Introduction

------------

XSLTools is a collection of modules and packages facilitating the development

of applications based on XML, XSL stylesheets and transformations, notably Web

applications involving complicated Web forms potentially consisting of

editable hierarchical structures and potentially involving "live" or "in-page"

dynamic updates to portions of those Web forms.

Quick Start

-----------

Try running the demo:

python tools/demo.py

An introductory guide to creating applications can be found in the docs

directory - see docs/index.html for the start page.

Contact, Copyright and Licence Information

------------------------------------------

The current Web page for XSLTools at the time of release is:

http://www.boddie.org.uk/python/XSLTools.html

Copyright and licence information can be found in the docs directory - see

docs/COPYING.txt, docs/LICENCE.txt and docs/LICENCE-Sarissa.txt for more

information.

Dependencies

------------

XSLTools has the following basic dependencies:

Package                     Release Information

-------                     -------------------

libxml2dom                  0.2.1

libxml2                     Tested with 2.6.17

libxslt                     Tested with 1.1.12

The example Web applications require WebStack (release 1.0 or later).

The example PyQt applications have been tested with PyQt 3.15.

New in XSLTools 0.2 (Changes since XSLTools 0.1)

------------------------------------------------

  * Made a new XSLTools package and moved XSLOutput into it.

  * Improved serialisation of transformation results so that output options

    are observed (in some cases, at least).

  * Fixed stylesheet and reference document paths so that libxslt should not

    now become confused by ambiguous relative paths.

  * Added expression parameters to XSLOutput.Processor so that in-document

    data can be used to, for example, initialise multiple-choice field values.

  * Added input/initialiser support so that input documents can be tidied or

    initialised using information from the template.

  * Added template:init for use with template:element in XSLForms to control

    element initialisation where necessary.

  * Added special high-level "macro" attributes (eg. template:attribute-field)

    which should make templates easier to write and maintain.

  * Added template:if to XSLForms, providing conditional output of annotated

    elements.

  * Added set_document to XSLForms.Fields.Form.

  * Added prepare_parameters to XSLForms.Resources.XSLFormsResource.

  * Added element-path and url-encode XSLForms extension functions.

  * Improved Unicode support in the XSLForms extension functions.

  * Changed in-page requests to contain proper POST data.

  * Updated the code to work with WebStack 1.0 changes and adopted the

    new-style WebStack demonstration mechanism.

  * Added XMLCalendar and XMLTable (to the XSLTools package).

  * Added a dictionary (or word lookup) example application.

  * Added a job candidate profile (or CV editor) example application.

  * Added a template attribute reference and an XSLFormsResource guide to the

    documentation.

  * Added Debian package support.

  * Added missing COPYING.txt file.

  * Renamed the scripts to avoid naming issues in system-wide installations.

  * Added a PyQt example based on the system configurator example, with the

    form prepared in Qt Designer.

Notes on In-Page Update Functionality

-------------------------------------

Special note #1: Konqueror seems in certain cases to remember replaced form

content (when replaceChild is used to replace regions of the page which

include form elements). This causes the browser to believe that more form

fields exist on the page than actually do so, and subsequent form submissions

thus include the values of such removed fields. A special hack is in place to

disable form fields by changing their names, thus causing Konqueror to not

associate such fields with the real, active fields; this hack does not seem to

cause problems for Mozilla. This needs some investigation to determine in

exactly which circumstances the problem arises.

Special note #2: Konqueror also seems to crash if asked to find elements using

an empty 'id' attribute string. This needs some investigation to see if it

really is the getElementById call that causes the crash.

Special note #3: Konqueror's XMLHttpRequest seems to append null characters to

the end of field values. Attempting to prune them before the request is sent

fails with a function like the following:

function fixValue(fieldValue) {

    if (fieldValue.length == 0) {

        return fieldValue;

    } else if (fieldValue[fieldValue.length - 1] == '\0') {

        return fieldValue.substr(0, fieldValue.length - 1);

    } else {

        return fieldValue;

    }

}

This may be because it is the entire message that is terminated with the null

character, and that this happens only upon sending the message. Consequently,

some frameworks (notably mod_python) do not support in-page functionality when

used from Konqueror.

Various browsers (eg. Mozilla/Firefox, Konqueror) will not allow the

XMLHttpRequest in-page updates to function unless the URL used in the

requestUpdate JavaScript function is compatible with the URL at which the

browser finds the application. Currently, relative URLs are in use to avoid

this issue of compatibility, but should an absolute URL be deduced using the

WebStack API and then used, it may be possible that the values returned by

that API do not match the actual addresses entered into the address bar of the

browser.

To check the behaviour of the applications, it is possible to view the

document source of the pages served by applications and to verify that the

URLs mentioned in the JavaScript function calls (to 'requestUpdate') either be

a relative link or involve a URL similar to that which appears in the

browser's address bar. In some environments, the use of 'localhost' addresses

often confuses the browser and server; one workaround is to use real host

names or addresses instead of 'localhost'.

Choosing an element-path:

When specifying the "context" of the in-page update, one must imagine which

element the template fragment should operate within. If the template:id

attribute marks a particular section, then the element-path should be a path

to the applicable context element for that section in the complete template

document. Note that if a template:element attribute appears on the same

element as the template:id attribute then the element-path should refer to the

element specified in the template:element attribute.

Choosing where to put template:attribute, template:id and id:

When specifying the extent of a template fragment, one must be sure not to put

the template:id attribute on the same element as a template:attribute

annotation; otherwise, the generated code will be improperly extracted as a

fragment producing two versions of the element - one for when the specified

attribute is present, and one for when it is not present. Generally,

template:id and id can be placed on the same node, however.

Stable element ordering and element-path:

Within the element-path, the numbering of the elements will start at 1.

Therefore it is vital to choose a region of the form data structure with the

element-path which is isolated from surrounding elements whose positions would

otherwise be dependent on a stable ordering of elements, and whose processing

would be disrupted if some new elements suddenly appeared claiming the same

positions in the document. For example:

  <item value="">         .../item$1/value

    <type value=""/>      .../item$1/type$1/value

    <comment value=""/>   .../item$1/comment$2/value

  </item>

  In-page update...

  <comment value=""/>     .../item$1/comment$1/value

Notes on XSL

------------

libxslt seems to be quite liberal on the definition of runtime parameters, in

that there is no apparent need to explicitly declare the corresponding global

variables in stylesheets. Whilst this is nice, we may eventually need to

detect such variables and add them in the preparation process.

Release Procedures

------------------

Update the XSLTools/__init__.py and XSLForms/__init__.py __version__

attributes.

Change the version number and package filename/directory in the documentation.

Change code examples in the documentation if appropriate.

Update the release notes (see above).

Check the setup.py file and ensure that all package directories are mentioned.

Check the release information in the PKG-INFO file and in the package

changelog (and other files).

Tag, export.

Generate the API documentation.

Remove generated .pyc files: rm `find . -name "*.pyc"`

Archive, upload.

Upload the introductory documentation.

Update PyPI, PythonInfo Wiki, Vaults of Parnassus entries.

Generating the API Documentation

--------------------------------

In order to prepare the API documentation, it is necessary to generate some

Web pages from the Python source code. For this, the epydoc application must

be available on your system. Then, inside the distribution directory, run the

apidocs.sh tool script as follows:

./tools/apidocs.sh

Some warnings may be generated by the script, but the result should be a new

apidocs directory within the distribution directory.

Making Packages

---------------

To make Debian-based packages:

  1. Create new package directories under packages if necessary.

  2. Make a symbolic link in the distribution's root directory to keep the

     Debian tools happy:

     ln -s packages/ubuntu-hoary/python2.4-xsltools/debian/

  3. Run the package builder:

     dpkg-buildpackage -rfakeroot

  4. Locate and tidy up the packages in the parent directory of the

     distribution's root directory.