GitLab

17 harvester logic

* Major redesign of the harvester logic.
* Rename the class `PublicationsTool` to `Automaton`.
* Refactoring of several methods by moving the code between the class `Record` and `CheckAndFix`, by renaing the method, *etc*.
* Detect errors when a record is rejected by the database engine.
* Close #17.

See merge request !26

CheckAndFix.conference_date.

Modify the method duplicate_article since JHEP 06 (2012) 115 and JHEP 06 (2015) 115 is valid (see cds 1443914 and 2003394).

- Non-conformity where the first author is in u100 and not in u700
  is now handle. it affects the  method authors_as_list, and all
  methods to find authors affiliations.
- Special treatment of the authors in RecordThesis since the field
  u700 contains the names of Thesis director.

16 reject on oai

* First step to improve the logic of the harvester
* The automate `PublicationTool` processes the requested URL. It starts with a list of record `id`. At that step, it looks for a publication in the database corresponding to that `id`. It moves to the next id when there is a match.
* The processing continue, the record is request to the store and decoded. If the `record.oai()` is not defined, the record is rejected.
* Several non conformities have been found when testing this new logic. Therefore, several protections and tests have been added
* Close #16

See merge request !25