replace bib2df

[SimonGreenhill]

https://github.com/ropensci/bibtex

[HedvigS]

Hi Simon :).

I discovered a fun bug in the original bib2df today. Apparently if you don't have at least one space before and after the = when assigning fields, things go badly. Just thought maybe that'd be useful for you to know. Maybe you already know :)

[HedvigS]

Like this.

[HedvigS]

i just tried the R-package just called "bibtex", and I think it's not optimal because if a given entry don't have certain fields it's ignored. For example, for the bib file in grambank I get

ignoring entry 'Tournadre2017' (line 68747) because :
	A bibentry of bibtype ‘InBook’ has to specify the field: chapter

I can't see an argument to tell it to chill and not be so strict.

[HedvigS]

If the cldf-dataset in question has been created with cldfbench, then the writing of the bib-file is with pybtex which in turn writes key = {value}, which is what bib2df expects (it doesn't parse key = "value"). So, bib2df is not so bad an idea for most cldf-datasets probably?

Given that... it seems to me like the disadvantages that remain with bib2df is mainly that

a) it gives this warning:

Column `YEAR` contains character strings.
              No coercion to numeric applied.

b) it gives this warning:

Warning message:
In bib2df_tidy(bib, separate_names) : NAs introduced by coercion

Both of these seem okay though, sometimes year is "n.d." etc and sometimes it struggles to parse names. Both fair warnings, but not that disruptive really, right?

[HedvigS]

and then there's the aforementioned spaces issue which.. Yeah. Maybe reading in with read.lines, replacing and outputting would help? Haha!

[HedvigS]

@SimonGreenhill what do you think about copying over some of the bib2df code into rcldf functions and solve the known issues? With proper accreditation to the original bib2df people of course.

[SimonGreenhill]

Don't want to have to maintain code for something that should be pretty standard!

[HedvigS]

Sure, agreed. Just a suggestion, based on the title of this issue.

I've been submitting PRs over at ropensci and am in conversation with the folks over there to see if they can push to CRAN soon.

[HedvigS]

Good news!

I'm in communication with the current caretaker maintainer of ropensci/bib2df, Prof Gianluca Baio. Several of the issues we've been having with the package have been fixed in the most recent version on GitHub. He hadn't considered a new CRAN release, but upon my request he said he'd look into it :)!