Help moving broken Movable Type database to WordPress?
June 29, 2007 8:25 AM Subscribe
I have a personal blog running under Movable Type. I want to move to WordPress. I've exported the existing entries to a text file, so that's good. However, a couple years ago my previous MT install tanked (similar to this). I cannot access it. I have several years worth of posts that are stuck in a database that I can't get to. I've been gradually moving these by hand to my existing MT install, but that's a pain in the ass. Is there some automated way to crawl/scrape the existing HTML files in order to retrieve these "lost" entries? Some way to break back into a corrupt MT database? Some way to repair it? (The install is BerkelyDB-based.)
Important note: I have FTP access to all the "lost" entries, too. If I were ambitious, I'd figure out how to write a script that would parse all this information for me. I'm not ambitious, and am hoping that somebody has already done such a thing...
Next on the agenda: converting an existing custom MT template to WP!
For the record: current MT install and previous MT install
Important note: I have FTP access to all the "lost" entries, too. If I were ambitious, I'd figure out how to write a script that would parse all this information for me. I'm not ambitious, and am hoping that somebody has already done such a thing...
Next on the agenda: converting an existing custom MT template to WP!
For the record: current MT install and previous MT install
Assuming that you published static pages of your old blog with the corrupt database, the information is all "out there" in a digestible format. One way of digesting it would be this:
1. Use something like wget or a site-ripper app to crawl your old blog and fetch all the entries to your desktop. (oh wait, you've got FTP access, just DL the whole thing).
2. Figure out the GREP patterns necessary to convert your old blog entries into MT's export format.
3. Run those patterns over your crawled files and cat them into a single file.
Make no mistake: this is a tedious process, but it can be done. I've done it on an old Blogger blog when the Blogger backend was unavailable.
posted by adamrice at 9:13 AM on June 29, 2007
1. Use something like wget or a site-ripper app to crawl your old blog and fetch all the entries to your desktop. (oh wait, you've got FTP access, just DL the whole thing).
2. Figure out the GREP patterns necessary to convert your old blog entries into MT's export format.
3. Run those patterns over your crawled files and cat them into a single file.
Make no mistake: this is a tedious process, but it can be done. I've done it on an old Blogger blog when the Blogger backend was unavailable.
posted by adamrice at 9:13 AM on June 29, 2007
For minimum effort you might try the Wordpress RSS import, since your RSS links seem fine.
posted by brool at 9:41 AM on June 29, 2007
posted by brool at 9:41 AM on June 29, 2007
Can you provide more details about the old DB that you "can't get to"? What kind of DB was/is it -- a MySQL DB, a Berkeley DB file, ...? What's preventing you from getting access to the data in it -- is it corrupt? You lost the username/password? More data is key.
posted by delfuego at 10:03 AM on June 29, 2007
posted by delfuego at 10:03 AM on June 29, 2007
Response by poster: As I mentioned in the question, the install that's giving me trouble is BerkelyDB-based. I don't know precisely what's preventing me from accessing the file. It's not a lost username/password. In my first post on this problem nearly two years ago, I noted that MT-Medic indicated that "MT-Medic does show existing weblogs and authors, though no weblogs are associated with any authors."
HA! In a delicious piece of irony, I seem to have deleted my comment script for that particular MT install last week when I removed a subdomain. "What is this here for?" I wondered. Now I know. It doesn't matter, though, because comments were broken too. I was going to suggest that people try to leave a comment to look at the error that resulted, but now I've created a new error in its place!
posted by jdroth at 10:16 AM on June 29, 2007
HA! In a delicious piece of irony, I seem to have deleted my comment script for that particular MT install last week when I removed a subdomain. "What is this here for?" I wondered. Now I know. It doesn't matter, though, because comments were broken too. I was going to suggest that people try to leave a comment to look at the error that resulted, but now I've created a new error in its place!
posted by jdroth at 10:16 AM on June 29, 2007
Response by poster: Here's some more info. I've been playing with MT-Medic, trying to get things to work. As I reported a couple years ago, it shows all the proper weblogs and authors, but it doesn't show any connection between them. Also, when I attempt to login to the MT install, it gives me "no such author" error for any author. When I attempt to re-create connections between blogs and users in MT-Medic, they do not "take". Very frustrating.
posted by jdroth at 10:28 AM on June 29, 2007
posted by jdroth at 10:28 AM on June 29, 2007
Hey JD, I work with the MT team, and I'm sorry you're having such a bad experience. I'd love to have us work with you to fix whatever's wrong with your install.
If you're game, drop me a line (or IM me at anildash) and I'll get you set up. It sounds like all the issues you're having are definitely fixable.
posted by anildash at 10:56 AM on June 29, 2007
If you're game, drop me a line (or IM me at anildash) and I'll get you set up. It sounds like all the issues you're having are definitely fixable.
posted by anildash at 10:56 AM on June 29, 2007
Response by poster: Anil, I tried to e-mail you a couple years ago about this problem but got no reply! I'll IM you, okay?
posted by jdroth at 10:57 AM on June 29, 2007
posted by jdroth at 10:57 AM on June 29, 2007
Try to avoid anything that uses the Movable Type export format, which is just broken and was the source of headaches every time I tried to use it (you may have no problem as long as your posts and comments don't contain troublesome pieces of text like "-----").
posted by jjwiseman at 11:10 AM on June 29, 2007
posted by jjwiseman at 11:10 AM on June 29, 2007
Assuming that the html pages exist, you should be able to do this. Another person and I did this for Wil Wheaton when his db got borked.
We went around the db issue and directly to the published archive pages themselves.
Basically, I used Teleport Pro to rip the html to local files.
Once we had the local finles, we ran a script based on this MT support thread to convert it to the MT export format.
Yoshi, my coworker did rewrite the code a bit to make it more to his liking, but we were able to create a functional file that was re-imported back into a clean MT and/or Typepad install.
I asked Yoshi if he still had the code, but it was back in December 05 or some time long ago. If he finds it, I will post linkage here.
That's the concept at least, it worked once, and hopefully is useful to you.
posted by Argyle at 11:24 AM on June 29, 2007
We went around the db issue and directly to the published archive pages themselves.
Basically, I used Teleport Pro to rip the html to local files.
Once we had the local finles, we ran a script based on this MT support thread to convert it to the MT export format.
Yoshi, my coworker did rewrite the code a bit to make it more to his liking, but we were able to create a functional file that was re-imported back into a clean MT and/or Typepad install.
I asked Yoshi if he still had the code, but it was back in December 05 or some time long ago. If he finds it, I will post linkage here.
That's the concept at least, it worked once, and hopefully is useful to you.
posted by Argyle at 11:24 AM on June 29, 2007
JD, this thread has some cool links for saving local copies of remote HTML files, as Argyle's methodology describes.
posted by misterbrandt at 10:59 AM on June 30, 2007
posted by misterbrandt at 10:59 AM on June 30, 2007
This thread is closed to new comments.
posted by tmcw at 8:35 AM on June 29, 2007