Book HomePerl & LWPSearch this book

5.3. Automating Form Analysis

Rather than searching through HTML hoping that you've found all the form components, you can automate the task. Example 5-2 contains a program, formpairs.pl, that extracts the names and values from GET or POST requests.

Example 5-2. formpairs.pl

#!/usr/local/bin/perl -w
# formpairs.pl - extract names and values from HTTP requests

use strict;
my $data;
if(! $ENV{'REQUEST_METHOD'}) { # not run as a CGI
  die "Usage: $0 \"url\"\n" unless $ARGV[0];
  $data = $ARGV[0];
  $data = $1 if $data =~ s/^\w+\:.*?\?(.+)//;
  print "Data from that URL:\n(\n";
} elsif($ENV{'REQUEST_METHOD'} eq 'POST') {
  read(STDIN, $data, $ENV{'CONTENT_LENGTH'});
  print "Content-type: text/plain\n\nPOST data:\n(\n";
} else {
  $data = $ENV{'QUERY_STRING'};
  print "Content-type: text/plain\n\nGET data:\n(\n";
}
for (split '&', $data, -1) {   # Assumes proper URLencoded input
  tr/+/ /;   s/"/\\"/g;   s/=/\" => \"/;   s/%20/ /g;
  s/%/\\x/g;  # so %0d => \x0d
  print "  \"$_\",\n";
}
print ")\n";

That program, when run as a command-line utility, takes a URL as its one argument, decodes the encoded GET query, and prints it in more Perlish terms:

% perl formpairs.pl "http://www.census.gov/cgi-bin/gazetteer?city=IEG
&state=&zip="
Data from that URL:
(
  "city" => "IEG",
  "state" => "",
  "zip" => "",
)

Using a more complex URL (wrapped here for readability) illustrates the benefit of it:

% perl -w formpairs.pl http://www.altavista.com/sites/search/web?q=
pie+AND+rhubarb+AND+strawberry%0D%0AAND+NOT+crumb&kl=en&r=&dt=tmperiod
&d2=0&d0=&d1=&sc=on&nbq=30&pg=aq&search=Search
Data from that URL:
(
  "q" => "pie AND rhubarb AND strawberry\x0D\x0AAND NOT crumb",
  "kl" => "en",
  "r" => "",
  "dt" => "tmperiod",
  "d2" => "0",
  "d0" => "",
  "d1" => "",
  "sc" => "on",
  "nbq" => "30",
  "pg" => "aq",
  "search" => "Search",
)

The same program also functions as a CGI, so if you want to see what data a given form ends up submitting, you can simply change the form element's action attribute to a URL where you've set up that program as a CGI. As a CGI, it accepts both GET and POST methods.

For example:

<form method="post" action="http://myhost.int/cgi-bin/formpairs.pl">
Kind of pie: <input name="what pie" size=15>
<input type="submit" value="Mmm pie">
</form>

When you fill the one blank out with "tasty pie!" and press the "Mmm pie" button, the CGI will print:

POST data:
(
  "what pie" => "tasty pie\x21",
)

A more ad hoc solution that doesn't involve bothering with a CGI is to take the local copy of the form, set the form tag's method attribute to get, set its action attribute to dummy.txt, and create a file dummy.txt consisting of the text "Look at my URL!" or the like. Then, when you submit the form, you will see only the "Look at my URL!" page, but the browser's "Location"/"Address"/"URL" window will show a URL like this:

file:///C%7C/form_work/dummy.txt?what+pie=tasty+pie%21

You can then copy that URL into a shell window as the argument to formpairs.pl:

% perl formpairs.pl "file:///C%7C/form_work/dummy.txt?what+pie=tasty+pie%21"
Data from that URL:
(
  "what pie" => "tasty pie\x21",
)


Library Navigation Links

Copyright © 2002 O'Reilly & Associates. All rights reserved.