Hello all, As regards multiple JSON records in a single file, I have written up a proposal here: http://en.wikipedia.org/wiki/User:Baxter.brad/Drafts/JSON_Document_Streaming_Proposal
So this is sort of an RFC. I've added a proof-of-concept routine that implements the proposal to this module (see each_record()): http://search.cpan.org/dist/MARC-Utils-MARC2MARC_in_JSON/ In the proposal, I say that I'm not a fan of newline-delimited JSON files, but really, I just couldn't figure out a way to advertise them that I liked. So I'd like to hear ideas about that. Anyway, I think it's a workable proposal, so I'd like to know what you all think. Thanks, Brad PS: Please don't read too much into the wikipedia address; I have no plans to write an article. It's just a nice place for allowing folks to edit the contents if they are so inclined. On Thu, Feb 10, 2011 at 9:19 AM, Galen Charlton <gmcha...@gmail.com> wrote: > Hi, > > On Thu, Feb 10, 2011 at 9:06 AM, Brad Baxter <b...@mail.libs.uga.edu> wrote: >> I had also thought about how to separate JSON records in a single file. >> Perhaps this has been solved already, but I haven't seen a reference >> to it. If it were YAML, you could simply put "\n---\n" between the JSON >> objects, and it would be simple to scan for that. > > A couple ways to go about it: > > [1] Simply define a collection of MARC-in-JSON records as an array and > use a progressive JSON parser to handle large collections. > [2] Use newline-delimited JSON, i.e., eliminate interior newlines and > just read or write MARC-in-JSON records under the assumption each > record occupies one line in the file. > > The first is slightly more time-consuming to implement but ensures > that MARC-in-JSON files are always conformant JSON. The latter is > easier and can Get Things Done quicker but produces output that can't > be handled correctly by just any random JSON parser. > > Regards, > > Galen > -- > Galen Charlton > gmcha...@gmail.com >