[epsi-coord] Update on site migration Fwd: http://epsi.reinvented.net/

Mon Apr 18 13:52:04 BST 2011

Hi all,

had a conversation w Peter Rukavina. He's had a look at the data, and we
discussed some of his findings. He says converting the content from the old
to the new site is very doable, with 2 or 3 things to choose the forward on
(things around tagging and categories, and things around relationships
between content items, and the way to redirect old links to migrated content
to prevent linkrot if we choose to create a new link structure.)
He'll give us an estimate of time/effort needed in the coming days, next
conversation with him I have on Monday next week. He's currently only
looking at the data migration effort.

Apart from that Tom and I need to really get going on the website part of
things, so we can know better what is next, what to do ourselves, and what
to let others do.

best,

Ton
-------------------------------------------
Interdependent Thoughts
Ton Zijlstra

ton at tonzijlstra.eu
+31-6-34489360

http://zylstra.org/blog
-------------------------------------------

---------- Forwarded message ----------
From: Peter Rukavina <peter at rukavina.net>
Date: Mon, Mar 28, 2011 at 7:02 PM
Subject: http://epsi.reinvented.net/
To: Ton Zijlstra <ton.zijlstra at gmail.com>

Hi Ton,

After spending more time with the EPSIplatform on EZPublish, I realized that
the "Export" functionality that was already present, that at first appeared
only to function for exporting users, can, in fact, be used to export
content as well. This functionality is obscured by the fact that when you
first click "Export", you're presented *only* with the option to export
Users; if, however, you click on the "Change" button in the "Data selection"
section, you can then click on the graphic "up arrow", select the "Home"
radio button, click "Select", then change the "Selected class" from "User"
to "Article", at which point it's possible to download a CSV file of all of
the articles under the "Home" navigation item and its children.

As a proof-of-concept, I did this, and then used the stock "Node Import"
module to import the resulting CSV file into a test Drupal instance, and you
can see the result at:

http://epsi.reinvented.net/

Obviously this was a test only, and isn't intended to represent anything
other than a first pass.  Some issues I encountered:

1. Mapping the EZPublish "countries," "themes," and "keywords" to Drupal
vocabularies works out of the box -- see
http://epsi.reinvented.net/taxonomy/term/2 for an example of the "Public
Sector Culture" theme.

2. EZPublish is considerably more flexible in terms of the ability to
attach, in parent-child relationships, content of different types -- these
are called "sub-items"  and they are, for example, what allows a page like
this:

http://epsiplatform.eu/contacts/european_public_sector_information_psi_platform_contacts/co_ordinating_team_european_public_sector_information_psi_platform

to be constructed like this:

http://epsiplatform-test.okfn.org/epsiplatform_admin/contacts/european_public_sector_information_psi_platform_contacts/co_ordinating_team_european_public_sector_information_psi_platform

While it's possible to create relationships between nodes in Drupal, there's
no direct mapping of the EZPublish object model to the Drupal object model,
so we'd have to develop some sort of way of mapping the various objects into
Drupal so that the data all ended up together but through some different
route.

3. The contents of the "introduction" and "body" fields uses proprietary
EZPublish XML, rather than standard HTML. This is why when, for example, you
look at this page:

http://epsi.reinvented.net/node/59

and compare it to this page:

http://epsiplatform.eu/topic_reports/topic_report_15_new_zealand_moves_to_embrace_psi_re_use_and_open_data/topic_report_15_new_zealand_moves_to_embrace_psi_re_use_and_open_data

you'll see that the EZPublish version contains considerably more
interlinkage. This is because links in EZPublish look like this:

<link xhtml:title="http://www.e.govt.nz/policy/information-and-data"
url_id="12197">

rather than something like <a href="..."> that one might be used to.

This isn't a deal-breaker, as we can transform the EZPublish XML into
standard HTML before doing the import, and doing so should preserve the
linkage.

Cheers,
Peter

--
Peter Rukavina, Charlottetown, Prince Edward Island, Canada
peter at rukavina.net
+ 1 (902) 892-2556
http://ruk.ca/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/mailman/private/epsi-coord/attachments/20110418/62a456cc/attachment-0001.htm>