Add regexes for `database_list`

[zachary-foster]

• [ ] ncbi
  • [ ] gbif
  • [ ] bold
  • [ ] col
  • [ ] eol
  • [ ] nbn
  • [ ] tps
  • [ ] itis

[sckott]

can you explain a bit more?

[zachary-foster]

Currently, database_list just has placeholders for the rergexes:

> database_list
$ncbi
<database> ncbi
  url: http://www.ncbi.nlm.nih.gov/taxonomy
  description: NCBI Taxonomy Database
  id regex: .*

$gbif
<database> gbif
  url: http://www.gbif.org/developer/species
  description: GBIF Taxonomic Backbone
  id regex: .*

$bold
<database> bold
  url: http://www.boldsystems.org
  description: Barcode of Life
  id regex: .*

$col
<database> col
  url: http://www.catalogueoflife.org
  description: Catalogue of Life
  id regex: .*

$eol
<database> eol
  url: http://eol.org
  description: Encyclopedia of Life
  id regex: .*

$nbn
<database> nbn
  url: https://nbn.org.uk
  description: UK National Biodiversity Network
  id regex: .*

$tps
<database> tps
  url: http://www.tropicos.org/
  description: Tropicos
  id regex: .*

$itis
<database> itis
  url: http://www.itis.gov
  description: Integrated Taxonomic Information System
  id regex: .*

We dont use database_list currently, but we probably will once some kind of taxon ID validity checking is added. So when a user does something like : taxon_id(id = "not valid", database = "ncbi") they get an error.

[sckott]

ah right, forgot about those

[zachary-foster]

What do you think about only using the taxon_database class for that list and forcing the user to pick a database name from that list instead of storing taxon_database objects? If they are using a custom database, they can either not specify a taxon database, or add their own taxon_database object to a local version of list.

It seems like up to three taxon_database objects per taxon objects (name, id, rank) would be a bit heavy on RAM, especially since 99% of the time they will all be the same database. In other instances where either a vector or object could be used, we decided (#139) to only allow objects for consistency. In this case, perhaps only allowing vectors might make more sense?

[sckott]

Yeah, it does make sense to try to reduce memory usage. agree that most cases will probably be each each of name, id and rank being from same database

[zachary-foster]

ok, so only store vectors for database in taxon_rank, taxon_id, and taxon_name objects then?

[sckott]

i think so, but the accessor methods/functions do construct a database object though ?