Date:         Fri, 15 Jun 2007 11:39:17 +0200
Reply-To:     SAS-L List <sas-l@listserv.uga.edu>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Robert Bardos <bardos2@ANSYS.CH>
Subject:      Re: how to import one txt file?
In-Reply-To:  <002101c7ae96$aa277af0$bd00a8c0@adriano>
Content-Type: text/plain; charset="iso-8859-1"

Adriano,

  Pointer to Adriano's post in SAS-L's archive:
  http://tinyurl.com/2qmuoa

it seems you want to read the content of emails that were
generated by people filling in some kind of form. Off the top of
my head I see two approaches

Approach one which relies on all emails having the exactly same
number and layout of lines [*1] would use an input statement
covering an exact number of lines somewhat like

  input #1 @'From:' ....
        #2 @'Date:' ....
        #3 @'To:'   ....
        ...
        ;

etc.

[*1] When using relative line numbers like #1, #2, #3 I'm assuming
that your original message was not interspersed with all those
empty lines. If the empty lines are part of your data, you will
have to adjust the line numbers accordingly. Anyway: do take your
time to study the powerful INPUT statement!

Approach two would read line by line, identify the 'record type',
parse it accordingly and write out an observation when an email
boundary is reached.
Somewhat like

 data ...;  /* <-- data step begins here */
  infile ...;
  input @;
  word_one = scan(_infile_,1,' ');
  select (word_one);
    when ('From:' ) link parse_from ;
    when ('Date:' ) link parse_date ;
    ...
    when ('1'     ) link process_question1 ;
    when ('2'     ) link process_question2 ;
    ...
    otherwise
      ...
  end;
 return;

 parse_date:
    input @'Date:' date_string $char40. ;
    /* more logic here */
 return;

 parse_from;
    /* potential place for boundary logic */
    input @'From:' from_string $char100. ;
    /* more logic here */
 return;

 process_question1:
    input #2 answer1 $8. ;
    /* more logic here */
 return;

 process_question2:
    input #2 answer2 $8. ;
    /* more logic here */
 return;

 run;  /* <-- data step ends here */


Adriano, all of this is untested and intended to give you an idea
only. There will be lots of problems that you encounter when
trying to parse your files. Tackle the problems one by one and try
to understand the various ERROR and NOTE messages that will most
certainly show up. Seems like a good learning exercise.


Kind regards

Robert Bardos
Ansys AG, Zurich, Switzerland


> -----Urspruengliche Nachricht-----
> Adriano Rodrigues - Instituto GPP
> Gesendet: Donnerstag, 14. Juni 2007 17:14
>
> Hi all,
>
> I have to import one file received from emails with
> this layout (here only 2 emails, but file have 5000+):
>
> Basically after all “:” I have data and after “?” also,
> only in time “:” is not true.
>
> Thanks in advance,
>
> Adriano
>
  Rest snipped as URL to original post has been provided above  /
rb.