=head1 NAME

CGI::Ex::Validate - The "Just Right" form validator with javascript in parallel

=head1 SYNOPSIS

use CGI::Ex::Validate;

    # THE SHORT

    my $errobj = CGI::Ex::Validate->new->validate($form, $val_hash);

    # THE LONG

    my $form = CGI->new;
    # OR #
    my $form = CGI::Ex->new; # OR CGI::Ex->get_form;
    # OR #
    my $form = {key1 => 'val1', key2 => 'val2'};


    # simplest
    my $val_hash = {
        'group order' => [qw(username email email2)],
        username => {
            required => 1,
            max_len  => 30,
            field    => 'username',
            # field is optional in this case - will use key name
        },
        email    => {
            required => 1,
            max_len  => 100,
            type     => 'email',
        },
        email2   => {
            equals   => 'email',
        },
    };

    # ordered
    my $val_hash = {
        'group order' => [{
            field    => 'username', # field is not optional in this case
            required => 1,
            max_len  => 30,
        }, {
            field    => 'email',
            required => 1,
            max_len  => 100,
        }, {
            field    => 'email2',
            equals   => 'email',
        }],
    };


    my $vob    = CGI::Ex::Validate->new;
    my $errobj = $vob->validate($form, $val_hash);
    if ($errobj) {
        # get errors back in any of several useful ways
        my $error_heading = $errobj->as_string; # OR "$errobj";
        my $error_list    = $errobj->as_array;  # ordered list of what when wrong
        my $error_hash    = $errobj->as_hash;   # hash of arrayrefs of errors
    } else {
        # the form passed validation
    }


    my $js_uri_path = '/js/';     # static or dynamic URI path to find CGI/Ex/validate.js
    my $form_name   = "the_form"; # name of the form to attach javascript to

    # generate javascript to validate an existing form
    my $javascript = $vob->generate_js($val_hash, {
        form_name   => $form_name,
        js_uri_path => $js_uri_path,
    });

    # OR let Validate create the form and javascript for you
    my $form = $vob->generate_form($val_hash, {
        form_name   => $form_name,   # will use a random name if not passed
        js_uri_path => $js_uri_path,
    });


=head1 DESCRIPTION

CGI::Ex::Validate is one of many validation modules.  It aims to have
all of the basic data validation functions, avoid adding all of
the millions of possible types, while still giving the capability
for the developer to add their own types for the rare cases that
the basic ones don't suffice.  Generally anything more than basic
validation probably needs programmatic or data based validation.

CGI::Ex::Validate also has full support for providing the same
validation in javascript.  It provides methods for attaching the
javascript to existing forms.  This ability is tightly integrated into
CGI::Ex::App, but it should be easy to add validation just about
anywhere using any type of controller.

As opposed to other kitchen sync validation modules, CGI::Ex::Validate
offers the simple types of validation, and makes it easy to add your
own custom types.  Asside from custom and custom_js, all validation
markup is declarative.

=head1 METHODS

=over 4

=item C<new>

Used to instantiate the object.  Arguments are either a hash, or hashref,
or nothing at all.  Keys of the hash become the keys of the object.

=item C<get_validation>

Uses CGI::Ex::Conf::conf_read to read in the hash.  conf_read will all passing
a filename or YAML string or a hashref.

=item C<get_validation_keys>

Takes the validation hashref returned from get_validation.  Will return all
of the possible keys found in the validation hashref.  This can be used to
check to see if extra items have been passed to validate.  If a second
argument contains a form hash is passed, get_validation_keys will only
return the keys of groups that were validated.

    my $key_hashref = $self->get_validation_keys($val_hash);

The keys of the hash are the names of the fields.

=item C<validate>

Arguments are a form hashref or cgi object, a validation hashref or
filename, and an optional what_was_validated arrayref (discussed
further later on).  If a CGI object is passed, CGI::Ex::get_form will
be called on that object to turn it into a hashref.  If a filename is
given for the validation, get_validation will be called on that
filename.  If the what_was_validated_arrayref is passed - it will be
populated (pushed) with the field hashes that were actually validated
(anything that was skipped because of validate_if will not be in the
array).

If the form passes validation, validate will return undef.  If it
fails validation, it will return a CGI::Ex::Validate::Error object.
If the 'raise_error' option has been set, validate will die with a
CGI::Ex::validate::Error object as the value.

    my $err_obj = $self->validate($form, $val_hash);

    # OR #

    $self->{raise_error} = 1; # can also be listed in the val_hash
    eval { $self->validate($form, $val_hash) };
    if ($@) { my $err_obj = $@; }

=item C<generate_form>

Takes a validation hash, and additional arguments and generates an HTML form suitable
for inclusion in a web based application.

    my $html = $self->generate_form($val_hash, {
        form_name   => 'my_form',
        js_uri_path => '/cgi-bin/js', # will be used by generate_js
    });

=item C<generate_js>

Works with CGI::Ex::JSONDump.

Takes a validation hash, a form name, and an optional javascript uri
path and returns Javascript that can be embedded on a page and will
perform identical validations as the server side.  The form name must be
the name of the form that the validation will act upon - the name is
used to register an onsubmit function.  The javascript uri path is
used to embed the locations of javascript source files included
with the CGI::Ex distribution.

The javascript uri path is highly dependent upon the server
configuration and therefore must be configured manually.  It may be
passed to generate_js, or it may be specified in $JS_URI_PATH.
There is one file included with this module that is needed -
CGI/Ex/validate.js.  When generating the js code, generate_js will
look in $JS_URI_PATH_VALIDATE.  If this is not set,
generate_js will use "$JS_URI_PATH/CGI/Ex/validate.js".

    my $js = $self->generate_js($val_hash, 'my_form', "/cgi-bin/js")
    # OR
    my $js = $self->generate_js($val_hash, {
        form_name   => 'my_form',
        js_uri_path => '/cgi-bin/js',
    });

    # would generate something like the following...

    <script src="/cgi-bin/js/CGI/Ex/validate.js"></script>
    ... more js follows ...

    $CGI::Ex::Validate::JS_URI_PATH = "/stock/js";
    $self->generate_js($val_hash, 'my_form')

    # would generate something like the following...

    <script src="/stock/js/CGI/Ex/validate.js"></script>
    ... more js follows ...

Referencing validate.js can be done in any of
several ways.  It can be copied to or symlinked to a fixed location
in the server's html directory.  It can also be printed out by a cgi.
The method C<-E<gt>print_js> has been provided in CGI::Ex for printing
js files found in the perl hierarchy.  See L<CGI::Ex> for more details.
The $JS_URI_PATH of "/cgi-bin/js" could contain the following:

    #!/usr/bin/perl -w

    use strict;
    use CGI::Ex;

    # path_info should contain something like /CGI/Ex/validate.js
    my $info = $ENV{PATH_INFO} || '';
    die "Invalid path" if $info !~ m|^(/\w+)+.js$|;
    $info =~ s|^/+||;

    CGI::Ex->new->print_js($info);
    exit;

The print_js method in CGI::Ex is designed to cache the javascript in
the browser.

=item C<-E<gt>cgix>

Returns a CGI::Ex object.  Used internally if a CGI object is
passed to validate rather than a straight form hash.

=back

=head1 VALIDATION HASH

The validation hash may be passed as a hashref or as a filename, or as
a YAML document string.  Experience has shown it to be better
programming to pass in a hashref.  If the validation "hash" is a
filename or a YAML string, it will be translated into a hash using
CGI::Ex::Conf.

Keys matching the regex m/^group \s+ (\w+)$/x such as "group
onevent" are reserved and are counted as GROUP OPTIONS.  Other keys
(if any, should be field names that need validation).

If the GROUP OPTION 'group validate_if' is set, the validation will
only be validated if the conditions of the validate_if are met.  If
'group validate_if' is not specified, then the validation will
proceed.  See the validate_if VALIDATION type for more information.

Each of the items listed in the validation will be validated.  The
validation order is determined the following ways:

=over 4

=item Specify 'group order' arrayref with hashrefs.

    # order will be (username, password, 'm/\w+_foo/', somethingelse)
    {
        'group title' => "User Information",
        'group order' => [
            {field => 'username',   required => 1},
            {field => 'password',   required => 1},
            {field => 'm/\w+_foo/', required => 1},
        ],
        somethingelse => {required => 1},
        }

=item Specify 'group order' arrayref with field key names.

    # order will be (username, password, 'm/\w+_foo/', somethingelse)
    {
        'group title' => "User Information",
        'group order' => [qw(username password), 'm/\w+_foo/'],
        username      => {required => 1},
        password      => {required => 1},
        'm/\w+_foo/'  => {required => 1},
        somethingelse => {required => 1},
    }

=item Do nothing - use sorted order.

    # order will be ('m/\w+_foo/', password, somethingelse, username)
    {
        'group title' => "User Information",
        username      => {required => 1},
        password      => {required => 1},
        'm/\w+_foo/'  => {required => 1},
        somethingelse => {required => 1},
    }

=back

Optionally the 'group order' may contain the word 'OR' as a special
keyword.  If the item preceding 'OR' fails validation the item after
'OR' will be tested instead.  If the item preceding 'OR' passes
validation the item after 'OR' will not be tested.

    'group order' => [qw(zip OR postalcode state OR region)],

At this time, only "group onevent" submit works with this option.
Using OR is not needed if testing for one or more values -- instead you
should use min_in_set or max_in_set (OR is still useful for other cases).

    'zip' => {
      max_in_set: '1 of zip, postalcode',
    },
    'state' => {
      max_in_set: '1 of state, region',
    },

Each individual field validation hashref will operate on the field contained
in the 'field' key.  This key may also be a regular expression in the
form of 'm/somepattern/'.  If a regular expression is used, all keys
matching that pattern will be validated.  If the field key is
not specified, the key from the top level hash will be used.

    foobar => {   # "foobar" is not used as key because field is specified
        field    => 'real_key_name',
        required => 1,
    },
    real_key_name2 => {
        required => 1,
    },
    'm/\w+/' => { # this will apply to all fields matching this regex
        required => 1,
    },

Each of the individual field validation hashrefs should contain the
types listed in VALIDATION TYPES.

=head1 VALIDATION TYPES

This section lists the available validation types.  Multiple instances
of the same type may be used for some validation types by adding a
number to the type (ie match, match2, match232).  Multiple instances
are validated in sorted order.  Types that allow multiple values are:
compare, custom, custom_js, equals, match, required_if, sql, validate_if,
and replace (replace is a MODIFICATION TYPE).

=over 4

=item C<compare>

Allows for custom comparisons.  Available types are
>, <, >=, <=, !=, ==, gt, lt, ge, le, ne, and eq.  Comparisons
also work in the JS.

    {
        field    => 'my_number',
        match    => 'm/^\d+$/',
        compare1 => '> 100',
        compare2 => '< 255',
        compare3 => '!= 150',
    }

=item C<custom>

Custom value - not available in JS.  Allows for extra programming types.
May be either a boolean value predetermined before calling validate, or may be
a coderef that will be called during validation.  If coderef is called, it will
be passed the field name, the form value for that name, and a reference to the
field validation hash.  If the custom type returns false the element fails
validation and an error is added.

    {
        field => 'username',
        custom => sub {
            my ($key, $val, $field_val_hash, $checktype, $form) = @_;
            # do something here
            return 0;
        },
        custom_error => '$name was not valid',
    }

Often it is desirable to specify a different message depending upon
the code passed to custom.  To use a custom error message simply die
with the error message.  Note that you will want to add a newline or
else perl will add the line number and file for you -
CGI::Ex::Validate will remove the trailing newline.

    {
        field => 'username',
        custom => sub {
            my ($key, $val) = @_;
            die "Custom error message 1\n" if $val eq '1';
            die "Custom error message 2\n" if $val eq '2';
            return 0;
        },
        custom_error => '$name default custom error message',
    }

=item C<custom_js>

Custom value - only available in JS.  Allows for extra programming
types.  May be a javascript function (if fully declared in
javascript), a string containing a javascript function (that will be
eval'ed into a real function), a boolean value pre-determined before
calling validate, or may be section of javascript that will be eval'ed
(the last value of the eval'ed javascript will determine if validation
passed).  A false response indicates the value did not pass
validation.  A true response indicates that it did.  See the
samples/validate_js_0_tests.html page for a sample of usages.

    {
        field => 'date',
        required => 1,
        match    => 'm|^\d\d\d\d/\d\d/\d\d$|',
        match_error => 'Please enter date in YYYY/MM/DD format',
        custom_js => "function (args) {
            var t = new Date();
            var y = t.getYear()+1900;
            var m = t.getMonth() + 1;
            var d = t.getDate();
            if (m < 10) m = '0'+m;
            if (d < 10) d = '0'+d;
            (args.value > ''+y+'/'+m+'/'+d) ? 1 : 0;
        }",
        custom_js_error => 'The date was not greater than today.',
    }

Often it is desirable to specify a different message depending upon
the function passed to custom_js.  To use a custom error message simply throw
the error message.

    {
        field => 'username',
        custom_js => 'function (args) {
            if (args.value == 1) throw "Custom error message 1";
            if (args.value == 2) throw "Custom error message 2";
            return 0;
        }',
        custom_js_error => '$name default custom error message',
    }

=item C<enum>

Allows for checking whether an item matches a set of options.  In perl
the value may be passed as an arrayref.  In the conf or in perl the
value may be passed of the options joined with ||.

    {
        field => 'password_type',
        enum  => 'plaintext||crypt||md5', # OR enum => [qw(plaintext crypt md5)],
    }

=item C<equals>

Allows for comparison of two form elements.  Can have an optional !.

    {
        field  => 'password',
        equals => 'password_verify',
    },
    {
        field  => 'domain1',
        equals => '!domain2', # make sure the fields are not the same
    }

=item C<had_error>

Typically used by a validate_if.  Allows for checking if this item has had
an error.

    {
        field => 'alt_password',
        validate_if => {field => 'password', had_error => 1},
    }

This is basically the opposite of was_valid.

=item C<match>

Allows for regular expression comparison.  Multiple matches may
be concatenated with ||.  Available in JS.

    {
        field   => 'my_ip',
        match   => 'm/^\d{1,3}(\.\d{1,3})3$/',
        match_2 => '!m/^0\./ || !m/^192\./',
    }

=item C<max_in_set> and C<min_in_set>

Somewhat like min_values and max_values except that you specify the
fields that participate in the count.  Also - entries that are not
defined or do not have length are not counted.  An optional "of" can
be placed after the number for human readability.

    min_in_set => "2 of foo bar baz",
      # two of the fields foo, bar or baz must be set
      # same as
    min_in_set => "2 foo bar baz",
      # same as
    min_in_set => "2 OF foo bar baz",

    validate_if => {field => 'whatever', max_in_set => '0 of whatever'},
      # only run validation if there were zero occurrences of whatever

=item C<max_len and min_len>

Allows for check on the length of fields

    {
        field   => 'site',
        min_len => 4,
        max_len => 100,
    }

=item C<max_values> and C<min_values>

Allows for specifying the maximum number of form elements passed.
max_values defaults to 1 (You must explicitly set it higher
to allow more than one item by any given name).

=item C<required>

Requires the form field to have some value.  If the field is not present,
no other checks will be run and an error will be given.

It has been common for code to try C<required => 0> which essentially has
no effect - instead use C<validate_if => 'fieldname', required => 1>.  This
results in the fieldname only being required if the fieldname is present.

=item C<required_if>

Requires the form field if the condition is satisfied.  The conditions
available are the same as for validate_if.  This is somewhat the same
as saying:

    validate_if => 'some_condition',
    required    => 1

    required_if => 'some_condition',

It is different in that other checks will run - whereas validate_if skips
all validation if some condition is not met.

If a regex is used for the field name, the required_if
field will have any match patterns swapped in.

    {
        field       => 'm/^(\w+)_pass/',
        required_if => '$1_user',
    }

This example would require the "foobar_pass" field to be set
if the "foobar_user" field was passed.

=item C<sql>

SQL query based - not available in JS.  The database handle will be looked
for in the value $self->{dbhs}->{foo} if sql_db_type is set to 'foo',
otherwise it will default to $self->{dbh}.  If $self->{dbhs}->{foo} or
$self->{dbh} is a coderef - they will be called and should return a dbh.

    {
        field => 'username',
        sql   => 'SELECT COUNT(*) FROM users WHERE username = ?',
        sql_error_if => 1, # default is 1 - set to 0 to negate result
        # sql_db_type  => 'foo', # will look for a dbh under $self->{dbhs}->{foo}
    }

=item C<type>

Allows for more strict type checking.  Currently supported types
include CC (credit card), EMAIL, DOMAIN, IP, URL, INT, UINT, and NUM.
Other types will be added upon request provided we can add a perl and
a javascript version (extra types often aren't necessary as the custom and
custom_js options give arbitrary checking).  If a type checks fails - other
compare, custom, or length checks will not be ran.

    {
        field => 'credit_card',
        type  => 'CC',
    }

=over 4

=item C<CC>

Simple Luhn-10 check.  Note that spaces and dashes are left in place.

=item C<EMAIL>

Very basic check to see if the value looks like an address.  The local part
must only contain [\w.~!\#\$%\^&*\-=+?] and the domain must be a domain or ip.
If you want full fledged RFC compliant checking consider something like:

    {
        field => 'email',
        custom => sub {
            my ($key, $val, $fv, $type, $form) = @_;
            require Mail::Address;
            my @a = Mail::Address->parse($val);
            die "Invalid address\n" if @a != 1;
            return $form->{$key} = $a[0]->address;
         },
     }

=item C<DOMAIN>

Checks for a valid domain name - does no lookup of the domain.  For that use
a custom sub.

=item C<IP>

Basic IPv4 check.

=item C<URL>

Basic check that string matches something resembling an http or https url.

=item C<INT>

Checks for an integer between -2147483648 and 2147483648

=item C<UINT>

Checks for an unsigned integer between 0 and 4294967295.

=item C<NUM>

Checks for something that looks like a number.  Scientic notation is not allowed.  No range enforced.

=back

=item C<validate_if>

If validate_if is specified, the field will only be validated
if the conditions are met.  Works in JS.

    validate_if => {field => 'name', required => 1, max_len => 30}
    # Will only validate if the field "name" is present and is less than 30 chars.

    validate_if => 'name',
    # SAME as
    validate_if => {field => 'name', required => 1},

    validate_if => '! name',
    # SAME as
    validate_if => {field => 'name', max_in_set => '0 of name'},

    validate_if => 'name was_valid',
    # SAME as
    validate_if => {field => 'name', was_valid => 1},

    validate_if => {field => 'country', compare => "eq US"},
    # only if country's value is equal to US

    validate_if => {field => 'country', compare => "ne US"},
    # if country doesn't equal US

    validate_if => {field => 'password', match => 'm/^md5\([a-z0-9]{20}\)$/'},
    # if password looks like md5(12345678901234567890)

    {
        field       => 'm/^(\w+)_pass/',
        validate_if => '$1_user',
        required    => 1,
    },
    {
        field       => 'm/^(\w+)_pass2/',
        validate_if => '$1_pass',
        equals      => '$1_pass',
        required    => 1,
    }

    # will validate foo_pass only if foo_user was present.

The validate_if may also contain an arrayref of validation items.  So that
multiple checks can be run.  They will be run in order.  validate_if will
return true only if all options returned true.

    validate_if => ['email', 'phone', 'fax']

Optionally, if validate_if is an arrayref, it may contain the word
'OR' as a special keyword.  If the item preceding 'OR' fails validation
the item after 'OR' will be tested instead.  If the item preceding 'OR'
passes validation the item after 'OR' will not be tested.

    validate_if => [qw(zip OR postalcode)],

=item C<was_valid>

Typically used by a validate_if.  Allows for checking if this item has successfully
been validated.

    {
        field => 'password2',
        validate_if => {field => 'password', was_valid => 1},
    }

This is basically the opposite of had_error.

=back

=head1 SPECIAL VALIDATION TYPES

=over 4

=item C<field>

Specify which field to work on.  Key may be a regex in the form
'm/\w+_user/'.  This key is required in a hashref passed to 'group
order'.  It can optionally be used with other types to specify a
different form element to operate on.  On errors, if a non-default
error is found, $field will be swapped with the value found in field.

The field name may also be a regular expression in the
form of 'm/somepattern/'.  If a regular expression is used, all keys
matching that pattern will be validated.

=item C<name>

Name to use for errors.  If a name is not specified, default errors will use
"The field $field" as the name.  If a non-default error is found, $name
will be swapped with this name.

=item C<delegate_error>

This option allows for any errors generated on a field to delegate to
a different field.  If the field name was a regex, any patterns will
be swapped into the delegate_error value. This option is generally only
useful with the as_hash method of the error object (for inline errors).

    {
        field => 'zip',
        match => 'm/^\d{5}/',
    },
    {
        field => 'zip_plus4',
        match => 'm/^\d{4}/',
        delegate_error => 'zip',
    },
    {
        field => 'm/^(id_[\d+])_user$/',
        delegate_error => '$1',
    },

=item C<exclude_js>

This allows the cgi to do checking while keeping the checks from
being run in JavaScript

    {
        field      => 'cgi_var',
        required   => 1,
        exclude_js => 1,
    }

=item C<exclude_cgi>

This allows the js to do checking while keeping the checks from
being run in the cgi

    {
        field       => 'js_var',
        required    => 1,
        exclude_cgi => 1,
    }

=item C<vif_disable>

Only functions in javascript.  Will mark set the form element to
disabled if validate_if fails.  It will mark it as enabled if
validate_if is successful.  This item should normally only be used
when onevent includes "change" or "blur".

=back

=head1 MODIFYING VALIDATION TYPES

The following types will modify the form value before it is processed.
They work in both the perl and in javascript as well.  The javascript
version changes the actual value in the form on appropriate form types.

=over 4

=item C<do_not_trim>

By default, validate will trim leading and trailing whitespace
from submitted values.  Set do_not_trim to 1 to allow it to
not trim.

    {field => 'foo', do_not_trim => 1}

=item C<trim_control_chars>

Off by default.  If set to true, removes characters in the
\x00 to \x31 range (Tabs are translated to a single space).

    {field => 'foo', trim_control_chars => 1}

=item C<replace>

Pass a swap pattern to change the actual value of the form.
Any perl regex can be passed but it is suggested that javascript
compatible regexes are used to make generate_js possible.

    {field => 'foo', replace => 's/(\d{3})(\d{3})(\d{3})/($1) $2-$3/'}

=item C<default>

Set item to default value if there is no existing value (undefined
or zero length string).

    {field => 'country', default => 'EN'}

=item C<to_upper_case> and C<to_lower_case>

Do what they say they do.

=item C<untaint>

Requires that the validated field has been also checked with
an enum, equals, match, compare, custom, or type check.  If the
field has been checked and there are no errors - the field is "untainted."

This is for use in conjunction with perl's -T switch.

=item C<clear_on_error>

Clears the form field should a validation error occur.  Only supported
on the Javascript side (no affect on the server side).

=back

=head1 ERROR OBJECT

Failed validation results in an error an error object created via the
new_error method.  The default error class is CGI::Ex::Validate::Error.

The error object has several methods for determining what the errors were.

=over 4

=item C<as_array>

Returns an array or arrayref (depending on scalar context) of errors that
occurred in the order that they occurred.  Individual groups may have a heading
and the entire validation will have a heading (the default heading can be changed
via the 'as_array_title' group option).  Each error that occurred is a separate
item and are pre-pended with 'as_array_prefix' (which is a group option - default
is '  ').  The as_array_ options may also be set via a hashref passed to as_array.
as_array_title defaults to 'Please correct the following items:'.

    # if this returns the following
    my $array = $err_obj->as_array;
    # $array looks like
    # ['Please correct the following items:', '  error1', '  error2']

    # then this would return the following
    my $array = $err_obj->as_array({
        as_array_prefix => '  - ',
        as_array_title  => 'Something went wrong:',
    });
    # $array looks like
    # ['Something went wrong:', '  - error1', '  - error2']

=item C<as_string>

Returns values of as_array joined with a newline.  This method is used as
the stringification for the error object.  Values of as_array are joined with
'as_string_join' which defaults to "\n".  If 'as_string_header' is set, it will
be pre-pended onto the error string.  If 'as_string_footer' is set, it will be
appended onto the error string.

    # if this returns the following
    my $string = $err_obj->as_string;
    # $string looks like
    # "Please correct the following items:\n  error1\n  error2"

    # then this would return the following
    my $string = $err_obj->as_string({
        as_array_prefix  => '  - ',
        as_array_title   => 'Something went wrong:',
        as_string_join   => '<br />',
        as_string_header => '<span class="error">',
        as_string_footer => '</span>',
    });
    # $string looks like
    # '<span class="error">Something went wrong:<br />  - error1<br />  - error2</span>'

=item C<as_hash>

Returns a hash or hashref (depending on scalar context) of errors that
occurred.  Each key is the field name of the form that failed
validation with 'as_hash_suffix' added on as a suffix.  as_hash_suffix
is available as a group option and may also be passed in via a
hashref as the only argument to as_hash.  The default value is
'_error'.  The values of the hash are arrayrefs of errors that
occurred to that form element.

By default as_hash will return the values of the hash as arrayrefs (a
list of the errors that occurred to that key).  It is possible to also
return the values as strings.  Three options are available for
formatting: 'as_hash_header' which will be pre-pended onto the error
string, 'as_hash_footer' which will be appended, and 'as_hash_join'
which will be used to join the arrayref.  The only argument required
to force the stringification is 'as_hash_join'.

    # if this returns the following
    my $hash = $err_obj->as_hash;
    # $hash looks like
    # {key1_error => ['error1', 'error2']}

    # then this would return the following
    my $hash = $err_obj->as_hash({
        as_hash_suffix => '_foo',
        as_hash_join   => '<br />',
        as_hash_header => '<span class="error">'
        as_hash_footer => '</span>'
    });
    # $hash looks like
    # {key1_foo => '<span class="error">error1<br />error2</span>'}

=back

=head1 GROUP OPTIONS

Any key in a validation hash matching the pattern
m/^group \s+ (\w+)$/x is considered a group option (the reason
that either group or general may be used is that CGI::Ex::Validate
used to have the concept of validation groups - these were not
commonly used so support has been removed as of the 2.10 release).
(the old name of 'general' vs 'group' is still supported but deprecated)

=over 4

=item C<title>

Used as a group section heading when as_array or as_string is called
by the error object.

    'group title' => 'Title of errors',

=item C<order>

Order in which to validate key/value pairs of group.

    'group order' => [qw(user pass email OR phone)],

    # OR

    'group order' => [{
        field    => 'field1',
        required => 1,
    }, {
        field    => 'field2',
        required => 1,
    }],

=item C<fields>

Alias for 'group order'.

=item C<validate_if>

If specified - the entire hashref will only be validated if
the "if" conditions are met.

    'group validate_if => {field => 'email', required => 1},

This group would only validate all fields if the email field
was present.

=item C<raise_error>

If raise_error is true, any call to validate that fails validation
will die with an error object as the value.

=item C<no_extra_fields>

If no_extra_fields is true, validate will add errors for any field found
in form that does not have a field_val hashref in the validation hash.
Default is false.  If no_extra_fields is set to 'used', it will check for
any keys that were not in a group that was validated.

An important exception to this is that field_val hashrefs or field names listed
in a validate_if or required_if statement will not be included.  You must
have an explicit entry for each key.

=item C<\w+_error>

These items allow for an override of the default errors.

    'group required_error' => '$name is really required',
    'group max_len_error'  => '$name must be shorter than $value characters',
      # OR #
    my $self = CGI::Ex::Validate->new({
        max_len_error => '$name must be shorter than $value characters',
    });

=item C<as_array_title>

Used as the section title for all errors that occur, when as_array
or as_string is called by the error object.

=item C<as_array_prefix>

Used as prefix to individual errors that occur, when as_array
or as_string is called by the error object.  Each individual error
will be prefixed with this string.  Headings will not be prefixed.
Default is '  '.

=item C<as_string_join>

When as_string is called, the values from as_array will be joined with
as_string_join.  Default value is "\n".

=item C<as_string_header>

If set, will be pre-pended onto the string when as_string is called.

=item C<as_string_footer>

If set, will be pre-pended onto the string when as_string is called.

=item C<as_hash_suffix>

Added on to key names during the call to as_hash.  Default is '_error'.

=item C<as_hash_join>

By default, as_hash will return hashref values that are errors joined with
the default as_hash_join value of <br />.  It can also return values that are
arrayrefs of the errors.  This can be done by setting as_hash_join to a non-true value
(for example '')

=item C<as_hash_header>

If as_hash_join has been set to a true value, as_hash_header may be set to
a string that will be pre-pended on to the error string.

=item C<as_hash_footer>

If as_hash_join has been set to a true value, as_hash_footer may be set to
a string that will be postpended on to the error string.

=item C<onevent>

Defaults to {submit => 1}.  This controls when the javascript validation
will take place.  May be passed any or all or load, submit, change, or blur.
Multiple events may be passed in the hash.

    'group onevent' => {submit => 1, change => 1}',

A comma separated string of types may also be passed:

    'group onevent' => 'submit,change,blur,load',

Currently, change and blur will not work for dynamically matched
field names such as 'm/\w+/'.  Support will be added.

=item C<set_hook>

Defaults document.validate_set_hook which defaults to nothing.  If
"group set_hook" or document.validate_set_hook are set to a function,
they will be passed the key name of a form element that had a
validation error and the error that will be set.  If a true value is
returned, then validate will not also the inline error.  If no value
or false is returned (default) the validate will continue setting the
inline error.  This gives full control over setting inline
errors. samples/validate_js_2_onchange.html has a good example of
using these hooks.

    'group set_hook' => "function (args) {
        alert("Setting error to field "+args.key);
    }",

The args parameter includes key, value, val_hash, and form.

The document.validate_set_hook option is probably the better option to use,
as it helps to separate display functionality out into your html templates
rather than storing too much html logic in your CGI.

=item C<clear_hook>

Similar to set_hook, but called when inline error is cleared.  Its
corresponding default is document.validate_clear_hook.  The clear hook
is also sampled in samples/validate_js_2_onchange.html

    'group clear_hook' => "function (args) {
        alert("Clear error on field "+args.key);
    }",

The args parameter includes key, val_hash, form, and was_valid.

=item C<no_inline>

If set to true, the javascript validation will not attempt to generate
inline errors when the only "group onevent" type is "submit".  Default
is true.  Inline errors are independent of confirm and alert errors.

    'group no_inline' => 1,

=item C<no_confirm>

If set to true, the javascript validation will try to use an alert
instead of a confirm to inform the user of errors when one of the
"group onevent" types is "submit".  Alert and confirm are independent
or inline errors.  Default is false.

    'group no_confirm' => 1,

=item C<no_alert>

If set to true, the javascript validation will not show an alert box
when errors occur.  Default is false.  This option only comes into
play if no_confirm is also set.  This option is only in effect if
"group onevent" includes "submit".  This option is independent of
inline errors.  Although it is possible to turn off all errors by
setting no_inline, no_confirm, and no_alert all to 1, it is suggested
that at least one of the error reporting facilities is left on.

    'group no_alert' => 1,

=back

=head1 JAVASCRIPT

CGI::Ex::Validate provides for having duplicate validation on the
client side as on the server side.  Errors can be shown in any
combination of inline and confirm, inline and alert, inline only,
confirm only, alert only, and none.  These combinations are controlled
by the group options no_inline, no_confirm, and no_alert.
Javascript validation can be generated for a page using the
C<-E<gt>generate_js> method of CGI::Ex::Validate.

(Note: It is also possible to store the validation inline with the
html as YAML and have it read in using the HTML conf handler - but
this feature has been deprecated - see the included html samples for
how to do this).

Generate JS will create something similar to the following (based on your validation):

    <script src="/cgi-bin/js/CGI/Ex/validate.js"></script>
    <script>
    document.validation = {
      'group no_confirm': 1,
      'group no_alert':   1,
      'group onevent':    'change,blur,submit',
      'group order': ['username', 'password'],
      username: {
        required: 1,
        max_len: 20
      },
      password: {
        required: 1,
        max_len: 30
      }
    };
    if (document.check_form) document.check_form('my_form_name');
    </script>

If inline errors are enabled (default), each error that occurs will attempt
to find an html element with its name as the id.  For example, if
the field "username" failed validation and created a "username_error",
the javascript would set the html of <span id="username_error"></span>
to the error message.

It is suggested to use something like the following so that you can
have inline javascript validation as well as report validation errors
from the server side as well.

   <span class=error id=password_error>[% password_error %]</span><br>

If the javascript fails for some reason, the form should still be able
to submit as normal (fail gracefully).

Additionally, there are two hooks that are called when ever an inline
error is set or cleared.  The following hooks are used in
samples/validate_js_2_onchange.html to highlight the row and set an icon.

    document.validate_set_hook = function (args) {
      document.getElementById(args.key+'_img').innerHTML
        = '<span style="font-weight:bold;color:red">!</span>';
      document.getElementById(args.key+'_row').style.background
        = '#ffdddd';
    };

    document.validate_clear_hook = function (args) {
      if (args.was_valid) {
        document.getElementById(args.key+'_img').innerHTML
            = '<span style="font-weight:bold;color:green">+</span>';
        document.getElementById(args.key+'_row').style.background
            = '#ddffdd';
      } else {
        document.getElementById(args.key+'_img').innerHTML = '';
        document.getElementById(args.key+'_row').style.background = '#fff';
      }
    };

If you have jquery that looks like:

    document.validate_set_hook = function (args) {
      $('#'+args.key+'_img').html('<span style="font-weight:bold;color:red">!</span>');
      $('#'+args.key+'_row').css('backgroundColor', '#ffdddd');
    };

    document.validate_clear_hook = function (args) {
      if (args.was_valid) {
        $('#'+args.key+'_img').html('<span style="font-weight:bold;color:green">+</span>');
        $('#'+args.key+'_row').css('backgroundColor', '#ddffdd');
      } else {
        $('#'+args.key+'_img').html('');
        $('#'+args.key+'_row').css('backgroundColor', '#fff');
      }
    };

These hooks can also be set as "group clear_hook" and "group set_hook"
    which are defined further above.

    If the confirm option is used ("group onevent" includes submit and
    "group no_confirm" is false), the errors will be displayed to the
    user.  If they choose OK they will be able to try and fix the errors.
    If they choose cancel, the form will submit anyway and will rely on
    the server to do the validation.  This is for fail safety to make sure
    that if the javascript didn't validate correctly, the user can still
submit the data.

=head1 LICENSE

This module may be distributed under the same terms as Perl itself.

=head1 AUTHOR

Paul Seamons <perl at seamons dot com>

=cut