--- parser3/src/main/compile.y 2001/09/18 16:05:42 1.166 +++ parser3/src/main/compile.y 2002/10/15 14:00:44 1.200 @@ -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 1.200 2002/10/15 14:00:44 paf Exp $ */ -%{ -static char *RCSId="$Id: compile.y,v 1.166 2001/09/18 16:05:42 parser Exp $"; /** @todo parser4: @@ -34,8 +34,8 @@ static char *RCSId="$Id: compile.y,v 1.1 #include "pa_vdouble.h" #include "pa_globals.h" #include "pa_vvoid.h" +#include "pa_vmethod_frame.h" -#define SELF_ELEMENT_NAME "self" #define USE_CONTROL_METHOD_NAME "USE" static int real_yyerror(parse_control *pc, char *s); @@ -59,15 +59,21 @@ static int yylex(YYSTYPE *lvalp, void *p %token BAD_STRING_COMPARISON_OPERATOR %token BAD_HEX_LITERAL +%token BAD_METHOD_DECL_START +%token BAD_METHOD_PARAMETER_NAME_CHARACTER +%token BAD_MATH_OPERATOR_CHARACTER %token LAND "&&" %token LOR "||" -%token LXOR "##" +%token LXOR "!||" +%token NXOR "!|" %token NLE "<=" %token NGE ">=" %token NEQ "==" %token NNE "!=" +%token NSL "<<" +%token NSR ">>" %token SLT "lt" %token SGT "gt" @@ -83,7 +89,7 @@ static int yylex(YYSTYPE *lvalp, void *p %token IS "is" /* logical */ -%left "##" +%left "!||" %left "||" %left "&&" %left '<' '>' "<=" ">=" "lt" "gt" "le" "ge" @@ -92,26 +98,28 @@ static int yylex(YYSTYPE *lvalp, void *p %left '!' /* bitwise */ -%left '#' +%left "!|" %left '|' %left '&' %left '~' +%left ">>" +%left "<<" /* numerical */ %left '-' '+' -%left '*' '/' '%' +%left '*' '/' '%' '\\' %left NEG /* negation: unary - */ %% -all: +all: one_big_piece { Method& method=*NEW Method(POOL, - *main_method_name, + PC.request->main_method_name, 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.request->main_method_name, method); } | methods; @@ -131,7 +139,7 @@ control_method: '@' STRING '\n' YYERROR; } if(command==CLASS_NAME) { - if(PC.cclass->base()) { // already changed from default? + 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, "'"); @@ -151,29 +159,30 @@ control_method: '@' STRING '\n' } } else if(command==USE_CONTROL_METHOD_NAME) { for(int i=0; isize(); i+=2) - PC.request->use_file(*LA2S(strings_code, i)); + PC.request->use_file(PC.request->main_class, *LA2S(strings_code, i)); } else if(command==BASE_NAME) { - if(PC.cclass->base()) { // already changed from default? + if(PC.cclass->base_class()) { // already changed from default? strcpy(PC.error, "class already have a base '"); - strncat(PC.error, PC.cclass->base()->name().cstr(), 100); + strncat(PC.error, PC.cclass->base_class()->name().cstr(), 100); strcat(PC.error, "'"); YYERROR; } if(strings_code->size()==1*2) { const String& base_name=*LA2S(strings_code); - VClass *base=static_cast( + Value *vbase_class=static_cast( PC.request->classes().get(base_name)); - if(!base) { + VStateless_class *base_class=vbase_class?vbase_class->get_class():0; + if(!base_class) { strcpy(PC.error, base_name.cstr()); strcat(PC.error, ": undefined class in @"BASE_NAME); YYERROR; } // @CLASS == @BASE sanity check - if(PC.cclass==base) { + if(PC.cclass==base_class) { strcpy(PC.error, "@"CLASS_NAME" equals @"BASE_NAME); YYERROR; } - PC.cclass->set_base(*base); + PC.cclass->set_base(base_class); } else { strcpy(PC.error, "@"BASE_NAME" must contain sole name"); YYERROR; @@ -238,8 +247,11 @@ 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, false, /*->*/OP_GET_ELEMENT__WRITE, + /*or */OP_WRITE_VALUE + ); /* value=pop; wcontext.write(value) */ }; get_value: '$' get_name_value { $$=$2 }; get_name_value: name_without_curly_rdive EON | name_in_curly_rdive; @@ -251,13 +263,24 @@ name_without_curly_rdive_read: name_with $$=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($$, OP_WITH_READ); /* stack: starting context */ + + // OP_VALUE+string+OP_GET_ELEMENT. -> OP_VALUE+string+...OR_OPERATOR/OR_JUNCTION_EXP. + if(diving_code->size()==3) // ELEMENT without .SUBELEMENT + diving_code->put_int(2, + PC.in_call_value? + OP_GET_ELEMENT_OR_OPERATOR// possibly ^if [search for operator] + :OP_GET_ELEMENT_OR_JUNCTION_EXPAND // possibly $junction [if junction, expand it] + ); + P($$, diving_code); } /* diving code; stack: current context */ @@ -279,11 +302,13 @@ 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($$, diving_code); @@ -309,15 +334,17 @@ construct_square: '[' any_constructor_co } ; construct_round: '(' expr_value ')' { + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); // stack: context, name - $$=$2; // stack: 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 '}' { // stack: context, name $$=N(POOL); - CCA($$, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ + OA($$, OP_CURLY_CODE__CONSTRUCT, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ }; any_constructor_code_value: @@ -327,9 +354,9 @@ any_constructor_code_value: ; constructor_code_value: constructor_code { $$=N(POOL); - O($$, OP_CREATE_EWPOOL); /* stack: empty write context */ - P($$, $1); /* some code that writes to that context */ - O($$, OP_REDUCE_EWPOOL); /* context=pop; stack: context.value() */ + OA($$, 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) }; @@ -338,22 +365,27 @@ codes__excluding_sole_str_literal: actio 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 } + changetail_or_append($$, + OP_CALL, true, /*->*/ OP_CALL__WRITE, + /*or */OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ +}; +call_value: '^' { + PC.in_call_value=true; + } + call_name { + PC.in_call_value=false; + } store_params EON { /* ^field.$method{vasya} */ $$=$3; /* with_xxx,diving code; stack: context,method_junction */ - O($$, OP_GET_METHOD_FRAME); /* stack: context,method_frame */ YYSTYPE params_code=$5; - if(params_code->size()==3) // probably [] case. [OP_VALUE + Void + STORE_PARAM] + if(params_code->size()==3) { // probably [] case. [OP_VALUE + Void + STORE_PARAM] if(Value *value=LA2V(params_code)) // it is OP_VALUE + value? if(!value->is_defined()) // value is VVoid? params_code=0; // ^zzz[] case. don't append lone empty param. - if(params_code) - P($$, params_code); // filling method_frame.store_params - O($$, OP_CALL); // method_frame=pop; ncontext=pop; call(ncontext,method_frame) stack: value + } + /* stack: context, method_junction */ + OA($$, OP_CALL, params_code); // method_frame=make frame(pop junction); ncontext=pop; call(ncontext,method_frame) stack: value }; call_name: name_without_curly_rdive; @@ -385,11 +417,11 @@ store_code_param_part: code_param_value }; store_expr_param_part: write_expr_value { $$=N(POOL); - PEA($$, $1); + OA($$, OP_EXPR_CODE__STORE_PARAM, $1); }; store_curly_param_part: maybe_codes { $$=N(POOL); - PCA($$, $1); + OA($$, OP_CURLY_CODE__STORE_PARAM, $1); }; code_param_value: void_value /* optimized [;...] case */ @@ -397,7 +429,9 @@ code_param_value: | constructor_code_value /* [something complex] */ ; write_expr_value: expr_value { - $$=$1; + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); + P($$, $1); O($$, OP_WRITE_EXPR_RESULT); }; @@ -408,6 +442,9 @@ name_expr_dive_code: name_expr_value | n name_path: name_step | name_path name_step { $$=$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($$, OP_GET_ELEMENT); /* name=pop; context=pop; stack: context.get_element(name) */ @@ -430,18 +467,19 @@ name_expr_subvar_value: '$' subvar_ref_n O($$, 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); + OA($$, OP_STRING_POOL, code); }; name_square_code_value: '[' codes ']' { $$=N(POOL); - O($$, OP_CREATE_EWPOOL); - P($$, $2); - O($$, OP_REDUCE_EWPOOL); + OA($$, OP_OBJECT_POOL, $2); /* stack: empty write context */ + /* some code that writes to that context */ + /* context=pop; stack: context.value() */ }; subvar_ref_name_rdive: STRING { $$=N(POOL); @@ -460,11 +498,19 @@ class_prefix: ; class_static_prefix: STRING ':' { $$=$1; // stack: class name string + if(*LA2S($$) == BASE_NAME) { // pseude BASE class + if(VStateless_class *base=PC.cclass->base_class()) { + change_string_literal_value($$, base->name()); + } else { + strcpy(PC.error, "no base class declared"); + YYERROR; + } + } O($$, OP_GET_CLASS); }; class_constructor_prefix: class_static_prefix ':' { $$=$1; - if(!PC.object_constructor_allowed) { + if(!PC.in_call_value) { strcpy(PC.error, ":: not allowed here"); YYERROR; } @@ -475,6 +521,7 @@ 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 }; @@ -500,12 +547,15 @@ 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($$, OP_BIN_SL) } +| expr ">>" expr { $$=$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; P($$, $3); O($$, OP_LOG_AND) } -| expr "||" expr { $$=$1; P($$, $3); O($$, OP_LOG_OR) } -| expr "##" expr { $$=$1; P($$, $3); O($$, OP_LOG_XOR) } +| 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($$, OP_NUM_LE) } @@ -523,9 +573,9 @@ expr: string_inside_quotes_value: maybe_codes { $$=N(POOL); - O($$, OP_CREATE_SWPOOL); /* stack: empty write context */ - P($$, $1); /* some code that writes to that context */ - O($$, OP_REDUCE_SWPOOL); /* context=pop; stack: context.get_string() */ + OA($$, OP_STRING_POOL, $1); /* stack: empty write context */ + /* some code that writes to that context */ + /* context=pop; stack: context.get_string() */ }; /* basics */ @@ -558,7 +608,7 @@ 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; @@ -584,14 +634,39 @@ 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); @@ -639,6 +714,8 @@ static int yylex(YYSTYPE *lvalp, void *p } break; } + break; + } // #comment start skipping if(c=='#' && PC.col==1) { if(end!=begin) { @@ -646,7 +723,8 @@ static int yylex(YYSTYPE *lvalp, void *p PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); } // fall into COMMENT lexical state [wait for \n] - push_LS(PC, LS_COMMENT); + push_LS(PC, LS_USER_COMMENT); + continue; } switch(PC.ls) { @@ -669,12 +747,6 @@ static int yylex(YYSTYPE *lvalp, void *p case '^': push_LS(PC, LS_METHOD_NAME); RC; - case '@': - if(PC.col==0+1) { - push_LS(PC, LS_DEF_NAME); - RC; - } - break; case ']': if(PC.ls==LS_NAME_SQUARE_PART) if(--lexical_brackets_nestage==0) {// $name.[co<]?>de<]?> @@ -690,7 +762,7 @@ static int yylex(YYSTYPE *lvalp, void *p break; // #COMMENT - case LS_COMMENT: + case LS_USER_COMMENT: if(c=='\n') { // skip comment begin=PC.source; @@ -737,6 +809,9 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_DEF_PARAMS: switch(c) { + case '$': // common error + result=BAD_METHOD_PARAMETER_NAME_CHARACTER; + goto break2; case ';': RC; case ']': @@ -770,19 +845,8 @@ 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) @@ -796,6 +860,15 @@ static int yylex(YYSTYPE *lvalp, void *p else // PC.ls==LS_VAR_ROUND // variable constructor ended pop_LS(PC); // return to normal life RC; + case '#': // comment start skipping + if(end!=begin) { + // append piece till # + PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + } + // fall into COMMENT lexical state [wait for \n] + push_LS(PC, LS_EXPRESSION_COMMENT); + lexical_brackets_nestage=1; + continue; case '$': push_LS(PC, LS_EXPRESSION_VAR_NAME_WITH_COLON); RC; @@ -815,34 +888,68 @@ static int yylex(YYSTYPE *lvalp, void *p skip_analized=1; result=DEXISTS; goto break2; - default: + default: // minus result=c; goto break2; } goto break2; - case '+': case '*': case '/': case '%': + case '+': case '*': case '/': case '%': case '\\': case '~': case ';': RC; - case '&': case '|': case '#': + case '&': case '|': if(*PC.source==c) { // && || - result=c=='#'?LXOR:c=='&'?LAND:LOR; + result=c=='&'?LAND:LOR; skip_analized=1; } else result=c; goto break2; - case '<': case '>': case '=': case '!': - if(*PC.source=='=') { // <= >= == != + case '!': + switch(PC.source[0]) { + case '|': // !| !|| skip_analized=1; - switch(c) { - case '<': result=NLE; break; - case '>': result=NGE; break; - case '=': result=NEQ; break; - case '!': result=NNE; break; - } - } else - result=c; + if(PC.source[1]=='|') { + skip_analized++; + result=LXOR; + } else + result=NXOR; + goto break2; + case '=': // != + skip_analized=1; + result=NNE; + goto break2; + } + RC; + + case '<': // <<, <=, < + switch(*PC.source) { + case '<': // <[<] + skip_analized=1; result=NSL; break; + case '=': // <[=] + skip_analized=1; result=NLE; break; + default: // <[] + result=c; break; + } + goto break2; + case '>': // >>, >=, > + switch(*PC.source) { + case '>': // >[>] + skip_analized=1; result=NSR; break; + case '=': // >[=] + skip_analized=1; result=NGE; break; + default: // >[] + result=c; break; + } goto break2; + case '=': // == + switch(*PC.source) { + case '=': // =[=] + skip_analized=1; result=NEQ; break; + default: // =[] + result=c; break; // not used now + } + goto break2; + case '"': push_LS(PC, LS_EXPRESSION_STRING_QUOTED); RC; @@ -905,6 +1012,24 @@ static int yylex(YYSTYPE *lvalp, void *p continue; } break; + case LS_EXPRESSION_COMMENT: + if(c=='(') + lexical_brackets_nestage++; + + switch(*PC.source) { + case '\n': case ')': + if(*PC.source==')') + if(--lexical_brackets_nestage!=0) + continue; + + // skip comment + begin=PC.source; + begin_line=PC.line; + + pop_LS(PC); + continue; + } + break; // VARIABLE GET/PUT/WITH case LS_VAR_NAME_SIMPLE_WITH_COLON: @@ -914,9 +1039,12 @@ static int yylex(YYSTYPE *lvalp, void *p 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 + // 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; @@ -946,7 +1074,7 @@ static int yylex(YYSTYPE *lvalp, void *p case '&': case '|': case '=': case '!': // common delimiters - case ',': case '?': + case ',': case '?': case '#': // before call case '^': pop_LS(PC); @@ -1080,6 +1208,8 @@ 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; @@ -1163,7 +1293,7 @@ break2: if(end!=begin && end[-1]=='\n') // allow one empty line before LS_DEF_NAME end--; } - if(end!=begin && PC.ls!=LS_COMMENT) { // last piece still alive and not comment? + if(end!=begin && PC.ls!=LS_USER_COMMENT) { // last piece still alive and not comment? // append it PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line/*, start_col*/); }