在野牛解析第一线

Question

我正在用python语言创建一个bison语法，我在运行我的testcode文件时得到的输出是这样的：

found identifier a at line 2
memory exhausted

Parsing completed successfully

我得到一些转移减少错误并减少减少，但我可以正常创建我的.exe文件，如果我运行它，它显示了这一点。

我已经尝试减少我的大部分班次/减少到无效。这真的是问题吗？因为我认为它不会给我一个.exe

.l文件

%{
#include <stdio.h>
#include <stdlib.h>
#include "sym_tab.h"
#include "define.h"

FILE *new_file;
int stringtoint;
int current_indent = 0;
void count();
void comment();
int count_indent();

%}



L                  [A-Za-z]
D                  [0-9]
N                  [1-9]
C   "%"|"!"|"@"|"$"|"%"|"^"|"&"|"_"



identifier          {L}({L}|{D})*
dec_const           (-|\+)*(0|{N}{D}*)
blank               [ \v\f]+
invalid_identifier  {D}|{C}(({L}|{D})*|{L})
invalid_keyword     {C}({L}|{D})+
block_count                 ^[\t]+


%%

"while"           {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(WHILE); }

"for"           {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(FOR); }
"in range"            {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(IN_RANGE); }

"input"           {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(INPUT); }
"print"           {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(PRINT); }

"if"              {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(IF); }
"elif"            {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(ELIF); }
"else"            {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(ELSE); }

"and"             {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(AND); }
"not"             {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(NOT); }
"or"              {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(OR); }

"return"          {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(RETURN); }
"exit"            {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(EXIT); }

"def"             {char *yycopy=strdup(yytext); count(); printf("found keyword %s at line %d\n" ,yycopy, line);
                        addsym( yycopy, block_num ); return(DEF); }

L?\"(\\.|[^\\"])*\" {char *yycopy=strdup(yytext); count(); printf("found literal string %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(STRING_LITERAL); }


{dec_const}               {char *yycopy=strdup(yytext); count(); stringtoint=atoi(yycopy);if(stringtoint<(-32768)|| stringtoint>32767){
              printf("dec_const %d in line %d not an acceptable value\n",stringtoint,line);}else{
              printf("found dec_constant %s at line %d\n", yycopy,line);
              addsym( yycopy, block_num ); return(DEC_CONST);}}


{identifier}              {char *yycopy=strdup(yytext); count(); if(strlen(yycopy)>20){
               printf("identifier %s in line %d not valid(longer than 20 characters)\n",yycopy,line);}
               else{printf("found identifier %s at line %d\n", yycopy,line);
                           addsym( yycopy, block_num ); return(IDENTIFIER);}}


"+"                       {char *yycopy=strdup(yytext); count(); printf("found symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(PLUS);}
"-"                       {char *yycopy=strdup(yytext); count(); printf("found symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(MINUS);}
"*"                       {char *yycopy=strdup(yytext); count(); printf("found symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(STAR);}
"/"                       {char *yycopy=strdup(yytext); count(); printf("found symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(DIV);}
"<"                       {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(L_THAN);}
">"                       {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(G_THAN);}
"=="                      {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(EQUAL);}
"<="                      {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(L_EQ_THAN);}
">="                      {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(G_EQ_THAN);}
"<>"                      {char *yycopy=strdup(yytext); count(); printf("found equation_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(NEQUAL);}
":="                        {char *yycopy=strdup(yytext); count(); printf("found asign_symbol %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(ASSIGN);}
"("                       {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(LPAREN);}
")"                       {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(RPAREN);}

"["                       {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(LSQUARE_BRACK);}
"]"                       {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(RSQUARE_BRACK);}

"\n"                        {char *yycopy=strdup(yytext); count(); printf("found new line at line %d\n" , line);
                           addsym( yycopy, block_num ); return(END_LINE);}
","                       {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(COMMA);}
":"                                             {char *yycopy=strdup(yytext); count(); printf("found %s at line %d\n" ,yycopy, line);
                           addsym( yycopy, block_num ); return(COLON);}

"#"                      { comment();}

{blank}                   { count();}



{invalid_keyword}    {char *yycopy=strdup(yytext); count(); printf("invalid keyword %s at line %d\n", yycopy, line);
              addsym( yycopy, block_num );}

{invalid_identifier}   {char *yycopy=strdup(yytext); count(); printf("invalid identifier %s at line %d\n", yycopy, line);
              addsym( yycopy, block_num );}


.              {char *yycopy=strdup(yytext); count(); printf("unexpected character %s at line %d\n", yycopy, line);
              addsym( yycopy, block_num );}

%%

int yywrap()
{
  return 1;
}

// void main(int argc, char *argv[]){
//   int ret_val=1;
//
//   if (argc!=2) printf("\nUsage: lexyy <input file name> \n");
//   else
//     if ((new_file=fopen(argv[1],"r"))==NULL)
//       printf("\n<%s> not found.\n",argv[1]);
//     else{
//       yyrestart(new_file);
//       while(ret_val!=0){
//         ret_val=yylex();
//       }
//       fclose(new_file);
//     }
//}

void count()
{
        int i;
        for(i=0;yytext[i]!='\0';i++)
        if(yytext[i]=='\n')
        {
                line++;
        }
}

int count_indent()
{
  int i;
  int tab_num = 0;
  for(i=0;yytext[i]=='\t';i++)
  {
    tab_num++;
  }
  return tab_num;
}


void comment()
{
  int c;
  while(c=input()!='\n' && c!=EOF)
  {

  }
  line++;
}

.y文件

%{
    #include<stdio.h>
    #include<stdlib.h>
    #include<string.h>
    #include <ctype.h>
    #include "y.tab.h"
    extern int yylex();
    extern FILE *yyin;
%}


%token L_THAN G_THAN EQUAL L_EQ_THAN G_EQ_THAN NEQUAL
%token ASSIGN
%token LPAREN RPAREN LSQUARE_BRACK RSQUARE_BRACK
%token END_LINE COMMA COLON
%token INDENT DEDENT


%start Program
%%

Program: Block Program
    | Empty
    ;

Empty: /* empty */
    ;

Block: Declarations
    | Subprograms
    | Sequence
    ;



%%
extern int column;

int main(int argc, char *argv[])
{
    yyin = fopen("test_code.sy", "r");
    if(yyparse()==1)
        printf("\nParsing failed\n\n");
       else
        printf("\nParsing completed successfully\n");
       fclose(yyin);
    return 0;
}

int yyerror(s)
char *s;
{
                printf("%s\n", s);
        fflush(stdout);
        return 1;
}

**编辑：** sym_tab.h文件

/*#include <iostream.h>*/
#include <stdio.h>
#include <malloc.h>
#include <string.h>

#define table_size          100

extern int line=1;
extern int end_file=1;
extern int block_num=0;
extern FILE *new_file; 

typedef struct hash_sym
    {
    struct hash_sym *prev, *next;
    char *nam;
    char *str_val;
    char *id_type;
    int  id_value;
    int  block_num;
} Hashing_table; 


Hashing_table *table[table_size];

int hash_funct( char str[], int hash_size);

addsym( sym, bloc_num )
register char sym[];
int bloc_num;

{   
    int hash_val = hash_funct( sym, table_size );
    register struct hash_sym *sym_sym = table[hash_val];
    register struct hash_sym *new_sym;
    register struct hash_sym *successor;

    while ( sym_sym!=0 )
    {

        if (  strcmp( sym, table[hash_val]->nam )==0 )
        {   

            printf("the entry %s at line %d already exists at symbol table\n", sym,line);   
            return -1;
        }   

        sym_sym = sym_sym->next;
    }


    new_sym = (struct hash_sym *)
    malloc( sizeof( struct hash_sym ) );


    if ( (successor = table[hash_val]) )
    { 

        new_sym->next = successor;
        successor->prev = new_sym;
    }
    else
        new_sym->next = NULL;

    new_sym->prev = NULL;
    new_sym->nam = sym;
    new_sym->block_num = bloc_num;

    table[hash_val] = new_sym;
    return 0;
}




int hash_funct( str, hash_size )
register char str[];
int hash_size;
{
    register int hashval;
    register int i;

    hashval = 0;
    i = 0;

    while ( str[i]!='\0' )
        {
        hashval = hashval +  str[i++]*(16+i);
        /*hashval %= hash_size;*/
        }
    return (hashval %= hash_size);

}

我希望解析器解析到文件结束。我还没有在我的.y文件中设置打印件，所以我不希望在那里打印。

Answer 1

基本问题是你的语法包括一个可以匹配空字符串的非终端的任意重复。

这总是模棱两可的，因为无法区分空字符串和两个连续的空字符串，或者实际上是一百万个连续的空字符串。因此，重复可空的非终端将始终产生冲突。大多数情况下，生成的解析器只是不正确，但它仍然终止。 Bison通过选择shift来解决移位/减少冲突，这保证了解析器将通过输入进行。实际上，它解决了“有多少空字符串”的问题，答案最小，通常为“一”。

但在你的情况下，重复有多个替代方案，其中几个可以为空。现在解析器有一个更难的问题：它必须弄清楚空字符串应匹配哪个非终端。这是一个减少/减少冲突，而野牛的决心是总是选择语法中第一个出现的非终端。如果特定输入的正确选择是某些其他非终端，那将是一个问题。

这是我所说的最简单的例子：

%%
list: %empty | unit list
unit: as | bs
as: %empty | as 'a'
bs: %empty | bs 'b'

在这里，unit可以是零或更多as，或零或更多bs。由于零as和零bs看起来完全相同，解析器实际上无法分辨哪一个从语法中选择，因此它总是选择零as（因为它在语法中首先出现）。输入包含b时出现问题。由于解析器从不使用规则bs: %empty（事实上，野牛警告你），它永远不会应用规则bs: bs 'b'。所以面对b，解析器减少空as，使其成为unit，将其添加到list，然后尝试解析另一个unit。然而，一切都没有改变;没有令牌被读取所以前瞻仍然是b。所以解析器将进入一个无限循环解析空units一遍又一遍地包含空as。

使用上面编写的list生成（右递归），需要将这些空的units添加到解析器堆栈中。因此，解析器最终会耗尽其堆栈的空间，并因“内存耗尽”错误而死亡。如果你将它改为左递归（list: %empty | list unit），那么解析器不需要使用堆栈空间，它可以继续解析空的units。

我建议您尝试使用bison非常有用的跟踪功能的上述简单示例（请参阅Bison手册中的"Debugging your parser"）。这比使用printf调用填充语法文件容易得多，而且它也提供了更多信息。

要解决这个问题，你只需要要求unit非空，这就避免了“重复一个空字符串”问题。如果每个unit需要匹配某些东西，那么语法匹配相同的语言，但它明确地这样做。