parsing problem with lex/yacc - yylval returns unintended value for $1

[rajatkmitra]

Hi Folks,
I am writing a simple parser to parse SPICE code ( Spice is a transistor level simulator ). The spice code looks like this -
=============== =========
R10 2 3 0.5e-3
C11 4 6 0.2e-6
L13 4 4 0.1e-9
=============== ==========
As you can see each line has 4 items -
a character string followed by an integer then another integer and finally a floating point number.
I need to return the appropriate token on a per line basis.

I am attaching a tar ball called foo.zip that has the following files -
1. netlist ( the source file to be parsed )
2. spice.lex ( the lexer code )
3. spice.y ( the grammar )
4. Makefile

To build the parser, simple type make ( on linux ). Then to run -
%parser < netlist

In the .y file I have the grammar rule where I expect $1 to return the character string, $2 to return the first integer, $3 to return the next integer value and finally $4 to return the floating point value on a per line basis. Mos of this works correctly, except $1 returns the entire line instead of the character string. Can anyone tell what's wrong in the token generator ( lex file ) or the grammar description ( .y file ) ?? Any help will be highly appreciated.

Thanks,

Rajat Mitra

[weaknessforcats]

I can't read you files, but if your data is as yuou say it is then:

I typed in: R10 2 3 0.5e-3

int main()
{
   char str[4];
   int var1;
   int var2;
   float var3;

   scanf("%s%d%d%f\n", str, &var1, &var2, &var3);

   printf("%s\n%d\n%d\n%f\n", str, var1, var2, var3);
}

works fine. Just use fscanf with your file.

[rajatkmitra]

I can't read you files, but if your data is as yuou say it is then:

I typed in: R10 2 3 0.5e-3

int main()
{
   char str[4];
   int var1;
   int var2;
   float var3;

   scanf("%s%d%d%f\n", str, &var1, &var2, &var3);

   printf("%s\n%d\n%d\n%f\n", str, var1, var2, var3);
}

works fine. Just use fscanf with your file.

=============== =============== =============== ===========
Hi weaknessforcats ,
Many thanks for your reply. Unfortunately while the fscanf works very reliably, this parser is going to have to grow to support additional ( optional ) parameters on a per line basis ( or even multiline )and needs to be written using lex/yacc.
Can you unzip the archive that I have attached the source files are in there for you to take a look at ???

Thanks again for your help,
Rajat

[donbock]

Are you required to implement this parser with lex/yacc?
I think it would be a lot easier to do this in perl.

[rajatkmitra]

Are you required to implement this parser with lex/yacc?
I think it would be a lot easier to do this in perl.

yeah it has to be lex/yacc to fit in with the rest of the software my colleague is developing. you are correct perl would be definitely the easier way yo to do...but we have legacy code that is hooked into the lex/yacc structure.

[shankarke]

Hi Folks,
I am writing a simple parser to parse SPICE code ( Spice is a transistor level simulator ). The spice code looks like this -
=============== =========
R10 2 3 0.5e-3
C11 4 6 0.2e-6
L13 4 4 0.1e-9
=============== ==========
As you can see each line has 4 items -
a character string followed by an integer then another integer and finally a floating point number.
I need to return the appropriate token on a per line basis.

I am attaching a tar ball called foo.zip that has the following files -
1. netlist ( the source file to be parsed )
2. spice.lex ( the lexer code )
3. spice.y ( the grammar )
4. Makefile

To build the parser, simple type make ( on linux ). Then to run -
%parser < netlist

In the .y file I have the grammar rule where I expect $1 to return the character string, $2 to return the first integer, $3 to return the next integer value and finally $4 to return the floating point value on a per line basis. Mos of this works correctly, except $1 returns the entire line instead of the character string. Can anyone tell what's wrong in the token generator ( lex file ) or the grammar description ( .y file ) ?? Any help will be highly appreciated.

Thanks,

Rajat Mitra

you would need to do a strdup(yytext) in the code to get it working

[rajatkmitra]

you would need to do a strdup(yytext) in the code to get it working

That most definitely worked !!! Thanks a ton...

-Rajat