this is useful for importing data over mnet, but also potentially in the
future when we can import leap data into existing user accounts

Signed-off-by: Penny Leach <penny@mjollnir.org>

- Breaking up of the validation into two stages:
    - Performed by the importtransport object, before moving files
      around
    - Performed by the import class, after the files are transported

To this end I changed the importtransport objects to take the
importqueue record (or a fake one) in their constructor, so they have
the same data available as the importer would.

Signed-off-by: Penny Leach <penny@mjollnir.org>

This commit includes:

- Moved a lot of logic out of admin/users/add.php to the various plugins
- Moved around unzipping and detecting files into the transport layers
- Refactored a lot of transport code into the base class

I tested:

- Adding a user from a leap2a xml file
- Adding a user from a leap2a zip file
- Importing non leap (straight files) data over mnet

Signed-off-by: Penny Leach <penny@mjollnir.org>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

The export includes the view type now, and we import the views having
the correct type.

On second thoughts, I'm guessing the importer will die horribly if
someone makes an export with two profile views, but that can be a
problem for another day.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

It turns out I was relying on a nasty behaviour quirk of json_decode in PHP,
namely that it'll just pass strings through - in some cases. This is horribly
out of touch with the JSON specification, see these PHP bugs:

http://bugs.php.net/bug.php?id=38680 (where they broke it)
http://bugs.php.net/bug.php?id=46518

 (people complaining about it)

The solution is to wrap all configuration fields when exporting as LEAP in an
array and then json_encode that. json_decode will handle that properly in all
versions of PHP.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

It takes stdClass objects, not arrays.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Set $CFG->log_file to whatever log file the importer should be using, before logging a trace message.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

This puts the right artefacts back into Views.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

The other base class methods had detailed descriptions, so this only
seems fair.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

We unconditionally json_encode it on the way out, so this is necessary.

Allows us to store arrays in the export, too.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

This is necessary so blocks can make decisions about importing based on
the owner.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Found this bug while doing View import. Using uksort passes the keys in
for sorting, which is not what the sorting function was actually
expecting. uasort is what I meant to use.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

This is so that we know all artefacts have been created, so that the
views they're in can also then be set up.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

The LEAP code now changes the XML generated at export time back into the
config to use for this method.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

This is much more reliable than looking at the entry content, and means
we can easily get at the configuration for the blockinstances.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Now we just have to get their configuration right!

Ah crap, I said "just", I guess I just volunteered myself.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Imports all content into one big text box.

Manages to import title, summary, ctime and mtime too. Still have
ownerformat to do.

The next step after this is to try and detect if we're importing a
Mahara view, and if so, create the right blocks and artefact links for
it.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

This should make us a little more able to find them in other exports.
Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Some other portfolio systems don't declare them properly. This really
harms interoperability, but it's not our fault.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Use it everywhere we're looking for category schemes.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Call it everywhere we're looking for rdf types.

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Signed-off-by: Nigel McNie <nigel@catalyst.net.nz>

Saves a bunch of copy+pasted code.

Should be useful in many places - in particular, the resume.

Use this hook instead of hard coding a call to the Internal plugin's
import_author_data method.

This allows the resume to also hook in and grab the personalinformation
fields out.