GitLab

Harvester logic

The logic implements in the Automaton class is the following

1. Ask to the store, all the record_id satisfying the user request.
2. Reject record_id matching the origin field of a database entry.
3. Request to the store, the XML description of the publication and decode it.
4. Check that the oai of the publication is defined and well formed. Recover it if it is not the case. From time to time, the id encoded in the oai field is different from the record id. This happens when an old record is redirected to new one for obscure reasons. The record is ignore if a database entry is found with the bad OAI.
5. Reject temporarily publication.
6. Check that authors are defined. Reject the publication if it is not the case.
7. Check that my institute is in the list of the institutes signing the publication. Reject the publication if it is not the case. When the affiliation are not defined, try to recover this case, by finding the author of my institute signing the publication. This recovery procedure uses the author rescue list. Reject the record when the recovery procedure failed.
8. Check that the collaboration, if defined, is well formed. Reject the publication if it is not the case
9. Several check are applied depending on the publication type.
10. At the end of this process, the publisher, the authors are formatted and the list of signatories of my institute extracted.