--- parser3/src/main/compile.y 2001/08/10 13:03:05 1.162 +++ parser3/src/main/compile.y 2002/04/15 08:13:09 1.184 @@ -1,12 +1,12 @@ +%{ /** @file Parser: compiler(lexical parser and grammar). Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) + + $Id: compile.y,v 1.184 2002/04/15 08:13:09 paf Exp $ */ -%{ -static char *RCSId="$Id: compile.y,v 1.162 2001/08/10 13:03:05 parser Exp $"; /** @todo parser4: @@ -59,6 +59,7 @@ static int yylex(YYSTYPE *lvalp, void *p %token BAD_STRING_COMPARISON_OPERATOR %token BAD_HEX_LITERAL +%token BAD_METHOD_DECL_START %token LAND "&&" %token LOR "||" @@ -99,7 +100,7 @@ static int yylex(YYSTYPE *lvalp, void *p /* numerical */ %left '-' '+' -%left '*' '/' '%' +%left '*' '/' '%' '\\' %left NEG /* negation: unary - */ %% @@ -238,10 +239,13 @@ action: get | put | call; /* get */ get: get_value { - $$=$1; /* stack: resulting value */ - O($$, OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ + $$=$1; /* stack: resulting value */ + changetail_or_append($$, + OP_GET_ELEMENT, OP_GET_ELEMENT__WRITE, + OP_WRITE_VALUE + ); /* value=pop; wcontext.write(value) */ }; -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 '}' { $$=$2 }; name_without_curly_rdive: @@ -251,13 +255,20 @@ name_without_curly_rdive_read: name_with $$=N(POOL); Array *diving_code=$1; const String *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->size()>2?3/*OP_+string+get_element*/:2/*OP_+string*/); + diving_code->size()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); } else { O($$, OP_WITH_READ); /* stack: starting context */ + + // ^if ELEMENT -> ^if ELEMENT_OR_OPERATOR + // OP_VALUE+string+OP_GET_ELEMENT. -> OP_VALUE+string+OP_GET_ELEMENT_OR_OPERATOR. + if(PC.in_call_value && diving_code->size()==3) + diving_code->put_int(2, OP_GET_ELEMENT_OR_OPERATOR); P($$, diving_code); } /* diving code; stack: current context */ @@ -279,11 +290,13 @@ name_expr_wdive_root: name_expr_dive_cod $$=N(POOL); Array *diving_code=$1; const String *first_name=LA2S(diving_code); + // $self.xxx... -> $xxx... + // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { O($$, OP_WITH_SELF); /* stack: starting context */ P($$, diving_code, /* skip over... */ - diving_code->size()>2?3/*OP_+string+get_element*/:2/*OP_+string*/); + diving_code->size()>=3?3/*OP_VALUE+string+OP_GET_ELEMENTx*/:2/*OP_+string*/); } else { O($$, OP_WITH_ROOT); /* stack: starting context */ P($$, diving_code); @@ -298,9 +311,9 @@ name_expr_wdive_write: '.' name_expr_div name_expr_wdive_class: class_prefix name_expr_dive_code { $$=$1; P($$, $2) }; construct: - construct_square | - construct_round | - construct_curly + construct_square +| construct_round +| construct_curly ; construct_square: '[' any_constructor_code_value ']' { // stack: context, name @@ -309,9 +322,11 @@ construct_square: '[' any_constructor_co } ; construct_round: '(' expr_value ')' { + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); // stack: context, name - $$=$2; // stack: context, name, value - O($$, OP_CONSTRUCT_EXPR); /* value=pop; name=pop; context=pop; construct(context,name,value) */ + P($$, $2); // stack: context, name, value + O($$, OP_CONSTRUCT_EXPR); /* value=pop->as_expr_result; name=pop; context=pop; construct(context,name,value) */ } ; construct_curly: '{' maybe_codes '}' { @@ -327,9 +342,9 @@ any_constructor_code_value: ; constructor_code_value: constructor_code { $$=N(POOL); - O($$, OP_CREATE_EWPOOL); /* stack: empty write context */ - P($$, $1); /* some code that writes to that context */ - O($$, OP_REDUCE_EWPOOL); /* context=pop; stack: context.value() */ + PJP($$, $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) }; @@ -340,8 +355,12 @@ call: call_value { $$=$1; /* stack: value */ O($$, OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ }; -call_value: '^' { PC.object_constructor_allowed=true } - call_name { PC.object_constructor_allowed=false } +call_value: '^' { + PC.in_call_value=true; + } + call_name { + PC.in_call_value=false; + } store_params EON { /* ^field.$method{vasya} */ $$=$3; /* with_xxx,diving code; stack: context,method_junction */ O($$, OP_GET_METHOD_FRAME); /* stack: context,method_frame */ @@ -397,7 +416,9 @@ code_param_value: | constructor_code_value /* [something complex] */ ; write_expr_value: expr_value { - $$=$1; + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); + P($$, $1); O($$, OP_WRITE_EXPR_RESULT); }; @@ -408,6 +429,9 @@ name_expr_dive_code: name_expr_value | n name_path: name_step | name_path name_step { $$=$1; P($$, $2) }; name_step: name_advance1 '.'; name_advance1: name_expr_value { + // we know that name_advance1 not called from ^xxx context + // so we'll not check for operator call possibility as we do in name_advance2 + /* stack: context */ $$=$1; /* stack: context,name */ O($$, OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ @@ -423,25 +447,26 @@ name_expr_value: STRING /* subname_is_const */ | name_expr_subvar_value /* $subname_is_var_value */ | name_expr_with_subvar_value /* xxx$part_of_subname_is_var_value */ -| name_curly_code_value /* (codes) */ +| name_square_code_value /* [codes] */ ; name_expr_subvar_value: '$' subvar_ref_name_rdive { $$=$2; O($$, OP_GET_ELEMENT); }; name_expr_with_subvar_value: STRING subvar_get_writes { + Array *code; + { + change_string_literal_to_write_string_literal(code=$1); + P(code, $2); + } $$=N(POOL); - O($$, OP_CREATE_EWPOOL); - P($$, $1); - O($$, OP_WRITE_VALUE); - P($$, $2); - O($$, OP_REDUCE_EWPOOL); + PSP($$, code); }; -name_curly_code_value: '[' codes ']' { +name_square_code_value: '[' codes ']' { $$=N(POOL); - O($$, OP_CREATE_EWPOOL); - P($$, $2); - O($$, OP_REDUCE_EWPOOL); + PJP($$, $2); /* stack: empty write context */ + /* some code that writes to that context */ + /* context=pop; stack: context.value() */ }; subvar_ref_name_rdive: STRING { $$=N(POOL); @@ -464,7 +489,7 @@ class_static_prefix: STRING ':' { }; class_constructor_prefix: class_static_prefix ':' { $$=$1; - if(!PC.object_constructor_allowed) { + if(!PC.in_call_value) { strcpy(PC.error, ":: not allowed here"); YYERROR; } @@ -475,6 +500,7 @@ class_constructor_prefix: class_static_p /* expr */ expr_value: expr { + // see OP_PREPARE_TO_EXPRESSION!! if(($$=$1)->size()==2) // only one string literal in there? change_string_literal_to_double_literal($$); // make that string literal Double }; @@ -487,6 +513,7 @@ expr: | '(' expr ')' { $$ = $2; } /* stack: operand // stack: @operand */ | '-' expr %prec NEG { $$=$2; O($$, OP_NEG) } +| '+' expr %prec NEG { $$=$2 } | '~' expr { $$=$2; O($$, OP_INV) } | '!' expr { $$=$2; O($$, OP_NOT) } | "def" expr { $$=$2; O($$, OP_DEF) } @@ -499,11 +526,12 @@ expr: | expr '*' expr { $$=$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_AND) } | expr '|' expr { $$=$1; P($$, $3); O($$, OP_BIN_OR) } | expr '#' expr { $$=$1; P($$, $3); O($$, OP_BIN_XOR) } -| expr "&&" expr { $$=$1; P($$, $3); O($$, OP_LOG_AND) } -| expr "||" expr { $$=$1; P($$, $3); O($$, OP_LOG_OR) } +| expr "&&" expr { $$=$1; PNC($$, $3); O($$, OP_LOG_AND) } +| expr "||" expr { $$=$1; PNC($$, $3); O($$, 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($$, OP_NUM_GT) } @@ -522,9 +550,9 @@ expr: string_inside_quotes_value: maybe_codes { $$=N(POOL); - O($$, OP_CREATE_SWPOOL); /* stack: empty write context */ - P($$, $1); /* some code that writes to that context */ - O($$, OP_REDUCE_SWPOOL); /* context=pop; stack: context.get_string() */ + PSP($$, $1); /* stack: empty write context */ + /* some code that writes to that context */ + /* context=pop; stack: context.get_string() */ }; /* basics */ @@ -557,7 +585,7 @@ empty: /* empty */ { $$=N(POOL) }; */ static int yylex(YYSTYPE *lvalp, void *pc) { - #define lexical_brackets_nestage PC.brackets_nestages[PC.sp] + #define lexical_brackets_nestage PC.brackets_nestages[PC.ls_sp] #define RC {result=c; goto break2; } register int c; @@ -575,6 +603,7 @@ static int yylex(YYSTYPE *lvalp, void *p 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); if(c=='\n') { PC.line++; @@ -582,25 +611,57 @@ static int yylex(YYSTYPE *lvalp, void *p } else PC.col++; - if(c=='^' && PC.ls!=LS_COMMENT && PC.ls!=LS_DEF_COMMENT) + if(c=='@' && PC.col==0+1) { + if(PC.ls==LS_DEF_SPECIAL_BODY) { + // @SPECIAL + // ... + // @APPEND_CLEAN(begin, end-begin, PC.file, begin_line); } // reset piece 'begin' position & line - begin=PC.source; // ^ + end=begin=PC.source; // ^ begin_line=PC.line; - // skip over ^ and _ - PC.source++; PC.col++; - // skip analysis = forced literal - continue; + if(PC.ls==LS_METHOD_AFTER) { + pop_LS(PC); + result=EON; + skip_analized=-1; // return to ^ afterwards to assure it's literality + goto break2; + } else { + // skip over _ after ^ + PC.source++; PC.col++; + // skip analysis = forced literal + continue; + } // converting ^#HH into char(hex(HH)) case '#': @@ -630,6 +691,8 @@ static int yylex(YYSTYPE *lvalp, void *p } break; } + break; + } // #comment start skipping if(c=='#' && PC.col==1) { if(end!=begin) { @@ -638,11 +701,13 @@ static int yylex(YYSTYPE *lvalp, void *p } // fall into COMMENT lexical state [wait for \n] push_LS(PC, LS_COMMENT); + continue; } switch(PC.ls) { // USER'S = NOT OURS case LS_USER: + case LS_NAME_SQUARE_PART: // name.[here].xxx if(PC.trim_bof) switch(c) { case '\n': case ' ': case '\t': @@ -654,25 +719,21 @@ static int yylex(YYSTYPE *lvalp, void *p } switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); RC; - case '@': - if(PC.col==0+1) { - push_LS(PC, LS_DEF_NAME); - RC; - } - break; - case ']': // $name.(code<)> - if(--lexical_brackets_nestage==0) { - pop_LS(PC); - RC; - } + case ']': + if(PC.ls==LS_NAME_SQUARE_PART) + if(--lexical_brackets_nestage==0) {// $name.[co<]?>de<]?> + pop_LS(PC); // $name.[co<]>de<]!> + RC; + } break; - case '[': - lexical_brackets_nestage++; + case '[': // $name.[co<[>de] + if(PC.ls==LS_NAME_SQUARE_PART) + lexical_brackets_nestage++; break; } break; @@ -703,7 +764,7 @@ static int yylex(YYSTYPE *lvalp, void *p } break; case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -758,19 +819,8 @@ static int yylex(YYSTYPE *lvalp, void *p break; case LS_DEF_SPECIAL_BODY: - // @todo in case - // ################ - // @next-method - // we are here with c=='@' - // which is wrong, and need action - if(c=='\n') { - switch(*PC.source) { - case '@': case 0: // end of special_code - pop_LS(PC); - break; - } + if(c=='\n') RC; - } break; // (EXPRESSION) @@ -785,7 +835,7 @@ static int yylex(YYSTYPE *lvalp, void *p pop_LS(PC); // return to normal life RC; case '$': - push_LS(PC, LS_EXPRESSION_VAR_NAME); + push_LS(PC, LS_EXPRESSION_VAR_NAME_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -808,7 +858,7 @@ static int yylex(YYSTYPE *lvalp, void *p goto break2; } goto break2; - case '+': case '*': case '/': case '%': + case '+': case '*': case '/': case '%': case '\\': case '~': case ';': RC; @@ -895,20 +945,28 @@ static int yylex(YYSTYPE *lvalp, void *p break; // VARIABLE GET/PUT/WITH - case LS_VAR_NAME_SIMPLE: - case LS_EXPRESSION_VAR_NAME: - case LS_VAR_NAME_NO_COLON: - if(PC.ls==LS_EXPRESSION_VAR_NAME) { - // name in expr ends also before binary operators + case LS_VAR_NAME_SIMPLE_WITH_COLON: + case LS_VAR_NAME_SIMPLE_WITHOUT_COLON: + 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) { + // 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; } result=EON; goto break2; } } - if(PC.ls==LS_VAR_NAME_NO_COLON) { + if( + 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 ':': @@ -939,10 +997,11 @@ static int yylex(YYSTYPE *lvalp, void *p case '[': // $name.<[>code] if(PC.col>1/*not first column*/ && ( - end[-1]=='.'/*was dot */ || - end[-1]=='$'/*was start of get*/ + end[-1]=='$'/*was start of get*/ || + end[-1]==':'/*was class name delim */ || + end[-1]=='.'/*was name delim */ )) { - push_LS(PC, LS_USER); + push_LS(PC, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; } @@ -965,7 +1024,12 @@ static int yylex(YYSTYPE *lvalp, void *p case '.': // name part delim case '$': // name part subvar case ':': // class<:>name - PC.ls=LS_VAR_NAME_NO_COLON; // stop before next ':' + // 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; + // ...stop before next ':' RC; } break; @@ -973,8 +1037,8 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_NAME_CURLY: switch(c) { case '[': - // ${name.<[>code)} - push_LS(PC, LS_USER); + // ${name.<[>code]} + push_LS(PC, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; case '}': // ${name} finished, restoring LS @@ -990,7 +1054,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1012,7 +1076,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1033,12 +1097,13 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_NAME: switch(c) { case '[': - // $name.<[>code) + // ^name.<[>code].xxx if(PC.col>1/*not first column*/ && ( - end[-1]=='.'/*was dot */ || - end[-1]=='^'/*was start of call*/ + end[-1]=='^'/*was start of call*/ || // never, ^[ is literal... + end[-1]==':'/*was class name delim */ || + end[-1]=='.'/*was name delim */ )) { - push_LS(PC, LS_USER); + push_LS(PC, LS_NAME_SQUARE_PART); lexical_brackets_nestage=1; RC; } @@ -1056,6 +1121,8 @@ static int yylex(YYSTYPE *lvalp, void *p case '.': // name part delim case '$': // name part subvar case ':': // ':name' or 'class:name' + case '^': // ^abc^xxx wrong. bailing out + case ']': case '}': case ')': // ^abc]}) wrong. bailing out RC; } break; @@ -1063,7 +1130,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1085,7 +1152,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME);