--- parser3/src/main/compile.y 2001/06/27 10:07:53 1.139 +++ parser3/src/main/compile.y 2001/10/10 12:41:29 1.171 @@ -1,11 +1,11 @@ +%{ /** @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 1.139 2001/06/27 10:07:53 parser Exp $ + $Id: compile.y,v 1.171 2001/10/10 12:41:29 parser Exp $ */ /** @@ -19,7 +19,6 @@ -#: in iis make up specialized Pool object for that */ -%{ #define YYSTYPE Array/**/ * #define YYPARSE_PARAM pc #define YYLEX_PARAM pc @@ -34,11 +33,10 @@ #include "pa_vobject.h" #include "pa_vdouble.h" #include "pa_globals.h" -#include "pa_vnothing.h" +#include "pa_vvoid.h" #define SELF_ELEMENT_NAME "self" #define USE_CONTROL_METHOD_NAME "USE" -#define END_CONTROL_METHOD_NAME "end" static int real_yyerror(parse_control *pc, char *s); static void yyprint(FILE *file, int type, YYSTYPE value); @@ -61,6 +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 "||" @@ -105,7 +104,7 @@ static int yylex(YYSTYPE *lvalp, void *p %left NEG /* negation: unary - */ %% -all: +all: one_big_piece { Method& method=*NEW Method(POOL, *main_method_name, @@ -126,8 +125,6 @@ control_method: '@' STRING '\n' maybe_control_strings { const String& command=*LA2S($2); YYSTYPE strings_code=$4; - if(command==END_CONTROL_METHOD_NAME && strings_code->size()==0) - break; if(strings_code->size()<1*2) { strcpy(PC.error, "@"); strcat(PC.error, command.cstr()); @@ -135,7 +132,7 @@ control_method: '@' STRING '\n' YYERROR; } if(command==CLASS_NAME) { - if(PC.cclass) { // already changed from default? + if(PC.cclass->base()) { // already changed from default? strcpy(PC.error, "class already have a name '"); strncat(PC.error, PC.cclass->name().cstr(), 100); strcat(PC.error, "'"); @@ -155,10 +152,9 @@ control_method: '@' STRING '\n' } } else if(command==USE_CONTROL_METHOD_NAME) { for(int i=0; isize(); i+=2) - PC.request->use_file( - PC.request->absolute(*LA2S(strings_code, i))); + PC.request->use_file(*LA2S(strings_code, i)); } else if(command==BASE_NAME) { - if(PC.cclass->base()!=&PC.request->OP) { // already changed from default? + if(PC.cclass->base()) { // already changed from default? strcpy(PC.error, "class already have a base '"); strncat(PC.error, PC.cclass->base()->name().cstr(), 100); strcat(PC.error, "'"); @@ -238,7 +234,7 @@ maybe_codes: empty | codes; codes: code | codes code { $$=$1; P($$, $2) }; code: write_string | action; -action: get | put | with | call; +action: get | put | call; /* get */ @@ -246,12 +242,11 @@ get: get_value { $$=$1; /* stack: resulting value */ O($$, 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 '}' { $$=$2 }; name_without_curly_rdive: name_without_curly_rdive_read -| name_without_curly_rdive_root | name_without_curly_rdive_class; name_without_curly_rdive_read: name_without_curly_rdive_code { $$=N(POOL); @@ -268,11 +263,6 @@ name_without_curly_rdive_read: name_with } /* diving code; stack: current context */ }; -name_without_curly_rdive_root: ':' name_without_curly_rdive_code { - $$=N(POOL); - O($$, OP_WITH_ROOT); /* stack: starting context */ - P($$, $2); /* diving code; stack: current context */ -}; name_without_curly_rdive_class: class_prefix name_without_curly_rdive_code { $$=$1; P($$, $2) }; name_without_curly_rdive_code: name_advance2 | name_path name_advance2 { $$=$1; P($$, $2) }; @@ -283,10 +273,10 @@ put: '$' name_expr_wdive construct { P($$, $3); /* stack: context,name,constructor_value */ }; name_expr_wdive: - name_expr_wdive_write -| name_expr_wdive_root + name_expr_wdive_root +| name_expr_wdive_write | name_expr_wdive_class; -name_expr_wdive_write: name_expr_dive_code { +name_expr_wdive_root: name_expr_dive_code { $$=N(POOL); Array *diving_code=$1; const String *first_name=LA2S(diving_code); @@ -296,31 +286,43 @@ name_expr_wdive_write: name_expr_dive_co /* skip over... */ diving_code->size()>2?3/*OP_+string+get_element*/:2/*OP_+string*/); } else { - O($$, OP_WITH_WRITE); /* stack: starting context */ + O($$, OP_WITH_ROOT); /* stack: starting context */ P($$, diving_code); } /* diving code; stack: current context */ }; -name_expr_wdive_root: ':' name_expr_dive_code { +name_expr_wdive_write: '.' name_expr_dive_code { $$=N(POOL); - O($$, OP_WITH_ROOT); /* stack: starting context */ + O($$, OP_WITH_WRITE); /* stack: starting context */ P($$, $2); /* diving code; stack: context,name */ }; name_expr_wdive_class: class_prefix name_expr_dive_code { $$=$1; P($$, $2) }; -construct: construct_by_code | construct_by_expr; -construct_by_code: '[' any_constructor_code_value ']' { - $$=$2; /* stack: context, name, value */ +construct: + construct_square +| construct_round +| construct_curly +; +construct_square: '[' any_constructor_code_value ']' { + // stack: context, name + $$=$2; // stack: context, name, value O($$, OP_CONSTRUCT_VALUE); /* value=pop; name=pop; context=pop; construct(context,name,value) */ } ; -construct_by_expr: '(' expr_value ')' { - $$=$2; /* stack: context, name, value */ - O($$, OP_CONSTRUCT_EXPR); /* value=pop; name=pop; context=pop; construct(context,name,value) */ +construct_round: '(' expr_value ')' { + // stack: context, name + $$=$2; // stack: context, name, value + O($$, OP_CONSTRUCT_EXPR); /* value=pop->as_expr_result; name=pop; context=pop; construct(context,name,value) */ } ; +construct_curly: '{' maybe_codes '}' { + // stack: context, name + $$=N(POOL); + CCA($$, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ +}; + any_constructor_code_value: - unknown_value /* optimized $var[] case */ + void_value /* optimized $var[] case */ | STRING /* optimized $var[STRING] case */ | constructor_code_value /* $var[something complex] */ ; @@ -339,14 +341,16 @@ call: call_value { $$=$1; /* stack: value */ O($$, OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ }; -call_value: '^' call_name store_params EON { /* ^field.$method{vasya} */ - $$=$2; /* with_xxx,diving code; stack: context,method_junction */ +call_value: '^' { PC.object_constructor_allowed=true } + call_name { PC.object_constructor_allowed=false } + store_params EON { /* ^field.$method{vasya} */ + $$=$3; /* with_xxx,diving code; stack: context,method_junction */ O($$, OP_GET_METHOD_FRAME); /* stack: context,method_frame */ - YYSTYPE params_code=$3; - if(params_code->size()==3) // probably [] case. [OP_VALUE + Nothing + STORE_PARAM] + YYSTYPE params_code=$5; + if(params_code->size()==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 VNothing? + if(!value->is_defined()) // value is VVoid? params_code=0; // ^zzz[] case. don't append lone empty param. if(params_code) P($$, params_code); // filling method_frame.store_params @@ -389,7 +393,7 @@ store_curly_param_part: maybe_codes { PCA($$, $1); }; code_param_value: - unknown_value /* optimized [;...] case */ + void_value /* optimized [;...] case */ | STRING /* optimized [STRING] case */ | constructor_code_value /* [something complex] */ ; @@ -419,7 +423,8 @@ name_advance2: name_expr_value { 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[$...] */ +| name_expr_with_subvar_value /* xxx$part_of_subname_is_var_value */ +| name_square_code_value /* [codes] */ ; name_expr_subvar_value: '$' subvar_ref_name_rdive { $$=$2; @@ -433,16 +438,16 @@ name_expr_with_subvar_value: STRING subv P($$, $2); O($$, OP_REDUCE_EWPOOL); }; -subvar_ref_name_rdive: subvar_ref_name_rdive_read | subvar_ref_name_rdive_root; -subvar_ref_name_rdive_read: STRING { +name_square_code_value: '[' codes ']' { $$=N(POOL); - O($$, OP_WITH_READ); - P($$, $1); + O($$, OP_CREATE_EWPOOL); + P($$, $2); + O($$, OP_REDUCE_EWPOOL); }; -subvar_ref_name_rdive_root: ':' STRING { +subvar_ref_name_rdive: STRING { $$=N(POOL); - O($$, OP_WITH_ROOT); - P($$, $2); + O($$, OP_WITH_READ); + P($$, $1); }; subvar_get_writes: subvar__get_write | subvar_get_writes subvar__get_write { $$=$1; P($$, $2) }; subvar__get_write: '$' subvar_ref_name_rdive { @@ -450,22 +455,24 @@ subvar__get_write: '$' subvar_ref_name_r O($$, OP_GET_ELEMENT__WRITE); }; -class_prefix: STRING ':' { +class_prefix: + class_static_prefix +| class_constructor_prefix +; +class_static_prefix: STRING ':' { $$=$1; // stack: class name string O($$, OP_GET_CLASS); }; - - -/* with */ - -with: '$' name_without_curly_rdive '{' maybe_codes '}' { - $$=$2; - O($$, OP_CREATE_RWPOOL); - P($$, $4); - O($$, OP_REDUCE_RWPOOL); - O($$, OP_WRITE_VALUE); +class_constructor_prefix: class_static_prefix ':' { + $$=$1; + if(!PC.object_constructor_allowed) { + strcpy(PC.error, ":: not allowed here"); + YYERROR; + } + O($$, OP_PREPARE_TO_CONSTRUCT_OBJECT); }; + /* expr */ expr_value: expr { @@ -477,9 +484,11 @@ expr: | get_value | call_value | '"' string_inside_quotes_value '"' { $$ = $2; } +| '\'' string_inside_quotes_value '\'' { $$ = $2; } | '(' 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) } @@ -527,7 +536,7 @@ write_string: STRING { change_string_literal_to_write_string_literal($$=$1) }; -unknown_value: /* empty */ { $$=VL(NEW VNothing(POOL)) }; +void_value: /* empty */ { $$=VL(NEW VVoid(POOL)) }; empty: /* empty */ { $$=N(POOL) }; %% @@ -568,6 +577,7 @@ 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++; @@ -575,9 +585,28 @@ 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_USER) { + push_LS(PC, LS_DEF_NAME); + RC; + } else // @ in first column inside some code + result=BAD_METHOD_DECL_START; + goto break2; + } else if(c=='^') + switch(PC.ls) { +case LS_EXPRESSION_VAR_NAME_WITH_COLON: +case LS_EXPRESSION_VAR_NAME_WITHOUT_COLON: +case LS_VAR_NAME_SIMPLE_WITH_COLON: +case LS_VAR_NAME_SIMPLE_WITHOUT_COLON: +case LS_VAR_NAME_CURLY: +case LS_METHOD_NAME: +case LS_COMMENT: +case LS_DEF_COMMENT: + // no literals in names, please + break; +default: switch(*PC.source) { - // escaping: ^^ & co + // ^escaping some punctuators case '^': case '$': case ';': case '(': case ')': case '[': case ']': @@ -588,12 +617,19 @@ static int yylex(YYSTYPE *lvalp, void *p PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); } // reset piece 'begin' position & line - begin=PC.source; // ^ + end=begin=PC.source; // ^ begin_line=PC.line; - // skip over ^ and _ - PC.source++; PC.col++; - // skip analysis = forced literal - continue; + if(PC.ls==LS_METHOD_AFTER) { + pop_LS(PC); + result=EON; + skip_analized=-1; // return to ^ afterwards to assure it's literality + goto break2; + } else { + // skip over _ after ^ + PC.source++; PC.col++; + // skip analysis = forced literal + continue; + } // converting ^#HH into char(hex(HH)) case '#': @@ -623,6 +659,8 @@ static int yylex(YYSTYPE *lvalp, void *p } break; } + break; + } // #comment start skipping if(c=='#' && PC.col==1) { if(end!=begin) { @@ -636,18 +674,33 @@ static int yylex(YYSTYPE *lvalp, void *p // USER'S = NOT OURS case LS_USER: + case LS_NAME_SQUARE_PART: // name.[here].xxx + if(PC.trim_bof) + switch(c) { + case '\n': case ' ': case '\t': + begin=PC.source; + begin_line=PC.line; + continue; // skip it + default: + PC.trim_bof=false; + } switch(c) { case '$': - 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; - } + 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 '[': // $name.[co<[>de] + if(PC.ls==LS_NAME_SQUARE_PART) + lexical_brackets_nestage++; break; } break; @@ -665,13 +718,20 @@ static int yylex(YYSTYPE *lvalp, void *p break; // STRING IN EXPRESSION - case LS_EXPRESSION_STRING: + case LS_EXPRESSION_STRING_QUOTED: + case LS_EXPRESSION_STRING_APOSTROFED: switch(c) { case '"': - pop_LS(PC); //"abc". - RC; + case '\'': + if( + PC.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || + PC.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { + pop_LS(PC); //"abc". | 'abc'. + RC; + } + break; case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -726,6 +786,11 @@ 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 @@ -748,7 +813,7 @@ 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); @@ -795,7 +860,10 @@ static int yylex(YYSTYPE *lvalp, void *p result=c; goto break2; case '"': - push_LS(PC, LS_EXPRESSION_STRING); + push_LS(PC, LS_EXPRESSION_STRING_QUOTED); + RC; + case '\'': + push_LS(PC, LS_EXPRESSION_STRING_APOSTROFED); RC; case 'l': case 'g': case 'e': case 'n': if(end==begin) // right after whitespace @@ -855,9 +923,13 @@ static int yylex(YYSTYPE *lvalp, void *p break; // VARIABLE GET/PUT/WITH - case LS_VAR_NAME_SIMPLE: - case LS_EXPRESSION_VAR_NAME: - if(PC.ls==LS_EXPRESSION_VAR_NAME) { + 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 binary operators switch(c) { case '-': @@ -867,6 +939,18 @@ static int yylex(YYSTYPE *lvalp, void *p goto break2; } } + 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 ':': + pop_LS(PC); + PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } + result=EON; + goto break2; + } + } switch(c) { case 0: case ' ': case '\t': case '\n': @@ -878,7 +962,7 @@ static int yylex(YYSTYPE *lvalp, void *p case '&': case '|': case '=': case '!': // common delimiters - case ',': + case ',': case '?': // before call case '^': pop_LS(PC); @@ -886,6 +970,16 @@ static int yylex(YYSTYPE *lvalp, void *p result=EON; goto break2; case '[': + // $name.<[>code] + if(PC.col>1/*not first column*/ && ( + end[-1]=='$'/*was start of get*/ || + end[-1]==':'/*was class name delim */ || + end[-1]=='.'/*was name delim */ + )) { + push_LS(PC, LS_NAME_SQUARE_PART); + lexical_brackets_nestage=1; + RC; + } PC.ls=LS_VAR_SQUARE; lexical_brackets_nestage=1; RC; @@ -904,13 +998,24 @@ static int yylex(YYSTYPE *lvalp, void *p RC; case '.': // name part delim case '$': // name part subvar - case ':': // ':name' or 'class:name' + 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; + // ...stop before next ':' RC; } break; case LS_VAR_NAME_CURLY: switch(c) { + case '[': + // ${name.<[>code]} + push_LS(PC, LS_NAME_SQUARE_PART); + lexical_brackets_nestage=1; + RC; case '}': // ${name} finished, restoring LS pop_LS(PC); RC; @@ -924,7 +1029,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -946,7 +1051,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -967,6 +1072,16 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_NAME: switch(c) { case '[': + // ^name.<[>code].xxx + if(PC.col>1/*not first column*/ && ( + end[-1]=='^'/*was start of call*/ || // never, ^[ is literal... + end[-1]==':'/*was class name delim */ || + end[-1]=='.'/*was name delim */ + )) { + push_LS(PC, LS_NAME_SQUARE_PART); + lexical_brackets_nestage=1; + RC; + } PC.ls=LS_METHOD_SQUARE; lexical_brackets_nestage=1; RC; @@ -981,6 +1096,8 @@ static int yylex(YYSTYPE *lvalp, void *p case '.': // name part delim case '$': // name part subvar case ':': // ':name' or 'class:name' + case '^': // ^abc^xxx wrong. bailing out + case ']': case '}': case ')': // ^abc]}) wrong. bailing out RC; } break; @@ -988,7 +1105,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1010,7 +1127,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME);