattnerf4601652007-11-22 20:49:04 +000044 // Keywords.
David Greenea1b1b792011-10-07 18:25:05 +000045 Bit, Bits, Class, Code, Dag, Def, Defm, Field, In, Int, Let, List,
Chris Lattnerf4601652007-11-22 20:49:04 +000046 MultiClass, String,
47
48 // !keywords.
Chris Lattnerc7252ce2010-10-06 00:19:21 +000049 XConcat, XSRA, XSRL, XSHL, XStrConcat, XCast, XSubst,
David Greene1434f662011-01-07 17:05:37 +000050 XForEach, XHead, XTail, XEmpty, XIf, XEq,
David Greenec7cafcd2009-04-22 20:18:10 +000051
Chris Lattnerf4601652007-11-22 20:49:04 +000052 // Integer value.
53 IntVal,
54
55 // String valued tokens.
56 Id, StrVal, VarName, CodeFragment
57 };
58}
Chris Lattnera8058742007-11-18 02:57:27 +000059
Chris Lattnerf4601652007-11-22 20:49:04 +000060/// TGLexer - TableGen Lexer class.
Chris Lattnera8058742007-11-18 02:57:27 +000061class TGLexer {
Chris Lattner8070ea32009-06-21 03:41:50 +000062 SourceMgr &SrcMgr;
Chris Lattneraa739d22009-03-13 07:05:43 +000063
Chris Lattnera8058742007-11-18 02:57:27 +000064 const char *CurPtr;
Chris Lattneraa739d22009-03-13 07:05:43 +000065 const MemoryBuffer *CurBuf;
Chris Lattnera8058742007-11-18 02:57:27 +000066
Chris Lattnerf4601652007-11-22 20:49:04 +000067 // Information about the current token.
68 const char *TokStart;
69 tgtok::TokKind CurCode;
70 std::string CurStrVal; // This is valid for ID, STRVAL, VARNAME, CODEFRAGMENT
Dan Gohman63f97202008-10-17 01:33:43 +000071 int64_t CurIntVal; // This is valid for INTVAL.
Chris Lattneraa739d22009-03-13 07:05:43 +000072
73 /// CurBuffer - This is the current buffer index we're lexing from as managed
74 /// by the SourceMgr object.
75 int CurBuffer;
Joerg Sonnenbergerdd137902011-06-01 13:10:15 +000076 /// Dependencies - This is the list of all included files.
77 std::vector<std::string> Dependencies;
Chris Lattnera8058742007-11-18 02:57:27 +000078
Chris Lattnera8058742007-11-18 02:57:27 +000079public:
Chris Lattner8070ea32009-06-21 03:41:50 +000080 TGLexer(SourceMgr &SrcMgr);
Chris Lattneraa739d22009-03-13 07:05:43 +000081 ~TGLexer() {}
Chris Lattnera8058742007-11-18 02:57:27 +000082
Chris Lattnerf4601652007-11-22 20:49:04 +000083 tgtok::TokKind Lex() {
84 return CurCode = LexToken();
85 }
Joerg Sonnenbergerdd137902011-06-01 13:10:15 +000086
87 const std::vector<std::string> &getDependencies() const {
88 return Dependencies;
89 }
Chris Lattnera8058742007-11-18 02:57:27 +000090
Chris Lattnerf4601652007-11-22 20:49:04 +000091 tgtok::TokKind getCode() const { return CurCode; }
92
93 const std::string &getCurStrVal() const {
94 assert((CurCode == tgtok::Id || CurCode == tgtok::StrVal ||
95 CurCode == tgtok::VarName || CurCode == tgtok::CodeFragment) &&
96 "This token doesn't have a string value");
97 return CurStrVal;
98 }
Dan Gohman63f97202008-10-17 01:33:43 +000099 int64_t getCurIntVal() const {
Chris Lattnerf4601652007-11-22 20:49:04 +0000100 assert(CurCode == tgtok::IntVal && "This token isn't an integer");
101 return CurIntVal;
102 }
103
Chris Lattner1e3a8a42009-06-21 03:39:35 +0000104 SMLoc getLoc() const;
Chris Lattnerf4601652007-11-22 20:49:04 +0000105
Chris Lattnera8058742007-11-18 02:57:27 +0000106private:
Chris Lattnerf4601652007-11-22 20:49:04 +0000107 /// LexToken - Read the next token and return its code.
108 tgtok::TokKind LexToken();
109
Benjamin Kramerd1e17032010-09-27 17:42:11 +0000110 tgtok::TokKind ReturnError(const char *Loc, const Twine &Msg);
Chris Lattnerc8a9bbc2007-11-19 07:38:58 +0000111
Chris Lattnera8058742007-11-18 02:57:27 +0000112 int getNextChar();
David Greenea761f922011-10-19 13:03:35 +0000113 int peekNextChar(int Index);
Chris Lattnera8058742007-11-18 02:57:27 +0000114 void SkipBCPLComment();
115 bool SkipCComment();
Chris Lattnerf4601652007-11-22 20:49:04 +0000116 tgtok::TokKind LexIdentifier();
Chris Lattnera8058742007-11-18 02:57:27 +0000117 bool LexInclude();
Chris Lattnerf4601652007-11-22 20:49:04 +0000118 tgtok::TokKind LexString();
119 tgtok::TokKind LexVarName();
120 tgtok::TokKind LexNumber();
121 tgtok::TokKind LexBracket();
122 tgtok::TokKind LexExclaim();
Chris Lattnera8058742007-11-18 02:57:27 +0000123};
124
125} // end namespace llvm
126
127#endif