XF 1.5 Minor Problems with IPB 3.4 Import

[Tiki Tiki]

I've got a couple of minor problems when trying to import my IPB 3.4.x database.

My IPB database says its collation is utf8_unicode_ci, it looks like a fresh installation of XF 1.5 is the same, but I've got some characters that are coming across as gibberish. A few examples are – instead of — em dash, •A instead of • bullet, and ≈ instead of ≈ almost equal to. (As an aside, I believe IPB 4 is now using utf8mb4, but, again, this is a 3.4.x database I'm importing.)

One IPB post, where the entire content was enclosed in < > (e.g., <rolling eyes at person above>) came through blank. The member name is there, the post date/time is there, but the content is blank. Weird.

The Import Polls step blows up with errors:

ErrorException: array_map(): Argument #2 should be an array - library/XenForo/Importer/IPBoard.php:2970

array(3) {
["url"] => string(59) "http://community.mysite.com/admin.php?import/start-step"
["_GET"] => array(1) {
["import/start-step"] => string(0) ""
}
["_POST"] => array(3) {
["options"] => array(1) {
["whichQuestion"] => string(5) "first"
}
["step"] => string(5) "polls"
["_xfToken"] => string(8) "********"
}
}

My site might have one or two ancient polls, if that, so this one is of little consequence.

Anyone have any suggestions as to what is going wrong and how to fix these issues?

 

[Mike]

In general, errors will require a reimport. That will mean restoring to a pre-import backup or doing a fresh install.

The character encoding issue likely relates to the DB connection using a different character set than it did. There's an option to force a character set on the DB details page. Try forcing one of these two:

latin1
utf8

(Likely the former.)

The post with the <...> is a tricky one, as posts are normally stored as HTML now, so the <> there is presumably not escaped, which is weird. Is it an older post?

With those ancient polls, it may be a case that they're using an old data format that we aren't fully handling. I'll need to investigate the code to see what would trigger that.

 

[Tiki Tiki]

Try forcing one of these two:

latin1
utf8

Will do.

The post with the <...> is a tricky one, as posts are normally stored as HTML now, so the <> there is presumably not escaped, which is weird. Is it an older post?

Jan 25, 2011. The actual post content is:

<throwing my wand into the ring...>

I noticed the post by happenstance as it's in a hidden Admin only forum and I was playing with XF's permissions to make that forum operate the way it was intended. It's one post, no biggie, but it made me wonder if this sort of thing could be happening elsewhere, and I figured I should ask about it.