grobidOrg / grobid / 29838452886

44%

push

github

Avoid assigning the same affiliation once per marker character (#1512)

* fix: stop assigning the same affiliation once per marker character

Authors carrying a repeated-symbol marker received the same affiliation
several times:
- A marker is now a run of identical characters taken whole: an alphanumeric run ("12", "1a"), a repeated symbol ("**", "§§§§"), or a repeated comma (",,") - a single comma still separates markers. 
- Person.addAffiliation appended unconditionally. The tiers in AuthorAffiliationAssigner run in sequence rather than exclusively, so one   author can be offered the same affiliation instance more than once; only preLinkByPrecedingAuthorCluster guarded against it. The guard now lives in addAffiliation, covering every call site, and holds independently of the split pattern.
 - Update benchmarks


---------

Signed-off-by: Luca Foppiano <luca@foppiano.org>

10124 of 25672 branches covered (39.44%)

Branch coverage included in aggregate %.

20976 of 46507 relevant lines covered (45.1%)

1.83 hits per line