NAME

WWW::CheckSite::Manual - A description of the metrics used in this package

SYNOPSIS

This document contains a description of modules and tools in this suite.

Kwalitee

checksite

DESCRIPTION

Kwalitee

The idea behind this package is to provide an analysis of items contained in a web-site. We use the word kwalitee because it looks and sounds like quality but just isn't. The metrics used to assess kwalitee only give an indication of the technical state a web-site is in, and do not reflect on the user experience of quality of that web-site.

At the heart of the package is the spider that fetches all the pages referred to within the web-site. For each page that is fetched a number of things is checked. Here is an explanation of the kwalitee metrics:

* return status

The most basic check for a web-page is to see if it can be fetched. The HTTP return-status should be 200 OK.

SCORE: 0 for return status other than 200; 1 for return status 200

* title

The next check is to see if the <title></title> tag-pair has content.

SCORE: 0 for not content; 1 for content

* valid

The next check is to see if the (X)HTML in the page validates. The default behaviour is to use the validator available on http://validator.w3.org

SCORE: 0 for not valid, 1 for valid or validation disabled

* links

The next check is to see if the web-page does not contain "dead links".

All hyperlinks (<a href=>, <area href=>) are checked with a HTTP HEAD request to see if they can be "followed". URLs that have the same origin as the primary url will also be put on the "to-fetch-list" of the spider.

MAX SCORE: 1 (do not count urls excluded by robot-rules/exclude pattern)

* images

The next check is to see if the web-page does not contain "dead images".

All images (<img src=>, <input type=image>) are checked with a HTTP HEAD request to see if they exist on the server. If the Image::Info module is available, the image is fetched from the server and a basic sanity test on the image is done.

MAX SCORE: 1 (do not count images excluded by robot-rules/exclude pattern)

* styles

The next check is to see if the web-page does not contain "dead style references".

All styles referenced in <link rel=stylesheet type=text/css> are fetched and if validation is switched on, they will be sent to the css-validator at: http://jigsaw.w3.org/validator

TODO: Extract inline styles, and send them of for validation.

MAX SCORE: 1

kwalitee

Every individual page can have a maximum of 6 kwalitee points that lead to a kwalitee of 1.00. For the complete web-site the mean of the page scores is taken and presented as a fraction of 1.

checksite

This script is a wrapper around WWW::CheckSite that supports some command-line options to tweak the behaviour of the module.

Here is an explanation of these options:

[--uri|-u] <uri> (mandatory unless --load)

This specifies the uri to be spidered. The --uri option-qualifier is optional. --uri can be abbreviated to -u.

--prefix|-p <prefix> (mandatory)

This option specifies a prefix that will be used as a subdirectory name which is used to store the saved spider data and the reports. --prefix can be abbreviated to -p.

The subdirectory is created the current directory, or in the directory specified with the --dir option. The data stored as a result of the --save option will be in this subdirectory with the name <prefix>.wcs

--dir|-d <directory>

This option specifies the base directory for storing the data. --dir can be abbreviated to -d.

--save or --nosave

This option specifies that the spider data should be saved. The default behaviour is to save the data, if you do not want that, use --nosave. The saved data can later be used to regenerate the reports with the --load option. The data is stored as <directory>/<prefix>/<prefix>.wcs with Storable::nstore(). --[no]save cannot be abbreviated.

See also: WWW::CheckSite Report-Templates

--load

This options specifies that you want to load the results of a previous run and not do an actual run of the programme. This option is useful to regenerate the reports. --load cannot be abbreviated.

See also: WWW::CheckSite Report-Templates

--validate or --novalidate

This option specifies that HTML- and CSS-validation should be done. The default behaviour is to validate by uri. That means the url that is to be validated is send to the validation service and should be accessible for that service. If you do not want the validation, use the --novalidate option. --[no]validation cannot be abbreviated.

See also: checksite --by_uri and --by_upload

--by_uri [validator-url-mask]

This option sets the validation method to use the uri interface (unless --novalidate is specified). You can optionally specify a mask for an alternative HTML-validator site. The default HTML validator url mask is http://validator.w3.org/check?uri=%s, where the %s is a sprintf() placeholder for the uri to be validated. The optional validator-url-mask can be used to accommodate a local copy of the W3 HTML validator (see http://validator.w3.org/source/).

NOTE: This option does only influence the form used for CSS-validation at http://jigsaw.w3.org/css-validator/, not the CSS-validation service used.

--by_uri cannot be abbreviated.

--by_upload [validator-url]

This option sets the validation method to use the upload interface (unless --novalidate is specified). All the content to be validated (HTML and CSS) is saved as a local file (using File::Temp). You can optionally specify an alternative HTML-validator site. The default HTML validator url is http://validator.w3.org/, and the second form on the page is used. The optional validator-url can be used to accommodate a local copy of the W3 HTML validator (see http://validator.w3.org/source/).

NOTE: This option does only influence the form used for CSS-validation at http://jigsaw.w3.org/css-validator/, not the CSS-validation service used.

--by_upload cannot be abbreviated.

--by_none

This option is another way to specify --novalidate.

--by_none cannot be abbreviated.

--lang|-l <accept-language>

This option can be used to force a web-server to return web-pages in the specified language (if applicable). The accept-language argument can be a simple two letter language code as specified in ISO 639, or a complete Accept-language: field as described in section 14.4 of RFC 2616.

NOTE: My apache config says:

  # Note 3: In the case of 'ltz' we violate the RFC by using a three
  # char specifier. There is 'work in progress' to fix this and get
  # the reference data for rfc1766 cleaned up.

So there may be more weird stuff out there, but since you are supposed to be using this on your own web-sites only, you should know about that!

--lang can be abbreviated to -l.

--ua_class <ua_class>

This option can be used to override the default user-agent class WWW::Mechanize. The new user-agent class could be a WWW::Mechanize descendant that caters for special needs:

    package BA_Mech;
    # This package sets credentials for basic authentication
    use base 'WWW::Mechanize';
    sub get_basic_credentials { ( 'abeltje', '********' ) }
    1;

and call checksite like

    checksite -p mysite --ua_class BA_Mech http://www.mysite.org

AUTHOR

Abe Timmerman, <abeltje@cpan.org>

$Id: Manual.pod 472 2006-04-02 12:16:16Z abeltje $

COPYRIGHT & LICENSE

Copyright MMV-MMVI Abe Timmerman, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.