Can I isolate the publisher-friendly aspects of biblatex in a file?

Question

The biblatex package with Biber is currently designed to be incapable of achieving one crucial thing that natbib does: that is to produce a .texsource file that works independently of outside sources of bibliographic information. Publishers who process .tex will never accept manuscripts for submission that require using these biblatex with Biber features (quoting from CTAN):

A tool mode for trans­form­ing bib­li­o­graphic data sources

Au­to­matic bib­li­og­ra­phy data re­cod­ing (UTF-8 -> lat­in1, LATEX macros -> UTF-8 etc)

Re­mote data sources

Even publishers who happily use natbib ask that you put the .bbl file in the manuscript so their printers are not involved with the sources of your bibliography. The .bbl files for biblatex are designed not to work this way.

But biblatex does other things many publishers would like as much as authors do:

Full Uni­code sup­port

Highly cus­tomis­able sort­ing us­ing the Uni­code Col­la­tion Al­go­rithm + CLDR tai­lor­ing

Highly cus­tomis­able bib­li­og­ra­phy la­bels

Poly­glos­sia and ba­bel support for au­to­matic lan­guage switch­ing for bib­li­o­graphic en­tries and ci­ta­tions

Highly so­phis­ti­cated au­to­matic name and name list dis­am­bigua­tion sys­tem

Is there a way I could use biblatex, and let it do all its sophisticated processing in runs as I create the document, and then at the end use it to produce something that will work like a natbib .bbl file? Call it a quasi-.bbl file. I mean a file I could include in the manuscript as I now include that .bbl file, so that publishers who use natbib would succeed at generating the bibliography.

Answer 1

At a technical level there are a few things that would need working on here. First, as biblatex is set up for typesetting the bibliography, each command would need to be altered to write to a file instead. For 'expandable' commands that's straight-forward (they will work directly inside \write) but for non-expandable ones (the majority) some care would be needed to ensure that the final result of the processing would be written to file, not some intermediate form. (Depending on the style in use, quite a lot of processing can be going on.)

In a normal biblatex run the data (.bbl) is read in the preamble, meaning that numbering/labelling is done 'on-the-fly' in a LaTeX-[backend]-LaTeX process, rather than LaTeX-BibTeX-LaTeX-LaTeX of 'traditional' BibTeX. Saving the 'pre-formatted' data in a traditional-style .bbl would mean this would need to be addressed, most likely by needing a separate code path for the 'pre-packed' version.

In a lot of classical physical science styles, the citation labels are simple (numbers or author-date) and there is no need to do anything else other than dump the bibliography. However, where more complex set ups are required, from the likes of \citeauthor through to the complex citation tracking in humanities styles, the bibliographic data is needed in the document body. Again, finding a way to support this and read a classical .bbl is going to be hard.

Taken together, these issues mean that supporting all of the functionality of biblatex with a 'traditional' .bbl file would be hard-to-very-hard, and would represent a rewrite of quite a bit of the internal code.

The technical issues above are not the only concern. It's difficult to generalise across all publishers, but broadly those using traditional BibTeX set ups are in the physical sciences or with similar 'straight-forward' bibliography styles. Those are already served as 'known quantities' by BibTeX, and so there is little need to provide new solutions that the publishers may then not accept. (At least in my subject area, it's only relatively recently that any attempt has been made to allow chars outside of the classical ASCII range in names, for example.)

Different publishers have different workflows. Some will take TeX input 'as-is' and use it for productions. Others will use a modified version: for example, I've seen .bst files that put an XML version of the bibliography as a comment in the .bbl, and that is then used by the publishers. Such set-ups would be very hard to match as a third-party. Then there are publishers who don't even use the TeX sources, and so for whom a .bbl file is not that useful. (These tend to do PDF-to-Word-to-XML.)

Thus the number of realistic target journals where doing the work to make biblatex 'fit' their workflow in a way they'd actually accept, and the amount of work this would need, mean that it's not a use of resources that seems particularly attractive.