Records But No Fields with OAH 2.3.0

[pennington]

We've been using Harvester 2.0.1 for several years and have really enjoyed it. We have several different institutional repositories, and it has served as a great way to search for objects across our repositories.

I recently installed the 2.3.0 version, and I like many of the improvements, but one thing that is not working is, once I have harvested items from a repository via OAI, I cannot see the different field data for a particular object. This is happening on DSpace, Fedora, and a different repository.

For example, this is our production 2.0.1 instance:

http://dam.rhodes.edu/harvester/index.p ... se/index/1

As you can see, there are objects with fields of data. However, the same browse command on the 2.3.0 version displays no data per object:

http://dam.rhodes.edu/harvester230/inde ... se/index/1

If you click on View Record, you can see that the data is there but not broken up by field.

Any idea what could be going on?

BTW, I've already set up the Sort Order and Crosswalks on the 2.3.0 version, and it didn't make any difference.

I did this with a new install of 2.3.0. Would I get any better luck if I upgraded the 2.0.1 system to 2.3.0?

Thanks in advance for the help.

Stacy

[asmecher]

Hi Stacy,

What metadata format are you using, and what version of PHP is your server running?

Regards,
Alec Smecher
Public Knowledge Project Team

[pennington]

All 3 of our repositories use Dublin Core, and we are running PHP 4.3.9, MySQL 4.1.22, and Apache 2.0.52.

[asmecher]

Hi pennington,

Do you have the option of trying this with a newer version of PHP? Version 4.3.9 had some strange reference handling, and I suspect that may be the culprit. If that's not possible, let me know and I'll look into it further.

Regards,
Alec Smecher
Public Knowledge Project Team

[pennington]

We are running RHEL AS 4 on that box, and if we want to stay in the supported channel, PHP 4.3.9 is the highest RPM available.

I've got a test Ubuntu box running a pretty recent version of PHP 5, so I'll give it try on that system.

Thankfully, the Open Harvester System software is well designed and easy to set up!

Stacy

[asmecher]

Hi Stacy,

Thanks -- let me know if it works OK on a newer PHP and I can probably find a work-around for your production environment.

Regards,
Alec Smecher
Public Knowledge Project Team

[pennington]

OK, I installed 2.3.0 on a server using PHP5, and the fields are showing up fine, so this must be a PHP 4.3.9 issue. You can see my test server output here:

http://173.203.87.138/harvester/index.p ... se/index/1

However, I am not able to parse all the objects on the site. I get through 1078 out of over 6000, and I get an "Invalid Character" error when harvesting via the command line. I'm sure that one of our objects has some sort of invalid character that needs to be removed, but how can I figure out which one? No identifying information about the object in question is given in the output.

Also, I just re-harvested this same OAI repository this morning using the 2.0.1 harvester, and I did not get the "Invalid Character" error. Is that version of PHP or the harvester more tolerant of invalid characters?

Thanks.

Stacy

[asmecher]

Hi Stacy,

Try turning off the charset_normalization option in config.inc.php and re-harvesting.

Regards,
Alec Smecher
Public Knowledge Project Team

[pennington]

Thanks! Changing the charset_normalization setting fixed it!

I appreciate all of the help. I'm going to upgrade to the 2.3.0 version on a production PHP5 system and see how that goes.

Thanks again!