--- parser3/src/main/compile.y 2001/05/28 06:10:06 1.138 +++ parser3/src/main/compile.y 2003/04/07 07:07:30 1.205 @@ -1,11 +1,11 @@ +%{ /** @file Parser: compiler(lexical parser and grammar). - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - + Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) Author: Alexander Petrosyan (http://design.ru/paf) - $Id: compile.y,v 1.138 2001/05/28 06:10:06 parser Exp $ + $Id: compile.y,v 1.205 2003/04/07 07:07:30 paf 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" +#include "pa_vmethod_frame.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); @@ -46,6 +44,7 @@ static int yylex(YYSTYPE *lvalp, void *p // local convinient inplace typecast & var +#undef PC #define PC (*(parse_control *)pc) #define POOL (*PC.pool) #undef NEW @@ -61,15 +60,21 @@ 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" @@ -85,35 +90,38 @@ static int yylex(YYSTYPE *lvalp, void *p %token IS "is" /* logical */ -%left "##" +%left "!||" %left "||" %left "&&" %left '<' '>' "<=" ">=" "lt" "gt" "le" "ge" %left "==" "!=" "eq" "ne" %left "is" "def" "in" "-f" "-d" -%left '!' /* bitwise */ -%left '#' +%left "!|" %left '|' %left '&' -%left '~' +%left "<<" ">>" /* numerical */ -%left '-' '+' -%left '*' '/' '%' -%left NEG /* negation: unary - */ +%left '+' '-' +%left '*' '/' '\\' '%' +%left NUNARY /* unary - + */ + +/* out-of-group */ +%left '~' /* bitwise */ +%left '!' /* logical */ %% 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; @@ -126,8 +134,6 @@ 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 +141,7 @@ control_method: '@' STRING '\n' YYERROR; } if(command==CLASS_NAME) { - if(PC.cclass) { // 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, "'"); @@ -155,30 +161,30 @@ 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(PC.request->main_class, *LA2S(strings_code, i)); } else if(command==BASE_NAME) { - if(PC.cclass->base()!=&PC.request->OP) { // 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; @@ -238,41 +244,45 @@ 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 */ 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 { $$=$2 } +get_value: '$' get_name_value { $$=$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); 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 */ }; -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($$, $2) }; @@ -283,52 +293,68 @@ put: '$' name_expr_wdive construct { P($$, $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 { $$=N(POOL); Array *diving_code=$1; const String *first_name=LA2S(diving_code); + // $self.xxx... -> $xxx... + // OP_VALUE+string+OP_GET_ELEMENT+... -> OP_WITH_SELF+... if(first_name && *first_name==SELF_ELEMENT_NAME) { O($$, 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_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($$, OP_CONSTRUCT_EXPR); /* value=pop; name=pop; context=pop; construct(context,name,value) */ +construct_round: '(' expr_value ')' { + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); + // stack: context, name + P($$, $2); // stack: context, name, value + O($$, OP_CONSTRUCT_EXPR); /* value=pop->as_expr_result; name=pop; context=pop; construct(context,name,value) */ } ; +construct_curly: '{' maybe_codes '}' { + // stack: context, name + $$=N(POOL); + OA($$, OP_CURLY_CODE__CONSTRUCT, $2); /* code=pop; name=pop; context=pop; construct(context,name,junction(code)) */ +}; + any_constructor_code_value: - unknown_value /* optimized $var[] case */ + void_value /* optimized $var[] case */ | STRING /* optimized $var[STRING] case */ | constructor_code_value /* $var[something complex] */ ; constructor_code_value: constructor_code { $$=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) }; @@ -337,20 +363,27 @@ codes__excluding_sole_str_literal: actio 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} */ - $$=$2; /* with_xxx,diving code; stack: context,method_junction */ - O($$, OP_GET_METHOD_FRAME); /* stack: context,method_frame */ + 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 */ - 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($$, params_code); // filling method_frame.store_params - O($$, OP_CALL); // method_frame=pop; ncontext=pop; call(ncontext,method_frame) stack: value + } + /* stack: context, method_junction */ + OA($$, OP_CALL, params_code); // method_frame=make frame(pop junction); ncontext=pop; call(ncontext,method_frame) stack: value }; call_name: name_without_curly_rdive; @@ -382,19 +415,21 @@ store_code_param_part: code_param_value }; store_expr_param_part: write_expr_value { $$=N(POOL); - PEA($$, $1); + OA($$, OP_EXPR_CODE__STORE_PARAM, $1); }; store_curly_param_part: maybe_codes { $$=N(POOL); - PCA($$, $1); + OA($$, OP_CURLY_CODE__STORE_PARAM, $1); }; code_param_value: - unknown_value /* optimized [;...] case */ + void_value /* optimized [;...] case */ | STRING /* optimized [STRING] case */ | constructor_code_value /* [something complex] */ ; write_expr_value: expr_value { - $$=$1; + $$=N(POOL); + O($$, OP_PREPARE_TO_EXPRESSION); + P($$, $1); O($$, OP_WRITE_EXPR_RESULT); }; @@ -405,6 +440,9 @@ 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) */ @@ -419,30 +457,32 @@ 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 { $$=$2; O($$, 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($$, OP_STRING_POOL, code); }; -subvar_ref_name_rdive: subvar_ref_name_rdive_read | subvar_ref_name_rdive_root; -subvar_ref_name_rdive_read: STRING { +name_square_code_value: '[' codes ']' { $$=N(POOL); - O($$, OP_WITH_READ); - P($$, $1); + OA($$, OP_OBJECT_POOL, $2); /* stack: empty write context */ + /* some code that writes to that context */ + /* context=pop; stack: context.value() */ }; -subvar_ref_name_rdive_root: ':' STRING { +subvar_ref_name_rdive: STRING { $$=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 { @@ -450,25 +490,36 @@ 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 + 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($$, OP_GET_CLASS); }; - - -/* with */ - -with: '$' name_without_curly_rdive '{' maybe_codes '}' { - $$=$2; - O($$, OP_CREATE_RWPOOL); - P($$, $4); - O($$, OP_REDUCE_RWPOOL); - O($$, OP_WRITE_VALUE); +class_constructor_prefix: class_static_prefix ':' { + $$=$1; + if(!PC.in_call_value) { + strcpy(PC.error, ":: not allowed here"); + YYERROR; + } + O($$, OP_PREPARE_TO_CONSTRUCT_OBJECT); }; + /* 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 }; @@ -477,9 +528,11 @@ 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($$, OP_NEG) } +| '-' expr %prec NUNARY { $$=$2; O($$, OP_NEG) } +| '+' expr %prec NUNARY { $$=$2 } | '~' expr { $$=$2; O($$, OP_INV) } | '!' expr { $$=$2; O($$, OP_NOT) } | "def" expr { $$=$2; O($$, OP_DEF) } @@ -492,12 +545,15 @@ 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($$, OP_NUM_LE) } @@ -515,9 +571,9 @@ expr: string_inside_quotes_value: maybe_codes { $$=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 */ @@ -527,7 +583,7 @@ 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) }; %% @@ -550,7 +606,7 @@ 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; @@ -568,6 +624,7 @@ 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++; @@ -575,22 +632,54 @@ 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; // ^ + begin=PC.source; // ->punctuation begin_line=PC.line; - // skip over ^ and _ + // skip over _ after ^ PC.source++; PC.col++; // skip analysis = forced literal continue; @@ -617,12 +706,16 @@ static int yylex(YYSTYPE *lvalp, void *p PC.source+=3; PC.col+=3; // reset piece 'begin' position & line - begin=PC.source; // ^ + begin=PC.source; // ->after ^#HH begin_line=PC.line; + // skip analysis = forced literal continue; } break; } + break; + } + } // #comment start skipping if(c=='#' && PC.col==1) { if(end!=begin) { @@ -630,30 +723,46 @@ 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) { // 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 '[': // $name.[co<[>de] + if(PC.ls==LS_NAME_SQUARE_PART) + lexical_brackets_nestage++; break; } break; // #COMMENT - case LS_COMMENT: + case LS_USER_COMMENT: if(c=='\n') { // skip comment begin=PC.source; @@ -665,13 +774,20 @@ static int yylex(YYSTYPE *lvalp, void *p break; // STRING IN EXPRESSION - case LS_EXPRESSION_STRING: + case LS_EXPRESSION_STRING_QUOTED: + case LS_EXPRESSION_STRING_APOSTROFED: switch(c) { case '"': - pop_LS(PC); //"abc". - RC; + case '\'': + if( + PC.ls == LS_EXPRESSION_STRING_QUOTED && c=='"' || + PC.ls == LS_EXPRESSION_STRING_APOSTROFED && c=='\'') { + pop_LS(PC); //"abc". | 'abc'. + RC; + } + break; case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -693,6 +809,9 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_DEF_PARAMS: switch(c) { + case '$': // common error + result=BAD_METHOD_PARAMETER_NAME_CHARACTER; + goto break2; case ';': RC; case ']': @@ -726,14 +845,8 @@ static int yylex(YYSTYPE *lvalp, void *p break; case LS_DEF_SPECIAL_BODY: - if(c=='\n') { - switch(*PC.source) { - case '@': case 0: // end of special_code - pop_LS(PC); - break; - } + if(c=='\n') RC; - } break; // (EXPRESSION) @@ -747,8 +860,17 @@ static int yylex(YYSTYPE *lvalp, void *p else // PC.ls==LS_VAR_ROUND // variable constructor ended pop_LS(PC); // return to normal life RC; + case '#': // comment start skipping + if(end!=begin) { + // append piece till # + PC.string->APPEND_CLEAN(begin, end-begin, PC.file, begin_line); + } + // fall into COMMENT lexical state [wait for \n] + push_LS(PC, LS_EXPRESSION_COMMENT); + lexical_brackets_nestage=1; + continue; case '$': - push_LS(PC, LS_EXPRESSION_VAR_NAME); + push_LS(PC, LS_EXPRESSION_VAR_NAME_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -766,36 +888,73 @@ 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); + 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 @@ -853,14 +1012,51 @@ 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: - case LS_EXPRESSION_VAR_NAME: - 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_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; @@ -873,17 +1069,29 @@ static int yylex(YYSTYPE *lvalp, void *p case ';': case ']': case '}': case ')': case '"': case '\'': - case '<': case '>': // these stand for HTML brackets and expression binary ops + case '<': case '>': // these stand for HTML brackets AND expression binary ops case '+': case '*': case '/': case '%': case '&': case '|': case '=': case '!': // common delimiters - case ',': + case ',': case '?': case '#': + // before call + case '^': pop_LS(PC); PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } 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; @@ -902,13 +1110,24 @@ 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 '}': // ${name} finished, restoring LS pop_LS(PC); RC; @@ -922,7 +1141,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -944,7 +1163,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_VAR_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -965,6 +1184,16 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_NAME: switch(c) { case '[': + // ^name.<[>code].xxx + if(PC.col>1/*not first column*/ && ( + end[-1]=='^'/*was start of call*/ || // never, ^[ is literal... + end[-1]==':'/*was class name delim */ || + end[-1]=='.'/*was name delim */ + )) { + push_LS(PC, LS_NAME_SQUARE_PART); + lexical_brackets_nestage=1; + RC; + } PC.ls=LS_METHOD_SQUARE; lexical_brackets_nestage=1; RC; @@ -979,6 +1208,8 @@ static int yylex(YYSTYPE *lvalp, void *p case '.': // name part delim case '$': // name part subvar case ':': // ':name' or 'class:name' + case '^': // ^abc^xxx wrong. bailing out + case ']': case '}': case ')': // ^abc]}) wrong. bailing out RC; } break; @@ -986,7 +1217,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_SQUARE: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1008,7 +1239,7 @@ static int yylex(YYSTYPE *lvalp, void *p case LS_METHOD_CURLY: switch(c) { case '$': - push_LS(PC, LS_VAR_NAME_SIMPLE); + push_LS(PC, LS_VAR_NAME_SIMPLE_WITH_COLON); RC; case '^': push_LS(PC, LS_METHOD_NAME); @@ -1062,7 +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*/); }