The xISBN problem you encountered is fixed in the trunk as of r2367.  Code has been moved around significantly in the trunk since the release of RC2, so you can't easily take advantage of this fix simply by copying files out of the trunk.  However, if you need to backport my fix to RC2, you can just replace the main processing loop of Record::getXISBN in web/services/Record/Record.php with this:


        // Fetch results

        if ($fp = @fopen($url, "r")) {

            $query = '';

            while (($data = fgetcsv($fp, 1000, ",")) !== FALSE) {

                // Filter out non-ISBN characters and validate the length of

                // whatever is left behind; this will prevent us from treating

                // error messages like "invalidId" or "overlimit" as ISBNs.

                $isbn = preg_replace('/[^0-9xX]/', '', $data[0]);

                if (strlen($isbn) < 10) {



                if ($query != '') {

                    $query .= ' OR ';


                $query .= 'isbn:' . $isbn;




Please let me know if you have any questions or problems!


- Demian


From: John Houser [] On Behalf Of John Houser
Sent: Thursday, March 25, 2010 4:11 PM
To: Demian Katz
Subject: Re: [VuFind-Tech] Other Editions


Thanks, Demian! 


ISBN clean-up is on our list of todos. No point in cleaning up the test data yet, however, since it will just get replaced when we've got our migration process finished. 





On Mar 25, 2010, at 3:59 PM, Demian Katz wrote:

Here's what's happening:


1.) VuFind calls WorldCat's xISBN service using the ISBN for Beginning programming for dummies (0764505953).

2.) Although this is a valid ISBN, the API doesn't like it for some reason -- it gives an "invalidId" error.

3.) VuFind isn't smart enough to detect this error.  It does a search in Solr for ISBNs matching "invalidId."

4.) Solr's data normalization for ISBNs strips all illegal ISBN characters from the search string, which happens to be all of them.  Thus, the search matches all other records in the index that have invalid values in their ISBN fields.


A little wacky, but hopefully that explanation makes sense.  I'll fix the VuFind bug that allows this to happen in the near future…  but in the meantime, you might want to take advantage of this bug to do a little bit of data cleanup, since it seems to point out a few problem records with missing ISBNs.


- Demian


From: John Houser [] 
Sent: Thursday, March 25, 2010 3:51 PM
Subject: [VuFind-Tech] Other Editions


We're getting some string responses from the Other Editions. I've included an example below. Anyone have any hints or pointers for me as I investigate?



Beginning programming for dummies /

Main Author: 






Foster City, CA : IDG Books Worldwide, 1999




Other Editions

John Houser
Technology Coordinator



John Houser
Technology Coordinator