Skip to main content

A Better View of Your Trees

I’ve been playing around with Trees over the last couple of days. One of the requirements dictates that I render a Tree based on a sub-set of a larger hierarchy of data. Luckily, the data is already in the correct format (ID, PARENT_ID, DESCRIPTION) for an Oracle HTML DB Tree, so I don’t have to make any changes to the data structure.

Due to the large size of the table, it was much more efficient to create an HTML DB Collection based on a subset of the data, and then use that Collection in my Trees. Sounds simple enough, but you cannot easily create a Tree based on a Collection, as the view HTMLDB_COLLECTIONS does not show up in the Select List of Tables/Views that you can base a Tree on.

The solution to this was almost too simple: Create your own View for HTMLDB_COLLECTIONS. This killed two birds with one stone: I could now select my view from the Select List when I created a new Tree, and I could also rename the columns to reflect those of the base table, rather than be forced to use the C00X nomenclature that HTML DB Collections use.

Here’s the code that I used to create my view on the HTMLDB_COLLECTIONS table:

create or replace view my_collection_view
(collection_name, seq_id, id, name, address)
select collection_name, seq_id, c001, c002, c003
from htmldb_collections
where collection_name = 'MY_COLLECTION';

Now that I have my View, I had better create a Collection so that my View will have some data. I chose to create a named procedure to create the Collection. This way, all of my code is managed in the database, making it easier to edit and debug. I simply call this procedure when the user clicks on the corresponding button to create the Collection:

procedure create_my_collection
l_sql varchar2(32767);
l_collection_exists boolean;
l_sql := 'select id, name, address
from my_collection_view';
l_collection_exists :=
htmldb_collection.collection_exists(p_collection_name =>
if l_collection_exists = true then
htmldb_collection.delete_collection(p_collection_name =>
end if;
p_collection_name => 'MY_COLLECTION',
p_query => l_sql);

Of course you can edit the SQL statement which I stuff into l_sql in the above code to reflect user-specified or other parameters. In my actual application, that’s exactly what I do. Remember to use the v('ITEM_NAME') syntax to refer to HTML DB Item values while in a named PL/SQL unit.

There you have it – a much more “civilized” way to work with HTML DB Trees and Collections.


Anonymous said…
Very useful. Just what I needed. Please keep blogging about HTMLDB.
Scott said…

Glad it helped. I have no plans on stopping this blog, and will try my best to keep it updated.


- Scott -
Anonymous said…
Scott, you are my new hero! Works like a champ! Thanks for all your help on the forum area!

Scott said…

I'm happy to be of assistance. I'm also working on an eBusiness Suite implementation w/BOM - perhaps we can compare notes at some point in the future!


- Scott -
Colin said…
Was this OTN forum posting the inspiration for this blog post?
Scott said…

Yes - I mentioned that in the post itself. :)

- Scott -
Colin Sheppard said…
I also noticed this related topic in the Tips and Tricks section of HTML DB Studio:

Popular posts from this blog

Logging APEX Report Downloads

A customer recently asked how APEX could track who clicked “download” from an Interactive Grid.  After some quick searching of the logs, I realized that APEX simply does not record this type of activity, aside from a simple page view type of “AJAX” entry.  This was not specific enough, and of course, led to the next question - can we prevent users from downloading data from a grid entirely?

I knew that any Javascript-based solution would fall short of their security requirements, since it is trivial to reconstruct the URL pattern required to initiate a download, even if the Javascript had removed the option from the menu.  Thus, I had to consider a PL/SQL-based approach - one that could not be bypassed by a malicious end user.

To solve this problem, I turned to APEX’s Initialization PL/SQL Code parameter.  Any PL/SQL code entered in this region will be executed before any other APEX-related process.  Thus, it is literally the first place that a developer can interact with an APEX page…

Custom Export to CSV

It's been a while since I've updated my blog. I've been quite busy lately, and just have not had the time that I used to. We're expecting our 1st child in just a few short weeks now, so most of my free time has been spent learning Lamaze breathing, making the weekly run to Babies R Us, and relocating my office from the larger room upstairs to the smaller one downstairs - which I do happen to like MUCH more than I had anticipated. I have everything I need within a short walk - a bathroom, beer fridge, and 52" HD TV. I only need to go upstairs to eat and sleep now, but alas, this will all change soon...

Recently, I was asked if you could change the way Export to CSV in ApEx works. The short answer is, of course, no. But it's not too difficult to "roll your own" CSV export procedure.

Why would you want to do this? Well, the customer's requirement was to manipulate some data when the Export link was clicked, and then export it to CSV in a format…

Refreshing PL/SQL Regions in APEX

If you've been using APEX long enough, you've probably used a PL/SQL Region to render some sort of HTML that the APEX built-in components simply can't handle. Perhaps a complex chart or region that has a lot of custom content and/or layout. While best practices may be to use an APEX component, or if not, build a plugin, we all know that sometimes reality doesn't give us that kind of time or flexibility.While the PL/SQL Region is quite powerful, it still lacks a key feature: the ability to be refreshed by a Dynamic Action. This is true even in APEX 5. Fortunately, there's a simple workaround that only requires a small change to your code: change your procedure to a function and call it from a Classic Report region.In changing your procedure to a function, you'll likely only need to make one type of change: converting and htp.prn calls to instead populate and return a variable at the end of the function. Most, if not all of the rest of the code can remain un…