move TEI idno identifiers under <analytics>

[lfoppiano]

See #1192 . The same treatment is applied to any identifier: PMCID, PMID, halID, etc

See example:

                                </address>
                            </affiliation>
                        </author>
                        <title level="a" type="main">Transgressive phenotypes from outbreeding between the Trichoderma reesei hyper producer RutC30 and a natural isolate</title>
                        <idno type="DOI">10.1128/spectrum.00441-24</idno>
                    </analytic>
                    <monogr>
                        <imprint>
                            <date type="published" when="2024-08-20">20 August 2024</date>
                        </imprint>
                    </monogr>
                    <idno type="MD5">9E9A05DAEBD10C49EB098AF73FA55CD1</idno>
                    <idno type="DOI" status="deprecatedLocation">10.1128/spectrum.00441-24</idno>
                    <note type="submission">Received 22 February 2024 Accepted 3 July 2024</note>
                </biblStruct>

[coveralls]

coverage: 40.766% (+0.01%) from 40.755%
when pulling 60d7a19 on bugfix/move-idno-under-analytics
into be44579 on master.

[kermitt2]

Hi Luca !

Thinking about it a second time, it might be more complicated than that and I think I understand the motivation for letting the identifiers under in the case of Grobid.

If we extract a raw DOI from a PDF, the position of the DOI in normal TEI depends on the type of document we process, either under if this is a part of a monograph or journal, or under if we have a report, standalone article, etc. As we can't know for sure about the document type, it make sense to put it under in Grobid, to avoid errors, as a relaxed rule.

Second, we have HAL ID which corresponds to standalone document, same for arXiv ID. We cannot put them under , but it does not make sense to put it under neither. is a valid default choice that explicitly capture the fact that we cant know the level of the document.

The idea behind this is: when we don't know the level associated to a extracted ID, we explicitly put it under .

@laurentromary for comments :)

[kermitt2]

Note to myself: check how the DOI/ids are positioned in the case of consolidated header. We can have 2 different DOIs, one for the part and one for the "hosting" document.

[lfoppiano]

@laurentromary I'm revamping this PR and I'm wondering if you have any further comment.

One option forward is to move identifier on conditions:

• Container detected (journal article, book chapter, proceedings paper): article-level ids — DOI, PMID, PMCID, PII, istexId — go under <analytic>, because they identify the part (the article), while the journal's own ids (ISSN) already sit in <monogr>.
• No container (preprint, report, thesis-like standalone): the same ids go under <monogr>, because there is no "part vs. container" split — the document is the monograph.
• halId, arXiv, ark: these identify a standalone deposit by nature. They'd stay at the biblStruct level, unmarked (no deprecatedLocation).

This has the disadvantage of being more difficult to parse and possibly taking years for downstream process to adapt.

[laurentromary]

Oui, c'est le plus cohérent. Les rapports et les thèses sont effectivement des objets monographiques. Pour les preprint, c'est moins clairs, mais c'est mieux de rester dans le même cadre (comme tu le proposes).