[pdftex] Experiment with a new TeX primitive

[Taco Hoekwater]

Hi,

How about this:

  begin
    scan_toks(false, true);
    str_toks(str_start[experiment(tokens_to_string(def_ref))]);
    ins_list(link(temp_head));
    delete_token_ref(def_ref);
  end

- I don't understand why these call_func's in the pdftex code exist.
  (If anybody knows and can explain that would be great ;))
- str_toks() expects a pool_pointer as argument, not a str_number. 
  this was the main problem with your code.
- ins_list() is somewhat more accurate than back_list().
- you need to call tokens_to_string() so that experiment receives
  a tex string (instead of a token list)

Good luck, Taco


On Wed, 04 Feb 2004 17:40:21 +0100
Hans Hagen <pragma at wxs.nl> wrote:

> At 12:03 31/01/2004, Nagy Bence wrote:
> >Dear developers,
> >
> >I made a small experiment to understand better how web2c works
> >(calling pascal<->C codes), so i defined a new TeX primitive
> >(\experiment). It calls an external C-function to print out its
> >parameter, and this function gives back a new string (like a search and
> >replace engine).
> >
> >The code from the C-file:
> >
> >strnumber experiment(integer t)
> >{
> >     tex_printf("<%s>",makecstring(tokenstostring(t)));
> >     return maketexstring("apple");
> >}
> >
> >The main inserted code from pdftex.ch:
> >
> >@ @<Implement \.{\\experiment}@>=
> >begin
> >     call_func(scan_toks(false, true));
> >     back_list(link(experiment(def_ref)));
> >     delete_token_ref(def_ref);
> >end
> >
> >All works fine except that I can not put back the new string ("apple")
> >to the token list. The function back_list causes a segmentation
> >fault, but it should work like the \lowercase or \uppercase primitive
> >(the experiment C-function gives back an integer value well). What is
> >wrong?
> 
> You need to simulate something:
> 
> \ifx\scratchread \undefined \newread \scratchread  \fi
> \ifx\scratchwrite\undefined \newwrite\scratchwrite \fi
> 
> \let\endofresultline\relax
> \let\resultofrun    \empty
> 
> \def\resultrunfile {\jobname.src.tmp}
> \def\resultrunerror{no result}
> 
> \def\runsystemcommand#1%
>    {\bgroup
>     % reset results
>     \global\let\endofresultline\relax
>     \global\let\resultofrun\empty
>     % erase resultfile
>     \immediate\openout\scratchwrite\resultrunfile
>     \immediate\write\scratchwrite{\resultrunerror}%
>     \immediate\closeout\scratchwrite
>     % run command
>     \immediate\write18{#1 > \resultrunfile}%
>     % define collector
>     \def\runsystemcommand
>       {\ifeof\scratchread \else
>          \endlinechar -1
>          \readline\scratchread to \next
>          \xdef\resultofrun{\resultofrun\next\endofresultline}%
>          \expandafter\runsystemcommand
>        \fi}%
>     % fetch result
>     \openin\scratchread\resultrunfile\relax
>     \runsystemcommand
>     \closein\scratchread
>     % done
>     \egroup}
> 
> \starttext
> 
> \runsystemcommand{dir /s} \let\endofresultline\par \tt \resultofrun
> 
> \stoptext
> 
> so, the best place to look is in the etex code and see how the \readline 
> results end up in tex
> 
> Hans  
> 
> _______________________________________________
> pdftex mailing list
> pdftex at tug.org
> http://tug.org/mailman/listinfo/pdftex