[ragel-users] Re: Help with extracting a list of numbers

Manoj Rajagopalan ma... at nanorex.com
Wed Mar 26 00:08:31 UTC 2008


Oh and in the program that I sent to this group in my previous email, 
the print_info() action to be executed when the last state ('\0') is 
reached never gets executed:

main :=
         space*
         'bond' %{cerr << "saw 'bond'" << endl;}
         (space+  whole_number %foundNumberAction)+
         space* 0
         @{ print_info();}
      ;

Input:  bond <number> <number>  <number> ...

Here is a twist: when I extract the user input into a character buffer, 
then replace the last NULL char with '\n' and make the next character 
NULL, incrementing pe by 1 and then changing the pattern's last 0 to 
'\n', the machine works perfectly.

Does ragel interpret line-ending NULL chars in some special way?


-- Manoj


Adrian Thurston wrote:
> Hi Manoj,
> 
> Are you sure the null is being sent to the parser? If it is then
> printNumber should be executed on the last character.
> 
> Adrian
> 
> Manoj Rajagopalan wrote:
>> Hi all,
>>
>>    I'm trying to write a pattern to extract a list of numbers following 
>> the keyword bond and I'm not able to figure out why my pattern below 
>> always skips the last number
>>
>>    My input is of the form:
>>
>>    bond <number> <number> ...
>>
>> %%{
>>      machine bond_parser;
>>
>>      whole_number = digit+    >to{intVal=0;}   ${intVal = intVal*10 + 
>> (fc-'0');}    ;
>>
>>      main :=
>>          space* 'bond'
>>          (space+ whole_number %printNumber)+
>>          space* 0;
>> }%%
>>
>> where printNumber is an action to print the number just updated into the 
>> integer variable intVal.
>>
>> The 0 at the end of the pattern is the NULL character that cin.getline() 
>> adds when reading upto the newline character from the console.
>>
>> I realize this may be due to the leaving action for the last number that 
>> never gets called but could someone tell me a way around this?
>>
>> Thanks,
>> Manoj
>>
> 
> > 



More information about the ragel-users mailing list