如果我使用 Flex/Bison (Lex/Yacc) 更改 2 行的顺序，则打印到输出文件不起作用
Printing to output file does not work if I change the order of 2 lines with Flex/Bison (Lex/Yacc)
我正在尝试为一种简单的格式化语言构建一个编译器，但我现在想做的是在 Bison 识别时向输出文件打印一条消息
beginDocument docProperties endDocument
我将首先展示一个 .y 文件，其中的功能与预期的一样。但是，我将仅更改 2 行代码的顺序并打印到输出文件不再有效！
这是工作正常的 .y 文件：
%{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "y.tab.h"
void yyerror(const char *);
int yylex(void);
/* "Connect" with the output file */
extern FILE *yyout;
extern int yylineno;
/* Integers related to user-defined document properties */
int tabSize, linesPerPage, charsPerLine, linesPrinted, pageNumber = 1;
char* title; /* char* title, author, date; is wrong. In C that would have been char *title, *author, *date because typically in C it's int *p while in C++ it's int* p */
char* author;
char* date;
/* An array with counters of how many times each of the 5 document properties appears in the input file. The order of the properties is defined in the enum below */
int docPropertyCounters[5];
/* An enumerated list with the 5 document properties */
typedef enum {PAGE_SETUP, TAB_SIZE, DOC_TITLE, DOC_AUTHOR, DOC_DATE} document_property;
/* Takes a document_property enum (basically an integer) and returns the corresponding document property as a string. The order is based on the enum. */
static inline char *stringFromDocPropertyEnum(const document_property indexOfProperty) {
static char *strings[] = { "\pagesetup{}", "\tabsize()", "\title{}", "\author{}", "\date{}"};
return strings[indexOfProperty];
}
/* Checks for all possible errors in document properties */
void dealWithDocPropertyErrors() {
for (int i = 0; i < sizeof(docPropertyCounters)/sizeof(docPropertyCounters[0]); i++) {
if (docPropertyCounters[i] < 1)
/* yyerror() is not used in this function because the line number does not need to be shown */
fprintf(stderr, "SYNTAX ERROR: Your source file does not contain the required document property %s", stringFromDocPropertyEnum(i));
else if (docPropertyCounters[i] > 1)
fprintf(stderr, "SYNTAX ERROR: Your source file contains more than one instance of the document property %s", stringFromDocPropertyEnum(i));
}
exit(-1);
}
/* Returns the string it takes as an argument excluding the first and last characters */
char* removeFirstAndLastChar(char* string) {
string = string + 1; // Removes the first character
int i = 0;
for (; string[i] != '[=13=]'; i++);
string[i - 1] = '[=13=]';
return string;
}
%}
%union {
int iValue; /* integer value */
char* sValue; /* C-String */
};
%error-verbose /* sometimes provides better error reporting. Useful sometimes when debugging */
%start file /* defining the start condition */
%token BSLASH LBRACE RBRACE LPAREN RPAREN COMMA
%token BEGIN_ END DOCUMENT /* BEGIN seems to be a reserved word so BEGIN_ was used instead */
%token PAGESETUP TABSIZE TITLE AUTHOR DATE
%token <iValue> INTEGER
%token <sValue> DDMMYYYYDATE STRING
%%
file: beginDocument docProperties endDocument
{
fprintf(yyout, "Message");
dealWithDocPropertyErrors();
}
| /* An empty document is parsed to an empty document, no errors generated */
;
beginDocument: BSLASH BEGIN_ LBRACE DOCUMENT RBRACE;
docProperties: docProperties docProperty
| /* empty */
;
/* required properties... there should be one instance of each in the input file */
docProperty: pageSetupProperty { docPropertyCounters[PAGE_SETUP]++; }
| tabSizeProperty { docPropertyCounters[TAB_SIZE]++; }
| titleProperty { docPropertyCounters[DOC_TITLE]++; }
| authorProperty { docPropertyCounters[DOC_AUTHOR]++; }
| dateProperty { docPropertyCounters[DOC_DATE]++; }
;
pageSetupProperty: BSLASH PAGESETUP LBRACE INTEGER COMMA INTEGER RBRACE
{
linesPerPage = ;
charsPerLine = ;
}
;
tabSizeProperty: BSLASH TABSIZE LPAREN INTEGER RPAREN
{
tabSize = ;
}
;
titleProperty: BSLASH TITLE LBRACE STRING RBRACE
{
/* is copied into title excluding the quotation marks at the beginning and end of the string */
title = removeFirstAndLastChar(); /* is a "pseudo-variable" and can be passed in like this */
}
;
authorProperty: BSLASH AUTHOR LBRACE STRING RBRACE
{
author = removeFirstAndLastChar();
}
;
dateProperty: BSLASH DATE LBRACE DDMMYYYYDATE RBRACE
{
date = removeFirstAndLastChar();
}
;
endDocument: BSLASH END LBRACE DOCUMENT RBRACE
{
// Freeing memory created by strdup()
free(title);
free(author);
free(date);
}
;
%%
int yywrap(void) {
return 1;
}
void yyerror(const char* str)
{
fprintf(stderr,"SYNTAX ERROR near line [%d]: %s\n", yylineno, str);
}
以此为输入文件，
\begin {document}
\tabsize( 5)
\pagesetup{30,100 }
\title{"Why I Love Compiler Design"}
\author{"COMP421 Student"}
\date{29/12/2016}
\end{document}
我得到这个输出：
Message
这正是我所期望的。
假设我想在处理文档属性错误后打印消息，我把打印消息的行和它后面的一行的顺序改一下，也就是我做的：
dealWithDocPropertyErrors();
fprintf(yyout, "Message");
我现在得到一个相同输入的空输出文件。我在编译期间没有得到任何 warnings/errors。
我正在使用 Windows 10 和 Flex 2.5.4a。我的 .l 文件在下面
%{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "y.tab.h"
void yyerror(const char *);
int yylex(void);
extern FILE *yyout; /* "Connect" with the output file */
extern int yyparse();
%}
/* %option noyywrap; Tells flex not to declare the function yywrap() which is only useful in the case of more than 1 input file */
/* Allows printing the line number (of an error) */
%option yylineno
/* Catches some errors */
%option nodefault
/* Tells flex not to generate code for the input() and unput() functions which I will not be using */
%option nounput
%option noinput
/* Prints the tokens flex recognizes to the console. Useful when debugging and avoids having to write printf() statements for that */
%option debug
%%
^\ { return BSLASH; } /* every backslash has to be at the start of a line */
\{ { return LBRACE; }
\} { return RBRACE; }
\( { return LPAREN; }
\) { return RPAREN; }
, { return COMMA; }
begin { return BEGIN_; }
end { return END; }
document { return DOCUMENT; }
pagesetup { return PAGESETUP; }
tabsize { return TABSIZE; }
title { return TITLE; }
author { return AUTHOR; }
date { return DATE; }
(((0[1-9]|[12][0-9]|30)[-/ ]?(0[13-9]|1[012])|31[-/ ]?(0[13578]|1[02])|(0[1-9]|1[0-9]|2[0-8])[-/ ]?02)[-/ ]?[0-9]{4}|29[-/ ]?02[-/ ]?([0-9]{2}(([2468][048]|[02468][48])|[13579][26])|([13579][26]|[02468][048]|0[0-9]|1[0-6])00)) { yylval.sValue = strdup(yytext); return DDMMYYYYDATE; }
[0-9]*[1-9][0-9]* { yylval.iValue = atoi(strdup(yytext)); return INTEGER; } /* strdup() is string duplicate. yytext must be copied because of its temporary nature */
\".*\" { yylval.sValue = strdup(yytext); return STRING; }
/* skip whitespace which is not part of a string. [ \t\r\n]+ is better than [ \t\r\n] performance-wise */
[ \t\r\n]+ ;
/* anything else is an error */
. yyerror("invalid character");
%%
int main(int argc, char *argv[]) {
if ( argc != 3)
yyerror("ERROR You need 2 args: inputFileName outputFileName");
else {
yyin = fopen(argv[1], "r");
yyout = fopen(argv[2], "w");
yyparse();
fclose(yyin);
fclose(yyout);
}
return 0;
}
这不是 lex/yacc 问题。
函数dealWithDocPropertyErrors()调用C库函数exit()，终止进程，永不returns。
我正在尝试为一种简单的格式化语言构建一个编译器，但我现在想做的是在 Bison 识别时向输出文件打印一条消息
beginDocument docProperties endDocument
我将首先展示一个 .y 文件，其中的功能与预期的一样。但是，我将仅更改 2 行代码的顺序并打印到输出文件不再有效！
这是工作正常的 .y 文件：
%{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "y.tab.h"
void yyerror(const char *);
int yylex(void);
/* "Connect" with the output file */
extern FILE *yyout;
extern int yylineno;
/* Integers related to user-defined document properties */
int tabSize, linesPerPage, charsPerLine, linesPrinted, pageNumber = 1;
char* title; /* char* title, author, date; is wrong. In C that would have been char *title, *author, *date because typically in C it's int *p while in C++ it's int* p */
char* author;
char* date;
/* An array with counters of how many times each of the 5 document properties appears in the input file. The order of the properties is defined in the enum below */
int docPropertyCounters[5];
/* An enumerated list with the 5 document properties */
typedef enum {PAGE_SETUP, TAB_SIZE, DOC_TITLE, DOC_AUTHOR, DOC_DATE} document_property;
/* Takes a document_property enum (basically an integer) and returns the corresponding document property as a string. The order is based on the enum. */
static inline char *stringFromDocPropertyEnum(const document_property indexOfProperty) {
static char *strings[] = { "\pagesetup{}", "\tabsize()", "\title{}", "\author{}", "\date{}"};
return strings[indexOfProperty];
}
/* Checks for all possible errors in document properties */
void dealWithDocPropertyErrors() {
for (int i = 0; i < sizeof(docPropertyCounters)/sizeof(docPropertyCounters[0]); i++) {
if (docPropertyCounters[i] < 1)
/* yyerror() is not used in this function because the line number does not need to be shown */
fprintf(stderr, "SYNTAX ERROR: Your source file does not contain the required document property %s", stringFromDocPropertyEnum(i));
else if (docPropertyCounters[i] > 1)
fprintf(stderr, "SYNTAX ERROR: Your source file contains more than one instance of the document property %s", stringFromDocPropertyEnum(i));
}
exit(-1);
}
/* Returns the string it takes as an argument excluding the first and last characters */
char* removeFirstAndLastChar(char* string) {
string = string + 1; // Removes the first character
int i = 0;
for (; string[i] != '[=13=]'; i++);
string[i - 1] = '[=13=]';
return string;
}
%}
%union {
int iValue; /* integer value */
char* sValue; /* C-String */
};
%error-verbose /* sometimes provides better error reporting. Useful sometimes when debugging */
%start file /* defining the start condition */
%token BSLASH LBRACE RBRACE LPAREN RPAREN COMMA
%token BEGIN_ END DOCUMENT /* BEGIN seems to be a reserved word so BEGIN_ was used instead */
%token PAGESETUP TABSIZE TITLE AUTHOR DATE
%token <iValue> INTEGER
%token <sValue> DDMMYYYYDATE STRING
%%
file: beginDocument docProperties endDocument
{
fprintf(yyout, "Message");
dealWithDocPropertyErrors();
}
| /* An empty document is parsed to an empty document, no errors generated */
;
beginDocument: BSLASH BEGIN_ LBRACE DOCUMENT RBRACE;
docProperties: docProperties docProperty
| /* empty */
;
/* required properties... there should be one instance of each in the input file */
docProperty: pageSetupProperty { docPropertyCounters[PAGE_SETUP]++; }
| tabSizeProperty { docPropertyCounters[TAB_SIZE]++; }
| titleProperty { docPropertyCounters[DOC_TITLE]++; }
| authorProperty { docPropertyCounters[DOC_AUTHOR]++; }
| dateProperty { docPropertyCounters[DOC_DATE]++; }
;
pageSetupProperty: BSLASH PAGESETUP LBRACE INTEGER COMMA INTEGER RBRACE
{
linesPerPage = ;
charsPerLine = ;
}
;
tabSizeProperty: BSLASH TABSIZE LPAREN INTEGER RPAREN
{
tabSize = ;
}
;
titleProperty: BSLASH TITLE LBRACE STRING RBRACE
{
/* is copied into title excluding the quotation marks at the beginning and end of the string */
title = removeFirstAndLastChar(); /* is a "pseudo-variable" and can be passed in like this */
}
;
authorProperty: BSLASH AUTHOR LBRACE STRING RBRACE
{
author = removeFirstAndLastChar();
}
;
dateProperty: BSLASH DATE LBRACE DDMMYYYYDATE RBRACE
{
date = removeFirstAndLastChar();
}
;
endDocument: BSLASH END LBRACE DOCUMENT RBRACE
{
// Freeing memory created by strdup()
free(title);
free(author);
free(date);
}
;
%%
int yywrap(void) {
return 1;
}
void yyerror(const char* str)
{
fprintf(stderr,"SYNTAX ERROR near line [%d]: %s\n", yylineno, str);
}
以此为输入文件，
\begin {document}
\tabsize( 5)
\pagesetup{30,100 }
\title{"Why I Love Compiler Design"}
\author{"COMP421 Student"}
\date{29/12/2016}
\end{document}
我得到这个输出：
Message
这正是我所期望的。
假设我想在处理文档属性错误后打印消息，我把打印消息的行和它后面的一行的顺序改一下，也就是我做的：
dealWithDocPropertyErrors();
fprintf(yyout, "Message");
我现在得到一个相同输入的空输出文件。我在编译期间没有得到任何 warnings/errors。
我正在使用 Windows 10 和 Flex 2.5.4a。我的 .l 文件在下面
%{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "y.tab.h"
void yyerror(const char *);
int yylex(void);
extern FILE *yyout; /* "Connect" with the output file */
extern int yyparse();
%}
/* %option noyywrap; Tells flex not to declare the function yywrap() which is only useful in the case of more than 1 input file */
/* Allows printing the line number (of an error) */
%option yylineno
/* Catches some errors */
%option nodefault
/* Tells flex not to generate code for the input() and unput() functions which I will not be using */
%option nounput
%option noinput
/* Prints the tokens flex recognizes to the console. Useful when debugging and avoids having to write printf() statements for that */
%option debug
%%
^\ { return BSLASH; } /* every backslash has to be at the start of a line */
\{ { return LBRACE; }
\} { return RBRACE; }
\( { return LPAREN; }
\) { return RPAREN; }
, { return COMMA; }
begin { return BEGIN_; }
end { return END; }
document { return DOCUMENT; }
pagesetup { return PAGESETUP; }
tabsize { return TABSIZE; }
title { return TITLE; }
author { return AUTHOR; }
date { return DATE; }
(((0[1-9]|[12][0-9]|30)[-/ ]?(0[13-9]|1[012])|31[-/ ]?(0[13578]|1[02])|(0[1-9]|1[0-9]|2[0-8])[-/ ]?02)[-/ ]?[0-9]{4}|29[-/ ]?02[-/ ]?([0-9]{2}(([2468][048]|[02468][48])|[13579][26])|([13579][26]|[02468][048]|0[0-9]|1[0-6])00)) { yylval.sValue = strdup(yytext); return DDMMYYYYDATE; }
[0-9]*[1-9][0-9]* { yylval.iValue = atoi(strdup(yytext)); return INTEGER; } /* strdup() is string duplicate. yytext must be copied because of its temporary nature */
\".*\" { yylval.sValue = strdup(yytext); return STRING; }
/* skip whitespace which is not part of a string. [ \t\r\n]+ is better than [ \t\r\n] performance-wise */
[ \t\r\n]+ ;
/* anything else is an error */
. yyerror("invalid character");
%%
int main(int argc, char *argv[]) {
if ( argc != 3)
yyerror("ERROR You need 2 args: inputFileName outputFileName");
else {
yyin = fopen(argv[1], "r");
yyout = fopen(argv[2], "w");
yyparse();
fclose(yyin);
fclose(yyout);
}
return 0;
}
这不是 lex/yacc 问题。
函数dealWithDocPropertyErrors()调用C库函数exit()，终止进程，永不returns。