Adding books requires identifiers

[gedankenstuecke]

It seems that the library currently requires specifying either an ISBN or LCCN identifier in order to be able to create a new book, which differs from the behaviour of the web form for the book creation, where the identifiers are optional.

To reproduce this should work just fine

from olclient.openlibrary import OpenLibrary
import olclient.common as common
ol = OpenLibrary()
book = common.Book(title="My test book title", authors=[common.Author(name="Testy McTestFace")], publisher=u"Publisher", publish_date=u"2018")
new_book = ol.create_book(book)

This will trigger the value error.

I'm not sure if this is expected behaviour that was designed to work like this, or whether the openlibrary-client is just out-of-sync with how the web-form behaves? If it's the latter I'd be happy to give it a shot to adjust it according to the web form🙂

[LeadSongDog]

Because auto imports happen much faster than manual ones, they should be made in a more trustworthy way, from reliable source metadata (not the crapfest that is BWB and AMZ).

That isn’t the only inconsistency to address. At

openlibrary-client/olclient/openlibrary.py

Line 51 in dc9195c

VALID_IDS = ['isbn_10', 'isbn_13', 'lccn', 'ocaid']

the VALID_IDS include ocaids, but at

openlibrary-client/olclient/openlibrary.py

Line 765 in dc9195c

raise ValueError("ISBN10/13 or LCCN required")

they do not.

DRY: let’s have one list of first-class identifiers in one place, please.

That said, it would make sense to support other identifiers, not just these four. This is part of the reason we’ve had such poor uptake of pre-ISBN books from libraries outside the anglosphere, where Worldcat and LCCN coverage is less common. There are six first-class and 66 second-class identifiers listed on the edition edit page drop-down for "Select one of many…" under ID Numbers. For starters, we should pick up the BSB-ID, the Biblioteca Nacional de Español Depósito Legal, and the BnF identifiers to be treated as first-class. They are reliably and neutrally curated, far more so than Goodreads, Library Thing, Google, AMZ, and other such newcomers operating with commercial purposes. The British Library and BNB ids could also be worthwhile inclusions in the first-class list.

[cclauss]

msg = f"{' or '.join(VALID_IDS)} required"
>>> msg
'isbn_10 or isbn_13 or lccn or ocaid required'

[LeadSongDog]

@cclauss
Right. That too needs to be DRYed.

[gedankenstuecke]

Thanks for chiming in folks!

Because auto imports happen much faster than manual ones, they should be made in a more trustworthy way, from reliable source metadata (not the crapfest that is BWB and AMZ).

I guess that means that the openlibrary-client is supposed to require some form of external ID for importing books, correct? 🙂

[cclauss]

msg = f"{' or '.join(VALID_IDS)} required" is dry.