| Prev | Chapter 7. Development Tutorial | Next |
|---|
This tutorial covers topics which are not essential to creating a minimal working package. Each section can be used independently of all of the others; all sections assume that you've completed the basic tutorial.
It's time to document. For the tutorial we'll use pre-written documentation. When creating a package from scratch, start by copying the documentation template from /var/lib/aolserver/openacs-dev/packages/acs-core-docs/xml/docs/xml/package-documentation-template.xml to myfirstpackage/www/docs/xml/index.xml.
You then edit that file with emacs to write the requirements and design sections, generate the html, and start coding. Store any supporting files, like page maps or schema diagrams, in the www/doc/xml directory, and store png or jpg versions of supporting files in the www/doc directory.
For this tutorial, you should instead install the pre-written documentation files for the tutorial app. Log in as service0, create the standard directories, and copy the prepared documentation:
[service0 service0]$ cd /var/lib/aolserver/service0/packages/myfirstpackage/ [service0 myfirstpackage]$ mkdir -p www/doc/xml [service0 myfirstpackage]$ cd www/doc/xml [service0 xml]$ cp /var/lib/aolserver/service0/packages/acs-core-docs/www/files/myfirstpackage/* . [service0 xml]$
OpenACS uses DocBook for documentation. DocBook is an XML standard for semantic markup of documentation. That means that the tags you use indicate meaning, not intended appearance. The style sheet will determine appearance. You will edit the text in an xml file, and then process the file into html for reading.
Open the file index.xml in emacs. Examine the file. Find the version history (look for the tag <revhistory>). Add a new record to the document version history. Look for the <authorgroup> tag and add yourself as a second author. Save and exit. For tips on editing SGML files in emacs, see OpenACS Documentation Guide
Process the xml file to create html documentation. The html documentation, including supporting files such as pictures, is stored in the www/docs/ directory. A Makefile is provided to generate html from the xml, and copy all of the supporting files. If Docbook is set up correctly, all you need to do is:
[service0 xml]$ make cd .. ; /usr/bin/xsltproc ../../../acs-core-docs/www/xml/openacs.xsl xml/index.xml Writing requirements-introduction.html for sect1(requirements-introduction) Writing requirements-overview.html for sect1(requirements-overview) Writing requirements-cases.html for sect1(requirements-cases) Writing sample-data.html for sect1(sample-data) Writing requirements.html for chapter(requirements) Writing design-data-model.html for sect1(design-data-model) Writing design-ui.html for sect1(design-ui) Writing design-config.html for sect1(design-config) Writing design-future.html for sect1(design-future) Writing filename.html for chapter(filename) Writing user-guide.html for chapter(user-guide) Writing admin-guide.html for chapter(admin-guide) Writing bi01.html for bibliography Writing index.html for book [service0 xml]$
Verify that the documentation was generated and reflects your changes by browsing to http://yoursite:8000/myfirstpackage/doc
Before you do any more work, make sure that your work is protected by putting it all into cvs. The cvs add command is not recursive, so you'll have to traverse the directory tree manually and add as you go. (More on CVS)
[service0 xml]$ cd .. [service0 doc]$ cd .. [service0 www]$ cd .. [service0 myfirstpackage]$ cd .. [service0 packages]$ cvs add myfirstpackage/ Directory /cvsroot/service0/packages/myfirstpackage added to the repository [service0 packages]$ cd myfirstpackage/ [service0 myfirstpackage]$ cvs add www Directory /cvsroot/service0/packages/myfirstpackage/www added to the repository [service0 myfirstpackage]$ cd www [service0 www]$ cvs add doc Directory /cvsroot/service0/packages/myfirstpackage/www/doc added to the repository [service0 www]$ cd doc [service0 doc]$ cvs add * cvs add: cannot add special file `CVS'; skipping cvs add: scheduling file `admin-guide.html' for addition cvs add: scheduling file `bi01.html' for addition cvs add: scheduling file `data-model.dia' for addition cvs add: scheduling file `data-model.png' for addition cvs add: scheduling file `design-config.html' for addition cvs add: scheduling file `design-data-model.html' for addition cvs add: scheduling file `design-future.html' for addition cvs add: scheduling file `design-ui.html' for addition cvs add: scheduling file `filename.html' for addition cvs add: scheduling file `index.html' for addition cvs add: scheduling file `page-map.dia' for addition cvs add: scheduling file `page-map.png' for addition cvs add: scheduling file `requirements-cases.html' for addition cvs add: scheduling file `requirements-introduction.html' for addition cvs add: scheduling file `requirements-overview.html' for addition cvs add: scheduling file `requirements.html' for addition cvs add: scheduling file `sample-data.html' for addition cvs add: scheduling file `sample.png' for addition cvs add: scheduling file `user-guide.html' for addition cvs add: scheduling file `user-interface.dia' for addition cvs add: scheduling file `user-interface.png' for addition Directory /cvsroot/service0/packages/myfirstpackage/www/doc/xml added to the repository cvs add: use 'cvs commit' to add these files permanently [service0 doc]$ cd xml [service0 xml]$ cvs add Makefile index.xml cvs add: scheduling file `Makefile' for addition cvs add: scheduling file `index.xml' for addition cvs add: use 'cvs commit' to add these files permanently [service0 xml]$ cd ../../.. [service0 myfirstpackage]$ cvs commit -m "new package" cvs commit: Examining . cvs commit: Examining www cvs commit: Examining www/doc cvs commit: Examining www/doc/xml RCS file: /cvsroot/service0/packages/myfirstpackage/www/doc/admin-guide.html,v done Checking in www/doc/admin-guide.html; /cvsroot/service0/packages/myfirstpackage/www/doc/admin-guide.html,v <-- admin-guide.html initial revision: 1.1 done (many lines omitted) [service0 myfirstpackage]$
You can track comments for any ACS Object. Here we'll track comments for notes. On the note-edit.tcl/adp pair, which is used to display individual notes, we want to put a link to add comments at the bottom of the screen. If there are any comments, we want to show them.
First, we need to generate a url for adding comments. In note-edit.tcl:
set comment_add_url "[general_comments_package_url]comment-add?[export_vars {
{ object_id $note_id }
{ object_name $title }
{ return_url "[ad_conn url]?[ad_conn query]"}
}]"
This calls a global, public tcl function that the general_comments package registered, to get its url. You then embed in that url the id of the note and its title, and set the return_url to the current url so that the user can return after adding a comment.
We need to create html that shows any existing comments. We do this with another general_comments function:
set comments_html [general_comments_get_comments
-print_content_p 1 $note_id]First, we pass in an optional parameter that that says to actually show the contents of the comments, instead of just the fact that there are comments. Then you pass the note id, which is also the acs_object id.
We put our two new variables in the note-edit.adp page.
<a href="@comment_add_url@">Add a comment</a> @comments_html@
There are at least two flavors of admin user interface:
Admins use same pages as all other users, except that they are offered admin links and buttons where appropriate. For example, if admins have privilege to bulk-delete items you could provide checkboxes next to every item seen on a list and the Delete Selected button on the bottom of the list.
Dedicated admin pages. If you want admins to have access to data that users aren't interested in or aren't allowed to see you will need dedicated admin pages. The conventional place to put those dedicated admin pages is in the /var/lib/aolserver/service0/packages/myfirstpackage/www/admin directory.
[service0 www]$ mkdir admin
[service0 www]$ cd admin
Even if your application doesn't need any admin pages of its own you will usually need at least one simple page with a bunch of links to existing administration UI such as Category Management or standard Parameters UI. Adding the link to Category Management is described in the section on categories. The listing below adds a link to the Parameters UI of our package.
[service0 admin]$ vi index.adp
<master> <property name="title">@title;noquote@</property> <property name="context">@context;noquote@</property> <ul class="action-links"> <li><a href="@parameters_url@" title="Set parameters" class="action_link">Set parameters</a></li> </ul>
[service0 admin]$ vi index.tcl
ad_page_contract {} {
} -properties {
context_bar
}
set package_id [ad_conn package_id]
set admin_p [ad_require_permission $package_id admin]
set context [list]
set title "Administration"
set parameters_url [export_vars -base "/shared/parameters" {
package_id { return_url [ad_return_url] }
}]
Now that you have the first admin page it would be nice to have a link to it somewhere in the system so that admins don't have to type in the /admin every time they need to reach it. You could put a static link to the toplevel index.adp but that might be distracting for people who are not admins. Besides, some people consider it impolite to first offer a link and then display a nasty "You don't have permission to access this page" message.
In order to display the link to the admin page only to users that have admin privileges add the following code near the top of /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.tcl:
set package_id [ad_conn package_id]
set admin_p [permission::permission_p -object_id $package_id \
-privilege admin -party_id [ad_conn untrusted_user_id]]
if { $admin_p } {
set admin_url "admin"
set admin_title Administration
}
In /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.adp put:
<if @admin_p@ ne nil> <a href="@admin_url@">@admin_title@</a> </if>
You can associate any ACS Object with one or more categories. In this tutorial we'll show how to equip your application with user interface to take advantage of the Categories service.
We'll start by installing the Categories service. Go to /acs/admin and install it. This step won't be necessary for the users of your applications because you'll create a dependency with the Package Manager which will take care that the Categories service always gets installed when your application gets installed.
Now that we have installed the Categories service we can proceed to modifying our application so that it can take advantage of it. We'll do it in three steps:
The Categories service provides a mechanism to associate one or more category trees that are relevant to your application. One example of such tree is a tree of geographical locations. Continents are on the top of such tree, each continent containing countries etc. Another tree might contain market segments etc. Before users of your application can take advantage of the Categories service there needs to be a way for administrators of your application to choose which category trees are applicable for the application.
The way to achieve this is is to provide a link to the Category Management pages. Add the following snippet to your /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.tcl file:
set category_map_url [export_vars -base \
"[site_node::get_package_url -package_key categories]cadmin/one-object" \
{ { object_id $package_id } }]
and the following snippet to your /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.adp file:
<li><a href="@category_map_url@"
class="action_link">Site-Wide Categories</a>
The link created by the above code will take the admin to the generic admin UI where he can pick category trees that make sense for this application. The same UI also includes facilities to build and edit category trees. Notice that the only parameter in this example is package_id so that category trees will be associated with the object identified by this package_id. The categorization service is actually more general than that: instead of package_id you could use an ID of some other object that serves as a "container" in your application. For example, if your discussion forums application supports multiple forums you would use forum_id to associate category trees with just that one forum rather than the entire application instance.
Once the category trees have been selected users need a way to categorize items. The easiest way to do this is by adding the category widget type of the form builder to note-edit.tcl. To achieve this we'll need to use the -extend switch to the ad_form command. Here's the "meat" of the note-edit.tcl page:
ad_form -name note -form {
{item_id:key}
{title:text {label Title}}
}
set package_id [ad_conn package_id]
set category_trees [category_tree::get_mapped_trees $package_id]
foreach tree $category_trees {
foreach { tree_id name subtree_id } $tree {}
ad_form -extend -name note -form \
[list [list category_id_${tree_id}:integer(category),optional \
{label $name} \
{html {single single}} \
{category_tree_id $tree_id} \
{category_subtree_id $subtree_id} \
{category_object_id {[value_if_exists entry_id]}}]]
}
ad_form -extend \
-name note \
-new_request {
permission::require_permission -object_id [ad_conn package_id] -privilege create
set page_title "Add a Note"
set context [list $page_title]
} -edit_request {
permission::require_write_permission -object_id $item_id
mfp::note::get \
-item_id $item_id \
-array note_array
set title $note_array(title)
set page_title "Edit a Note"
set context [list $page_title]
} -new_data {
mfp::note::add \
-title $title
} -after_submit {
ad_returnredirect "."
ad_script_abort
}This page requires a note_id to determine which record should be deleted. It also looks for a confirmation variable, which should initially be absert. If it is absent, we create a form to allow the user to confirm the deletion. Note that in entry-edit.tcl we used ad_form to access the Form Template commands; here, we call them directly because we don't need the extra features of ad_form. The form calls itself, but with hidden variables carrying both note_id and confirm_p. If confirm_p is present, we delete the record, set redirection back to the index, and abort script execution.
The database commands:
[service0@yourserver www]$ emacs note-delete.xql
<?xml version="1.0"?>
<queryset>
<fullquery name="do_delete">
<querytext>
select samplenote__delete(:note_id)
</querytext>
</fullquery>
<fullquery name="get_name">
<querytext>
select samplenote__name(:note_id)
</querytext>
</fullquery>
</queryset>And the adp page:
[service0@yourserver www]$ emacs note-delete.adp
<master>
<property name="title">@title@</property>
<property name="context">{@title@}</property>
<h2>@title@</h2>
<formtemplate id="note-del-confirm"></formtemplate>
</form>The ADP is very simple. The formtemplate tag outputs the HTML form generated by the ad_form command with the matching name. Test it by adding the new files in the APM and then deleting a few samplenotes.
You can track comments for any ACS Object. Here we'll track comments for notes. On the notes.tcl/adp pair, which is used to display individual notes, we want to put a link to add comments at the bottom of the screen. If there are any comments, we want to show them.
First, we need to generate a url for adding comments. In notes.tcl:
set comment_add_url "[general_comments_package_url]comment-add?[export_vars {
{ object_id $note_id }
{ object_name $title }
{ return_url "[ad_conn url]?[ad_conn query]"}
}]"
This calls a global, public tcl function that the general_comments package registered, to get its url. You then embed in that url the id of the note and its title, and set the return_url to the current url so that the user can return after adding a comment.
We need to create html that shows any existing comments. We do this with another general_comments function:
set comments_html [general_comments_get_comments
-print_content_p 1 $note_id]First, we pass in an optional parameter that that says to actually show the contents of the comments, instead of just the fact that there are comments. Then you pass the note id, which is also the acs_object id.
We put our two new variables in the notes.adp page.
<a href="@comment_add_url@">Add a comment</a> @comments_html@
There are at least two flavors of admin user interface:
Admins use same pages as all other users, except that they are offered admin links and buttons where appropriate. For example, if admins have privilege to bulk-delete items you could provide checkboxes next to every item seen on a list and the Delete Selected button on the bottom of the list.
Dedicated admin pages. If you want admins to have access to data that users aren't interested in or aren't allowed to see you will need dedicated admin pages. The conventional place to put those dedicated admin pages is in the /var/lib/aolserver/service0/packages/myfirstpackage/www/admin directory.
[service0@yourserver www]$ mkdir admin
[service0@yourserver www]$ cd admin
Even if your application doesn't need any admin pages of its own you will usually need at least one simple page with a bunch of links to existing administration UI such as Category Management or standard Parameters UI. Adding the link to Category Management is described in the section on categories. The listing below adds a link to the Parameters UI of our package.
[service0@yourserver admin]$ vi index.adp
<master> <property name="title">@title;noquote@</property> <property name="context">@context;noquote@</property> <ul class="action-links"> <li><a href="@parameters_url@" title="Set parameters" class="action_link">Set parameters</a></li> </ul>
[service0@yourserver admin]$ vi index.tcl
ad_page_contract {} {
} -properties {
context_bar
}
set package_id [ad_conn package_id]
set admin_p [ad_require_permission $package_id admin]
set context [list]
set title "Administration"
set parameters_url [export_vars -base "/shared/parameters" {
package_id { return_url [ad_return_url] }
}]
Now that you have the first admin page it would be nice to have a link to it somewhere in the system so that admins don't have to type in the /admin every time they need to reach it. You could put a static link to the toplevel index.adp but that might be distracting for people who are not admins. Besides, some people consider it impolite to first offer a link and then display a nasty "You don't have permission to access this page" message.
In order to display the link to the admin page only to users that have admin privileges add the following code near the top of /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.tcl:
set package_id [ad_conn package_id]
set admin_p [permission::permission_p -object_id $package_id \
-privilege admin -party_id [ad_conn untrusted_user_id]]
if { $admin_p } {
set admin_url "admin"
set admin_title Administration
}
In /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.adp put:
<if @admin_p@ ne nil> <a href="@admin_url@">@admin_title@</a> </if>
You can associate any ACS Object with one or more categories. In this tutorial we'll show how to equip your application with user interface to take advantage of the Categories service.
We'll start by installing the Categories service. Go to /acs/admin and install it. This step won't be necessary for the users of your applications because you'll create a dependency with the Package Manager which will take care that the Categories service always gets installed when your application gets installed.
Now that we have installed the Categories service we can proceed to modifying our application so that it can take advantage of it. We'll do it in three steps:
The Categories service provides a mechanism to associate one or more category trees that are relevant to your application. One example of such tree is a tree of geographical locations. Continents are on the top of such tree, each continent containing countries etc. Another tree might contain market segments etc. Before users of your application can take advantage of the Categories service there needs to be a way for administrators of your application to choose which category trees are applicable for the application.
The way to achieve this is is to provide a link to the Category Management pages. Add the following snippet to your /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.tcl file:
set category_map_url [export_vars -base \
"[site_node::get_package_url -package_key categories]cadmin/one-object" \
{ { object_id $package_id } }]
and the following snippet to your /var/lib/aolserver/service0/packages/myfirstpackage/www/admin/index.adp file:
<li><a href="@category_map_url@"
class="action_link">Site-Wide Categories</a>
The link created by the above code will take the admin to the generic admin UI where he can pick category trees that make sense for this application. The same UI also includes facilities to build and edit category trees. Notice that the only parameter in this example is package_id so that category trees will be associated with the object identified by this package_id. The categorization service is actually more general than that: instead of package_id you could use an ID of some other object that serves as a "container" in your application. For example, if your discussion forums application supports multiple forums you would use forum_id to associate category trees with just that one forum rather than the entire application instance.
Once the category trees have been selected users need a way to categorize items. The easiest way to do this is by adding the category widget type of the form builder to note-edit.tcl. To achieve this we'll need to use the -extend switch to the ad_form command. Here's the "meat" of the note-edit.tcl page:
ad_form -name note -form {
{item_id:key}
{title:text {label Title}}
}
set package_id [ad_conn package_id]
set category_trees [category_tree::get_mapped_trees $package_id]
foreach tree $category_trees {
foreach { tree_id name subtree_id } $tree {}
ad_form -extend -name note -form \
[list [list category_id_${tree_id}:integer(category),optional \
{label $name} \
{html {single single}} \
{category_tree_id $tree_id} \
{category_subtree_id $subtree_id} \
{category_object_id {[value_if_exists entry_id]}}]]
}
ad_form -extend \
-name note \
-new_request {
permission::require_permission -object_id [ad_conn package_id] -privilege create
set page_title "Add a Note"
set context [list $page_title]
} -edit_request {
permission::require_write_permission -object_id $item_id
mfp::note::get \
-item_id $item_id \
-array note_array
set title $note_array(title)
set page_title "Edit a Note"
set context [list $page_title]
} -new_data {
mfp::note::add \
-title $title
} -after_submit {
ad_returnredirect "."
ad_script_abort
}
Browse to the package manager. Click on tutorialapp.
Click on Generate a distribution file for this package from the filesystem.
Click on the file size (37.1KB) after the label Distribution File: and save the file to /tmp.
The notifications package allows you to send notifications through any defined communications medium (e.g. email, sms) upon some event occuring within the system.
This tutorial steps through the process of integrating the notifications package with your package.
First step is to create the notification types. To do this a script similar to the one below needs to be loaded into Postgresql. I create this script in a package-name/sql/postgresql/package-name-notifications-init.sql file. I then load this file from my create sql file. The following code snippet is taken from Weblogger. It creates a lars_blogger_notif notification type (which was created above).
create function inline_0() returns integer as '
declare
impl_id integer;
v_foo integer;
begin
-- the notification type impl
impl_id := acs_sc_impl__new (
''NotificationType'',
''lars_blogger_notif_type'',
''lars-blogger''
);
v_foo := acs_sc_impl_alias__new (
''NotificationType'',
''lars_blogger_notif_type'',
''GetURL'',
''lars_blogger::notification::get_url'',
''TCL''
);
v_foo := acs_sc_impl_alias__new (
''NotificationType'',
''lars_blogger_notif_type'',
''ProcessReply'',
''lars_blogger::notification::process_reply'',
''TCL''
);
PERFORM acs_sc_binding__new (
''NotificationType'',
''lars_blogger_notif_type''
);
v_foo:= notification_type__new (
NULL,
impl_id,
''lars_blogger_notif'',
''Blog Notification'',
''Notifications for Blog'',
now(),
NULL,
NULL,
NULL
);
-- enable the various intervals and delivery methods
insert into notification_types_intervals
(type_id, interval_id)
select v_foo, interval_id
from notification_intervals where name in (''instant'',''hourly'',''daily'');
insert into notification_types_del_methods
(type_id, delivery_method_id)
select v_foo, delivery_method_id
from notification_delivery_methods where short_name in (''email'');
return (0);
end;
' language 'plpgsql';
select inline_0();
drop function inline_0();
The next step is to setup our notification creation. A new notification must be added to the notification table for each blog entry added. We do this using the notification::new procedure
notification::new \
-type_id [notification::type::get_type_id \
-short_name lars_blogger_notif] \
-object_id $blog(package_id) \
-response_id $blog(entry_id) \
-notif_subject $blog(title) \
-notif_text $new_content
This code is placed in the tcl procedure that creates blog entries, right after the entry gets created in the code. The $blog(package_id) is the OpenACS object_id of the Weblogger instance to which the entry has been posted to and the $new_content is the content of the entry.
The final step is to setup the notification subscription process. In this example we want to let a user find out when a new entry has been posted to the blog. To do this we put a link on the blog that allows them to subscribe to notifications of new entries. The notifications/requests-new page is very handy in this situation.
Such a link can be created using the notification::display::request_widget proc:
set notification_chunk [notification::display::request_widget \
-type lars_blogger_notif \
-object_id $package_id \
-pretty_name [lars_blog_name] \
-url [lars_blog_public_package_url] \
]
which will return something like
You may <a href="/notifications/request-new?...">request notification</a> for Weblogger.which can be readily put on the blog index page. The pretty_name parameter is what appears at the end of the text returned (i.e. "... request notification</a> for pretty_name"), The url parameter should be set to the address we want the user to be redirected to after they have finished the subscription process.
This should be all you need to implement a notification system. For more examples look at the forums package.
Future Topics:
How to enforce security so that users can't change other users records
How to use the content management tables so that ... what?
How to change the default stylesheets for Form Builder HTML forms.
How to make your package searchable with OpenFTS/Oracle
How to prepare pagelets for inclusion in other pages
How and when to put procedures in a tcl procedure library
More on ad_form - data validation, other stuff. (plan to draw from Jon Griffin's doc)
How and when to implement caching
partialquery in xql
How to use the html/text entry widget to get the "does this look right" confirm page
APM package dependencies
See also the OpenACS Programming FAQ