--- parser3/src/main/compile.y 2001/03/15 09:37:55 1.99 +++ parser3/src/main/compile.y 2001/03/26 09:59:38 1.116 @@ -1,20 +1,22 @@ -/* - Parser +/** @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 1.99 2001/03/15 09:37:55 paf Exp $ + $Id: compile.y,v 1.116 2001/03/26 09:59:38 paf Exp $ */ -/* - TODO.parser4: - cache compiled code from request to request. to do that... - 1: make method definitions, @CLASS, @BASE, @USE instructions, +/** + @todo parser4: + - cache compiled code from request to request. to do that... + -#: make method definitions, @CLASS, @BASE, @USE instructions, which would be executed afterwards, and actions now performed at compile time would be delayed to run time. - 2: make cache expiration on time and on disk-change of class source - 3: in apache use subpools for compiled class storage - 4: in iis make up specialized Pool object for that + -#: make cache expiration on time and on disk-change of class source + -#: in apache use subpools for compiled class storage + -#: in iis make up specialized Pool object for that */ %{ @@ -22,12 +24,12 @@ #define YYPARSE_PARAM pc #define YYLEX_PARAM pc #define YYDEBUG 1 -#define YYERROR_VERBOSE +#define YYERROR_VERBOSE 1 #define yyerror(msg) real_yyerror((parse_control *)pc, msg) #define YYPRINT(file, type, value) yyprint(file, type, value) +#include "pa_config_includes.h" #include -#include #include #include "compile_tools.h" @@ -46,10 +48,11 @@ int yylex(YYSTYPE *lvalp, void *pc); // local convinient inplace typecast & var -#define PC ((parse_control *)pc) -#define POOL *PC->pool +#define PC (*(parse_control *)pc) +#define POOL (*PC.pool) #undef NEW #define NEW new(POOL) +#ifndef DOXYGEN %} %pure_parser @@ -59,6 +62,7 @@ int yylex(YYSTYPE *lvalp, void *pc); %token BOGUS %token BAD_STRING_COMPARISON_OPERATOR +%token BAD_HEX_LITERAL %token LAND "&&" %token LOR "||" @@ -110,7 +114,7 @@ all: 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(*main_method_name, method); } | methods; @@ -124,65 +128,67 @@ control_method: '@' STRING '\n' const String& command=*SLA2S($2); YYSTYPE strings_code=$4; if(strings_code->size()<1*2) { - strcpy(PC->error, "@"); - strcat(PC->error, command.cstr()); - strcat(PC->error, " is empty"); + strcpy(PC.error, "@"); + strcat(PC.error, command.cstr()); + strcat(PC.error, " is empty"); YYERROR; } if(command==CLASS_NAME) { - if(PC->cclass!=&PC->request->root_class) { // already changed from default? - strcpy(PC->error, "class already have a name '"); - strncat(PC->error, PC->cclass->name().cstr(), 100); - strcat(PC->error, "'"); + if(PC.cclass!=&PC.request->ROOT) { // already changed from default? + strcpy(PC.error, "class already have a name '"); + strncat(PC.error, PC.cclass->name().cstr(), 100); + strcat(PC.error, "'"); YYERROR; } if(strings_code->size()==1*2) { // new class' name const String *name=SLA2S(strings_code); // creating the class - PC->cclass=NEW VClass(POOL); - PC->cclass->set_name(*name); + PC.cclass=NEW VClass(POOL); + PC.cclass->set_name(*name); // defaulting base. may change with @BASE - PC->cclass->set_base(PC->request->root_class); + PC.cclass->set_base(PC.request->ROOT); // append to request's classes - PC->request->classes().put(*name, PC->cclass); + PC.request->classes().put(*name, PC.cclass); } else { - strcpy(PC->error, "@"CLASS_NAME" must contain sole name"); + strcpy(PC.error, "@"CLASS_NAME" must contain sole name"); YYERROR; } } else { if(command==USE_CONTROL_METHOD_NAME) { - for(int i=0; isize(); i+=2) { - String file(*SLA2S(strings_code, i)); - file.APPEND_CONST(".p"); - PC->request->use_file(file.cstr()); - } + for(int i=0; isize(); i+=2) + PC.request->use_file( + PC.request->absolute(*SLA2S(strings_code, i))); } else if(command==BASE_NAME) { - if(PC->cclass->base()!=&PC->request->root_class) { // already changed from default? - strcpy(PC->error, "class already have a base '"); - strncat(PC->error, PC->cclass->base()->name().cstr(), 100); - strcat(PC->error, "'"); + if(PC.cclass->base()!=&PC.request->ROOT) { // already changed from default? + strcpy(PC.error, "class already have a base '"); + strncat(PC.error, PC.cclass->base()->name().cstr(), 100); + strcat(PC.error, "'"); YYERROR; } if(strings_code->size()==1*2) { - // TODO: преодолеть self и циклические base const String& base_name=*SLA2S(strings_code); VClass *base=static_cast( - PC->request->classes().get(base_name)); + PC.request->classes().get(base_name)); if(!base) { - strcpy(PC->error, base_name.cstr()); - strcat(PC->error, ": undefined class in @"BASE_NAME); + strcpy(PC.error, base_name.cstr()); + strcat(PC.error, ": undefined class in @"BASE_NAME); YYERROR; } - PC->cclass->set_base(*base); + // @CLASS == @BASE sanity check + if(PC.cclass==base) { + strcpy(PC.error, "@"CLASS_NAME" equals @"BASE_NAME); + YYERROR; + } + PC.cclass->set_base(*base); } else { - strcpy(PC->error, "@"BASE_NAME" must contain sole name"); + strcpy(PC.error, "@"BASE_NAME" must contain sole name"); YYERROR; } } else { - strcpy(PC->error, "'"); - strncat(PC->error, command.cstr(), MAX_STRING/2); - strcat(PC->error, "' invalid special name. valid names are " + strcpy(PC.error, "'"); + strncat(PC.error, command.cstr(), MAX_STRING/2); + strcat(PC.error, "' invalid special name. valid names are " "'"CLASS_NAME"', '"USE_CONTROL_METHOD_NAME"' and '"BASE_NAME"'"); YYERROR; } @@ -217,7 +223,7 @@ code_method: '@' STRING bracketed_maybe_ 0, 0/*min,max numbered_params_count*/, params_names, locals_names, $7, 0); - PC->cclass->add_method(*name, method); + PC.cclass->add_method(*name, method); }; maybe_bracketed_strings: empty | bracketed_maybe_strings; @@ -239,7 +245,7 @@ action: get | put | with | call; get: get_value { $$=$1; /* stack: resulting value */ - O($$, OP_WRITE); /* value=pop; wcontext.write(value) */ + O($$, OP_WRITE_VALUE); /* value=pop; wcontext.write(value) */ }; get_value: '$' get_name_value { $$=$2 } get_name_value: name_without_curly_rdive EON | name_in_curly_rdive; @@ -332,7 +338,7 @@ codes__excluding_sole_str_literal: actio call: call_value { $$=$1; /* stack: value */ - O($$, OP_WRITE); /* value=pop; wcontext.write(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 */ @@ -353,6 +359,10 @@ store_square_param: '[' store_code_param store_round_param: '(' store_expr_param_parts ')' {$$=$2}; store_curly_param: '{' store_curly_param_parts '}' {$$=$2}; store_code_param_parts: + empty /* optimized [] case */ +| store_code_param_parts_not_empty +; +store_code_param_parts_not_empty: store_code_param_part | store_code_param_parts ';' store_code_param_part { $$=$1; P($$, $3) } ; @@ -365,8 +375,7 @@ store_curly_param_parts: | store_curly_param_parts ';' store_curly_param_part { $$=$1; P($$, $3) } ; store_code_param_part: - empty /* optimized [] case */ -| STRING { /* optimized [STRING] case */ + STRING { /* optimized [STRING] case */ $$=$1; O($$, OP_STORE_PARAM); } @@ -377,7 +386,7 @@ store_code_param_part: ; store_expr_param_part: write_expr_value { $$=N(POOL); - PCA($$, $1); + PEA($$, $1); }; store_curly_param_part: maybe_codes { $$=N(POOL); @@ -385,7 +394,7 @@ store_curly_param_part: maybe_codes { }; write_expr_value: expr_value { $$=$1; - O($$, OP_WRITE); + O($$, OP_WRITE_EXPR_RESULT); }; /* name */ @@ -419,7 +428,7 @@ name_expr_with_subvar_value: STRING subv $$=N(POOL); O($$, OP_CREATE_EWPOOL); P($$, $1); - O($$, OP_WRITE); + O($$, OP_WRITE_VALUE); P($$, $2); O($$, OP_REDUCE_EWPOOL); }; @@ -453,7 +462,7 @@ with: '$' name_without_curly_rdive '{' c O($$, OP_CREATE_RWPOOL); P($$, $4); O($$, OP_REDUCE_RWPOOL); - O($$, OP_WRITE); + O($$, OP_WRITE_VALUE); }; /* expr */ @@ -512,7 +521,7 @@ string_inside_quotes_value: maybe_codes /* basics */ write_string: STRING { - // optimized from OP_STRING+OP_WRITE to OP_STRING__WRITE + // optimized from OP_STRING+OP_WRITE_VALUE to OP_STRING__WRITE change_string_literal_to_write_string_literal($$=$1) }; @@ -520,6 +529,7 @@ empty_string_value: /* empty */ { $$=VL( empty: /* empty */ { $$=N(POOL) }; %% +#endif /* 000$111(2222)00 @@ -537,54 +547,89 @@ empty: /* empty */ { $$=N(POOL) }; 4:[^({]=pop */ -int yylex(YYSTYPE *lvalp, void *pc) { - #define lexical_brackets_nestage PC->brackets_nestages[PC->sp] +static int yylex(YYSTYPE *lvalp, void *pc) { + #define lexical_brackets_nestage PC.brackets_nestages[PC.sp] #define RC {result=c; goto break2; } 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; const char *end; - int begin_line=PC->line; + int begin_line=PC.line; int skip_analized=0; while(true) { - c=*(end=(PC->source++)); + c=*(end=(PC.source++)); if(c=='\n') { - PC->line++; - PC->col=0; + PC.line++; + PC.col=0; } else - PC->col++; + PC.col++; - // todo: # in 0+1 column comment - - // escaping: ^^ ^$ ^; ^) ^} ^( ^{ ^" if(c=='^') - switch(*PC->source) { + switch(*PC.source) { + // escaping: ^^ ^$ ^; ^) ^} ^( ^{ ^" case '^': case '$': case ';': case '[': case ']': case '{': case '}': - case '"': + case '"': if(end!=begin) { // append piece till ^ - PC->string->APPEND(begin, end-begin, PC->file, begin_line); + PC.string->APPEND(begin, end-begin, PC.file, begin_line); } // reset piece 'begin' position & line - begin=PC->source; // ^ - begin_line=PC->line; + begin=PC.source; // ^ + begin_line=PC.line; // skip over ^ and _ - PC->source++; PC->col++; + PC.source++; PC.col++; // skip analysis = forced literal continue; + + // converting ^#HH into char(hex(HH)) + case '#': + if(end!=begin) { + // append piece till ^ + PC.string->APPEND(begin, end-begin, PC.file, begin_line); + } + // #HH ? + if(PC.source[0]=='#' && PC.source[1] && PC.source[2]) { + char *hex=(char *)POOL.malloc(1); + hex[0]= + hex_value[(unsigned char)PC.source[1]]*0x10+ + hex_value[(unsigned char)PC.source[2]]; + if(hex[0]==0) { + result=BAD_HEX_LITERAL; + goto break2; // wrong hex value[no ^#00 chars allowed]: bail out + } + // append char(hex(HH)) + PC.string->APPEND(hex, 1, PC.file, begin_line); + // skip over ^#HH + PC.source+=3; + PC.col+=3; + // reset piece 'begin' position & line + begin=PC.source; // ^ + begin_line=PC.line; + continue; + } + break; + } + // #comment start skipping + if(c=='#' && PC.col==1) { + if(end!=begin) { + // append piece till # + PC.string->APPEND(begin, end-begin, PC.file, begin_line); } - switch(PC->ls) { + // fall into COMMENT lexical state [wait for \n] + push_LS(PC, LS_COMMENT); + } + switch(PC.ls) { // USER'S = NOT OURS case LS_USER: @@ -596,7 +641,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { push_LS(PC, LS_METHOD_NAME); RC; case '@': - if(PC->col==0+1) { + if(PC.col==0+1) { push_LS(PC, LS_DEF_NAME); RC; } @@ -604,6 +649,18 @@ int yylex(YYSTYPE *lvalp, void *pc) { } break; + // #COMMENT + case LS_COMMENT: + if(c=='\n') { + // skip comment + begin=PC.source; + begin_line=PC.line; + + pop_LS(PC); + continue; + } + break; + // STRING IN EXPRESSION case LS_EXPRESSION_STRING: switch(c) { @@ -623,10 +680,10 @@ int yylex(YYSTYPE *lvalp, void *pc) { 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; @@ -636,7 +693,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { 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); @@ -650,7 +707,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { case ';': RC; case ']': - PC->ls=LS_DEF_COMMENT; + PC.ls=LS_DEF_COMMENT; RC; case '\n': // wrong. bailing out pop_LS(PC); @@ -667,7 +724,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { case LS_DEF_SPECIAL_BODY: if(c=='\n') { - switch(*PC->source) { + switch(*PC.source) { case '@': case 0: // end of special_code pop_LS(PC); break; @@ -682,9 +739,9 @@ int yylex(YYSTYPE *lvalp, void *pc) { 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 + 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 '$': @@ -697,7 +754,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { lexical_brackets_nestage++; RC; case '-': - if(*PC->source=='f') { // -f + if(*PC.source=='f') { // -f skip_analized=1; result=FEXISTS; } else @@ -708,14 +765,14 @@ int yylex(YYSTYPE *lvalp, void *pc) { case ';': RC; case '&': case '|': case '#': - if(*PC->source==c) { // && || + if(*PC.source==c) { // && || result=c=='#'?LXOR:c=='&'?LAND:LOR; skip_analized=1; } else result=c; goto break2; case '<': case '>': case '=': case '!': - if(*PC->source=='=') { // <= >= == != + if(*PC.source=='=') { // <= >= == != skip_analized=1; switch(c) { case '<': result=NLE; break; @@ -731,7 +788,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { RC; case 'l': case 'g': case 'e': case 'n': if(end==begin) // right after whitespace - switch(*PC->source) { + 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; @@ -749,7 +806,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { break; case 'i': if(end==begin) // right after whitespace - switch(PC->source[0]) { + switch(PC.source[0]) { case 'n': { // in skip_analized=1; @@ -766,7 +823,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { 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; @@ -780,8 +837,8 @@ int yylex(YYSTYPE *lvalp, void *pc) { // 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_line=PC.line; continue; } break; @@ -789,12 +846,12 @@ int yylex(YYSTYPE *lvalp, void *pc) { // VARIABLE GET/PUT/WITH case LS_VAR_NAME_SIMPLE: case LS_EXPRESSION_VAR_NAME: - if(PC->ls==LS_EXPRESSION_VAR_NAME) { + if(PC.ls==LS_EXPRESSION_VAR_NAME) { // name in expr ends also before binary operators switch(c) { case '-': pop_LS(PC); - PC->source--; if(--PC->col<0) { PC->line--; PC->col=-1; } + PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } result=EON; goto break2; } @@ -809,26 +866,26 @@ int yylex(YYSTYPE *lvalp, void *pc) { case '&': case '|': case '=': case '!': // common delimiters - case '\'': + case '\'': case ',': pop_LS(PC); - PC->source--; if(--PC->col<0) { PC->line--; PC->col=-1; } + PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } result=EON; goto break2; case '[': - 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 @@ -896,15 +953,15 @@ int yylex(YYSTYPE *lvalp, void *pc) { case LS_METHOD_NAME: switch(c) { case '[': - 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 @@ -926,7 +983,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { RC; case ']': if(--lexical_brackets_nestage==0) { - PC->ls=LS_METHOD_AFTER; + PC.ls=LS_METHOD_AFTER; RC; } break; @@ -948,7 +1005,7 @@ int yylex(YYSTYPE *lvalp, void *pc) { RC; case '}': if(--lexical_brackets_nestage==0) { - PC->ls=LS_METHOD_AFTER; + PC.ls=LS_METHOD_AFTER; RC; } break; @@ -960,22 +1017,22 @@ int yylex(YYSTYPE *lvalp, void *pc) { 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; } + PC.source--; if(--PC.col<0) { PC.line--; PC.col=-1; } result=EON; goto break2; } @@ -993,38 +1050,29 @@ break2: } if(end!=begin) { // last piece still alive? // append it - PC->string->APPEND(begin, end-begin, PC->file, begin_line/*, start_col*/); + PC.string->APPEND(begin, end-begin, PC.file, begin_line/*, start_col*/); } } - if(PC->string->size()) { // something accumulated? + if(PC.string->size()) { // something accumulated? // create STRING value: array of OP_VALUE+vstring - *lvalp=VL(NEW VString(*PC->string)); + *lvalp=VL(NEW VString(*PC.string)); // new pieces storage - PC->string=NEW String(POOL); + PC.string=NEW String(POOL); // 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.col+=skip_analized; } return result; } -int real_yyerror(parse_control *pc, char *s) /* Called by yyparse on error */ - { - //fprintf(stderr, "[%s]\n", s); - - strncpy(pc->error, s, MAX_STRING); // TODO: перепроверить с треклятым последним байтом +static int real_yyerror(parse_control *pc, char *s) { // Called by yyparse on error + strncpy(PC.error, s, MAX_STRING); return 1; - } - -static void - yyprint( - FILE *file, - int type, - YYSTYPE value) - { - if(type==STRING) - fprintf(file, " \"%s\"", SLA2S(value)->cstr()); - } +} +static void yyprint(FILE *file, int type, YYSTYPE value) { + if(type==STRING) + fprintf(file, " \"%s\"", SLA2S(value)->cstr()); +}