NAME
HTTP::Request::Form - Construct HTTP::Request objects for form
processing
SYNOPSIS
use the following as a tool to query Altavista for "perl" from
the commandline:
use URI::URL;
use LWP::UserAgent;
use HTTP::Request;
use HTTP::Request::Common;
use HTTP::Request::Form;
use HTML::TreeBuilder 3.0;
my $ua = LWP::UserAgent->new;
my $url = url 'http://www.altavista.digital.com/';
my $res = $ua->request(GET $url);
my $tree = HTML::TreeBuilder->new;
$tree->parse($res->content);
$tree->eof();
my @forms = $tree->find_by_tag_name('FORM');
die "What, no forms in $url?" unless @forms;
my $f = HTTP::Request::Form->new($forms[0], $url);
$f->field("q", "perl");
my $response = $ua->request($f->press("search"));
print $response->content if $response->is_success;
DESCRIPTION
This is an extension of the HTTP::Request suite. It allows easy
processing of forms in a user agent by filling out fields,
querying fields, selections and buttons and pressing buttons. It
uses HTML::TreeBuilder generated parse trees of documents
(especially the forms parts extracted with extract_links) and
generates it's own internal representation of forms from which
it then generates the request objects to process the form
application.
CLASS METHODS
new($form [, $base [, $debug]])
The `new-method' constructs a new form processor. It get's
an HTML::Element object that contains a FORM element or
ISINDEX element as the single parameter. If an base-url is
given as an additional parameter, this is used to make the
form-url absolute in regard to the given URL.
If debugging is true, the following functions will be a bit
"talky" on stdio.
new_many($tree_part [, $base [, $debug]])
The `new_many' method returns a list of newly constructed
form processors. It's just like the `new' method except that
it can apply to any part of an HTML::Element tree, including
the root; it constructs a new form processor for each FORM
element at or under `$tree_part'.
Note that the return list might have zero, one or many new
objects in it, depending on how many FORM (or ISINDEX)
elements were found.
Form elements (like INPUT, etc.) found outside of FORM
elements are counted as being part of the preceding FORM
element. (And if there is no preceding FORM element, they
are ignored.) This feature is useful with the odd parse
trees that can result from basd HTML in or around FORM
elements. If you need to override that feature, then instead
call:
map HTTP::Request::Form->new($_), $tree->find_by_tag_name('FORM');
INSTANCE METHODS
base()
This returns the parameter $base to the "new" constructor.
link()
This returns the action attribute of the original form
structure. This value is cached within the form processor,
so you can safely delete the form structure after you
created the form processor.
method()
This returns the method attribute of the original form
structure. This value is cached within the form processor,
so you can safely delete the form structure as soon as you
created the form processor.
isindex()
This returns true if this came from an original form
structure that was actually an ISINDEX element. In that
case, the form will hagve only one field, an input/text
field named "keywords".
name()
This returns the name attribute of the original form
structure. This value is cached within the form processor,
so you can safely delete the form structure after you
created the form processor.
fields()
This method delivers a list of fieldnames that are of "open"
type. This excludes the "hidden" and "submit" elements,
because they are already filled with a value (and as such
declared as "closed") or as in the case of "submit" are
buttons, of which only one must be used.
allfields()
This delivers a list of all fieldnames in the order as they
occured in the form-source excluding the submit fields.
field($name [, $value])
This method retrieves or sets a field-value. The field is
identified by its name. You have to be sure that you only
put a allowed value into the field.
field_type($name)
This method gives you the type of the named field, so that
you can distinguish on this type. (this is the only way to
distinguish selections and radio buttons).
is_selection($name)
This tests if a field is a selection or an input. Radio-
Buttons are used in the same way as standard selection
fields, so is_selection returns a true value for radio
buttons, too! (Of course, only one value is submitted for a
radio button)
field_selection($name)
This delivers the array of the options of a selection. The
element that is marked with selected in the source is given
as the default value. This works in the same way for radio
buttons, as they are just handled as a special case of
selections!
is_checkbox($name)
This tells you if a field is a checkbox. If it is, there are
several support methods to make use of the special features
of checkboxes, for example the fact that it is only sent if
it is checked.
checkboxes()
This method delivers a list of all checkbox fields much in
the same way as the buttons method.
checkbox_check($name)
checkbox_uncheck($name)
checkbox_toggle($name)
These methods set, unset or toggle the checkbox checked
state. Checkbox values are only added to the result if they
are checked.
checkbox_ischecked($name)
This methods tells you wether a checkbox is checked or not.
This is important if you want to analyze the state of fields
directly after the parse.
buttons()
This delivers a list of all defined and named buttons of a
form.
button($button [, $value])
This gets or sets the value of a button. Normally only
getting a button value is needed. The value of a button is a
reference to an array of values (because a button can exist
multiple times).
button_type($button)
This gives you the type of a button (submit/reset/image).
The result is an array of type names, as a button with one
name can exist multiple times.
button_exists($button)
This gives true if the named button exists, false (undef)
otherwise.
referer([$value])
This returns or sets the referer header for an request. This
is useful if a CGI needs a set referer for authentication.
press([$name [, $coord ] [, $number]])
This method creates a HTTP::Request object (via
HTTP::Request::Common) that sends the formdata to the server
with the requested method. If you give a button-name, that
button is used. If you give no button name, it assumes a
button without a name and just leaves out this last
parameter. If the number of the button is given, that button
value is delivered. If the number is not given, 0 (the first
button of this name) is assumed.
The "coord" parameter comes in handy if you have an image
button. If this is the case, the button press will simulate
a press at coordinates [2,2] unless you provide an anonymous
array with different coordinates.
dump()
This method dumps the form-data on stdio for debugging
purpose.
SEE ALSO
the HTTP::Request manpage, the HTTP::Request::Common manpage,
the LWP::UserAgent manpage, the HTML::Element manpage, the
URI::URL manpage
INSTALLATION
perl Makefile.PL
make install
or see the perlmodinstall manpage
REQUIRES
Perl version 5.004 or later
HTTP::Request::Common
HTML::TreeBuilder
LWP::UserAgent
VERSION
HTTP::Request::Form version 0.9, February 8th, 2001
RESTRICTIONS
Only a subset of all possible form elements are currently
supported. The list of supported tags as of this version
includes:
INPUT/CHECKBOX
INPUT/HIDDEN
INPUT/IMAGE
INPUT/RADIO
INPUT/RESET
INPUT/SUBMIT
INPUT/FILE
INPUT/* (are all handled as simple text entry)
OPTION
SELECT
TEXTAREA
ISINDEX
BUGS
There is currently no support for multiple selections (you can
do them yourself by setting a selection to a comma-delimited
list of values).
Multiple fields are not properly handled, only the last value is
available. Exception are buttons, they are handled in the right
way.
If there are several fields with the same name, you can only set
the value of the first of this fields (this is especially
problematic with checkboxes). This does work with buttons that
have the same name, though (you can press each instance
identified by number).
Error-Checking is currently very limited (not to say
nonexistant).
Support for HTML 4.0 optgroup tags is missing (as is with
allmost all current browsers, so that is not a great loss).
The button tag (HTML 4.0) is just handled as an alias for the
input tag - this is of course incorrect, but sufficient for
support of the usual button types.
COPYRIGHT
Copyright 1998, 1999, 2000 Georg Bauer
<Georg_Bauer@muensterland.org>
This library is free software; you can redistribute it and/or
modify it under the same terms as Perl itself.
MAJOR CONTRIBUTORS
Sean M. Burke (ISINDEX, new_many)