880 alternate script handling

[hornc]

Closes #7264

• Adds test for Don't import MARC 250$6 as part of edition name #7617
• Extracts alternate script titles to Edition title, and transliterations to other_titles (justification: edition record is for the item as printed -- the alternate script is likely what is printed on the book, this is the most unambiguous identifier)
• Extracts alternate script Author names to alternate_names, with the transliteration as name (justification: the author record is an abstraction representing an individual who may have used multiple names and forms over time, which stands independently of how their name was presented on a specific book. We need to pick an identifier, and using the catalog language is not inappropriate)
• Uses standalone 880 fields to obtain a publisher name if no transliterated one exists. (part of Alternate script fields (880) not extracted from MARC imports #7264)
• De-duplicates series after stripping punctuation
• Refactoring around the MarcBinary and MarcXml classes
• Adds functionality, tests, and documentation, while deleting lots of code ;)

• Test MARC record for edition_name taken from the example in Don't import MARC 250$6 as part of edition name #7617
  • https://openlibrary.org/books/OL27062719M
  • https://openlibrary.org/show-records/ia:isbn_9787508617725

To add:

• MARC XML support for 880 linkages -- use existing XML test input: nybc200247_marc.xml

• Table-Of-Contents 880s example from Don't import MARC 250$6 as part of edition name #7617 : https://openlibrary.org/show-records/marc_miami_univ_ohio/allbibs0193.out:11791288:963

  • !! This record looks potentially broken -- the target 880 fields are missing -- I'd expect Cyrillic script in the record somewhere. There isn't any. For the purposes of this test case, we are only checking that the string 880-04 or similar does not appear in the generated TOC, so it 'works' for that purpose.

Examples from #7264

• Publisher only in 880 field: https://openlibrary.org/show-records/harvard_bibliographic_metadata/ab.bib.13.20150123.full.mrc:49430:858 RTL Hebrew.
• RTL example; RTL Arabic and French example, a recent import from Columbia. Many 880 fields, 880 in edition_name
  • https://openlibrary.org/books/OL31187713M https://openlibrary.org/show-records/marc_columbia/Columbia-extract-20221130-017.mrc:84193714:3643
• Japanese text from https://openlibrary.org/show-records/marc_columbia/Columbia-extract-20221130-008.mrc:340428848:1828

Technical

Testing

Screenshot

Stakeholders

@tfmorris

[hornc]

RESOLVED

@cclauss I'm not sure about these Ruff PLC1901 messages, e.g.

PLC1901 `i.attrib['code'] == ''` can be simplified to `not i.attrib['code']` as an empty string is falsey

I don't want to change those blindly as 0, False, and None are all falsey too, but aren't empty strings.

[tfmorris]

@mekarpeles @hornc This PR shouldn't have been closed when the other PR mentioning it was merged.

[tfmorris]

This looks great. I'm sorry I don't have time to do the review justice, but I tried to comment on a few things that caught my eye. I'll try to circle back later and do a more thorough review of the new test cases (which look awesome!)

[cclauss]

Placate mypy...

[cclauss]

LGTM

[hornc]

@tfmorris, thanks for your reviews so far. I've tried to address everything, and split of separate pieces into their own issues to be worked on as follow ups. Is there anything else I should address immediately of this PR?

[tfmorris]

I was going to dig up some additional test data to cover s.l., etc but I think it's fine to merge as is and handle that later. It's a great improvement and I really appreciate all the work.

[tfmorris]

Here are a couple of test cases for the latinized publish notation, in case they're still useful
Archive.zip