Rev 5631 | Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 5598 | pavelyakov | 1 | #ifndef INCLUDE_LEXER_H |
| 2 | #define INCLUDE_LEXER_H |
||
| 3 | |||
| 4 | #ifndef INCLUDE_STRING_H |
||
| 5 | #include "../lib/strings.h" |
||
| 6 | #endif |
||
| 7 | /** Splits text into tokens |
||
| 8 | * Author :Pavel Yakovlev |
||
| 9 | * Homepage:https://vk.com/pavelyakov39 |
||
| 10 | */ |
||
| 11 | |||
| 12 | /** Example: |
||
| 13 | * lexer lex; |
||
| 14 | * lex.load("var a=123;"); |
||
| 15 | * lex.next(); |
||
| 16 | * lex.token; //TOKEN == 'var' |
||
| 17 | * lex.type ; //TYPE == LEX_VAR |
||
| 18 | * |
||
| 19 | * lex.next(); |
||
| 20 | * lex.token; //TOKEN == 'a' |
||
| 21 | * lex.type ; //TYPE == LEX_VAR |
||
| 22 | * |
||
| 23 | * lex.next(); |
||
| 24 | * lex.token; //TOKEN == '=' |
||
| 25 | * lex.type ; //TYPE == LEX_IND |
||
| 26 | * |
||
| 27 | * lex.next(); |
||
| 28 | * lex.token; //TOKEN == '123' |
||
| 29 | * lex.type ; //TYPE == LEX_DEC |
||
| 30 | * |
||
| 31 | * lex.next(); |
||
| 32 | * lex.token; //TOKEN == ';' |
||
| 33 | * lex.type ; //TYPE == LEX_IND |
||
| 34 | * |
||
| 35 | * lex.next(); |
||
| 36 | * lex.token; //TOKEN == '' |
||
| 37 | * lex.type ; //TYPE == LEX_END |
||
| 38 | */ |
||
| 39 | |||
| 40 | #define LEX_END 1 |
||
| 41 | #define LEX_STR 2 |
||
| 42 | #define LEX_DEC 3 |
||
| 43 | #define LEX_VAR 4 |
||
| 44 | #define LEX_FNC 5 |
||
| 45 | #define LEX_IND 6 |
||
| 46 | #define LEX_NUL 0 |
||
| 47 | |||
| 48 | :char const_token_lexer[1024]; |
||
| 49 | :struct lexer |
||
| 50 | { |
||
| 51 | byte cmd; |
||
| 52 | dword token,text; |
||
| 53 | byte type; |
||
| 54 | char quote; |
||
| 55 | signed count,length; |
||
| 56 | dword next(void); |
||
| 57 | dword back(void); |
||
| 58 | void load(dword _text); |
||
| 59 | void expected(dword _text); |
||
| 60 | }; |
||
| 61 | |||
| 62 | |||
| 63 | :void expected(dword _text) |
||
| 64 | { |
||
| 65 | notify(_text); |
||
| 66 | ExitProcess(); |
||
| 67 | } |
||
| 68 | |||
| 69 | :void lexer::load(dword _text) |
||
| 70 | { |
||
| 71 | text = _text; |
||
| 72 | } |
||
| 73 | |||
| 74 | :dword lexer::next(void) |
||
| 75 | { |
||
| 76 | char s; |
||
| 77 | dword pos,in; |
||
| 78 | pos = #const_token_lexer; |
||
| 79 | in = text; |
||
| 80 | |||
| 81 | NEXT_TOKEN: |
||
| 82 | length = 0; |
||
| 83 | loop() |
||
| 84 | { |
||
| 85 | s = DSBYTE[in]; |
||
| 86 | if(s!=9)&&(s!=10)&&(s!=13)&&(s!=32)break; |
||
| 87 | in++; |
||
| 88 | text++; |
||
| 89 | } |
||
| 90 | |||
| 91 | if(s==0){type=LEX_END;DSBYTE[pos]=0;token="";return token;} |
||
| 92 | |||
| 93 | if(s=='/') |
||
| 94 | { |
||
| 95 | in++; |
||
| 96 | s = DSBYTE[in]; |
||
| 97 | |||
| 98 | // Line comments |
||
| 99 | if(s=='/') |
||
| 100 | { |
||
| 101 | loop() |
||
| 102 | { |
||
| 103 | in++; |
||
| 104 | s = DSBYTE[in]; |
||
| 105 | if(s==10)||(s==13)||(s==0)goto NEXT_TOKEN; |
||
| 106 | /* Add comments*/ |
||
| 107 | } |
||
| 108 | } |
||
| 109 | if(s=='*') |
||
| 110 | { |
||
| 111 | loop() |
||
| 112 | { |
||
| 113 | in++; |
||
| 114 | s = DSBYTE[in]; |
||
| 115 | if(s=='*')if(DSBYTE[in+1]=='/') |
||
| 116 | { |
||
| 117 | in+=2; |
||
| 118 | goto NEXT_TOKEN; |
||
| 119 | } |
||
| 120 | } |
||
| 121 | } |
||
| 122 | } |
||
| 123 | |||
| 124 | if (strchr("=<>!~&|#",s)) |
||
| 125 | { |
||
| 126 | loop() |
||
| 127 | { |
||
| 128 | if (!strchr("=<>!~&|#",s)) break; |
||
| 129 | |||
| 130 | DSBYTE[pos] = s; |
||
| 131 | pos++; |
||
| 132 | |||
| 133 | in++; |
||
| 134 | s = DSBYTE[in]; |
||
| 135 | } |
||
| 136 | type = LEX_IND; |
||
| 137 | } |
||
| 138 | else if (strchr(";(,)}{[]+-.*/:^%?$@№`",s)) |
||
| 139 | { |
||
| 140 | DSBYTE[pos] = s; |
||
| 141 | pos++; |
||
| 142 | type = LEX_IND; |
||
| 143 | in++; |
||
| 144 | } |
||
| 145 | else if(s>='0')&&(s<='9') |
||
| 146 | { |
||
| 147 | loop() |
||
| 148 | { |
||
| 149 | if(s<'0')||(s>'9')if(s!='.')break; |
||
| 150 | |||
| 151 | DSBYTE[pos] = s; |
||
| 152 | pos++; |
||
| 153 | |||
| 154 | in++; |
||
| 155 | s = DSBYTE[in]; |
||
| 156 | } |
||
| 157 | type = LEX_DEC; |
||
| 158 | } |
||
| 159 | else if(s>='A')&&(s<='z')&&(!strchr("[]\\^`",s)) |
||
| 160 | { |
||
| 161 | loop() |
||
| 162 | { |
||
| 163 | if(s<'A')||(s>'z')if(s<'0')||(s>'9')break; |
||
| 164 | if(strchr("[]\\^`",s))break; |
||
| 165 | |||
| 166 | DSBYTE[pos] = s; |
||
| 167 | pos++; |
||
| 168 | |||
| 169 | in++; |
||
| 170 | s = DSBYTE[in]; |
||
| 171 | } |
||
| 172 | |||
| 173 | loop() |
||
| 174 | { |
||
| 175 | s = DSBYTE[in]; |
||
| 176 | if(s!=9)if(s!=10)if(s!=13)if(s!=32)break; |
||
| 177 | in++; |
||
| 178 | text++; |
||
| 179 | } |
||
| 180 | type = LEX_VAR; |
||
| 181 | if(s=='(')type = LEX_FNC; |
||
| 182 | } |
||
| 183 | else if(s=='"')||(s=='\'') |
||
| 184 | { |
||
| 185 | quote = s; |
||
| 186 | in++; |
||
| 187 | s = DSBYTE[in]; |
||
| 188 | loop() |
||
| 189 | { |
||
| 190 | if(s=='\\') |
||
| 191 | { |
||
| 192 | in++; |
||
| 193 | s = DSBYTE[in]; |
||
| 194 | if(!s){type = LEX_STR;goto GOTO_LEX_END;} |
||
| 195 | if(!cmd)switch(s) |
||
| 196 | { |
||
| 197 | case 'n':s='\n';break; |
||
| 198 | case 'r':s='\r';break; |
||
| 199 | case 't':s='\t';break; |
||
| 200 | } |
||
| 201 | else { |
||
| 202 | DSBYTE[pos] = '\\'; |
||
| 203 | pos++; |
||
| 204 | } |
||
| 205 | goto LEX_STEP_1; |
||
| 206 | } |
||
| 207 | if(!s){type = LEX_STR;goto GOTO_LEX_END;} |
||
| 208 | else if(s==quote)break; |
||
| 209 | LEX_STEP_1: |
||
| 210 | DSBYTE[pos] = s; |
||
| 211 | pos++; |
||
| 212 | in++; |
||
| 213 | s = DSBYTE[in]; |
||
| 214 | } |
||
| 215 | in++; |
||
| 216 | type = LEX_STR; |
||
| 217 | } |
||
| 218 | else { |
||
| 219 | in++; |
||
| 220 | type = LEX_NUL; |
||
| 221 | DSBYTE[pos] = s; |
||
| 222 | pos++; |
||
| 223 | } |
||
| 224 | GOTO_LEX_END: |
||
| 225 | length = in-text; |
||
| 226 | text = in; |
||
| 227 | DSBYTE[pos] = 0; |
||
| 228 | token = #const_token_lexer; |
||
| 229 | return token; |
||
| 230 | } |
||
| 231 | |||
| 232 | #endif |