/vm/tinyrb/grammar.leg
Unknown | 334 lines | 282 code | 52 blank | 0 comment | 0 complexity | 4dcd747ac6b099c7cba2309e4786b4fa MD5 | raw file
1%{ 2#include <stdlib.h> 3#include "tr.h" 4#include "internal.h" 5 6/*#define YY_DEBUG 1*/ 7 8#define YYSTYPE OBJ 9#define YYMALLOC TR_MALLOC 10#define YYREALLOC TR_REALLOC 11#define yyvm compiler->vm 12 13static char *charbuf; 14static char *sbuf; 15static size_t nbuf; 16static TrCompiler *compiler; 17 18#define YY_INPUT(buf, result, max_size) { \ 19 int yyc; \ 20 if (charbuf && *charbuf != '\0') \ 21 yyc= *charbuf++; \ 22 else \ 23 yyc= EOF; \ 24 result= (EOF == yyc) ? 0 : (*(buf)= yyc, 1); \ 25} 26 27/* TODO grow buffer */ 28#define STRING_MAX 4096 29#define STRING_START sbuf = TR_ALLOC_N(char, STRING_MAX); nbuf = 0 30#define STRING_PUSH(P,L) \ 31 assert(nbuf + (L) < 4096); \ 32 TR_MEMCPY_N(sbuf + nbuf, (P), char, (L)); \ 33 nbuf += (L) 34 35%} 36 37Root = s:Stmts EOF { compiler->node = NODE(ROOT, s) } 38 39Stmts = SEP* 40 - head:Stmt Comment? { head = NODES(head) } 41 ( SEP - tail:Stmt Comment? { PUSH_NODE(head, tail) } 42 | SEP - Comment 43 )* SEP? { $$ = head } 44 | SEP+ { $$ = NODES_N(0) } 45 46OptStmts = Stmts 47 | - SEP? { $$ = NODES_N(0) } 48 49Stmt = While 50 | Until 51 | If 52 | Unless 53 | Def 54 | Class 55 | Module 56 | Expr 57 58Expr = Assign 59 | AsgnCall 60 | UnaryOp 61 | BinOp 62 | SpecCall 63 | Call 64 | Range 65 | Yield 66 | Return 67 | Break 68 | Value 69 70Comment = - '#' (!EOL .)* 71 72Call = { block = rcv = 0 } 73 ( rcv:Value '.' 74 )? ( rmsg:Message '.' { rcv = NODE2(SEND, rcv, rmsg) } 75 )* msg:Message 76 - block:Block? { $$ = NODE3(SEND, rcv, msg, block) } 77 78# TODO refactor head part w/ Call maybe eh? 79AsgnCall = { rcv = 0 } 80 ( rcv:Value '.' 81 )? ( rmsg:Message '.' { rcv = NODE2(SEND, rcv, rmsg) } 82 )* msg:ID - asg:ASSIGN 83 - val:Stmt { VM = yyvm; $$ = NODE2(SEND, rcv, NODE2(MSG, SYMCAT(msg, asg), NODES(NODE(ARG, val)))) } 84 85Receiver = ( { rcv = 0 } 86 rcv:Call 87 | rcv:Value 88 ) { $$ = rcv } 89 90SpecCall = rcv:Receiver '[' args:Args ']' 91 - ASSIGN - val:Stmt { PUSH_NODE(args, NODE(ARG, val)); $$ = NODE2(SEND, rcv, NODE2(MSG, TrSymbol_new(yyvm, "[]="), args)) } 92 | rcv:Receiver '[' args:Args ']' { $$ = NODE2(SEND, rcv, NODE2(MSG, TrSymbol_new(yyvm, "[]"), args)) } 93 94BinOp = ( rcv:SpecCall | rcv:Receiver ) 95 - 96 ( 97 '&&' - arg:Expr { $$ = NODE2(AND, rcv, arg) } 98 | '||' - arg:Expr { $$ = NODE2(OR, rcv, arg) } 99 | '+' - arg:Expr { $$ = NODE2(ADD, rcv, arg) } 100 | '-' - arg:Expr { $$ = NODE2(SUB, rcv, arg) } 101 | '<' - arg:Expr { $$ = NODE2(LT, rcv, arg) } 102 | op:BINOP - arg:Expr { $$ = NODE2(SEND, rcv, NODE2(MSG, op, NODES(NODE(ARG, arg)))) } 103 ) 104 105UnaryOp = '-' rcv:Expr { $$ = NODE(NEG, rcv) } 106 | '!' rcv:Expr { $$ = NODE(NOT, rcv) } 107 108Message = name:ID { args = 0 } 109 ( '(' args:Args? ')' 110 | SPACE args:Args 111 )? { $$ = NODE2(MSG, name, args) } 112 113Args = - head:Expr - { head = NODES(NODE(ARG, head)) } 114 ( ',' - tail:Expr - { PUSH_NODE(head, NODE(ARG, tail)) } 115 )* ( ',' - '*' splat:Expr - { PUSH_NODE(head, NODE2(ARG, splat, 1)) } 116 )? { $$ = head } 117 | - '*' splat:Expr - { $$ = NODES(NODE2(ARG, splat, 1)) } 118 119Block = 'do' SEP 120 - body:OptStmts - 121 'end' { $$ = NODE(BLOCK, body) } 122 | 'do' - '|' params:Params '|' SEP 123 - body:OptStmts - 124 'end' { $$ = NODE2(BLOCK, body, params) } 125 # FIXME this might hang the parser and is very slow. 126 # Clash with Hash for sure. 127 #| '{' - body:OptStmts - '}' { $$ = NODE(BLOCK, body) } 128 #| '{' - '|' params:Params '|' 129 # - body:OptStmts - '}' { $$ = NODE2(BLOCK, body, params) } 130 131Assign = name:ID - ASSIGN - val:Stmt { $$ = NODE2(ASSIGN, name, val) } 132 | name:CONST - ASSIGN - val:Stmt { $$ = NODE2(SETCONST, name, val) } 133 | name:IVAR - ASSIGN - val:Stmt { $$ = NODE2(SETIVAR, name, val) } 134 | name:CVAR - ASSIGN - val:Stmt { $$ = NODE2(SETCVAR, name, val) } 135 | name:GLOBAL - ASSIGN - val:Stmt { $$ = NODE2(SETGLOBAL, name, val) } 136 137While = 'while' SPACE cond:Expr SEP 138 body:Stmts - 139 'end' { $$ = NODE2(WHILE, cond, body) } 140 141Until = 'until' SPACE cond:Expr SEP 142 body:Stmts - 143 'end' { $$ = NODE2(UNTIL, cond, body) } 144 145If = 'if' SPACE cond:Expr SEP { else_body = 0 } 146 body:Stmts - 147 else_body:Else? 148 'end' { $$ = NODE3(IF, cond, body, else_body) } 149 | body:Expr - 'if' - cond:Expr { $$ = NODE2(IF, cond, NODES(body)) } 150 151Unless = 'unless' SPACE cond:Expr SEP { else_body = 0 } 152 body:Stmts - 153 else_body:Else? 154 'end' { $$ = NODE3(UNLESS, cond, body, else_body) } 155 | body:Expr - 156 'unless' - cond:Expr { $$ = NODE2(UNLESS, cond, NODES(body)) } 157 158Else = 'else' SEP - body:Stmts - { $$ = body } 159 160Method = rcv:ID '.' name:METHOD { $$ = NODE2(METHOD, NODE2(SEND, 0, NODE(MSG, rcv)), name) } 161 | rcv:Value '.' name:METHOD { $$ = NODE2(METHOD, rcv, name) } 162 | name:METHOD { $$ = NODE2(METHOD, 0, name) } 163 164Def = 'def' SPACE method:Method { params = 0 } 165 (- '(' params:Params? ')')? SEP 166 body:OptStmts - 167 'end' { $$ = NODE3(DEF, method, params ? params : NODES_N(0), body) } 168 169Params = head:Param { head = NODES(head) } 170 ( ',' tail:Param { PUSH_NODE(head, tail) } 171 )* { $$ = head } 172 173Param = - name:ID - '=' - def:Expr { $$ = NODE3(PARAM, name, 0, def) } 174 | - name:ID - { $$ = NODE(PARAM, name) } 175 | - '*' name:ID - { $$ = NODE2(PARAM, name, 1) } 176 177Class = 'class' SPACE name:CONST { super = 0 } 178 (- '<' - super:CONST)? SEP 179 body:OptStmts - 180 'end' { $$ = NODE3(CLASS, name, super, body) } 181 182Module = 'module' SPACE name:CONST SEP 183 body:OptStmts - 184 'end' { $$ = NODE3(MODULE, name, 0, body) } 185 186Range = s:Receiver - '..' - e:Expr { $$ = NODE3(RANGE, s, e, 0) } 187 | s:Receiver - '...' - e:Expr { $$ = NODE3(RANGE, s, e, 1) } 188 189Yield = 'yield' SPACE args:AryItems { $$ = NODE(YIELD, args) } 190 | 'yield' '(' args:AryItems ')' { $$ = NODE(YIELD, args) } 191 | 'yield' { $$ = NODE(YIELD, NODES_N(0)) } 192 193Return = 'return' SPACE arg:Expr - !',' { $$ = NODE(RETURN, arg) } 194 | 'return' '(' arg:Expr ')' - !','{ $$ = NODE(RETURN, arg) } 195 | 'return' SPACE args:AryItems { $$ = NODE(RETURN, NODE(ARRAY, args)) } 196 | 'return' '(' args:AryItems ')' { $$ = NODE(RETURN, NODE(ARRAY, args)) } 197 | 'return' { $$ = NODE(RETURN, 0) } 198 199Break = 'break' { $$ = NODE(BREAK, 0) } 200 201Value = v:NUMBER { $$ = NODE(VALUE, v) } 202 | v:SYMBOL { $$ = NODE(VALUE, v) } 203 | v:REGEXP { $$ = NODE(VALUE, v) } 204 | v:STRING1 { $$ = NODE(STRING, v) } 205 | v:STRING2 { $$ = NODE(STRING, v) } 206 | v:CONST { $$ = NODE(CONST, v) } 207 | 'nil' { $$ = NODE(NIL, 0) } 208 | 'true' { $$ = NODE(BOOL, TR_TRUE) } 209 | 'false' { $$ = NODE(BOOL, TR_FALSE) } 210 | 'self' { $$ = NODE(SELF, 0) } 211 | name:IVAR { $$ = NODE(GETIVAR, name) } 212 | name:CVAR { $$ = NODE(GETCVAR, name) } 213 | name:GLOBAL { $$ = NODE(GETGLOBAL, name) } # TODO 214 | '[' - ']' { $$ = NODE(ARRAY, NODES_N(0)) } 215 | '[' - items:AryItems - ']' { $$ = NODE(ARRAY, items) } 216 | '{' - '}' { $$ = NODE(HASH, NODES_N(0)) } 217 | '{' - items:HashItems - '}' { $$ = NODE(HASH, items) } 218 | '(' - Expr - ')' 219 220AryItems = - head:Expr - { head = NODES(head) } 221 ( ',' - tail:Expr - { PUSH_NODE(head, tail) } 222 )* { $$ = head } 223 224HashItems = head:Expr - '=>' - val:Expr { head = NODES_N(2, head, val) } 225 ( - ',' - key:Expr - { PUSH_NODE(head, key) } 226 '=>' - val:Expr { PUSH_NODE(head, val) } 227 )* { $$ = head } 228 229KEYWORD = 'while' | 'until' | 'do' | 'end' | 230 'if' | 'unless' | 'else' | 231 'true' | 'false' | 'nil' | 'self' | 232 'class' | 'module' | 'def' | 233 'yield' | 'return' | 'break' 234 235NAME = [a-zA-Z0-9_]+ 236ID = !'self' # self is special, can never be a method name 237 < KEYWORD > &('.' | '(' | '[') { $$ = TrSymbol_new(yyvm, yytext) } # hm, there's probably a better way 238 | < KEYWORD NAME > { $$ = TrSymbol_new(yyvm, yytext) } 239 | !KEYWORD 240 < [a-z_] NAME? 241 ( '=' &'(' | '!'| '?' )? > { $$ = TrSymbol_new(yyvm, yytext) } 242CONST = < [A-Z] NAME? > { $$ = TrSymbol_new(yyvm, yytext) } 243BINOP = < ( '**' | '^' | '&' | '|' | '~' | 244 '+' | '-' | '*' | '/' | '%' | '<=>' | 245 '<<' | '>>' | '==' | '=~' | '!=' | '===' | 246 '<' | '>' | '<=' | '>=' 247 ) > { $$ = TrSymbol_new(yyvm, yytext) } 248UNOP = < ( '-@' | '!' ) > { $$ = TrSymbol_new(yyvm, yytext) } 249METHOD = ID | UNOP | BINOP 250ASSIGN = < '=' > &(!'=') { $$ = TrSymbol_new(yyvm, yytext) } 251IVAR = < '@' NAME > { $$ = TrSymbol_new(yyvm, yytext) } 252CVAR = < '@@' NAME > { $$ = TrSymbol_new(yyvm, yytext) } 253GLOBAL = < '$' NAME > { $$ = TrSymbol_new(yyvm, yytext) } 254NUMBER = < [0-9]+ > { $$ = TR_INT2FIX(atoi(yytext)) } 255SYMBOL = ':' < (NAME | KEYWORD) > { $$ = TrSymbol_new(yyvm, yytext) } 256 257STRING1 = '\'' { STRING_START } 258 ( 259 '\\\'' { STRING_PUSH("'", 1) } 260 | < [^\'] > { STRING_PUSH(yytext, yyleng) } 261 )* '\'' { $$ = TrString_new2(yyvm, sbuf) } 262 263ESC_CHAR = '\\n' { STRING_PUSH("\n", 1) } 264 | '\\b' { STRING_PUSH("\b", 1) } 265 | '\\f' { STRING_PUSH("\f", 1) } 266 | '\\r' { STRING_PUSH("\r", 1) } 267 | '\\t' { STRING_PUSH("\t", 1) } 268 | '\\\"' { STRING_PUSH("\"", 1) } 269 | '\\\\' { STRING_PUSH("\\", 1) } 270 271STRING2 = '"' { STRING_START } 272 ( 273 ESC_CHAR 274 | < [^\"] > { STRING_PUSH(yytext, yyleng) } #" for higlighting 275 )* 276 '"' { $$ = TrString_new2(yyvm, sbuf) } 277 278REGEXP = '/' { STRING_START } 279 ( 280 ESC_CHAR 281 | < [^/] > { STRING_PUSH(yytext, yyleng) } 282 )* 283 '/' { $$ = TrRegexp_new(yyvm, sbuf, 0) } 284 285- = [ \t]* 286SPACE = [ ]+ 287EOL = ( '\n' | '\r\n' | '\r' ) { compiler->line++ } 288EOF = !. 289SEP = ( - Comment? (EOL | ';') )+ 290 291%% 292 293/* Raise a syntax error. */ 294OBJ yyerror() { 295 VM = yyvm; 296 OBJ msg = tr_sprintf(vm, "SyntaxError in %s at line %d", TR_STR_PTR(compiler->filename), compiler->line); 297 /* Stupid ugly code, just to build a string... I suck... */ 298 if (yytext[0]) TrString_push(vm, msg, tr_sprintf(vm, " near token '%s'", yytext)); 299 if (yypos < yylimit) { 300 yybuf[yylimit]= '\0'; 301 TrString_push(vm, msg, tr_sprintf(vm, " before text \"")); 302 while (yypos < yylimit) { 303 if ('\n' == yybuf[yypos] || '\r' == yybuf[yypos]) break; 304 char c[2] = { yybuf[yypos++], '\0' }; 305 TrString_push(vm, msg, tr_sprintf(vm, c)); 306 } 307 TrString_push(vm, msg, tr_sprintf(vm, "\"")); 308 } 309 /* TODO msg should not be a String object */ 310 tr_raise(SyntaxError, TR_STR_PTR(msg)); 311} 312 313/* Compiles code to a TrBlock. 314 Returns NULL on error, error is stored in TR_EXCEPTION. */ 315TrBlock *TrBlock_compile(VM, char *code, char *fn, size_t lineno) { 316 assert(!compiler && "parser not reentrant"); 317 charbuf = code; 318 compiler = TrCompiler_new(vm, fn); 319 compiler->line += lineno; 320 compiler->filename = TrString_new2(vm, fn); 321 TrBlock *b = NULL; 322 323 if (!yyparse()) { 324 yyerror(); 325 goto error; 326 } 327 328 TrCompiler_compile(compiler); 329 b = compiler->block; 330error: 331 charbuf = 0; 332 compiler = 0; 333 return b; 334}