--- parser3/src/main/compile.y 2003/02/04 09:44:27 1.204.2.6 +++ parser3/src/main/compile.y 2012/05/30 02:25:47 1.263 @@ -2,12 +2,14 @@ /** @file Parser: compiler(lexical parser and grammar). - Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexander Petrosyan (http://design.ru/paf) - $Id: compile.y,v 1.204.2.6 2003/02/04 09:44:27 paf Exp $ + */ +volatile const char * IDENT_COMPILE_Y = "$Id: compile.y,v 1.263 2012/05/30 02:25:47 misha Exp $"; + /** @todo parser4: - cache compiled code from request to request. to do that... @@ -19,35 +21,47 @@ -#: in iis make up specialized Pool object for that */ -#define YYSTYPE ArrayOperationPtr +#define YYSTYPE ArrayOperation* #define YYPARSE_PARAM pc #define YYLEX_PARAM pc #define YYDEBUG 1 #define YYERROR_VERBOSE 1 -#define yyerror(msg) real_yyerror((parse_control *)pc, msg) +#define yyerror(msg) real_yyerror((Parse_control *)pc, msg) #define YYPRINT(file, type, value) yyprint(file, type, value) +// includes + #include "compile_tools.h" #include "pa_value.h" #include "pa_request.h" #include "pa_vobject.h" #include "pa_vdouble.h" #include "pa_globals.h" -#include "pa_vvoid.h" #include "pa_vmethod_frame.h" -#define USE_CONTROL_METHOD_NAME "USE" - -static int real_yyerror(parse_control *pc, char *s); -static void yyprint(FILE *file, int type, YYSTYPE value); -static int yylex(YYSTYPE *lvalp, void *pc); +// 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 + +static int real_yyerror(Parse_control* pc, char* s); +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 VString vempty; // local convinient inplace typecast & var -#define PC (*(parse_control *)pc) +#undef PC +#define PC (*(Parse_control *)pc) +#undef POOL #define POOL (*PC.pool) -#undef NEW -#define NEW new(POOL) #ifndef DOXYGEN %} @@ -61,7 +75,7 @@ static int yylex(YYSTYPE *lvalp, void *p %token BAD_HEX_LITERAL %token BAD_METHOD_DECL_START %token BAD_METHOD_PARAMETER_NAME_CHARACTER -%token BAD_MATH_OPERATOR_CHARACTER +%token BAD_NONWHITESPACE_CHARACTER_IN_EXPLICIT_RESULT_MODE %token LAND "&&" %token LOR "||" @@ -88,6 +102,9 @@ static int yylex(YYSTYPE *lvalp, void *p %token DEXISTS "-d" %token IS "is" +%token LITERAL_TRUE "true" +%token LITERAL_FALSE "false" + /* logical */ %left "!||" %left "||" @@ -114,11 +131,11 @@ static int yylex(YYSTYPE *lvalp, void *p %% all: one_big_piece { - MethodPtr method(new Method(main_method_name, Method::CT_ANY, + Method* method=new Method(Method::CT_ANY, 0, 0, /*min, max numbered_params_count*/ - ArrayStringPtr(0)/*param_names*/, ArrayStringPtr(0)/*local_names*/, - $1/*parser_code*/, 0/*native_code*/)); - PC.cclass->add_method(main_method_name, method); + 0/*param_names*/, 0/*local_names*/, + $1/*parser_code*/, 0/*native_code*/); + PC.cclass->set_method(PC.alias_method(main_method_name), method); } | methods; @@ -129,47 +146,48 @@ method: control_method | code_method; control_method: '@' STRING '\n' maybe_control_strings { - StringPtr command=LA2S(*$2); + const String& command=LA2S(*$2)->trim(String::TRIM_END); YYSTYPE strings_code=$4; - if(strings_code->count()<1*2) { + if(strings_code->count()<1*OPERATIONS_PER_OPVALUE) { strcpy(PC.error, "@"); - strcat(PC.error, command->cstr()); + strcat(PC.error, command.cstr()); strcat(PC.error, " is empty"); 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*2) { + if(command==CLASS_NAME) { + if(strings_code->count()==1*OPERATIONS_PER_OPVALUE) { + PC.class_add(); // new class' name - StringPtr name=LA2S(*strings_code); + const String& name=LA2S(*strings_code)->trim(String::TRIM_END); // creating the class - VStateless_classPtr cclass(new VClass); - PC.cclass=cclass.get(); - PC.cclass->set_name(name); - // append to request's classes - PC.request->classes().put(name, cclass); + VStateless_class* cclass=new VClass; + PC.cclass_new=cclass; + PC.cclass_new->set_name(name); } else { - strcpy(PC.error, "@"CLASS_NAME" must contain sole name"); + strcpy(PC.error, "@"CLASS_NAME" must contain only one line with class name (contains more then one)"); + YYERROR; + } + } else if(command==USE_CONTROL_METHOD_NAME) { + PC.class_add(); + for(size_t i=0; icount(); i+=OPERATIONS_PER_OPVALUE) + PC.request.use_file(PC.request.main_class, LA2S(*strings_code, i)->trim(String::TRIM_END), PC.request.get_used_filename(PC.file_no)); + } 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; } - } else if(*command==USE_CONTROL_METHOD_NAME) { - for(int i=0; icount(); i+=2) - PC.request->use_file(*PC.request->main_class, LA2S(*strings_code, i)); - } else if(*command==BASE_NAME) { + 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; } - if(strings_code->count()==1*2) { - StringPtr base_name=LA2S(*strings_code); - if(ValuePtr base_class_value=PC.request->classes().get(base_name)) { + if(strings_code->count()==1*OPERATIONS_PER_OPVALUE) { + const String& base_name=LA2S(*strings_code)->trim(String::TRIM_END); + if(Value* base_class_value=PC.request.get_class(base_name)) { // @CLASS == @BASE sanity check if(VStateless_class *base_class=base_class_value->get_class()) { if(PC.cclass==base_class) { @@ -178,12 +196,12 @@ control_method: '@' STRING '\n' } PC.cclass->get_class()->set_base(base_class); } else { // they asked to derive from a class without methods ['env' & co] - strcpy(PC.error, base_name->cstr()); + strcpy(PC.error, base_name.cstr()); strcat(PC.error, ": you can not derive from this class in @"BASE_NAME); YYERROR; } } else { - strcpy(PC.error, base_name->cstr()); + strcpy(PC.error, base_name.cstr()); strcat(PC.error, ": undefined class in @"BASE_NAME); YYERROR; } @@ -191,52 +209,107 @@ 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)->trim(String::TRIM_END); + 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 { + // marks the new class as partial. we will be able to add methods here later. + PC.cclass_new->set_partial(); + } + } else { + strcpy(PC.error, "'"OPTION_PARTIAL_CLASS"' option should be used straight after @"CLASS_NAME); + YYERROR; + } + } else if(option==method_call_type_static){ + PC.set_methods_call_type(Method::CT_STATIC); + } else if(option==method_call_type_dynamic){ + PC.set_methods_call_type(Method::CT_DYNAMIC); + } else { + strcpy(PC.error, "'"); + strncat(PC.error, option.cstr(), MAX_STRING/2); + strcat(PC.error, "' invalid option. valid options are " + "'"OPTION_PARTIAL_CLASS"', '"OPTION_ALL_VARS_LOCAL_NAME"'" + ", '"METHOD_CALL_TYPE_STATIC"' and '"METHOD_CALL_TYPE_DYNAMIC"'" + ); + YYERROR; + } + } } else { strcpy(PC.error, "'"); - strncat(PC.error, command->cstr(), MAX_STRING/2); + 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; } }; maybe_control_strings: empty | control_strings; -control_strings: control_string | control_strings control_string { $$=$1; P(*$$, *$2) }; +control_strings: control_string | control_strings control_string { $$=$1; P(*$$, *$2); }; control_string: maybe_string '\n'; maybe_string: empty | STRING; -code_method: '@' STRING bracketed_maybe_strings maybe_bracketed_strings maybe_comment '\n' - maybe_codes { - StringPtr name=LA2S(*$2); +code_method: '@' STRING bracketed_maybe_strings maybe_bracketed_strings maybe_comment '\n' { + PC.class_add(); + PC.explicit_result=false; YYSTYPE params_names_code=$3; - ArrayStringPtr params_names; + ArrayString* params_names=0; if(int size=params_names_code->count()) { - params_names=ArrayStringPtr(new ArrayString); - for(int i=0; icount()) { - locals_names=ArrayStringPtr(new ArrayString); - for(int i=0; iis_vars_local()) + all_vars_local=true; - MethodPtr method(new Method( - name, - Method::CT_ANY, + Method* method=new Method( + //name, + GetMethodCallType(PC, *$2), 0, 0/*min,max numbered_params_count*/, params_names, locals_names, - $7, 0)); - PC.cclass->add_method(name, method); + 0/*to be filled later in next {} */, 0, all_vars_local); + + *reinterpret_cast(&$$)=method; +} maybe_codes { + Method* method=reinterpret_cast($7); + // fill in the code + method->parser_code=$8; + + // register in class + const String& name=*LA2S(*$2); + PC.cclass->set_method(PC.alias_method(name), method); }; maybe_bracketed_strings: empty | bracketed_maybe_strings; -bracketed_maybe_strings: '[' maybe_strings ']' {$$=$2}; +bracketed_maybe_strings: '[' maybe_strings ']' {$$=$2;}; maybe_strings: empty | strings; -strings: STRING | strings ';' STRING { $$=$1; P(*$$, *$3) }; +strings: STRING | strings ';' STRING { $$=$1; P(*$$, *$3); }; maybe_comment: empty | STRING; @@ -244,55 +317,128 @@ maybe_comment: empty | STRING; maybe_codes: empty | codes; -codes: code | codes code { $$=$1; P(*$$, *$2) }; +codes: code | codes code { $$=$1; P(*$$, *$2); }; code: write_string | action; action: get | put | call; /* get */ get: get_value { - $$=$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; + size_t count=code->count(); + +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + if( + count!=3 + || !maybe_change_first_opcode(*code, OP::OP_VALUE__GET_ELEMENT, /*=>*/OP::OP_VALUE__GET_ELEMENT__WRITE) + ) +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_SELF_ELEMENT + if( + count!=3 + || !maybe_change_first_opcode(*code, OP::OP_WITH_SELF__VALUE__GET_ELEMENT, /*=>*/OP::OP_WITH_SELF__VALUE__GET_ELEMENT__WRITE) + ) +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT + if( + count!=5 + || !maybe_change_first_opcode(*code, OP::OP_GET_OBJECT_ELEMENT, /*=>*/OP::OP_GET_OBJECT_ELEMENT__WRITE) + ) +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT + if( + count!=5 + || !maybe_change_first_opcode(*code, OP::OP_GET_OBJECT_VAR_ELEMENT, /*=>*/OP::OP_GET_OBJECT_VAR_ELEMENT__WRITE) + ) +#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(*$$, *code); }; -get_value: '$' get_name_value { $$=$2 }; +get_value: '$' get_name_value { $$=$2; }; get_name_value: name_without_curly_rdive EON | name_in_curly_rdive; -name_in_curly_rdive: '{' name_without_curly_rdive '}' { $$=$2 }; +name_in_curly_rdive: '{' name_without_curly_rdive '}' { $$=$2; }; name_without_curly_rdive: name_without_curly_rdive_read | name_without_curly_rdive_class; name_without_curly_rdive_read: name_without_curly_rdive_code { $$=N(); - ArrayOperationPtr diving_code=$1; - StringPtr first_name=LA2S(*diving_code); - // self.xxx... -> xxx... - // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... - if(first_name && *first_name==SELF_ELEMENT_NAME) { - O(*$$, OP_WITH_SELF); /* stack: starting context */ - P(*$$, *diving_code, - /* skip over... */ - diving_code->count()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); + YYSTYPE diving_code=$1; + size_t count=diving_code->count(); + + if(maybe_make_self(*$$, *diving_code, count)) { + // $self. + } else + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_ELEMENT + if(maybe_make_get_object_element(*$$, *diving_code, count)){ + // optimization for $object.field + ^object.method[ + } else +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_OBJECT_VAR_ELEMENT + if(maybe_make_get_object_var_element(*$$, *diving_code, count)){ + // optimization for $object.$var + } else +#endif + +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + if( + count>=4 + && (*diving_code)[0].code==OP::OP_VALUE + && (*diving_code)[3].code==OP::OP_GET_ELEMENT + ){ + // optimization + O(*$$, + (PC.in_call_value && count==4) + ? 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(*$$, *diving_code, 1/*offset*/, 2/*limit*/); // copy origin+value + if(count>4) + P(*$$, *diving_code, 4); // copy tail } else { - O(*$$, OP_WITH_READ); /* stack: starting context */ + O(*$$, OP::OP_WITH_READ); /* stack: starting context */ + P(*$$, *diving_code); + } +#else + { + O(*$$, OP::OP_WITH_READ); /* stack: starting context */ - // ^if ELEMENT -> ^if ELEMENT_OR_OPERATOR - // OP_VALUE+string+OP_GET_ELEMENT. -> OP_VALUE+string+OP_GET_ELEMENT_OR_OPERATOR. - if(PC.in_call_value && diving_code->count()==3) - (*diving_code)[2].code=OP_GET_ELEMENT_OR_OPERATOR; + // ^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(*$$, *$2) }; -name_without_curly_rdive_code: name_advance2 | name_path name_advance2 { $$=$1; P(*$$, *$2) }; +name_without_curly_rdive_class: class_prefix name_without_curly_rdive_code { $$=$1; P(*$$, *$2); }; +name_without_curly_rdive_code: name_advance2 | name_path name_advance2 { $$=$1; P(*$$, *$2); }; /* put */ put: '$' name_expr_wdive construct { - $$=$2; /* stack: context,name */ - P(*$$, *$3); /* stack: context,name,constructor_value */ + $$=N(); +#ifdef OPTIMIZE_BYTECODE_CONSTRUCT + if(maybe_optimize_construct(*$$, *$2, *$3)){ + // $a(expr), $.a(expr), $a[value], $.a[value], $self.a[value], $self.a(expr) + } else +#endif + { + P(*$$, *$2); /* stack: context,name */ + P(*$$, *$3); /* stack: context,name,constructor_value */ + } }; name_expr_wdive: name_expr_wdive_root @@ -300,74 +446,94 @@ name_expr_wdive: | name_expr_wdive_class; name_expr_wdive_root: name_expr_dive_code { $$=N(); - ArrayOperationPtr diving_code=$1; - StringPtr first_name=LA2S(*diving_code); - // $self.xxx... -> $xxx... - // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... - if(first_name && *first_name==SELF_ELEMENT_NAME) { - O(*$$, OP_WITH_SELF); /* stack: starting context */ - P(*$$, *diving_code, - /* skip over... */ - diving_code->count()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); - } else { - O(*$$, OP_WITH_ROOT); /* stack: starting context */ + YYSTYPE diving_code=$1; + size_t count=diving_code->count(); + + if(maybe_make_self(*$$, *diving_code, count)) { + // $self. + } else +#ifdef OPTIMIZE_BYTECODE_GET_ELEMENT + if( + count>=4 + && (*diving_code)[0].code==OP::OP_VALUE + && (*diving_code)[3].code==OP::OP_GET_ELEMENT + ){ + O(*$$, OP::OP_WITH_ROOT__VALUE__GET_ELEMENT); + P(*$$, *diving_code, 1/*offset*/, 2/*limit*/); // copy origin+value + if(count>4) + P(*$$, *diving_code, 4); // tail + } else +#endif + { + O(*$$, 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(*$$, *$2) }; +name_expr_wdive_class: class_prefix name_expr_dive_code { $$=$1; P(*$$, *$2); }; -construct: +construct: construct_square | construct_round | construct_curly ; -construct_square: '[' any_constructor_code_value ']' { +construct_square: '[' { + // allow $result_or_other_variable[ letters here any time ] + *reinterpret_cast(&$$)=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(*$$, OP::OP_CURLY_CODE__CONSTRUCT, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ }; any_constructor_code_value: - void_value /* optimized $var[] case */ + empty_value /* optimized $var[] case */ | STRING /* optimized $var[STRING] case */ | constructor_code_value /* $var[something complex] */ ; constructor_code_value: constructor_code { $$=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() */ }; constructor_code: codes__excluding_sole_str_literal; -codes__excluding_sole_str_literal: action | code codes { $$=$1; P(*$$, *$2) }; +codes__excluding_sole_str_literal: action | code codes { $$=$1; P(*$$, *$2); }; /* 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 + { + $$=$1; /* stack: value */ + if(!maybe_change_first_opcode(*$$, OP::OP_CONSTRUCT_OBJECT, /*=>*/OP::OP_CONSTRUCT_OBJECT__WRITE)) + changetail_or_append(*$$, + OP::OP_CALL, true, /*=>*/ OP::OP_CALL__WRITE, + /*or */OP::OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ + } }; call_value: '^' { PC.in_call_value=true; @@ -376,70 +542,107 @@ call_value: '^' { PC.in_call_value=false; } store_params EON { /* ^field.$method{vasya} */ - $$=$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){ + $$=N(); + } else +#endif + { + YYSTYPE params_code=$5; + if(params_code->count()==3) { // probably [] case. [OP::OP_VALUE+origin+Void] + if(Value* value=LA2V(*params_code)) // it is OP_VALUE+origin+value? + if(const String * string=value->get_string()) + if(string->is_empty()) // value is empty string? + params_code=0; // ^zzz[] case. don't append lone empty param. + } + /* stack: context, method_junction */ - YYSTYPE params_code=$5; - if(params_code->count()==3) { // probably [] case. [OP_VALUE + Void + STORE_PARAM] - if(ValuePtr value=LA2V(*params_code)) // it is OP_VALUE + value? - if(!value->is_defined()) // value is VVoid? - params_code=YYSTYPE(0); // ^zzz[] case. don't append lone empty param. - } - /* stack: context, method_junction */ - OA(*$$, OP_CALL, params_code); // method_frame=make frame(pop junction); ncontext=pop; call(ncontext,method_frame) stack: value + YYSTYPE var_code=$3; + if( + var_code->count()==8 + && (*var_code)[0].code==OP::OP_VALUE__GET_CLASS + && (*var_code)[3].code==OP::OP_PREPARE_TO_CONSTRUCT_OBJECT + && (*var_code)[4].code==OP::OP_VALUE + && (*var_code)[7].code==OP::OP_GET_ELEMENT + ){ + yyval=N(); + O(*$$, OP::OP_CONSTRUCT_OBJECT); + P(*$$, *var_code, 1/*offset*/, 2/*limit*/); // class name + P(*$$, *var_code, 5/*offset*/, 2/*limit*/); // constructor name + OA(*$$, params_code); + } else + { + $$=var_code; /* with_xxx,diving code; stack: context,method_junction */ + OA(*$$, 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; -store_params: store_param | store_params store_param { $$=$1; P(*$$, *$2) }; +store_params: store_param | store_params store_param { $$=$1; P(*$$, *$2); }; store_param: store_square_param | store_round_param | store_curly_param ; -store_square_param: '[' store_code_param_parts ']' {$$=$2}; -store_round_param: '(' store_expr_param_parts ')' {$$=$2}; -store_curly_param: '{' store_curly_param_parts '}' {$$=$2}; +store_square_param: '[' { + // allow ^call[ letters here any time ] + *reinterpret_cast(&$$)=PC.explicit_result; PC.explicit_result=false; +} store_code_param_parts { + PC.explicit_result=*reinterpret_cast(&$2); +} ']' {$$=$3;}; +store_round_param: '(' store_expr_param_parts ')' {$$=$2;}; +store_curly_param: '{' store_curly_param_parts '}' {$$=$2;}; store_code_param_parts: store_code_param_part -| store_code_param_parts ';' store_code_param_part { $$=$1; P(*$$, *$3) } +| store_code_param_parts ';' store_code_param_part { $$=$1; P(*$$, *$3); } ; store_expr_param_parts: store_expr_param_part -| store_expr_param_parts ';' store_expr_param_part { $$=$1; P(*$$, *$3) } +| store_expr_param_parts ';' store_expr_param_part { $$=$1; P(*$$, *$3); } ; store_curly_param_parts: store_curly_param_part -| store_curly_param_parts ';' store_curly_param_part { $$=$1; P(*$$, *$3) } +| store_curly_param_parts ';' store_curly_param_part { $$=$1; P(*$$, *$3); } ; store_code_param_part: code_param_value { $$=$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(*$$, OP::OP_CURLY_CODE__STORE_PARAM, $1); }; code_param_value: - void_value /* optimized [;...] case */ + empty_value /* optimized [;...] case */ | STRING /* optimized [STRING] case */ | constructor_code_value /* [something complex] */ ; -write_expr_value: expr_value { - $$=N(); - O(*$$, OP_PREPARE_TO_EXPRESSION); - P(*$$, *$1); - O(*$$, OP_WRITE_EXPR_RESULT); -}; /* name */ -name_expr_dive_code: name_expr_value | name_path name_expr_value { $$=$1; P(*$$, *$2) }; +name_expr_dive_code: name_expr_value | name_path name_expr_value { $$=$1; P(*$$, *$2); }; -name_path: name_step | name_path name_step { $$=$1; P(*$$, *$2) }; +name_path: name_step | name_path name_step { $$=$1; P(*$$, *$2); }; name_step: name_advance1 '.'; name_advance1: name_expr_value { // we know that name_advance1 not called from ^xxx context @@ -447,12 +650,12 @@ name_advance1: name_expr_value { /* stack: context */ $$=$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(*$$, OP::OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ } | STRING BOGUS ; @@ -464,32 +667,37 @@ name_expr_value: ; name_expr_subvar_value: '$' subvar_ref_name_rdive { $$=$2; - O(*$$, OP_GET_ELEMENT); + O(*$$, OP::OP_GET_ELEMENT); }; name_expr_with_subvar_value: STRING subvar_get_writes { - ArrayOperationPtr 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(*$$, *$2) }; +subvar_get_writes: subvar__get_write | subvar_get_writes subvar__get_write { $$=$1; P(*$$, *$2); }; subvar__get_write: '$' subvar_ref_name_rdive { $$=$2; - O(*$$, OP_GET_ELEMENT__WRITE); + O(*$$, OP::OP_GET_ELEMENT__WRITE); }; class_prefix: @@ -498,7 +706,7 @@ class_prefix: ; class_static_prefix: STRING ':' { $$=$1; // stack: class name string - if(*LA2S(*$$) == BASE_NAME) { // pseude BASE class + if(*LA2S(*$$) == BASE_NAME) { // pseudo BASE class if(VStateless_class* base=PC.cclass->base_class()) { change_string_literal_value(*$$, base->name()); } else { @@ -506,7 +714,8 @@ class_static_prefix: STRING ':' { YYERROR; } } - O(*$$, OP_GET_CLASS); + // optimized OP_VALUE+origin+string+OP_GET_CLASS => OP_VALUE__GET_CLASS+origin+string + maybe_change_first_opcode(*$$, OP::OP_VALUE, OP::OP_VALUE__GET_CLASS); }; class_constructor_prefix: class_static_prefix ':' { $$=$1; @@ -514,66 +723,81 @@ class_constructor_prefix: class_static_p strcpy(PC.error, ":: not allowed here"); YYERROR; } - O(*$$, 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 '\'' { $$ = $2; } | '(' expr ')' { $$ = $2; } /* stack: operand // stack: @operand */ -| '-' expr %prec NUNARY { $$=$2; O(*$$, 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(*$$, OP_DEXISTS) } +| '-' expr %prec NUNARY { $$=$2; O(*$$, OP::OP_NEG); } +| '+' expr %prec NUNARY { $$=$2; } +| '~' 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(); + 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_STRING_POOL, $1); /* stack: empty write context */ + OA(*$$, OP::OP_STRING_POOL, $1); /* stack: empty write context */ +#endif /* some code that writes to that context */ /* context=pop; stack: context.get_string() */ }; @@ -581,12 +805,15 @@ string_inside_quotes_value: maybe_codes /* basics */ write_string: STRING { - // optimized from OP_STRING+OP_WRITE_VALUE to OP_STRING__WRITE - change_string_literal_to_write_string_literal(*($$=$1)) + // optimized OP_STRING+OP_WRITE_VALUE => OP_STRING__WRITE + change_string_literal_to_write_string_literal(*($$=$1)); }; -void_value: /* empty */ { $$=VL(ValuePtr(new VVoid())) }; -empty: /* empty */ { $$=N() }; +empty_value: /* empty */ { $$=VL(/*we know that we will not change it*/const_cast(&vempty), 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 */ { $$=N(); }; %% #endif @@ -607,55 +834,66 @@ empty: /* empty */ { $$=N() }; 4:[^({]=pop */ -static int yylex(YYSTYPE *lvalp, void *pc) { - #define lexical_brackets_nestage PC.brackets_nestages[PC.ls_sp] +inline void ungetc(Parse_control& pc, uint last_line_end_col) { + pc.source--; + if(pc.pos.col==0) { + --pc.pos.line; pc.pos.col=last_line_end_col; + } else + --pc.pos.col; + +} +static int yylex(YYSTYPE *lvalp, void *apc) { + register Parse_control& pc=*static_cast(apc); + + #define lexical_brackets_nestage pc.brackets_nestages[pc.ls_sp] #define RC {result=c; goto break2; } - register int c; - int result; + register int c; + int result; - if(PC.pending_state) { - result=PC.pending_state; - PC.pending_state=0; + if(pc.pending_state) { + result=pc.pending_state; + pc.pending_state=0; return result; } - const char *begin=PC.source; + const char *begin=pc.source; + Pos begin_pos=pc.pos; const char *end; - int begin_line=PC.line; int skip_analized=0; while(true) { - c=*(end=(PC.source++)); -// fprintf(stderr, "\nchar: %c %02X; nestage: %d, sp=%d", c, c, lexical_brackets_nestage, PC.sp); + c=*(end=(pc.source++)); +// fprintf(stderr, "\nchar: %c %02X; nestage: %d, sp=%d", c, c, lexical_brackets_nestage, pc.sp); - if(c=='\n') { - PC.line++; - PC.col=0; - } else - PC.col++; + if(c=='\n') + pc.pos_next_line(); + else + pc.pos_next_c(c); +// fprintf(stderr, "\nchar: %c file(%d:%d)", c, pc.pos.line, pc.pos.col); - if(c=='@' && PC.col==0+1) { - if(PC.ls==LS_DEF_SPECIAL_BODY) { + if(pc.pos.col==0+1 && c=='@') { + if(pc.ls==LS_DEF_SPECIAL_BODY) { // @SPECIAL // ... // @APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + pc.string.append_strdup_know_length(begin, end-begin); } // reset piece 'begin' position & line - begin=PC.source; // ->punctuation - begin_line=PC.line; + begin=pc.source; // ->punctuation + begin_pos=pc.pos; // skip over _ after ^ - PC.source++; PC.col++; + pc.source++; pc.pos.col++; // skip analysis = forced literal continue; // converting ^#HH into char(hex(HH)) case '#': if(end!=begin) { + if(!pc.string_start) + pc.string_start=begin_pos; // append piece till ^ - PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + pc.string.append_strdup_know_length(begin, end-begin); } // #HH ? - if(PC.source[0]=='#' && PC.source[1] && PC.source[2]) { - char *hex=(char *)POOL.malloc(1); - hex[0]= - hex_value[(unsigned char)PC.source[1]]*0x10+ - hex_value[(unsigned char)PC.source[2]]; - if(hex[0]==0) { + 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]]); + if(c==0) { result=BAD_HEX_LITERAL; goto break2; // wrong hex value[no ^#00 chars allowed]: bail out } // append char(hex(HH)) - PC.string->APPEND_CLEAN(hex, 1, PC.file, begin_line); + pc.string.append(c); // skip over ^#HH - PC.source+=3; - PC.col+=3; + pc.source+=3; + pc.pos.col+=3; // reset piece 'begin' position & line - begin=PC.source; // ->after ^#HH - begin_line=PC.line; + begin=pc.source; // ->after ^#HH + begin_pos=pc.pos; // 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; } } // #comment start skipping - if(c=='#' && PC.col==1) { + if(c=='#' && pc.pos.col==1) { if(end!=begin) { + if(!pc.string_start) + pc.string_start=begin_pos; // append piece till # - PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + pc.string.append_strdup_know_length(begin, end-begin); } // fall into COMMENT lexical state [wait for \n] - push_LS(PC, LS_USER_COMMENT); + push_LS(pc, LS_USER_COMMENT); continue; } - switch(PC.ls) { + switch(pc.ls) { // USER'S = NOT OURS case LS_USER: - case LS_NAME_SQUARE_PART: // name.[here].xxx - if(PC.trim_bof) + case LS_NAME_SQUARE_PART: // name.[here].xxx + if(pc.trim_bof) switch(c) { case '\n': case ' ': case '\t': - begin=PC.source; - begin_line=PC.line; + begin=pc.source; + begin_pos=pc.pos; continue; // skip it default: - PC.trim_bof=false; + pc.trim_bof=false; } switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ']': - if(PC.ls==LS_NAME_SQUARE_PART) + if(pc.ls==LS_NAME_SQUARE_PART) if(--lexical_brackets_nestage==0) {// $name.[co<]?>de<]?> - pop_LS(PC); // $name.[co<]>de<]!> + pop_LS(pc); // $name.[co<]>de<]!> RC; } break; case '[': // $name.[co<[>de] - if(PC.ls==LS_NAME_SQUARE_PART) + if(pc.ls==LS_NAME_SQUARE_PART) lexical_brackets_nestage++; break; } + if(pc.explicit_result && c) + switch(c) { + case '\n': case ' ': case '\t': + begin=pc.source; + begin_pos=pc.pos; + continue; // skip it + default: + result=BAD_NONWHITESPACE_CHARACTER_IN_EXPLICIT_RESULT_MODE; + goto break2; + } break; // #COMMENT case LS_USER_COMMENT: if(c=='\n') { // skip comment - begin=PC.source; - begin_line=PC.line; + begin=pc.source; + begin_pos=pc.pos; - pop_LS(PC); + pop_LS(pc); continue; } break; @@ -782,17 +1042,17 @@ default: case '"': case '\'': if( - PC.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || - PC.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { - pop_LS(PC); //"abc". | 'abc'. + pc.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || + pc.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { + pop_LS(pc); //"abc". | 'abc'. RC; } break; case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; } break; @@ -801,10 +1061,10 @@ default: case LS_DEF_NAME: switch(c) { case '[': - PC.ls=LS_DEF_PARAMS; + pc.ls=LS_DEF_PARAMS; RC; case '\n': - PC.ls=LS_DEF_SPECIAL_BODY; + pc.ls=LS_DEF_SPECIAL_BODY; RC; } break; @@ -817,10 +1077,10 @@ default: case ';': RC; case ']': - PC.ls=*PC.source=='['?LS_DEF_LOCALS:LS_DEF_COMMENT; + pc.ls=*pc.source=='['?LS_DEF_LOCALS:LS_DEF_COMMENT; RC; case '\n': // wrong. bailing out - pop_LS(PC); + pop_LS(pc); RC; } break; @@ -831,17 +1091,17 @@ default: case ';': RC; case ']': - PC.ls=LS_DEF_COMMENT; + pc.ls=LS_DEF_COMMENT; RC; case '\n': // wrong. bailing out - pop_LS(PC); + pop_LS(pc); RC; } break; case LS_DEF_COMMENT: if(c=='\n') { - pop_LS(PC); + pop_LS(pc); RC; } break; @@ -857,31 +1117,33 @@ default: switch(c) { case ')': if(--lexical_brackets_nestage==0) - if(PC.ls==LS_METHOD_ROUND) // method round param ended - PC.ls=LS_METHOD_AFTER; // look for method end - else // PC.ls==LS_VAR_ROUND // variable constructor ended - pop_LS(PC); // return to normal life + if(pc.ls==LS_METHOD_ROUND) // method round param ended + pc.ls=LS_METHOD_AFTER; // look for method end + else // pc.ls==LS_VAR_ROUND // variable constructor ended + pop_LS(pc); // return to normal life RC; case '#': // comment start skipping if(end!=begin) { + if(!pc.string_start) + pc.string_start=begin_pos; // append piece till # - PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + pc.string.append_strdup_know_length(begin, end-begin); } // fall into COMMENT lexical state [wait for \n] - push_LS(PC, LS_EXPRESSION_COMMENT); + push_LS(pc, LS_EXPRESSION_COMMENT); lexical_brackets_nestage=1; continue; case '$': - push_LS(PC, LS_EXPRESSION_VAR_NAME_WITH_COLON); + push_LS(pc, LS_EXPRESSION_VAR_NAME_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case '(': lexical_brackets_nestage++; RC; case '-': - switch(*PC.source) { + switch(*pc.source) { case 'f': // -f skip_analized=1; result=FEXISTS; @@ -900,17 +1162,17 @@ default: case ';': RC; case '&': case '|': - if(*PC.source==c) { // && || + if(*pc.source==c) { // && || result=c=='&'?LAND:LOR; skip_analized=1; } else result=c; goto break2; case '!': - switch(PC.source[0]) { + switch(pc.source[0]) { case '|': // !| !|| skip_analized=1; - if(PC.source[1]=='|') { + if(pc.source[1]=='|') { skip_analized++; result=LXOR; } else @@ -924,7 +1186,7 @@ default: RC; case '<': // <<, <=, < - switch(*PC.source) { + switch(*pc.source) { case '<': // <[<] skip_analized=1; result=NSL; break; case '=': // <[=] @@ -934,7 +1196,7 @@ default: } goto break2; case '>': // >>, >=, > - switch(*PC.source) { + switch(*pc.source) { case '>': // >[>] skip_analized=1; result=NSR; break; case '=': // >[=] @@ -944,7 +1206,7 @@ default: } goto break2; case '=': // == - switch(*PC.source) { + switch(*pc.source) { case '=': // =[=] skip_analized=1; result=NEQ; break; default: // =[] @@ -953,15 +1215,15 @@ default: goto break2; case '"': - push_LS(PC, LS_EXPRESSION_STRING_QUOTED); + push_LS(pc, LS_EXPRESSION_STRING_QUOTED); RC; case '\'': - push_LS(PC, LS_EXPRESSION_STRING_APOSTROFED); + push_LS(pc, LS_EXPRESSION_STRING_APOSTROFED); RC; case 'l': case 'g': case 'e': case 'n': if(end==begin) // right after whitespace - if(isspace(PC.source[1])) { - switch(*PC.source) { + if(isspace(pc.source[1])) { + switch(*pc.source) { // case '?': // ok [and bad cases, yacc would bark at them] case 't': // lt gt [et nt] result=c=='l'?SLT:c=='g'?SGT:BAD_STRING_COMPARISON_OPERATOR; @@ -980,8 +1242,8 @@ default: break; case 'i': if(end==begin) // right after whitespace - if(isspace(PC.source[1])) { - switch(PC.source[0]) { + if(isspace(pc.source[1])) { + switch(pc.source[0]) { case 'n': // in skip_analized=1; result=IN; @@ -995,9 +1257,29 @@ default: break; case 'd': if(end==begin) // right after whitespace - if(PC.source[0]=='e' && PC.source[1]=='f') { // def - skip_analized=2; - result=DEF; + if(pc.source[0]=='e' && pc.source[1]=='f') { // 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; @@ -1009,8 +1291,8 @@ default: // that's a leading|traling space or after-operator-space // ignoring it // reset piece 'begin' position & line - begin=PC.source; // after whitespace char - begin_line=PC.line; + begin=pc.source; // after whitespace char + begin_pos=pc.pos; continue; } break; @@ -1018,17 +1300,17 @@ default: if(c=='(') lexical_brackets_nestage++; - switch(*PC.source) { + switch(*pc.source) { case '\n': case ')': - if(*PC.source==')') + if(*pc.source==')') if(--lexical_brackets_nestage!=0) continue; // skip comment - begin=PC.source; - begin_line=PC.line; + begin=pc.source; + begin_pos=pc.pos; - pop_LS(PC); + pop_LS(pc); continue; } break; @@ -1039,28 +1321,28 @@ default: case LS_EXPRESSION_VAR_NAME_WITH_COLON: case LS_EXPRESSION_VAR_NAME_WITHOUT_COLON: if( - PC.ls==LS_EXPRESSION_VAR_NAME_WITH_COLON || - PC.ls==LS_EXPRESSION_VAR_NAME_WITHOUT_COLON) { + pc.ls==LS_EXPRESSION_VAR_NAME_WITH_COLON || + pc.ls==LS_EXPRESSION_VAR_NAME_WITHOUT_COLON) { // name in expr ends also before switch(c) { // expression minus case '-': // expression integer division case '\\': - pop_LS(PC); - PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } + pop_LS(pc); + pc.ungetc(); result=EON; goto break2; } } if( - PC.ls==LS_VAR_NAME_SIMPLE_WITHOUT_COLON || - PC.ls==LS_EXPRESSION_VAR_NAME_WITHOUT_COLON) { + pc.ls==LS_VAR_NAME_SIMPLE_WITHOUT_COLON || + pc.ls==LS_EXPRESSION_VAR_NAME_WITHOUT_COLON) { // name already has ':', stop before next switch(c) { case ':': - pop_LS(PC); - PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } + pop_LS(pc); + pc.ungetc(); result=EON; goto break2; } @@ -1072,52 +1354,54 @@ default: case ']': case '}': case ')': case '"': case '\'': case '<': case '>': // these stand for HTML brackets AND expression binary ops - case '+': case '*': case '/': case '%': + case '+': case '*': case '/': case '\\': case '%': case '&': case '|': case '=': case '!': // common delimiters case ',': case '?': case '#': + // mysql column separators + case '`': // before call case '^': - pop_LS(PC); - PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } + pop_LS(pc); + pc.ungetc(); result=EON; goto break2; case '[': // $name.<[>code] - if(PC.col>1/*not first column*/ && ( + if(pc.pos.col>1/*not first column*/ && ( end[-1]=='$'/*was start of get*/ || end[-1]==':'/*was class name delim */ || end[-1]=='.'/*was name delim */ )) { - push_LS(PC, LS_NAME_SQUARE_PART); + push_LS(pc, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; } - PC.ls=LS_VAR_SQUARE; + pc.ls=LS_VAR_SQUARE; lexical_brackets_nestage=1; RC; case '{': if(begin==end) { // ${name}, no need of EON, switching LS - PC.ls=LS_VAR_NAME_CURLY; + pc.ls=LS_VAR_NAME_CURLY; } else { - PC.ls=LS_VAR_CURLY; + pc.ls=LS_VAR_CURLY; lexical_brackets_nestage=1; } RC; case '(': - PC.ls=LS_VAR_ROUND; + pc.ls=LS_VAR_ROUND; lexical_brackets_nestage=1; RC; case '.': // name part delim case '$': // name part subvar case ':': // class<:>name // go to _WITHOUT_COLON state variant... - if(PC.ls==LS_VAR_NAME_SIMPLE_WITH_COLON) - PC.ls=LS_VAR_NAME_SIMPLE_WITHOUT_COLON; - else if(PC.ls==LS_EXPRESSION_VAR_NAME_WITH_COLON) - PC.ls=LS_EXPRESSION_VAR_NAME_WITHOUT_COLON; + if(pc.ls==LS_VAR_NAME_SIMPLE_WITH_COLON) + pc.ls=LS_VAR_NAME_SIMPLE_WITHOUT_COLON; + else if(pc.ls==LS_EXPRESSION_VAR_NAME_WITH_COLON) + pc.ls=LS_EXPRESSION_VAR_NAME_WITHOUT_COLON; // ...stop before next ':' RC; } @@ -1127,11 +1411,11 @@ default: switch(c) { case '[': // ${name.<[>code]} - push_LS(PC, LS_NAME_SQUARE_PART); + push_LS(pc, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; case '}': // ${name} finished, restoring LS - pop_LS(PC); + pop_LS(pc); RC; case '.': // name part delim case '$': // name part subvar @@ -1143,14 +1427,14 @@ default: case LS_VAR_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ']': if(--lexical_brackets_nestage==0) { - pop_LS(PC); + pop_LS(pc); RC; } break; @@ -1165,14 +1449,14 @@ default: case LS_VAR_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case '}': if(--lexical_brackets_nestage==0) { - pop_LS(PC); + pop_LS(pc); RC; } break; @@ -1187,24 +1471,24 @@ default: switch(c) { case '[': // ^name.<[>code].xxx - if(PC.col>1/*not first column*/ && ( + if(pc.pos.col>1/*not first column*/ && ( end[-1]=='^'/*was start of call*/ || // never, ^[ is literal... end[-1]==':'/*was class name delim */ || end[-1]=='.'/*was name delim */ )) { - push_LS(PC, LS_NAME_SQUARE_PART); + push_LS(pc, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; } - PC.ls=LS_METHOD_SQUARE; + pc.ls=LS_METHOD_SQUARE; lexical_brackets_nestage=1; RC; case '{': - PC.ls=LS_METHOD_CURLY; + pc.ls=LS_METHOD_CURLY; lexical_brackets_nestage=1; RC; case '(': - PC.ls=LS_METHOD_ROUND; + pc.ls=LS_METHOD_ROUND; lexical_brackets_nestage=1; RC; case '.': // name part delim @@ -1212,6 +1496,7 @@ default: case ':': // ':name' or 'class:name' case '^': // ^abc^xxx wrong. bailing out case ']': case '}': case ')': // ^abc]}) wrong. bailing out + case ' ': // ^if ( wrong. bailing out RC; } break; @@ -1219,16 +1504,16 @@ default: case LS_METHOD_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ';': // param delim RC; case ']': if(--lexical_brackets_nestage==0) { - PC.ls=LS_METHOD_AFTER; + pc.ls=LS_METHOD_AFTER; RC; } break; @@ -1241,16 +1526,16 @@ default: case LS_METHOD_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ';': // param delim RC; case '}': if(--lexical_brackets_nestage==0) { - PC.ls=LS_METHOD_AFTER; + pc.ls=LS_METHOD_AFTER; RC; } break; @@ -1258,26 +1543,36 @@ default: lexical_brackets_nestage++; break; } + if(pc.explicit_result && c) + switch(c) { + case '\n': case ' ': case '\t': + begin=pc.source; + begin_pos=pc.pos; + continue; // skip it + default: + result=BAD_NONWHITESPACE_CHARACTER_IN_EXPLICIT_RESULT_MODE; + goto break2; + } break; case LS_METHOD_AFTER: if(c=='[') {/* ][ }[ )[ */ - PC.ls=LS_METHOD_SQUARE; + pc.ls=LS_METHOD_SQUARE; lexical_brackets_nestage=1; RC; } if(c=='{') {/* ]{ }{ ){ */ - PC.ls=LS_METHOD_CURLY; + pc.ls=LS_METHOD_CURLY; lexical_brackets_nestage=1; RC; } if(c=='(') {/* ]( }( )( */ - PC.ls=LS_METHOD_ROUND; + pc.ls=LS_METHOD_ROUND; lexical_brackets_nestage=1; RC; } - pop_LS(PC); - PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } + pop_LS(pc); + pc.ungetc(); result=EON; goto break2; } @@ -1289,37 +1584,39 @@ default: break2: if(end!=begin) { // there is last piece? - if((c=='@' || c==0) && end[-1]=='\n') { // we are before LS_DEF_NAME or EOF? - // strip last \n - end--; - if(end!=begin && end[-1]=='\n') // allow one empty line before LS_DEF_NAME + if(c=='@' || c==0) // we are before LS_DEF_NAME or EOF? + while(end!=begin && end[-1]=='\n') // trim all empty lines before LS_DEF_NAME and EOF end--; - } - if(end!=begin && PC.ls!=LS_USER_COMMENT) { // last piece still alive and not comment? + if(end!=begin && pc.ls!=LS_USER_COMMENT) { // last piece still alive and not comment? + if(!pc.string_start) + pc.string_start=begin_pos; // append it - PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line/*, start_col*/); + pc.string.append_strdup_know_length(begin, end-begin); } } - if(PC.string->size()) { // something accumulated? - // create STRING value: array of OP_VALUE+vstring - *lvalp=VL(ValuePtr(new VString(PC.string))); + if(!pc.string.is_empty()) { // something accumulated? + // create STRING value: array of OP_VALUE+origin+vstring + *lvalp=VL( + new VString(*new String(pc.string, String::L_CLEAN)), + pc.file_no, pc.string_start.line, pc.string_start.col); // new pieces storage - PC.string=StringPtr(new String); + pc.string.clear(); + pc.string_start.clear(); // make current result be pending for next call, return STRING for now - PC.pending_state=result; result=STRING; + pc.pending_state=result; result=STRING; } if(skip_analized) { - PC.source+=skip_analized; PC.col+=skip_analized; + pc.source+=skip_analized; pc.pos.col+=skip_analized; } return result; } -static int real_yyerror(parse_control *pc, char *s) { // Called by yyparse on error +static int real_yyerror(Parse_control *pc, char *s) { // Called by yyparse on error strncpy(PC.error, s, MAX_STRING); return 1; } static void yyprint(FILE *file, int type, YYSTYPE value) { if(type==STRING) - fprintf(file, " \"%s\"", LA2S(*value)->cstr().get()); + fprintf(file, " \"%s\"", LA2S(*value)->cstr()); }