I did try both of those .. without success.

I got a pdf I've created with openoffice and pdftotext is able to extract text from it, whereas CAM::PDF (or File::Extract::PDF) gives me messy characters.

[jerome@saab pdf]$ getpdftext.pl -v ~/faxTaxHabitation2005.pdf

       





 
  
                 
                      
    ! " #  $ 

%  # & '     (
"  ) *

        + + +
...
[download]

And pdftotext:

[jerome@saab pdf]$ pdftotext ~/faxTaxHabitation2005.pdf txt
[jerome@saab pdf]$ tail txt

Merci de bien vouloir me confirmer ces informations par retour de fax 
+afin que je puisse proceder au paiment le plus rapidement possible au
+ numero suivant : *************

Cordiales salutations.
...
[download]

The ideal would be a perl module linked to the xpdf C code .. :)