[OJS 3.3.0-4]: OAI earliestDatestamp problem

Eder_Sotto · March 23, 2021, 5:31am

Dear all,

I’ve migrated OJS from 2.4.8.5 to 3.3.0-4 this week.

I found problems when indexing articles with OAI. When I access /oai?Verb=ListRecords&metadataPrefix=oai_dc, all articles appear with the same current timestamp. I saw that others have already reported this problem in previous versions, and I verified that all the solutions already proposed were implemented in this version, including filling in the “last_modified” column on the tables “issues” and “submissions”.

The driver plugin is disabled. My code and database version is 3.3.0-4.

print

Can someone help me?

Thank you!

asmecher · March 30, 2021, 5:08pm

Hi @Eder_Sotto,

Are all dates for the following columns “valid” (not just set, but something other than the UNIX epoch date)?

submissions.last_modified
issues.last_modified
data_object_tombstones.date_deleted

Regards,
Alec Smecher
Public Knowledge Project Team

Eder_Sotto · March 31, 2021, 11:11am

Hello @asmecher, thank you for your reply.

I checked all the records in the three tables, and got the following result.

submissions.last_modified: all with the last_modified column
filled and all UNIX timestamp dates valid. 1249 records. The oldest last_modified 2011-09-27 17:24:48 and the most recent last_modified 2021-02-09 16:09:23.

issues.last_modified: all with the last_modified column
filled and all UNIX timestamp dates valid. 55 records. The oldest last_modified 2016-10-05 21:05:51 and the most recent last_modified 2021-03-15 07:43:10.

data_object_tombstones.date_deleted: this table is empty, with no records.

asmecher · March 31, 2021, 7:26pm

Hi @Eder_Sotto,

Hmm, can you verify that this change is in place (in the lib/pkp subdirectory)? pkp/pkp-lib#6803 Fix return of earliest datestamp · pkp/pkp-lib@aa7ee24 · GitHub

Regards,
Alec Smecher
Public Knowledge Project Team

Eder_Sotto · April 5, 2021, 3:11am

Hello @asmecher, I´ve verified. Yes, I have this bugfix applyed in the code. My database/code version is the newest 3.3.0-4. The problem is ocurring since i’ve updated from OJS 2.4.8. I’ve checked and applied all the suggested fixes, but the problem remains. Thank you!

asmecher · April 7, 2021, 12:24am

Hi @Eder_Sotto,

Hmm, I think you’re probably going to have to do some debugging in your code to track this down. The query that is responsible for getting the dates (among other things) is here:

github.com

pkp/ojs/blob/stable-3_3_0/classes/oai/ojs/OAIDAO.inc.php#L235


		if (isset($journalId)) $params[] = (int) $journalId;
		if (isset($sectionId)) $params[] = (int) $sectionId;
		if ($submissionId) $params[] = (int) $submissionId;
		if (isset($journalId)) $params[] = (int) $journalId;
		if (isset($sectionId)) $params[] = (int) $sectionId;
		if (isset($set)) {
			$params[] = $set;
			$params[] = $set . ':%';
		}
		if ($submissionId) $params[] = (int) $submissionId;
		return $this->retrieve(
			'SELECT	GREATEST(a.last_modified, i.last_modified) AS last_modified,
				a.submission_id AS submission_id,
				j.journal_id AS journal_id,
				s.section_id AS section_id,
				i.issue_id,
				NULL AS tombstone_id,
				NULL AS set_spec,
				NULL AS oai_identifier
			FROM
				submissions a

This is used by the following code to get the earliest datestamp:

github.com

pkp/pkp-lib/blob/stable-3_3_0/classes/oai/PKPOAIDAO.inc.php#L173


			$records[] = $this->_returnIdentifierFromRow((array) $row);
			$total++;
			$result->next();
		}
		while ($result->current()) {
			$total++;
			$result->next();
		}
		return $records;
	}

	/**
	 * Return the *nix timestamp of the earliest published submission.
	 * @param $setIds array optional Objects ids that specify an OAI set,
	 * in hierarchical order. If empty, all records from
	 * all sets will be included.
	 * @return int
	 */
	function getEarliestDatestamp($setIds = array()) {
		$result = $this->_getRecordsRecordSet($setIds, null, null, null, null, 'last_modified ASC');
		if ($row = $result->current()) {

If you use e.g. error_log / print_r to dump the $record that this last function is returning, it’ll help. Alternately, if it’s actually following through to the return 0; that’ll be helpful to know too.

Regards,
Alec Smecher
Public Knowledge Project Team

Eder_Sotto · April 7, 2021, 2:23am

Hi @asmecher ,

Thank you very much for your help, it was essential to resolve this. Through the query I found that a publication_settings record pointed to a non-existent issue, generating that problem.

It’s solved now. Thank you!

Regards.