--- parser3/src/main/compile.y 2004/04/06 14:08:41 1.210 +++ parser3/src/main/compile.y 2009/05/23 05:23:19 1.242 @@ -2,10 +2,10 @@ /** @file Parser: compiler(lexical parser and grammar). - Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) Author: Alexander Petrosyan (http://design.ru/paf) - $Id: compile.y,v 1.210 2004/04/06 14:08:41 paf Exp $ + $Id: compile.y,v 1.242 2009/05/23 05:23:19 misha Exp $ */ /** @@ -41,6 +41,10 @@ // defines #define USE_CONTROL_METHOD_NAME "USE" +#define OPTIONS_CONTROL_METHOD_NAME "OPTIONS" +#define OPTION_ALL_VARS_LOCAL_NAME "locals" +#define OPTION_PARTIAL_CLASS "partial" +#define REM_OPERATOR_NAME "rem" // forwards @@ -48,6 +52,10 @@ static int real_yyerror(Parse_control* p static void yyprint(FILE* file, int type, YYSTYPE value); static int yylex(YYSTYPE* lvalp, void* pc); +static const VBool vfalse(false); +static const VBool vtrue(true); +static const VVoid vvoid; + // local convinient inplace typecast & var #undef PC #define PC (*(Parse_control *)pc) @@ -93,6 +101,9 @@ static int yylex(YYSTYPE* lvalp, void* p %token DEXISTS "-d" %token IS "is" +%token LITERAL_TRUE "true" +%token LITERAL_FALSE "false" + /* logical */ %left "!||" %left "||" @@ -143,21 +154,13 @@ control_method: '@' STRING '\n' YYERROR; } if(command==CLASS_NAME) { - if(PC.cclass->base_class()) { // already changed from default? - strcpy(PC.error, "class already have a name '"); - strncat(PC.error, PC.cclass->name().cstr(), 100); - strcat(PC.error, "'"); - YYERROR; - } if(strings_code->count()==1*OPERATIONS_PER_OPVALUE) { // new class' name const String& name=*LA2S(*strings_code); // creating the class VStateless_class* cclass=new VClass; - PC.cclass=cclass; - PC.cclass->set_name(name); - // append to request's classes - PC.request.classes().put(name, cclass); + PC.cclass_new=cclass; + PC.cclass_new->set_name(name); } else { strcpy(PC.error, "@"CLASS_NAME" must contain only one line with class name (contains more then one)"); YYERROR; @@ -166,9 +169,16 @@ control_method: '@' STRING '\n' for(size_t i=0; icount(); i+=OPERATIONS_PER_OPVALUE) PC.request.use_file(PC.request.main_class, *LA2S(*strings_code, i)); } else if(command==BASE_NAME) { + if(PC.append){ + strcpy(PC.error, "can't set base while appending methods to class '"); + strncat(PC.error, PC.cclass->name().cstr(), MAX_STRING/2); + strcat(PC.error, "'"); + YYERROR; + } + PC.class_add(); if(PC.cclass->base_class()) { // already changed from default? strcpy(PC.error, "class already have a base '"); - strncat(PC.error, PC.cclass->base_class()->name().cstr(), 100); + strncat(PC.error, PC.cclass->base_class()->name().cstr(), MAX_STRING/2); strcat(PC.error, "'"); YYERROR; } @@ -196,11 +206,41 @@ control_method: '@' STRING '\n' strcpy(PC.error, "@"BASE_NAME" must contain sole name"); YYERROR; } + } else if(command==OPTIONS_CONTROL_METHOD_NAME) { + for(size_t i=0; icount(); i+=OPERATIONS_PER_OPVALUE) { + const String& option=*LA2S(*strings_code, i); + if(option==OPTION_ALL_VARS_LOCAL_NAME){ + PC.set_all_vars_local(); + } else if(option==OPTION_PARTIAL_CLASS){ + if(PC.cclass_new){ + if(VStateless_class* existed=PC.get_existed_class(PC.cclass_new)){ + if(!PC.reuse_existed_class(existed)){ + strcpy(PC.error, "can't append methods to '"); + strncat(PC.error, PC.cclass_new->name().cstr(), MAX_STRING/2); + strcat(PC.error, "' - the class wasn't marked as partial"); + YYERROR; + } + } else { + // mark new class as partial. we can add methods to it later. + PC.cclass_new->set_partial(); + } + } else { + strcpy(PC.error, "'"OPTION_PARTIAL_CLASS"' option should be used straight after @"CLASS_NAME); + YYERROR; + } + } else { + strcpy(PC.error, "'"); + strncat(PC.error, option.cstr(), MAX_STRING/2); + strcat(PC.error, "' invalid option. valid options are " + "'"OPTION_PARTIAL_CLASS"' and '"OPTION_ALL_VARS_LOCAL_NAME"'"); + YYERROR; + } + } } else { strcpy(PC.error, "'"); strncat(PC.error, command.cstr(), MAX_STRING/2); strcat(PC.error, "' invalid special name. valid names are " - "'"CLASS_NAME"', '"USE_CONTROL_METHOD_NAME"' and '"BASE_NAME"'"); + "'"CLASS_NAME"', '"USE_CONTROL_METHOD_NAME"', '"BASE_NAME"' and '"OPTIONS_CONTROL_METHOD_NAME"'."); YYERROR; } }; @@ -210,8 +250,8 @@ control_string: maybe_string '\n'; maybe_string: empty | STRING; code_method: '@' STRING bracketed_maybe_strings maybe_bracketed_strings maybe_comment '\n' { + PC.class_add(); PC.explicit_result=false; - const String& name=*LA2S(*$2); YYSTYPE params_names_code=$3; ArrayString* params_names=0; @@ -223,30 +263,40 @@ code_method: '@' STRING bracketed_maybe_ YYSTYPE locals_names_code=$4; ArrayString* locals_names=0; + bool all_vars_local=false; if(int size=locals_names_code->count()) { locals_names=new ArrayString; for(int i=0; iis_vars_local()) + all_vars_local=true; Method* method=new Method( //name, Method::CT_ANY, 0, 0/*min,max numbered_params_count*/, params_names, locals_names, - 0/*to be filled later in next {} */, 0); - PC.cclass->add_method(PC.alias_method(name), *method); + 0/*to be filled later in next {} */, 0, all_vars_local); + *reinterpret_cast(&$$)=method; // todo: check [][;result;] } maybe_codes { - // fill in the code - reinterpret_cast($7)->parser_code=$8; + Method& method=*reinterpret_cast($7); + // fill in the code + method.parser_code=$8; + + // register in class + const String& name=*LA2S(*$2); + PC.cclass->add_method(PC.alias_method(name), method); }; maybe_bracketed_strings: empty | bracketed_maybe_strings; @@ -267,11 +317,41 @@ action: get | put | call; /* get */ get: get_value { - $$=$1; /* stack: resulting value */ - changetail_or_append(*$$, - OP_GET_ELEMENT, false, /*->*/OP_GET_ELEMENT__WRITE, - /*or */OP_WRITE_VALUE - ); /* value=pop; wcontext.write(value) */ + $$=N(); + YYSTYPE code=$1; +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + if(!maybe_change_first_opcode(*code, OP::OP_VALUE__GET_ELEMENT, /*=>*/OP::OP_VALUE__GET_ELEMENT__WRITE)) +#endif + { + size_t count=code->count(); + size_t len=6; + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT + if( + count==len + && maybe_change_first_opcode(*code, OP::OP_GET_OBJECT_ELEMENT, OP::OP_GET_ELEMENT, /*=>*/OP::OP_GET_OBJECT_ELEMENT__WRITE) + ){ + //P(*$$, *code, 0/*offset*/, count-1/*limit*/); // someday skip last OP_GET_ELEMENT + //break; + } else +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT + if( + count==len + && maybe_change_first_opcode(*code, OP::OP_GET_OBJECT_VAR_ELEMENT, OP::OP_GET_ELEMENT, /*=>*/OP::OP_GET_OBJECT_VAR_ELEMENT__WRITE) + ){ + //P(*$$, *code, 0/*offset*/, count-1/*limit*/); // someday skip last OP_GET_ELEMENT + //break; + } else +#endif + + changetail_or_append(*code, + OP::OP_GET_ELEMENT, false, /*=>*/OP::OP_GET_ELEMENT__WRITE, + /*or */OP::OP_WRITE_VALUE + ); /* value=pop; wcontext.write(value) */ + } + P(*$$, *code); }; get_value: '$' get_name_value { $$=$2 }; get_name_value: name_without_curly_rdive EON | name_in_curly_rdive; @@ -281,24 +361,56 @@ name_without_curly_rdive: | name_without_curly_rdive_class; name_without_curly_rdive_read: name_without_curly_rdive_code { $$=N(); - ArrayOperation* diving_code=$1; + YYSTYPE diving_code=$1; const String* first_name=LA2S(*diving_code); - // self.xxx... -> xxx... + size_t count=diving_code->count(); + // self.xxx... => xxx... // OP_VALUE+origin+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { - O(*yyval, OP_WITH_SELF); /* stack: starting context */ - P(*yyval, *diving_code, + O(*$$, OP::OP_WITH_SELF); /* stack: starting context */ + P(*$$, *diving_code, /* skip over... */ - diving_code->count()>=4?4/*OP_VALUE+origin+string+OP_GET_ELEMENTx*/:3/*OP_+origin+string*/); - } else { - O(*yyval, OP_WITH_READ); /* stack: starting context */ + count>=4?4/*OP_VALUE+origin+string+OP_GET_ELEMENTx*/:3/*OP::OP_+origin+string*/); + } + + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT + else if(maybe_make_get_object_element(*$$, *diving_code, count)){ + // optimisation for $object.field + ^object.method[ + } +#endif - // ^if ELEMENT -> ^if ELEMENT_OR_OPERATOR - // OP_VALUE+origin+string+OP_GET_ELEMENT. -> OP_VALUE+origin+string+OP_GET_ELEMENT_OR_OPERATOR. - if(PC.in_call_value && diving_code->count()==4) - diving_code->put(4-1, OP_GET_ELEMENT_OR_OPERATOR); + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT + else if(maybe_make_get_object_var_element(*$$, *diving_code, count)){ + // optimisation for $object.$var + } +#endif + + +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + else if(count==4){ // optimization + O(*$$, + (PC.in_call_value) + ? OP::OP_VALUE__GET_ELEMENT_OR_OPERATOR // ^object[ : OP_VALUE+origin+string+OP_GET_ELEMENT => OP_VALUE__GET_ELEMENT_OR_OPERATOR+origin+string + : OP::OP_VALUE__GET_ELEMENT // $object : OP_VALUE+origin+string+OP_GET_ELEMENT => OP_VALUE__GET_ELEMENT+origin+string + ); + P(*$$, *diving_code, 1/*offset*/, 2/*limit*/); // copy origin+value + } else { + O(*$$, OP::OP_WITH_READ); /* stack: starting context */ P(*$$, *diving_code); } +#else + else { + O(*$$, OP::OP_WITH_READ); /* stack: starting context */ + + // ^if OP_ELEMENT => ^if OP_ELEMENT_OR_OPERATOR + // optimized OP_VALUE+origin+string+OP_GET_ELEMENT. => OP_VALUE+origin+string+OP_GET_ELEMENT_OR_OPERATOR. + if(PC.in_call_value && count==4) + diving_code->put(count-1, OP::OP_GET_ELEMENT_OR_OPERATOR); + P(*$$, *diving_code); + } +#endif /* diving code; stack: current context */ }; name_without_curly_rdive_class: class_prefix name_without_curly_rdive_code { $$=$1; P(*$$, *$2) }; @@ -307,8 +419,16 @@ name_without_curly_rdive_code: name_adva /* put */ put: '$' name_expr_wdive construct { - $$=$2; /* stack: context,name */ - P(*$$, *$3); /* stack: context,name,constructor_value */ + $$=N(); +#ifdef OPTIMIZE_BYTECODE_CONSTRUCT + if(maybe_make_root_or_write_construct(*$$, *$2, *$3)){ + // $a(1), $.a(1), $a[b], $.a[b], $a($b), $.a($b), $a[$b], $.a[$b] + } else +#endif + { + P(*$$, *$2); /* stack: context,name */ + P(*$$, *$3); /* stack: context,name,constructor_value */ + } }; name_expr_wdive: name_expr_wdive_root @@ -316,24 +436,24 @@ name_expr_wdive: | name_expr_wdive_class; name_expr_wdive_root: name_expr_dive_code { $$=N(); - ArrayOperation* diving_code=$1; + YYSTYPE diving_code=$1; const String* first_name=LA2S(*diving_code); - // $self.xxx... -> $xxx... - // OP_VALUE+origin+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... + // $self.xxx... => $xxx... + // OP_VALUE+origin+string+OP_GET_ELEMENT+... => OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { - O(*$$, OP_WITH_SELF); /* stack: starting context */ + O(*$$, OP::OP_WITH_SELF); /* stack: starting context */ P(*$$, *diving_code, /* skip over... */ - diving_code->count()>=4?4/*OP_VALUE+origin+string+OP_GET_ELEMENTx*/:3/*OP_+origin+string*/); + diving_code->count()>=4?4/*OP::OP_VALUE+origin+string+OP::OP_GET_ELEMENTx*/:3/*OP::OP_+origin+string*/); } else { - O(*$$, OP_WITH_ROOT); /* stack: starting context */ + O(*$$, OP::OP_WITH_ROOT); /* stack: starting context */ P(*$$, *diving_code); } /* diving code; stack: current context */ }; name_expr_wdive_write: '.' name_expr_dive_code { $$=N(); - O(*$$, OP_WITH_WRITE); /* stack: starting context */ + O(*$$, OP::OP_WITH_WRITE); /* stack: starting context */ P(*$$, *$2); /* diving code; stack: context,name */ }; name_expr_wdive_class: class_prefix name_expr_dive_code { $$=$1; P(*$$, *$2) }; @@ -343,24 +463,29 @@ construct: | construct_round | construct_curly ; -construct_square: '[' any_constructor_code_value ']' { +construct_square: '[' { + // allow $result_or_other_variable[ letters here any time ] + *reinterpret_cast(&$$)=PC.explicit_result; PC.explicit_result=false; +} any_constructor_code_value { + PC.explicit_result=*reinterpret_cast(&$2); +} ']' { // stack: context, name - $$=$2; // stack: context, name, value - O(*$$, OP_CONSTRUCT_VALUE); /* value=pop; name=pop; context=pop; construct(context,name,value) */ + $$=$3; // stack: context, name, value + O(*$$, OP::OP_CONSTRUCT_VALUE); /* value=pop; name=pop; context=pop; construct(context,name,value) */ } ; construct_round: '(' expr_value ')' { $$=N(); - O(*$$, OP_PREPARE_TO_EXPRESSION); + O(*$$, OP::OP_PREPARE_TO_EXPRESSION); // stack: context, name P(*$$, *$2); // stack: context, name, value - O(*$$, OP_CONSTRUCT_EXPR); /* value=pop->as_expr_result; name=pop; context=pop; construct(context,name,value) */ + O(*$$, OP::OP_CONSTRUCT_EXPR); /* value=pop->as_expr_result; name=pop; context=pop; construct(context,name,value) */ } ; construct_curly: '{' maybe_codes '}' { // stack: context, name $$=N(); - OA(*$$, OP_CURLY_CODE__CONSTRUCT, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ + OA(*$$, OP::OP_CURLY_CODE__CONSTRUCT, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ }; any_constructor_code_value: @@ -370,7 +495,7 @@ any_constructor_code_value: ; constructor_code_value: constructor_code { $$=N(); - OA(*$$, OP_OBJECT_POOL, $1); /* stack: empty write context */ + OA(*$$, OP::OP_OBJECT_POOL, $1); /* stack: empty write context */ /* some code that writes to that context */ /* context=pop; stack: context.value() */ }; @@ -380,10 +505,15 @@ codes__excluding_sole_str_literal: actio /* call */ call: call_value { - $$=$1; /* stack: value */ - changetail_or_append(*$$, - OP_CALL, true, /*->*/ OP_CALL__WRITE, - /*or */OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ +#ifdef OPTIMIZE_BYTECODE_CUT_REM_OPERATOR + if((*$1).count()) +#endif + { + $$=$1; /* stack: value */ + changetail_or_append(*$$, + OP::OP_CALL, true, /*=>*/ OP::OP_CALL__WRITE, + /*or */OP::OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ + } }; call_value: '^' { PC.in_call_value=true; @@ -392,16 +522,28 @@ call_value: '^' { PC.in_call_value=false; } store_params EON { /* ^field.$method{vasya} */ - $$=$3; /* with_xxx,diving code; stack: context,method_junction */ +#ifdef OPTIMIZE_BYTECODE_CUT_REM_OPERATOR +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + const String* operator_name=LA2S(*$3, 0, OP::OP_VALUE__GET_ELEMENT_OR_OPERATOR); +#else + const String* operator_name=LA2S(*$3, 1); +#endif + if(operator_name && *operator_name == REM_OPERATOR_NAME){ + $$=N(); + } else +#endif + { + $$=$3; /* with_xxx,diving code; stack: context,method_junction */ - YYSTYPE params_code=$5; - if(params_code->count()==4) { // probably [] case. [OP_VALUE+origin+Void+STORE_PARAM] - if(Value* value=LA2V(*params_code)) // it is OP_VALUE+origin+value? - if(!value->is_defined()) // value is VVoid? - params_code=0; // ^zzz[] case. don't append lone empty param. - } - /* stack: context, method_junction */ - OA(*$$, OP_CALL, params_code); // method_frame=make frame(pop junction); ncontext=pop; call(ncontext,method_frame) stack: value + YYSTYPE params_code=$5; + if(params_code->count()==3) { // probably [] case. [OP::OP_VALUE+origin+Void] + if(Value* value=LA2V(*params_code)) // it is OP_VALUE+origin+value? + if(value->is_void()) // value is VVoid? + params_code=0; // ^zzz[] case. don't append lone empty param. + } + /* stack: context, method_junction */ + OA(*$$, OP::OP_CALL, params_code); // method_frame=make frame(pop junction); ncontext=pop; call(ncontext,method_frame) stack: value + } }; call_name: name_without_curly_rdive; @@ -413,10 +555,10 @@ store_param: | store_curly_param ; store_square_param: '[' { - // allow ^call[ whitespace here any time ] + // allow ^call[ letters here any time ] *reinterpret_cast(&$$)=PC.explicit_result; PC.explicit_result=false; } store_code_param_parts { - PC.explicit_result=reinterpret_cast($2); + PC.explicit_result=*reinterpret_cast(&$2); } ']' {$$=$3}; store_round_param: '(' store_expr_param_parts ')' {$$=$2}; store_curly_param: '{' store_curly_param_parts '}' {$$=$2}; @@ -434,27 +576,30 @@ store_curly_param_parts: ; store_code_param_part: code_param_value { $$=$1; - O(*$$, OP_STORE_PARAM); }; -store_expr_param_part: write_expr_value { - $$=N(); - OA(*$$, OP_EXPR_CODE__STORE_PARAM, $1); +store_expr_param_part: expr_value { + YYSTYPE expr_code=$1; + if(expr_code->count()==3 + && (*expr_code)[0].code==OP::OP_VALUE) { // optimizing (double/bool/incidently 'string' too) case. [OP::OP_VALUE+origin+Double]. no evaluating + $$=expr_code; + } else { + YYSTYPE code=N(); + O(*code, OP::OP_PREPARE_TO_EXPRESSION); + P(*code, *expr_code); + O(*code, OP::OP_WRITE_EXPR_RESULT); + $$=N(); + OA(*$$, OP::OP_EXPR_CODE__STORE_PARAM, code); + } }; store_curly_param_part: maybe_codes { $$=N(); - OA(*$$, OP_CURLY_CODE__STORE_PARAM, $1); + OA(*$$, OP::OP_CURLY_CODE__STORE_PARAM, $1); }; code_param_value: void_value /* optimized [;...] case */ | STRING /* optimized [STRING] case */ | constructor_code_value /* [something complex] */ ; -write_expr_value: expr_value { - $$=N(); - O(*$$, OP_PREPARE_TO_EXPRESSION); - P(*$$, *$1); - O(*$$, OP_WRITE_EXPR_RESULT); -}; /* name */ @@ -468,12 +613,12 @@ name_advance1: name_expr_value { /* stack: context */ $$=$1; /* stack: context,name */ - O(*$$, OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ + O(*$$, OP::OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ }; name_advance2: name_expr_value { /* stack: context */ $$=$1; /* stack: context,name */ - O(*$$, OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ + O(*$$, OP::OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ } | STRING BOGUS ; @@ -485,32 +630,37 @@ name_expr_value: ; name_expr_subvar_value: '$' subvar_ref_name_rdive { $$=$2; - O(*$$, OP_GET_ELEMENT); + O(*$$, OP::OP_GET_ELEMENT); }; name_expr_with_subvar_value: STRING subvar_get_writes { - ArrayOperation* code; + YYSTYPE code; { change_string_literal_to_write_string_literal(*(code=$1)); P(*code, *$2); } $$=N(); - OA(*$$, OP_STRING_POOL, code); + OA(*$$, OP::OP_STRING_POOL, code); }; -name_square_code_value: '[' codes ']' { +name_square_code_value: '[' { + // allow $result_or_other_variable[ letters here any time ] + *reinterpret_cast(&$$)=PC.explicit_result; PC.explicit_result=false; +} codes { + PC.explicit_result=*reinterpret_cast(&$2); +} ']' { $$=N(); - OA(*$$, OP_OBJECT_POOL, $2); /* stack: empty write context */ + OA(*$$, OP::OP_OBJECT_POOL, $3); /* stack: empty write context */ /* some code that writes to that context */ /* context=pop; stack: context.value() */ }; subvar_ref_name_rdive: STRING { $$=N(); - O(*$$, OP_WITH_READ); + O(*$$, OP::OP_WITH_READ); P(*$$, *$1); }; subvar_get_writes: subvar__get_write | subvar_get_writes subvar__get_write { $$=$1; P(*$$, *$2) }; subvar__get_write: '$' subvar_ref_name_rdive { $$=$2; - O(*$$, OP_GET_ELEMENT__WRITE); + O(*$$, OP::OP_GET_ELEMENT__WRITE); }; class_prefix: @@ -527,7 +677,12 @@ class_static_prefix: STRING ':' { YYERROR; } } - O(*$$, OP_GET_CLASS); +#ifdef OPTIMIZE_BYTECODE_GET_CLASS + // optimized OP_VALUE+origin+string+OP_GET_CLASS => OP_VALUE__GET_CLASS+origin+string + maybe_change_first_opcode(*$$, OP::OP_VALUE, OP::OP_VALUE__GET_CLASS, true/*assert if top opcode != OP_VALUE*/) +#else + O(*$$, OP::OP_GET_CLASS); +#endif }; class_constructor_prefix: class_static_prefix ':' { $$=$1; @@ -535,66 +690,81 @@ class_constructor_prefix: class_static_p strcpy(PC.error, ":: not allowed here"); YYERROR; } - O(*$$, OP_PREPARE_TO_CONSTRUCT_OBJECT); + O(*$$, OP::OP_PREPARE_TO_CONSTRUCT_OBJECT); }; /* expr */ -expr_value: expr { - // see OP_PREPARE_TO_EXPRESSION!! - if(($$=$1)->count()==2) // only one string literal in there? - change_string_literal_to_double_literal(*$$); // make that string literal Double -}; +expr_value: expr; expr: - STRING + double_or_STRING +| true_value +| false_value | get_value | call_value -| '"' string_inside_quotes_value '"' { $$ = $2; } -| '\'' string_inside_quotes_value '\'' { $$ = $2; } +| '"' string_inside_quotes_value '"' { $$ = $2 } +| '\'' string_inside_quotes_value '\'' { $$ = $2 } | '(' expr ')' { $$ = $2; } /* stack: operand // stack: @operand */ -| '-' expr %prec NUNARY { $$=$2; O(*$$, OP_NEG) } +| '-' expr %prec NUNARY { $$=$2; O(*$$, OP::OP_NEG) } | '+' expr %prec NUNARY { $$=$2 } -| '~' expr { $$=$2; O(*$$, OP_INV) } -| '!' expr { $$=$2; O(*$$, OP_NOT) } -| "def" expr { $$=$2; O(*$$, OP_DEF) } -| "in" expr { $$=$2; O(*$$, OP_IN) } -| "-f" expr { $$=$2; O(*$$, OP_FEXISTS) } -| "-d" expr { $$=$2; O(*$$, OP_DEXISTS) } +| '~' expr { $$=$2; O(*$$, OP::OP_INV) } +| '!' expr { $$=$2; O(*$$, OP::OP_NOT) } +| "def" expr { $$=$2; O(*$$, OP::OP_DEF) } +| "in" expr { $$=$2; O(*$$, OP::OP_IN) } +| "-f" expr { $$=$2; O(*$$, OP::OP_FEXISTS) } +| "-d" expr { $$=$2; O(*$$, OP::OP_DEXISTS) } /* stack: a,b // stack: a@b */ -| expr '-' expr { $$=$1; P(*$$, *$3); O(*$$, OP_SUB) } -| expr '+' expr { $$=$1; P(*$$, *$3); O(*$$, OP_ADD) } -| expr '*' expr { $$=$1; P(*$$, *$3); O(*$$, OP_MUL) } -| expr '/' expr { $$=$1; P(*$$, *$3); O(*$$, OP_DIV) } -| expr '%' expr { $$=$1; P(*$$, *$3); O(*$$, OP_MOD) } -| expr '\\' expr { $$=$1; P(*$$, *$3); O(*$$, OP_INTDIV) } -| expr "<<" expr { $$=$1; P(*$$, *$3); O(*$$, OP_BIN_SL) } -| expr ">>" expr { $$=$1; P(*$$, *$3); O(*$$, OP_BIN_SR) } -| expr '&' expr { $$=$1; P(*$$, *$3); O(*$$, OP_BIN_AND) } -| expr '|' expr { $$=$1; P(*$$, *$3); O(*$$, OP_BIN_OR) } -| expr "!|" expr { $$=$1; P(*$$, *$3); O(*$$, OP_BIN_XOR) } -| expr "&&" expr { $$=$1; OA(*$$, OP_NESTED_CODE, $3); O(*$$, OP_LOG_AND) } -| expr "||" expr { $$=$1; OA(*$$, OP_NESTED_CODE, $3); O(*$$, OP_LOG_OR) } -| expr "!||" expr { $$=$1; P(*$$, *$3); O(*$$, OP_LOG_XOR) } -| expr '<' expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_LT) } -| expr '>' expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_GT) } -| expr "<=" expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_LE) } -| expr ">=" expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_GE) } -| expr "==" expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_EQ) } -| expr "!=" expr { $$=$1; P(*$$, *$3); O(*$$, OP_NUM_NE) } -| expr "lt" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_LT) } -| expr "gt" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_GT) } -| expr "le" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_LE) } -| expr "ge" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_GE) } -| expr "eq" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_EQ) } -| expr "ne" expr { $$=$1; P(*$$, *$3); O(*$$, OP_STR_NE) } -| expr "is" expr { $$=$1; P(*$$, *$3); O(*$$, OP_IS) } +| expr '-' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_SUB) } +| expr '+' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_ADD) } +| expr '*' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_MUL) } +| expr '/' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_DIV) } +| expr '%' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_MOD) } +| expr '\\' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_INTDIV) } +| expr "<<" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_BIN_SL) } +| expr ">>" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_BIN_SR) } +| expr '&' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_BIN_AND) } +| expr '|' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_BIN_OR) } +| expr "!|" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_BIN_XOR) } +| expr "&&" expr { $$=$1; OA(*$$, OP::OP_NESTED_CODE, $3); O(*$$, OP::OP_LOG_AND) } +| expr "||" expr { $$=$1; OA(*$$, OP::OP_NESTED_CODE, $3); O(*$$, OP::OP_LOG_OR) } +| expr "!||" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_LOG_XOR) } +| expr '<' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_LT) } +| expr '>' expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_GT) } +| expr "<=" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_LE) } +| expr ">=" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_GE) } +| expr "==" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_EQ) } +| expr "!=" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_NUM_NE) } +| expr "lt" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_LT) } +| expr "gt" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_GT) } +| expr "le" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_LE) } +| expr "ge" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_GE) } +| expr "eq" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_EQ) } +| expr "ne" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_STR_NE) } +| expr "is" expr { $$=$1; P(*$$, *$3); O(*$$, OP::OP_IS) } ; +double_or_STRING: STRING { + // optimized OP_STRING => OP_VALUE for doubles + maybe_change_string_literal_to_double_literal(*($$=$1)); +}; + string_inside_quotes_value: maybe_codes { +#ifdef OPTIMIZE_BYTECODE_STRING_POOL + // it brakes ^if(" 09 "){...} + YYSTYPE code=$1; $$=N(); - OA(*$$, OP_STRING_POOL, $1); /* stack: empty write context */ + if(code->count()==3 && maybe_change_first_opcode(*code, OP::OP_STRING__WRITE, OP::OP_VALUE)){ + // optimized OP_STRING__WRITE+origin+value => OP_VALUE+origin+value without starting OP_STRING_POOL + P(*$$, *code); + } else { + OA(*$$, OP::OP_STRING_POOL, code); /* stack: empty write context */ + } +#else + $$=N(); + OA(*$$, OP::OP_STRING_POOL, $1); /* stack: empty write context */ +#endif /* some code that writes to that context */ /* context=pop; stack: context.get_string() */ }; @@ -602,11 +772,14 @@ string_inside_quotes_value: maybe_codes /* basics */ write_string: STRING { - // optimized from OP_STRING+OP_WRITE_VALUE to OP_STRING__WRITE + // optimized OP_STRING+OP_WRITE_VALUE => OP_STRING__WRITE change_string_literal_to_write_string_literal(*($$=$1)) }; -void_value: /* empty */ { $$=VL(new VVoid(), 0, 0, 0) }; +void_value: /* empty */ { $$=VL(/*we know that we will not change it*/const_cast(&vvoid), 0, 0, 0) } +true_value: "true" { $$ = VL(/*we know that we will not change it*/const_cast(&vtrue), 0, 0, 0) } +false_value: "false" { $$ = VL(/*we know that we will not change it*/const_cast(&vfalse), 0, 0, 0) } + empty: /* empty */ { $$=N() }; %% @@ -701,7 +874,7 @@ case LS_DEF_COMMENT: default: switch(*pc.source) { // ^escaping some punctuators - case '^': case '$': case ';': + case '^': case '$': case ';': case '@': case '(': case ')': case '[': case ']': case '{': case '}': @@ -729,10 +902,10 @@ default: pc.string.append_strdup_know_length(begin, end-begin); } // #HH ? - if(pc.source[0]=='#' && pc.source[1] && pc.source[2]) { - char c= + if(pc.source[1] && isxdigit(pc.source[1]) && pc.source[2] && isxdigit(pc.source[2])) { + char c=(char)( hex_value[(unsigned char)pc.source[1]]*0x10+ - hex_value[(unsigned char)pc.source[2]]; + hex_value[(unsigned char)pc.source[2]]); if(c==0) { result=BAD_HEX_LITERAL; goto break2; // wrong hex value[no ^#00 chars allowed]: bail out @@ -748,7 +921,14 @@ default: // skip analysis = forced literal continue; } - break; + // just escaped char + // reset piece 'begin' position & line + begin=pc.source; + begin_pos=pc.pos; + // skip over _ after ^ + pc.source++; pc.pos.col++; + // skip analysis = forced literal + continue; } break; } @@ -1045,8 +1225,28 @@ default: case 'd': if(end==begin) // right after whitespace if(pc.source[0]=='e' && pc.source[1]=='f') { // def - skip_analized=2; - result=DEF; + switch(pc.source[2]){ + case ' ': case '\t': case '\n': case '"': case '\'': case '^': case '$': // non-quoted string without whitespace after 'def' is not allowed + skip_analized=2; + result=DEF; + goto break2; + } + // error: incorrect char after 'def' + } + break; + case 't': + if(end==begin) // right after whitespace + if(pc.source[0]=='r' && pc.source[1]=='u' && pc.source[2]=='e') { // true + skip_analized=3; + result=LITERAL_TRUE; + goto break2; + } + break; + case 'f': + if(end==begin) // right after whitespace + if(pc.source[0]=='a' && pc.source[1]=='l' && pc.source[2]=='s' && pc.source[3]=='e') { // false + skip_analized=4; + result=LITERAL_FALSE; goto break2; } break; @@ -1126,6 +1326,8 @@ default: case '=': case '!': // common delimiters case ',': case '?': case '#': + // mysql column separators + case '`': // before call case '^': pop_LS(pc);