Rev 5631 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 5598 | pavelyakov | 1 | #ifndef INCLUDE_LEXER_H |
| 2 | #define INCLUDE_LEXER_H |
||
| 5676 | pavelyakov | 3 | #print "[include |
| 5598 | pavelyakov | 4 | |
| 5 | #ifndef INCLUDE_STRING_H |
||
| 6 | #include "../lib/strings.h" |
||
| 7 | #endif |
||
| 5631 | pavelyakov | 8 | |
| 9 | #ifndef INCLUDE_MEM_H |
||
| 10 | #include "../lib/mem.h" |
||
| 11 | #endif |
||
| 5598 | pavelyakov | 12 | /** Splits text into tokens |
| 13 | * Author :Pavel Yakovlev |
||
| 14 | * Homepage:https://vk.com/pavelyakov39 |
||
| 5631 | pavelyakov | 15 | * Ver. : 1.51 |
| 5598 | pavelyakov | 16 | */ |
| 17 | |||
| 18 | /** Example: |
||
| 19 | * lexer lex; |
||
| 20 | * lex.load("var a=123;"); |
||
| 21 | * lex.next(); |
||
| 22 | * lex.token; //TOKEN == 'var' |
||
| 23 | * lex.type ; //TYPE == LEX_VAR |
||
| 24 | * |
||
| 25 | * lex.next(); |
||
| 26 | * lex.token; //TOKEN == 'a' |
||
| 27 | * lex.type ; //TYPE == LEX_VAR |
||
| 28 | * |
||
| 29 | * lex.next(); |
||
| 30 | * lex.token; //TOKEN == '=' |
||
| 31 | * lex.type ; //TYPE == LEX_IND |
||
| 32 | * |
||
| 33 | * lex.next(); |
||
| 34 | * lex.token; //TOKEN == '123' |
||
| 35 | * lex.type ; //TYPE == LEX_DEC |
||
| 36 | * |
||
| 37 | * lex.next(); |
||
| 38 | * lex.token; //TOKEN == ';' |
||
| 39 | * lex.type ; //TYPE == LEX_IND |
||
| 40 | * |
||
| 41 | * lex.next(); |
||
| 42 | * lex.token; //TOKEN == '' |
||
| 43 | * lex.type ; //TYPE == LEX_END |
||
| 44 | */ |
||
| 45 | |||
| 46 | #define LEX_END 1 |
||
| 47 | #define LEX_STR 2 |
||
| 48 | #define LEX_DEC 3 |
||
| 49 | #define LEX_VAR 4 |
||
| 50 | #define LEX_FNC 5 |
||
| 51 | #define LEX_IND 6 |
||
| 52 | #define LEX_NUL 0 |
||
| 53 | |||
| 54 | :char const_token_lexer[1024]; |
||
| 55 | :struct lexer |
||
| 56 | { |
||
| 57 | byte cmd; |
||
| 5631 | pavelyakov | 58 | dword token,text,mem_list,count,buffer_loading; |
| 59 | dword str_buffer; |
||
| 5598 | pavelyakov | 60 | byte type; |
| 61 | char quote; |
||
| 5631 | pavelyakov | 62 | signed length; |
| 5598 | pavelyakov | 63 | dword next(void); |
| 64 | dword back(void); |
||
| 5631 | pavelyakov | 65 | dword list(void); |
| 66 | void free(void); |
||
| 67 | dword position(dword __); |
||
| 5598 | pavelyakov | 68 | void load(dword _text); |
| 69 | void expected(dword _text); |
||
| 70 | }; |
||
| 5631 | pavelyakov | 71 | :dword back(void) |
| 5598 | pavelyakov | 72 | { |
| 5631 | pavelyakov | 73 | |
| 74 | } |
||
| 75 | :dword lexer::list(void) |
||
| 76 | { |
||
| 77 | dword count_mem,buf_loop,pos; |
||
| 78 | count_mem = 0; |
||
| 79 | buf_loop = 5000; // на тыс элементов. |
||
| 80 | count = 0; |
||
| 81 | buffer_loading = malloc(buf_loop); |
||
| 82 | pos = buffer_loading; |
||
| 83 | while(type!=LEX_END) |
||
| 84 | { |
||
| 85 | pos+=count_mem; |
||
| 86 | next(); |
||
| 87 | DSDWORD[pos] = strndup(token,length); |
||
| 88 | pos+=4; |
||
| 89 | DSBYTE [pos] = type; |
||
| 90 | pos++; |
||
| 91 | count++; |
||
| 92 | if(pos-buffer_loading>buf_loop) |
||
| 93 | { |
||
| 94 | buf_loop*=2; |
||
| 95 | buffer_loading = realloc(buffer_loading,buf_loop); |
||
| 96 | } |
||
| 97 | } |
||
| 98 | return buffer_loading; |
||
| 99 | } |
||
| 100 | :void lexer::free(void) |
||
| 101 | { |
||
| 102 | dword z; |
||
| 103 | z = count; |
||
| 104 | while(z) |
||
| 105 | { |
||
| 106 | z--; |
||
| 107 | position(z); |
||
| 108 | ::free(token); |
||
| 109 | } |
||
| 110 | count = 0; |
||
| 111 | ::free(buffer_loading); |
||
| 112 | } |
||
| 113 | :dword lexer::position(dword __) |
||
| 114 | { |
||
| 115 | dword pos1; |
||
| 116 | if(!count)list(); |
||
| 117 | if(__>=count)__=count-1; |
||
| 118 | else if(__<0)__=0; |
||
| 119 | pos1 = __*5; |
||
| 120 | pos1 += buffer_loading; |
||
| 121 | token = DSDWORD[pos1]; |
||
| 122 | pos1++; |
||
| 123 | type = DSBYTE[pos1]; |
||
| 124 | return token; |
||
| 125 | } |
||
| 126 | :void lexer::expected(dword _text) |
||
| 127 | { |
||
| 5598 | pavelyakov | 128 | notify(_text); |
| 129 | ExitProcess(); |
||
| 130 | } |
||
| 131 | |||
| 132 | :void lexer::load(dword _text) |
||
| 133 | { |
||
| 134 | text = _text; |
||
| 5631 | pavelyakov | 135 | count = 0; |
| 136 | str_buffer = 0; |
||
| 5598 | pavelyakov | 137 | } |
| 138 | |||
| 139 | :dword lexer::next(void) |
||
| 140 | { |
||
| 141 | char s; |
||
| 5631 | pavelyakov | 142 | dword len_str_buf,tmp; |
| 5598 | pavelyakov | 143 | dword pos,in; |
| 144 | pos = #const_token_lexer; |
||
| 145 | in = text; |
||
| 5631 | pavelyakov | 146 | //len_str_buf = 1024; |
| 147 | if(str_buffer)::free(str_buffer); |
||
| 5598 | pavelyakov | 148 | NEXT_TOKEN: |
| 149 | length = 0; |
||
| 150 | loop() |
||
| 151 | { |
||
| 152 | s = DSBYTE[in]; |
||
| 153 | if(s!=9)&&(s!=10)&&(s!=13)&&(s!=32)break; |
||
| 154 | in++; |
||
| 155 | text++; |
||
| 156 | } |
||
| 157 | |||
| 158 | if(s==0){type=LEX_END;DSBYTE[pos]=0;token="";return token;} |
||
| 159 | |||
| 160 | if(s=='/') |
||
| 161 | { |
||
| 162 | in++; |
||
| 163 | s = DSBYTE[in]; |
||
| 164 | |||
| 165 | // Line comments |
||
| 166 | if(s=='/') |
||
| 167 | { |
||
| 168 | loop() |
||
| 169 | { |
||
| 170 | in++; |
||
| 171 | s = DSBYTE[in]; |
||
| 172 | if(s==10)||(s==13)||(s==0)goto NEXT_TOKEN; |
||
| 173 | /* Add comments*/ |
||
| 174 | } |
||
| 175 | } |
||
| 176 | if(s=='*') |
||
| 177 | { |
||
| 178 | loop() |
||
| 179 | { |
||
| 180 | in++; |
||
| 181 | s = DSBYTE[in]; |
||
| 182 | if(s=='*')if(DSBYTE[in+1]=='/') |
||
| 183 | { |
||
| 184 | in+=2; |
||
| 185 | goto NEXT_TOKEN; |
||
| 186 | } |
||
| 187 | } |
||
| 188 | } |
||
| 189 | } |
||
| 190 | |||
| 191 | if (strchr("=<>!~&|#",s)) |
||
| 192 | { |
||
| 193 | loop() |
||
| 194 | { |
||
| 195 | if (!strchr("=<>!~&|#",s)) break; |
||
| 196 | |||
| 197 | DSBYTE[pos] = s; |
||
| 198 | pos++; |
||
| 199 | |||
| 200 | in++; |
||
| 201 | s = DSBYTE[in]; |
||
| 202 | } |
||
| 203 | type = LEX_IND; |
||
| 204 | } |
||
| 205 | else if (strchr(";(,)}{[]+-.*/:^%?$@№`",s)) |
||
| 206 | { |
||
| 207 | DSBYTE[pos] = s; |
||
| 208 | pos++; |
||
| 209 | type = LEX_IND; |
||
| 210 | in++; |
||
| 211 | } |
||
| 212 | else if(s>='0')&&(s<='9') |
||
| 213 | { |
||
| 214 | loop() |
||
| 215 | { |
||
| 216 | if(s<'0')||(s>'9')if(s!='.')break; |
||
| 217 | |||
| 218 | DSBYTE[pos] = s; |
||
| 219 | pos++; |
||
| 220 | |||
| 221 | in++; |
||
| 222 | s = DSBYTE[in]; |
||
| 223 | } |
||
| 224 | type = LEX_DEC; |
||
| 225 | } |
||
| 226 | else if(s>='A')&&(s<='z')&&(!strchr("[]\\^`",s)) |
||
| 227 | { |
||
| 228 | loop() |
||
| 229 | { |
||
| 230 | if(s<'A')||(s>'z')if(s<'0')||(s>'9')break; |
||
| 231 | if(strchr("[]\\^`",s))break; |
||
| 232 | |||
| 233 | DSBYTE[pos] = s; |
||
| 234 | pos++; |
||
| 235 | |||
| 236 | in++; |
||
| 237 | s = DSBYTE[in]; |
||
| 238 | } |
||
| 239 | |||
| 240 | loop() |
||
| 241 | { |
||
| 242 | s = DSBYTE[in]; |
||
| 243 | if(s!=9)if(s!=10)if(s!=13)if(s!=32)break; |
||
| 244 | in++; |
||
| 245 | text++; |
||
| 246 | } |
||
| 247 | type = LEX_VAR; |
||
| 248 | if(s=='(')type = LEX_FNC; |
||
| 249 | } |
||
| 250 | else if(s=='"')||(s=='\'') |
||
| 251 | { |
||
| 252 | quote = s; |
||
| 253 | in++; |
||
| 5631 | pavelyakov | 254 | tmp = in; |
| 5598 | pavelyakov | 255 | s = DSBYTE[in]; |
| 256 | loop() |
||
| 257 | { |
||
| 258 | if(s=='\\') |
||
| 259 | { |
||
| 260 | in++; |
||
| 261 | s = DSBYTE[in]; |
||
| 262 | if(!s){type = LEX_STR;goto GOTO_LEX_END;} |
||
| 263 | if(!cmd)switch(s) |
||
| 264 | { |
||
| 265 | case 'n':s='\n';break; |
||
| 266 | case 'r':s='\r';break; |
||
| 267 | case 't':s='\t';break; |
||
| 268 | } |
||
| 269 | else { |
||
| 270 | DSBYTE[pos] = '\\'; |
||
| 271 | pos++; |
||
| 272 | } |
||
| 273 | goto LEX_STEP_1; |
||
| 274 | } |
||
| 275 | if(!s){type = LEX_STR;goto GOTO_LEX_END;} |
||
| 276 | else if(s==quote)break; |
||
| 277 | LEX_STEP_1: |
||
| 278 | DSBYTE[pos] = s; |
||
| 279 | pos++; |
||
| 280 | in++; |
||
| 5631 | pavelyakov | 281 | /*if(in-tmp>len_str_buf) |
| 282 | { |
||
| 283 | if(str_buffer) |
||
| 284 | { |
||
| 285 | tmp = len_str_buf; |
||
| 286 | len_str_buf+=1024; |
||
| 287 | str_buffer = realloc(str_buffer,len_str_buf+1); |
||
| 288 | strlcpy(str_buffer+tmp,#const_token_lexer,1024); |
||
| 289 | pos = #const_token_lexer; |
||
| 290 | } |
||
| 291 | else { |
||
| 292 | len_str_buf+=1024; |
||
| 293 | str_buffer = malloc(len_str_buf+1); |
||
| 294 | strlcpy(str_buffer,#const_token_lexer,1024); |
||
| 295 | pos = #const_token_lexer; |
||
| 296 | } |
||
| 297 | }*/ |
||
| 5598 | pavelyakov | 298 | s = DSBYTE[in]; |
| 299 | } |
||
| 300 | in++; |
||
| 5631 | pavelyakov | 301 | /*tmp = pos-in; |
| 302 | if(str_buffer) |
||
| 303 | { |
||
| 304 | if(tmp) |
||
| 305 | { |
||
| 306 | str_buffer = realloc(str_buffer,tmp+1); |
||
| 307 | strlcpy(str_buffer+len_str_buf,#const_token_lexer,tmp); |
||
| 308 | } |
||
| 309 | type = LEX_STR; |
||
| 310 | length = len_str_buf+tmp; |
||
| 311 | text = in; |
||
| 312 | tmp = str_buffer+length; |
||
| 313 | DSBYTE[tmp] = 0; |
||
| 314 | token = str_buffer; |
||
| 315 | return token; |
||
| 316 | }*/ |
||
| 5598 | pavelyakov | 317 | type = LEX_STR; |
| 318 | } |
||
| 319 | else { |
||
| 320 | in++; |
||
| 321 | type = LEX_NUL; |
||
| 322 | DSBYTE[pos] = s; |
||
| 323 | pos++; |
||
| 324 | } |
||
| 325 | GOTO_LEX_END: |
||
| 326 | length = in-text; |
||
| 327 | text = in; |
||
| 328 | DSBYTE[pos] = 0; |
||
| 329 | token = #const_token_lexer; |
||
| 330 | return token; |
||
| 331 | } |
||
| 332 | |||
| 333 | #endif |