diff options
author | Mike Buland <eichlan@xagasoft.com> | 2010-10-12 07:35:08 +0000 |
---|---|---|
committer | Mike Buland <eichlan@xagasoft.com> | 2010-10-12 07:35:08 +0000 |
commit | 0981b9d6a12bd7aadbf9286459e033ac1a2ba910 (patch) | |
tree | fa461761086f182e43f94637a9c8df040333e8a4 | |
parent | 1ee5f374ed986333d5cdbbf41390f1c4c755a8e3 (diff) | |
download | libbu++-0981b9d6a12bd7aadbf9286459e033ac1a2ba910.tar.gz libbu++-0981b9d6a12bd7aadbf9286459e033ac1a2ba910.tar.bz2 libbu++-0981b9d6a12bd7aadbf9286459e033ac1a2ba910.tar.xz libbu++-0981b9d6a12bd7aadbf9286459e033ac1a2ba910.zip |
Ok, libbu++ compiles again, the basic parser system is getting there, I think,
it's still a little tricky becasue you have to do the non-terminal prime
seperation yourself (I forget what it's really called), but it's going quite
well. After a few tweaks to the core of it, we should be able to do some
math :)
-rw-r--r-- | src/lexer.cpp | 2 | ||||
-rw-r--r-- | src/lexer.h | 8 | ||||
-rw-r--r-- | src/parser.cpp | 93 | ||||
-rw-r--r-- | src/parser.h | 70 | ||||
-rw-r--r-- | src/tools/parser.cpp | 79 |
5 files changed, 246 insertions, 6 deletions
diff --git a/src/lexer.cpp b/src/lexer.cpp index c7a6fcb..af9a23f 100644 --- a/src/lexer.cpp +++ b/src/lexer.cpp | |||
@@ -13,7 +13,7 @@ Bu::Lexer::Token::Token() : | |||
13 | { | 13 | { |
14 | } | 14 | } |
15 | 15 | ||
16 | Bu::Lexer::Token::Token( int iToken ) : | 16 | Bu::Lexer::Token::Token( Bu::Lexer::TokenType iToken ) : |
17 | iToken( iToken ) | 17 | iToken( iToken ) |
18 | { | 18 | { |
19 | } | 19 | } |
diff --git a/src/lexer.h b/src/lexer.h index 37d268f..5847269 100644 --- a/src/lexer.h +++ b/src/lexer.h | |||
@@ -19,19 +19,21 @@ namespace Bu | |||
19 | Lexer(); | 19 | Lexer(); |
20 | virtual ~Lexer(); | 20 | virtual ~Lexer(); |
21 | 21 | ||
22 | typedef int32_t TokenType; | ||
23 | |||
22 | class Token | 24 | class Token |
23 | { | 25 | { |
24 | public: | 26 | public: |
25 | Token(); | 27 | Token(); |
26 | Token( int iToken ); | 28 | Token( TokenType iToken ); |
27 | 29 | ||
28 | template<class t> | 30 | template<class t> |
29 | Token( int iToken, const t &v ) : | 31 | Token( TokenType iToken, const t &v ) : |
30 | iToken( iToken ), | 32 | iToken( iToken ), |
31 | vExtra( v ) | 33 | vExtra( v ) |
32 | { | 34 | { |
33 | } | 35 | } |
34 | int iToken; | 36 | TokenType iToken; |
35 | Bu::Variant vExtra; | 37 | Bu::Variant vExtra; |
36 | }; | 38 | }; |
37 | 39 | ||
diff --git a/src/parser.cpp b/src/parser.cpp index 7015070..e8e8ff2 100644 --- a/src/parser.cpp +++ b/src/parser.cpp | |||
@@ -41,3 +41,96 @@ void Bu::Parser::parse() | |||
41 | } | 41 | } |
42 | } | 42 | } |
43 | 43 | ||
44 | void Bu::Parser::setRootNonTerminal( int iRoot ) | ||
45 | { | ||
46 | iRootNonTerminal = iRoot; | ||
47 | } | ||
48 | |||
49 | void Bu::Parser::setRootNonTerminal( const Bu::FString &sRoot ) | ||
50 | { | ||
51 | setRootNonTerminal( hNonTerminalName.get( sRoot ) ); | ||
52 | } | ||
53 | |||
54 | int Bu::Parser::addNonTerminal( const Bu::FString &sName, NonTerminal &nt ) | ||
55 | { | ||
56 | int iId = aNonTerminal.getSize(); | ||
57 | aNonTerminal.append( nt ); | ||
58 | hNonTerminalName.insert( sName, iId ); | ||
59 | return iId; | ||
60 | } | ||
61 | |||
62 | int Bu::Parser::addNonTerminal( const Bu::FString &sName ) | ||
63 | { | ||
64 | int iId = aNonTerminal.getSize(); | ||
65 | aNonTerminal.append( NonTerminal() ); | ||
66 | hNonTerminalName.insert( sName, iId ); | ||
67 | return iId; | ||
68 | } | ||
69 | |||
70 | void Bu::Parser::setNonTerminal( const Bu::FString &sName, NonTerminal &nt ) | ||
71 | { | ||
72 | aNonTerminal[hNonTerminalName.get(sName)] = nt; | ||
73 | } | ||
74 | |||
75 | int Bu::Parser::getNonTerminalId( const Bu::FString &sName ) | ||
76 | { | ||
77 | return hNonTerminalName.get( sName ); | ||
78 | } | ||
79 | |||
80 | int Bu::Parser::addReduction( const Bu::FString &sName, const Reduction &r ) | ||
81 | { | ||
82 | int iId = aReduction.getSize(); | ||
83 | aReduction.append( r ); | ||
84 | hReductionName.insert( sName, iId ); | ||
85 | return iId; | ||
86 | } | ||
87 | |||
88 | int Bu::Parser::addReduction( const Bu::FString &sName ) | ||
89 | { | ||
90 | int iId = aReduction.getSize(); | ||
91 | aReduction.append( Reduction() ); | ||
92 | hReductionName.insert( sName, iId ); | ||
93 | return iId; | ||
94 | } | ||
95 | |||
96 | void Bu::Parser::setReduction( const Bu::FString &sName, const Reduction &r ) | ||
97 | { | ||
98 | aReduction[hReductionName.get(sName)] = r; | ||
99 | } | ||
100 | |||
101 | int Bu::Parser::getReductionId( const Bu::FString &sName ) | ||
102 | { | ||
103 | return hReductionName.get( sName ); | ||
104 | } | ||
105 | |||
106 | // | ||
107 | // Bu::Parser::State | ||
108 | // | ||
109 | |||
110 | Bu::Parser::State::State( Bu::Parser::State::Type eType, int iIndex ) : | ||
111 | eType( eType ), | ||
112 | iIndex( iIndex ) | ||
113 | { | ||
114 | } | ||
115 | |||
116 | Bu::Parser::State::~State() | ||
117 | { | ||
118 | } | ||
119 | |||
120 | // | ||
121 | // Bu::Parser::NonTerminal | ||
122 | // | ||
123 | |||
124 | Bu::Parser::NonTerminal::NonTerminal() | ||
125 | { | ||
126 | } | ||
127 | |||
128 | Bu::Parser::NonTerminal::~NonTerminal() | ||
129 | { | ||
130 | } | ||
131 | |||
132 | void Bu::Parser::NonTerminal::addProduction( Production p ) | ||
133 | { | ||
134 | lProduction.append( p ); | ||
135 | } | ||
136 | |||
diff --git a/src/parser.h b/src/parser.h index 26b15a6..e720e54 100644 --- a/src/parser.h +++ b/src/parser.h | |||
@@ -1,8 +1,12 @@ | |||
1 | #ifndef BU_PARSER_H | 1 | #ifndef BU_PARSER_H |
2 | #define BU_PARSER_H | 2 | #define BU_PARSER_H |
3 | 3 | ||
4 | #include "bu/list.h" | ||
5 | #include "bu/fstring.h" | 4 | #include "bu/fstring.h" |
5 | #include "bu/list.h" | ||
6 | #include "bu/array.h" | ||
7 | #include "bu/hash.h" | ||
8 | #include "bu/signals.h" | ||
9 | |||
6 | #include "bu/lexer.h" | 10 | #include "bu/lexer.h" |
7 | 11 | ||
8 | namespace Bu | 12 | namespace Bu |
@@ -35,13 +39,75 @@ namespace Bu | |||
35 | */ | 39 | */ |
36 | void parse(); | 40 | void parse(); |
37 | 41 | ||
42 | void setRootNonTerminal( int iRoot ); | ||
43 | void setRootNonTerminal( const Bu::FString &sRoot ); | ||
44 | |||
45 | typedef Bu::Signal1<void, Parser &> Reduction; | ||
46 | |||
47 | /** | ||
48 | * Represents a possible state, either a terminal or non-terminal symbol | ||
49 | * in a Production. | ||
50 | */ | ||
51 | class State | ||
52 | { | ||
53 | public: | ||
54 | enum Type | ||
55 | { | ||
56 | typeTerminal, | ||
57 | typeTerminalPush, | ||
58 | typeNonTerminal, | ||
59 | typeReduction | ||
60 | }; | ||
61 | |||
62 | State( Type eType, int iIndex ); | ||
63 | virtual ~State(); | ||
64 | |||
65 | private: | ||
66 | Type eType; | ||
67 | int iIndex; | ||
68 | }; | ||
69 | |||
70 | typedef Bu::List<State> Production; | ||
71 | |||
72 | class NonTerminal | ||
73 | { | ||
74 | public: | ||
75 | NonTerminal(); | ||
76 | virtual ~NonTerminal(); | ||
77 | |||
78 | void addProduction( Production p ); | ||
79 | |||
80 | private: | ||
81 | typedef Bu::List<Production> ProductionList; | ||
82 | ProductionList lProduction; | ||
83 | }; | ||
84 | |||
85 | int addNonTerminal( const Bu::FString &sName, NonTerminal &nt ); | ||
86 | int addNonTerminal( const Bu::FString &sName ); | ||
87 | void setNonTerminal( const Bu::FString &sName, NonTerminal &nt ); | ||
88 | int getNonTerminalId( const Bu::FString &sName ); | ||
89 | |||
90 | int addReduction( const Bu::FString &sName, const Reduction &r ); | ||
91 | int addReduction( const Bu::FString &sName ); | ||
92 | void setReduction( const Bu::FString &sName, const Reduction &r ); | ||
93 | int getReductionId( const Bu::FString &sName ); | ||
94 | |||
38 | private: | 95 | private: |
39 | typedef Bu::List<Lexer *> LexerStack; | 96 | typedef Bu::List<Lexer *> LexerStack; |
40 | typedef Bu::List<Lexer::Token *> TokenStack; | 97 | typedef Bu::List<Lexer::Token *> TokenStack; |
41 | typedef Bu::List<State *> StateStack; | 98 | typedef Bu::List<Production::const_iterator> StateStack; |
99 | typedef Bu::Array<Reduction> ReductionArray; | ||
100 | typedef Bu::Hash<Bu::FString,int> NameIndexHash; | ||
101 | typedef Bu::Array<NonTerminal> NonTerminalArray; | ||
102 | |||
42 | LexerStack sLexer; | 103 | LexerStack sLexer; |
43 | TokenStack sToken; | 104 | TokenStack sToken; |
44 | StateStack sState; | 105 | StateStack sState; |
106 | ReductionArray aReduction; | ||
107 | NameIndexHash hReductionName; | ||
108 | NonTerminalArray aNonTerminal; | ||
109 | NameIndexHash hNonTerminalName; | ||
110 | int iRootNonTerminal; | ||
45 | }; | 111 | }; |
46 | }; | 112 | }; |
47 | 113 | ||
diff --git a/src/tools/parser.cpp b/src/tools/parser.cpp index a70dfa4..e4fc95f 100644 --- a/src/tools/parser.cpp +++ b/src/tools/parser.cpp | |||
@@ -149,12 +149,91 @@ private: | |||
149 | QueueBuf qbIn; | 149 | QueueBuf qbIn; |
150 | }; | 150 | }; |
151 | 151 | ||
152 | void redAdd( Bu::Parser &p ) | ||
153 | { | ||
154 | } | ||
155 | |||
156 | void redPrint( Bu::Parser &p ) | ||
157 | { | ||
158 | } | ||
159 | |||
152 | int main( int argc, char *argv[] ) | 160 | int main( int argc, char *argv[] ) |
153 | { | 161 | { |
154 | File fIn( argv[1], File::Read ); | 162 | File fIn( argv[1], File::Read ); |
155 | 163 | ||
156 | Parser p; | 164 | Parser p; |
157 | 165 | ||
166 | { | ||
167 | Parser::NonTerminal nt; | ||
168 | int iSelf = p.addNonTerminal("expr"); | ||
169 | nt.addProduction( | ||
170 | Parser::Production( | ||
171 | Parser::State( | ||
172 | Parser::State::typeTerminal, | ||
173 | tokPlus | ||
174 | ) | ||
175 | ).append( | ||
176 | Parser::State( | ||
177 | Parser::State::typeNonTerminal, | ||
178 | iSelf | ||
179 | ) | ||
180 | ).append( | ||
181 | Parser::State( | ||
182 | Parser::State::typeReduction, | ||
183 | p.addReduction("add") | ||
184 | ) | ||
185 | ) | ||
186 | ); | ||
187 | nt.addProduction( | ||
188 | Parser::Production() | ||
189 | ); | ||
190 | p.addNonTerminal( "expr", nt ); | ||
191 | } | ||
192 | { | ||
193 | Parser::NonTerminal nt; | ||
194 | nt.addProduction( | ||
195 | Parser::Production( | ||
196 | Parser::State( | ||
197 | Parser::State::typeTerminalPush, | ||
198 | tokNumber | ||
199 | ) | ||
200 | ).append( | ||
201 | Parser::State( | ||
202 | Parser::State::typeNonTerminal, | ||
203 | p.getNonTerminalId("expr") | ||
204 | ) | ||
205 | ) | ||
206 | ); | ||
207 | p.addNonTerminal( "expr'", nt ); | ||
208 | } | ||
209 | { | ||
210 | Parser::NonTerminal nt; | ||
211 | nt.addProduction( | ||
212 | Parser::Production( | ||
213 | Parser::State( | ||
214 | Parser::State::typeNonTerminal, | ||
215 | p.getNonTerminalId("expr'") | ||
216 | ) | ||
217 | ).append( | ||
218 | Parser::State( | ||
219 | Parser::State::typeTerminal, | ||
220 | tokCompute | ||
221 | ) | ||
222 | ).append( | ||
223 | Parser::State( | ||
224 | Parser::State::typeReduction, | ||
225 | p.addReduction("print") | ||
226 | ) | ||
227 | ) | ||
228 | ); | ||
229 | p.addNonTerminal("input", nt ); | ||
230 | } | ||
231 | |||
232 | p.setRootNonTerminal("input"); | ||
233 | |||
234 | p.setReduction("add", Bu::slot( &redAdd ) ); | ||
235 | p.setReduction("print", Bu::slot( &redPrint ) ); | ||
236 | |||
158 | p.pushLexer( new MathLexer( fIn ) ); | 237 | p.pushLexer( new MathLexer( fIn ) ); |
159 | 238 | ||
160 | p.parse(); | 239 | p.parse(); |