It’s my very last Plone 2.5 site and I’m very happy to see the back of it. Using the joy that is quintagroup.transmogrifier, I was able to, without much manual effort, migrate my content from a Plone 2.5 site (constructed with instancemanager, of all things!) straight into a shiny new Plone 4 site on Buildout.
My only key need was to port the content and structure, and I was happy to make big allowances for the sake of certain content and time. So, my thoughts here mightn’t (probably won’t!) suit everyone.My requirements and allowances were these:
- Content and structure only needed to be migrated. Plone settings, theme, user accounts, and everything else would either come from my newly-created Plone 4 theme or be manually re-added.
- There were/are only a handful of users on the site.
- My Plone 2.5 site used a lot of custom content types (IronicWiki,
Wicked, Quills, PloneSoftwareCentre, Poi and more).
- I want to avoid ever going through migrating custom ‘document’ types again. Thankfully, the ‘wiki’-based content types translate easily to pure Plone documents.
- I was happy to jettison content in PSC, Poi, and Quills (Weblogs). This is a big issue but it wasn’t worth my time since the content was already 5+ years out of date and never used.
- Default folder view selections don’t appear to migrate. I was able to manually fix things up.
That out of the way, here’s my process.
For Plone 2.5
- Follow Step 1 of QuintaGroup’s instructions for migrating a Plone 2.1 site to Plone 3. I had a Plone 2.5 site and this worked well. Essentially, you want to use ZopeSkel (paster) to create yourself a Plone 2.5 buildout and then checkout the relevant packages (as per the instructions). I created mine on my local Ubuntu machine for my sanity as the old server the site was running on was ages out of date.
- Figure out what you need to actually get the Plone 2.5 instance to
start up. Now, my site had a huge amount of 3rd party packages that
caused me grief, and I had to add these into the buildout before the
instance would even start successfully. In my case, it was a matter
of copying out the old product web links from my instancemanager
script and putting them in the productdistros section of my Plone
2.5 buildout.
- Note that if your Plone instance was entirely manually managed, then I’m guess you either on a hunt, or you could try copying the files into your relevant directory.
- I also had a number of SVN-based products too so I employed my fried, mr.developer, to check them out from the right SVN location to the buildout src directory.
- Once you’ve got your instance started (in the foreground), then you
should be able to access
http://localhost:8080/<your_site_name>/@@pipeline-config which is
the configuration for Transmogrifier.
- You can modify your configuration to suit you. There’s a little learning involved here, but there are a good number of Transmogrifier resources around.
- My configuration looks like this: transmogrifier-export.txt, where I explicitly ignore a whole bunch of content types (read: problematic for export/import), and export my content out to a temporary directory
- Make changes to the export configuration and save the result.
- Under the ZMI, head into portal_setup and click the ‘Properties’ tab. Change the active site configuration to be ‘Transmogrifier’ and save this.
- Click the ‘Export’ tab at the top, and then export the ‘Content’ step you see listed. Watch your console output for progress and anything that might be a problem.
- If successful, then you’ll get presented with a generic setup .tar.gz to download. Save this somewhere useful; you’ll need it shortly. If you were like me and exported to a directory, then you’ll get an ‘empty’ .tar.gz download and your content will be in the directory you specified.
Manually modify content before import
The reason I outputted to a directory so was I could hack a few things around in my content. Transmogrifier is powerful, but a bit complicated. I couldn’t find out how to rename a field to another in the pipeline (eg IronicWiki content use the “body” field rather than “text” as do normal Document content), and likewise I couldn’t successfully translate content type names.
So, find and replace it is. WickedDocs and IronicWiki are essentially modern Plone Documents in disguise (as wiki markup is the same), so we can just translate them over. In addition, I go ahead and rename anything that’s called “index_html” as this would otherwise be displayed as the actual page on a given folder automagically (and not through being selected in the Display drop down menu):
cd /path/to/our/temp/directory grep -lr -e 'WickedDoc' * | xargs sed -i 's/WickedDoc/Document/g' grep -lr -e 'IronicWiki' * | xargs sed -i 's/name=\"body\"/name=\"text\"/g' grep -lr -e 'IronicWiki' * | xargs sed -i 's/IronicWiki/Document/g' find . -name index_html | xargs rename -v 's/index_html/main_page/g' grep -lr -e 'index_html' * | xargs sed -i 's/index_html/main_page/g' grep -lr -e 'copy_of_main_page' * | xargs sed -i 's/copy_of_main_page/copy_of_index_html/g'
Once done with this, I copied the ‘structure’ directory (inside the temp directory) into the .tar.gz you have from earlier, ready for upload to the new site.
For Plone 4
- Set up your Plone 4 like normal. You should (are?) using Buildout, so follow the instructions here, again from QuintaGroup, to install the same quintagroup.transmogrifier product into your Plone 4 instance (eg just add it into your eggs and zcml for your instance).
- Create a Plone site if you haven’t already.
- Like before, you can use the configuration page at http://localhost:8080/<your_site_name>/@@pipeline-config , except this time, configure the import options. My config looks like this: transmogrifier-import.txt - pretty much the same as the default.
- When you’re ready, head into the ZMI and into ‘portal_setup’, this time clicking on the ‘Import’ tab.
- Browse down to the very bottom of the page, and use the file field (hint: Browse button) to locate your .tar.gz file you’ve prepared.
- Click ‘Import uploaded tarball’ and watch your console output for progress.
Cleaning up
This process wasn’t without manual work, unfortunately. But, it certainly took care of 95% of the pain of copying content across! Here’s an overview of the manual work I had to do:
- Set default pages of folders. This could have probably been carried out by a script but I didn’t write one because I only had a 20 folders to fix. In Plone 2.5, I believe the default view was when you set a page to be called ‘index_html’ (happy for corrections), but AFAIK, there weren’t options to set a certain page as the view.
- Set site settings. This involved going through all control panels to pick and manually port settings, but because of so many changes from Plone 2.5 to 4, there’s probably not an obvious migration path anyway.
- Convert legacy portlets. These settings do get copied across, but you’ll need to go to the ‘Manage portlets’ link on your main homepage and click the ‘Convert’ button. At time of writing (Plone 4.0.2) this migration is actually broken (ticket logged already!) but you can just as easily migrate manually — look in the ZMI properties at the site root for your portlet listings.
- Fix some workflow states. My old site used a non-default workflow, so workflow states needed correcting. Thankfully, I was able to push most content around just using the Types control panel.
And the rest of the actions I did were simply boilerplate for setting up a Plone 4 site (theme installation, etc).
This isn’t the most straightforward path, but it did work and get my content across. Hopefully moving to Plone 5 and above won’t be too much of a problem.