--- parser3/src/main/compile.y 2003/03/21 14:27:52 1.204.2.8.2.2 +++ parser3/src/main/compile.y 2005/12/16 14:04:20 1.219 @@ -2,10 +2,10 @@ /** @file Parser: compiler(lexical parser and grammar). - Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexander Petrosyan (http://design.ru/paf) - $Id: compile.y,v 1.204.2.8.2.2 2003/03/21 14:27:52 paf Exp $ + $Id: compile.y,v 1.219 2005/12/16 14:04:20 paf Exp $ */ /** @@ -24,9 +24,11 @@ #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" @@ -36,16 +38,23 @@ #include "pa_vvoid.h" #include "pa_vmethod_frame.h" +// defines + #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); +// 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 VVoid vvoid; // local convinient inplace typecast & var #undef PC -#define PC (*(parse_control *)pc) +#define PC (*(Parse_control *)pc) #undef POOL #define POOL (*PC.pool) #ifndef DOXYGEN @@ -61,7 +70,7 @@ 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 +97,9 @@ 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 +126,11 @@ static int yylex(YYSTYPE *lvalp, void *p %% all: one_big_piece { - Method& method=*new Method(/*main_method_name, */Method::CT_ANY, + Method& method=*new Method(Method::CT_ANY, 0, 0, /*min, max numbered_params_count*/ 0/*param_names*/, 0/*local_names*/, $1/*parser_code*/, 0/*native_code*/); - PC.cclass->add_method(main_method_name, method); + PC.cclass->add_method(PC.alias_method(main_method_name), method); } | methods; @@ -131,7 +143,7 @@ control_method: '@' STRING '\n' maybe_control_strings { const String& command=*LA2S(*$2); 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, " is empty"); @@ -144,7 +156,7 @@ control_method: '@' STRING '\n' strcat(PC.error, "'"); YYERROR; } - if(strings_code->count()==1*2) { + if(strings_code->count()==1*OPERATIONS_PER_OPVALUE) { // new class' name const String& name=*LA2S(*strings_code); // creating the class @@ -154,11 +166,11 @@ control_method: '@' STRING '\n' // append to request's classes PC.request.classes().put(name, cclass); } 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) { - for(int i=0; icount(); i+=2) + 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.cclass->base_class()) { // already changed from default? @@ -167,7 +179,7 @@ control_method: '@' STRING '\n' strcat(PC.error, "'"); YYERROR; } - if(strings_code->count()==1*2) { + if(strings_code->count()==1*OPERATIONS_PER_OPVALUE) { const String& base_name=*LA2S(*strings_code); if(Value* base_class_value=PC.request.classes().get(base_name)) { // @CLASS == @BASE sanity check @@ -204,33 +216,44 @@ control_strings: control_string | contro control_string: maybe_string '\n'; maybe_string: empty | STRING; -code_method: '@' STRING bracketed_maybe_strings maybe_bracketed_strings maybe_comment '\n' - maybe_codes { +code_method: '@' STRING bracketed_maybe_strings maybe_bracketed_strings maybe_comment '\n' { + PC.explicit_result=false; const String& name=*LA2S(*$2); YYSTYPE params_names_code=$3; - ArrayString* params_names; + ArrayString* params_names=0; if(int size=params_names_code->count()) { params_names=new ArrayString; - for(int i=0; icount()) { locals_names=new ArrayString; - for(int i=0; iadd_method(name, method); + 0/*to be filled later in next {} */, 0); + PC.cclass->add_method(PC.alias_method(name), *method); + *reinterpret_cast(&$$)=method; + + // todo: check [][;result;] +} maybe_codes { + // fill in the code + reinterpret_cast($7)->parser_code=$8; }; maybe_bracketed_strings: empty | bracketed_maybe_strings; @@ -268,19 +291,19 @@ name_without_curly_rdive_read: name_with ArrayOperation* diving_code=$1; const String* first_name=LA2S(*diving_code); // self.xxx... -> xxx... - // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... + // OP_VALUE+origin+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { - O(*$$, OP_WITH_SELF); /* stack: starting context */ - P(*$$, *diving_code, + O(*yyval, OP_WITH_SELF); /* stack: starting context */ + P(*yyval, *diving_code, /* skip over... */ - diving_code->count()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); + diving_code->count()>=4?4/*OP_VALUE+origin+string+OP_GET_ELEMENTx*/:3/*OP_+origin+string*/); } else { - O(*$$, OP_WITH_READ); /* stack: starting context */ + O(*yyval, 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; + // 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); P(*$$, *diving_code); } /* diving code; stack: current context */ @@ -303,12 +326,12 @@ name_expr_wdive_root: name_expr_dive_cod ArrayOperation* diving_code=$1; const String* first_name=LA2S(*diving_code); // $self.xxx... -> $xxx... - // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... + // OP_VALUE+origin+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { O(*$$, OP_WITH_SELF); /* stack: starting context */ P(*$$, *diving_code, /* skip over... */ - diving_code->count()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); + diving_code->count()>=4?4/*OP_VALUE+origin+string+OP_GET_ELEMENTx*/:3/*OP_+origin+string*/); } else { O(*$$, OP_WITH_ROOT); /* stack: starting context */ P(*$$, *diving_code); @@ -327,9 +350,14 @@ construct: | construct_round | construct_curly ; -construct_square: '[' any_constructor_code_value ']' { +construct_square: '[' { + // allow $result_or_other_variable[ letters here any time ] + *reinterpret_cast(&$$)=PC.explicit_result; PC.explicit_result=false; +} any_constructor_code_value { + PC.explicit_result=*reinterpret_cast(&$2); +} ']' { // stack: context, name - $$=$2; // stack: context, name, value + $$=$3; // stack: context, name, value O(*$$, OP_CONSTRUCT_VALUE); /* value=pop; name=pop; context=pop; construct(context,name,value) */ } ; @@ -379,9 +407,9 @@ call_value: '^' { $$=$3; /* with_xxx,diving code; stack: context,method_junction */ YYSTYPE params_code=$5; - if(params_code->count()==3) { // probably [] case. [OP_VALUE + Void + STORE_PARAM] - if(Value* value=LA2V(*params_code)) // it is OP_VALUE + value? - if(!value->is_defined()) // value is VVoid? + 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_void()) // value is VVoid? params_code=0; // ^zzz[] case. don't append lone empty param. } /* stack: context, method_junction */ @@ -396,7 +424,12 @@ store_param: | store_round_param | store_curly_param ; -store_square_param: '[' store_code_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 '}' {$$=$2}; store_code_param_parts: @@ -415,9 +448,20 @@ store_code_param_part: code_param_value $$=$1; O(*$$, OP_STORE_PARAM); }; -store_expr_param_part: write_expr_value { - $$=N(); - OA(*$$, OP_EXPR_CODE__STORE_PARAM, $1); +store_expr_param_part: expr_value { + YYSTYPE expr_code=$1; + if(expr_code->count()==3 + && (*expr_code)[0].code==OP_VALUE) { // optimizing (double/bool/incidently 'string' too) case. [OP_VALUE+origin+Double] + $$=expr_code; + O(*$$, OP_STORE_PARAM); // no evaluating + } else { + ArrayOperation* code=N(); + O(*code, OP_PREPARE_TO_EXPRESSION); + P(*code, *expr_code); + O(*code, OP_WRITE_EXPR_RESULT); + $$=N(); + OA(*$$, OP_EXPR_CODE__STORE_PARAM, code); + } }; store_curly_param_part: maybe_codes { $$=N(); @@ -428,12 +472,6 @@ code_param_value: | 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 */ @@ -475,9 +513,14 @@ name_expr_with_subvar_value: STRING subv $$=N(); OA(*$$, 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_OBJECT_POOL, $3); /* stack: empty write context */ /* some code that writes to that context */ /* context=pop; stack: context.value() */ }; @@ -498,7 +541,7 @@ 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 { @@ -520,17 +563,15 @@ class_constructor_prefix: class_static_p /* 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(*$$, OP_NEG) } @@ -571,6 +612,11 @@ expr: | expr "is" expr { $$=$1; P(*$$, *$3); O(*$$, OP_IS) } ; +double_or_STRING: STRING { + // optimized from OP_STRING->OP_VALUE for doubles + maybe_change_string_literal_to_double_literal(*($$=$1)); +}; + string_inside_quotes_value: maybe_codes { $$=N(); OA(*$$, OP_STRING_POOL, $1); /* stack: empty write context */ @@ -585,7 +631,10 @@ write_string: STRING { change_string_literal_to_write_string_literal(*($$=$1)) }; -void_value: /* empty */ { $$=VL(new VVoid()) }; +void_value: /* empty */ { $$=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 */ { $$=N() }; %% @@ -607,55 +656,66 @@ empty: /* empty */ { $$=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 // ... // @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(begin, end-begin); + pc.string.append_strdup_know_length(begin, end-begin); } // #HH ? - if(PC.source[0]=='#' && PC.source[1] && PC.source[2]) { - char c= - hex_value[(unsigned char)PC.source[1]]*0x10+ - hex_value[(unsigned char)PC.source[2]]; + if(pc.source[0]=='#' && pc.source[1] && 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(c); + 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; } @@ -718,58 +782,70 @@ default: } } // #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(begin, end-begin); + 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 '[': // $name.[co<[>de] - if(PC.ls==LS_NAME_SQUARE_PART) + if(pc.ls==LS_NAME_SQUARE_PART) lexical_brackets_nestage++; break; } + if(pc.explicit_result && c) + switch(c) { + case '\n': case ' ': case '\t': + begin=pc.source; + begin_pos=pc.pos; + continue; // skip it + default: + result=BAD_NONWHITESPACE_CHARACTER_IN_EXPLICIT_RESULT_MODE; + goto break2; + } break; // #COMMENT case LS_USER_COMMENT: if(c=='\n') { // skip comment - begin=PC.source; - begin_line=PC.line; + begin=pc.source; + begin_pos=pc.pos; - pop_LS(PC); + pop_LS(pc); continue; } break; @@ -781,17 +857,17 @@ default: case '"': case '\'': if( - PC.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || - PC.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { - pop_LS(PC); //"abc". | 'abc'. + pc.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || + pc.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { + pop_LS(pc); //"abc". | 'abc'. RC; } break; case '$': - 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; @@ -800,10 +876,10 @@ 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; @@ -816,10 +892,10 @@ 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; @@ -830,17 +906,17 @@ 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; @@ -856,31 +932,33 @@ 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(begin, end-begin); + 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; @@ -899,17 +977,17 @@ 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 @@ -923,7 +1001,7 @@ default: RC; case '<': // <<, <=, < - switch(*PC.source) { + switch(*pc.source) { case '<': // <[<] skip_analized=1; result=NSL; break; case '=': // <[=] @@ -933,7 +1011,7 @@ default: } goto break2; case '>': // >>, >=, > - switch(*PC.source) { + switch(*pc.source) { case '>': // >[>] skip_analized=1; result=NSR; break; case '=': // >[=] @@ -943,7 +1021,7 @@ default: } goto break2; case '=': // == - switch(*PC.source) { + switch(*pc.source) { case '=': // =[=] skip_analized=1; result=NEQ; break; default: // =[] @@ -952,15 +1030,15 @@ 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; @@ -979,8 +1057,8 @@ 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; @@ -994,12 +1072,28 @@ default: break; case 'd': if(end==begin) // right after whitespace - if(PC.source[0]=='e' && PC.source[1]=='f') { // def + if(pc.source[0]=='e' && pc.source[1]=='f') { // def skip_analized=2; result=DEF; goto break2; } break; + case 't': + if(end==begin) // right after whitespace + if(pc.source[0]=='r' && pc.source[1]=='u' && pc.source[2]=='e') { // def + 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') { // def + skip_analized=4; + result=LITERAL_FALSE; + goto break2; + } + break; case ' ': case '\t': case '\n': if(end!=begin) { // there were a string after previous operator? result=0; // return that string @@ -1008,8 +1102,8 @@ 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; @@ -1017,17 +1111,17 @@ 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; @@ -1038,28 +1132,28 @@ 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; } @@ -1078,45 +1172,45 @@ default: case ',': case '?': 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; } @@ -1126,11 +1220,11 @@ 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 @@ -1142,14 +1236,14 @@ default: case LS_VAR_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ']': if(--lexical_brackets_nestage==0) { - pop_LS(PC); + pop_LS(pc); RC; } break; @@ -1164,14 +1258,14 @@ default: case LS_VAR_CURLY: switch(c) { case '$': - 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; @@ -1186,24 +1280,24 @@ 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 @@ -1211,6 +1305,7 @@ 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; @@ -1218,16 +1313,16 @@ default: case LS_METHOD_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); + push_LS(pc, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': - push_LS(PC, LS_METHOD_NAME); + push_LS(pc, LS_METHOD_NAME); RC; case ';': // param delim RC; case ']': if(--lexical_brackets_nestage==0) { - PC.ls=LS_METHOD_AFTER; + pc.ls=LS_METHOD_AFTER; RC; } break; @@ -1240,16 +1335,16 @@ default: case LS_METHOD_CURLY: switch(c) { case '$': - 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; @@ -1257,26 +1352,36 @@ 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; } @@ -1294,26 +1399,31 @@ break2: if(end!=begin && end[-1]=='\n') // allow one empty line before LS_DEF_NAME 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(begin, end-begin); + pc.string.append_strdup_know_length(begin, end-begin); } } - if(PC.string!=CORD_EMPTY) { // something accumulated? - // create STRING value: array of OP_VALUE+vstring - *lvalp=VL(new VString(*new String(PC.string, String::L_CLEAN))); + 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.clear(); + 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; }